Home

Performance Analysis of Wireless-enabled PROFIBUS

1. Figure 7 5 Configuration file related to the Controller module instance of the BHW2PNetSim excerpt 70 Figure 7 6 Domain module NED definition of the BHW2PNetSim crono no cncnnoninons 71 Figure 7 7 Configuration file related to the Domain module instance of the BHW2PNetSim excerpt 71 Figure 7 8 OMNeT Master module composition of the BHW2PNetSim 72 Figure 7 9 Master module NED definition of the BHW2PNetSiM oooccicncnncnonoccnnnonnnnnnnnnononononnnnnnnnonnnn canon nn anar cnn an anar nnan an annncnans 72 Figure 7 10 Part of configuration file related to Master module instance ereta 73 Figure 7 11 OMNeT Master DLL module composition Figure 7 12 OMNeT DLL module NED definition anireo ceeseeeeseseeseseseseeeeseseeeescscseecsesseeeacseseeesseseeesaeseeeeevseseneeasates Figure 7 13 OMNe T BM module NED definitions icono eiii reis Figure 7 14 OMNeT DMM module NED definition Figure 7 15 OMNeT GMM module NED definition esesesceseeenesescsenesseessesseasassceasescsevasnesseasesssessasaeassavessesseaveese Figure 7 16 Interconnection schema between two BMS uc ceeeceseeeeseseseescseseeseseseeescseseeecseseeesacseseeesaesesseesseseeesasseseeesaeseeeeaeaee Figure 7 17 GMM state machine eee Figure 7 18 BM state machine Figure 7 19 DMM state machine r
2. h DMM BM DMM GMM DMM BM DMM M7 M10 M9 M6 M6 M5 M8 Token VA Phase 4 starts Token Phase 4 starts OE d a Discovery 5 M3 gt iscovery A lt ma a ea Token Discovery Token L M4 E Discovery Token S6 RU RU E RE RU RUE gt a Token Phase 4 ends Phase 4 ends D y D D D M Y v v v v Y Figure 3 8 Simplified timeline of the Tree like topology mechanism for Phase 4 3 3 3 Timer Based Mechanism A solution to handle errors in the IDMP can also be based in a set of timers enabling the error detection and handling during the evolution of the IDMP This error handling procedure is based in four timers which are assigned to the GMM one to each BM and one to each DMM present in the network Two of the timers associated to the GMM are used to detect and handle the errors during Phase 1 while the other two are related to Phase 2 The timers associated to Phase 1 are designated as GMM Phase 1 Alert Timer Temm PlAler and GMM Phase 1 Abort Timer TemmPlabori The timers associated to Phase 2 are designated as GMM Phase 2 Alert Timer Tomm p2alert and GMM Phase 2 Abort Timer Tomm p2Abort The timers associated to BMs and DMMs are designated as BM_IDMP_Abort_Timer Tpm1DMPAbor and DMM_IDMP _Abort_Timer Tbum DMPabor respectively The purpose of each timer is detailed next The Toumpi ter and Toumpiadon are started the Tomm piater lt Tomm piabort When
3. Master PHY Slave PHY A LOS Domain f O 1O Figure C 12 SRD transaction schema between Master and Slave Figure C 13 presents the send SDR procedure The first step is to evolve the Master state machine from the USE_TOKEN to the AWAIT_DATA_RESPONSE state and set the retry_ counter variable to zero After that the Master pops a message from its message output queue builds a frame and transmits it After it waits for the reception of a response frame If an invalid response frame is received with bit errors then it is discarded and the retry counter variable is increased The retry counter variable is also increased if no frame is received within the Tsz When the retry_counter Variable reaches the max_retry limit a Master parameter limit then the Master state machine returns to the USE_TOKEN state Receive SRD Procedure Figure C 14 presents the receive SRD procedure The frame is discarded either if it is an invalid frame with bit errors or if it is not addressed to it a master or a slave Otherwise if it is a valid frame addressed to the station then it tries to find a match with one pre configured message stream If a 144 Simulation Models Implementation match is found then it builds a response frame using the value of the internal variable After waiting Tspr 1t transmits the response frame Send SRD Procedure state AWAIT_DATA_RESPONSE retry_counter 0 Build frame re
4. GOOD BAD MIN MAX MEAN STD DVT N REG MIN MAX MEAN STD DVT N REG 0 0 019 0 002 0 001 4444624 0 0 007 0 001 O 4444624 1 0 002 0 794758 3532399 0 001 0 892978 3968951 2 0 004 0 163154 725160 0 7 0 002 0 072008 320050 1 3 0 006 0033459 148712 o 0 002 0 023561 104719 og 4 0 008 0 005969 26532 0 003 0 007716 34293 08 5 0 01 0 002105 9358 le 0 003 0 002549 11331 07 6 0 012 0 000439 1949 04 0 004 0 00106 4710 o 7 0013 81 05 360 o 0 005 8 37E 05 372 08 8 0015 286E 05 127 0 005 29E 05 129 05 9 0 017 5 17E 06 23 A 0 006 1 03E 05 46 04 10 0 019 9E 07 4 0 007 5 17E 06 23 03 02 o1 o o Ea 0 O 0 01 0 01 0 01 0 01 0 01 0 02 0 02 0 02 o 0 o 0 o o 0 01 0 01 0 01 0 01 Figure D 19 Screenshot of spreadsheet created by the Bit Error Model Channel State Quality option Amnex E Acronyms and Symbols This annex presents two lists one containing the acronyms and another containing the symbols used in this dissertation E 1 Acronyms Acronyms Description AGV Automatic Guided Vehicle AL Application Layer BEM Bit Error Mode BER Bit Error Rate BHW2PNetSim Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BM Bridge Master BM mi First bridge master in the path from the initiator to the responder of a transaction BM yes Last bridge master in the path from the initiator to the responder of a transaction BS Base Station BT Beacon Trigger CSRD Cyclic Send and Receive
5. 167 Figure D 18 Output channel state quality file excerpt oie eesseseeeeseseeseseseseseescsceecsescseserscsesececsesseesaeseeseesseaeeeeatseeseeees 167 Figure D 19 Screenshot of spreadsheet created by the Bit Error Model Channel State Quality option 167 VII List of Tables Table 2 1 Mapping between standard PROFIBUS frames and IDFSs ereta nnrnonn anar onanancnons 20 Table 2 2 Format of the IDMP messages requests Table 5 1 Summary of the output data information Table 8 1 Path Loss Exponents for Different Environments Source Vignaux and Muller 2006 oooicnicninononiccncncnnnonanacnnns 84 Table 8 2 Standard Deviation for Different Environments Source Vignaux and Muller 2006 Table 9 1 Physical media parameters Table 9 2 Station s address 93 Table 9 3 Repeater b sed domain parameters Li is 93 Table 9 4 Master SAS ii e ae are aoe Table 9 5 Bridge based domain parameters Table 9 0 Message SIIC AMS toos adan atrio aitor ei Table 9 7 Response time of the message stream SIMI oes ceeseeeeseseeeceeseseeceesesseecseseeecscseseescacsesseesseseesasseseeesseeeeeanate Table 9 8 Response time histogram for the message stream S1M2 Table 9 9 Response time of the message stream S3M3 Table 10 1 Parameters for Gilbert Elliot Channel Model Fable 10 2 Master s address ata ta Table A 1
6. eee 160 Figure D 2 Screenshot of Timeline Visualisation Tool RHW2PNetSim ereta 160 Figure D 3 Screenshot of the Output Data Analysis Tool Figure D 4 Output response time file excerpt Figure D 5 Screenshot of spread sheet created by Message Stream Response Time Analysis option 162 Figure D 6 Screenshot of spreadsheet created by Message Stream Response Time Central Limit Theorem option 163 Figure D 7 Output state machine file excerpt ae Figure D 8 Screenshot of spreadsheet created by State machine Analysis option Figure D 9 Output PDF file excerpt nnn a a n S ER Figure D 10 Screenshot of spreadsheet created by the Probability Distribution Functions Analysis option 164 Figure D 11 Output frame accounting file excerpt Figure D 12 Information about invalid frames relayed by a Domain module instance Figure D 13 Screenshot of spreadsheet created by the Bit Error Model Frame Accounting option 165 Figure D 14 Output deleted IDTs file excerpt eretas 166 Figure D 15 Screenshot of spreadsheet created by the Bit Error Model IDP Timeout option 166 Figure D 16 Output IDMP alerts and aborts file excerpt eeseeeeeeeeeeeeeeseteeeeeeneees 166 Figure D 17 Screenshot of spreadsheet created by the Bit Error Model IDMP Timeout option
7. Detect Detect a valid frame No Ts expired Ts pia No a valid frame 2 Yes Yes Notify DMM Notify DMM J state DISCOVERY state INQUIRY_MODE Figure C 33 Mobility procedure Amnex D Tools for Simulation Output Analysis This annex presents a description of the output data files and the tools which support the analysis of the output data files generated by the Repeater Based Hybrid Wired Wireless Network Simulator and the Bridge Based Hybrid Wired Wireless Network Simulator D 1 Introduction Output data analysis is the examination of the data generated by a simulator and this examination has two purposes Firstly it is used to verify and validate the simulator and its simulation model Secondly it is used for testing evaluating the performance of different scenarios and different systems configurations Additionally when the input variables are random values the output data exhibits random variability Therefore the output data is used to estimate the confidence level or to determine the number of observation required to achieve a desired precision The objective of this annex is to present and describe the information produced by the RHW2PNetSim and by the BHW2PNetSim It describes the output data files generated by both simulators from which it is possible to extract results To help on the analysis of the results some specific tools have been developed The Timeline Visualization Tool
8. M8 TGMM P1Abort pee e M8 z 1 Do v v D Y v Y p v Y p O Start timer s O Expire timer a Stop timer s g Transmission error Figure 3 9 Simplified timeline of the Timer Based mechanism for Phase 1 A similar mechanism is used to control the Phase 2 When the GMM broadcasts a PBT message Phase 2 starts and the Toum p241err and Temm P24bort timers are started the duration of the Temm P24lert must be smaller than Toum p24bor At reception of the PBT message the DMMs start the Tpmm IDMPAbor And a RBT message is transmitted when it holds the token frame The GMM stops the Temm Pz4tert and Temm P24bort if it receives a RBT message from all DMMs in the network and the IDMP evolves to Phase 3 Otherwise the GMM broadcasts again a PBT message 34 Error Handling Improvements for the Bridge based Architecture at expiration of the Tgyy prdier If the TGmm P24bort expires before the reception of all RBT messages then the IDMP is aborted Otherwise the Temm P24bort 18 Stopped and the IDMP evolves to Phase 3 Considering the system scenario illustrated in Figure 2 12 Figure 3 10 represents a simplified timeline regarding Phase 2 Assuming that DMM MO is holding the token frame a RBT message is immediately sent to the GMM and the PBT message is forward to domain D at the start of Phase 3 Supposing a transmission error when DMM M9 is sending the PBT message to domain D this means that the DMM M10 will not receive the PBT message Wh
9. 335 039187 23473 4897 0 087112 6218 045 6 434 0 094875 5683 04 797 0212337 12719 035 9508 0 1856260 11119 0 11045 0 02818 1688 02 N REG 59900 AS rH 3 36 4 897 6 434 7 971 9 508 11 0 Figure D 8 Screenshot of spreadsheet created by State machine Analysis option TRIANG 11 000000 50 000000 70 000000 18 103736 54 119785 62 908148 51 248531 26 079176 43 094876 66 415956 Figure D 9 Output PDF file excerpt Probability Distribution Function Statistical Analysis This output of the Probability Distribution Function Analysis option is similar to the previous It computes some statistical elements like mimimum MIN maximum MAX mean MEAN standard deviation STD DVT values as well as a histogram Figure D 10 shows a screenshot of a spreadsheet created by this option FUNCTION PAR1 PAR2 PAR3 TRIANG 41 50 70 PDF MIN MAX MEAN STD DVT N REG 11596 69 696 43 66 12 222 8750 4 17 406 0 019543 171 2 23 216 0 044343 388 02 3 29026 0071657 7 ois 4 34 836 0 114171 g9 I 5 40 646 0 130057 1138 04 6 46 456 0 168229 1472 DE fl Ed 52 266 0 185486 1623 8 58 076 0 149029 1304 04 4 L 9 63 886 0 086857 760 174232 29 348 405 46 5 52 3 58 1 63 9 69 7 10 69 696 0 030629 268 Figure D 10 Screenshot of spreadsheet created by the Probability Dist
10. 7 3 Simulator Architecture The BHW2PNetSim was developed using the OMNeT discrete event simulation platform It is composed by six main components modules HW2PNet Domain ComFunc Slave Master and Controller Except for the Hw2PNet and Controller modules the others can be clearly mapped on the main devices of the bridge based architecture The main features of the Hw2PNet Controller Master Slave and Domain modules were already detailed in Chapter 5 and the ComFunc module was detailed in Chapter 6 Therefore only the differences and extensions to these modules will be described in this chapter Figure 7 2 shows how the main modules are interconnected Besides the domain con and ctrl con connections described in Chapter 5 there is another kind of connection called bridge con This connection is used to model a bridge In this simulator architecture there is no module called bridge A bridge is an abstraction which is composed by two Master module instances connected by a ComFunc module instance Figure 7 3 shows a graphical representation used by the simulator to represent the network scenario shown in Figure 2 12 This network scenario is composed by three bridges The ComFunc module instance of each bridge is labelled bridge x where x can be 0 1 or 2 7 3 1 Controller Figure 7 4 presents the NED definition of the Controller simple module To define the GMM it is necessary to assign the name of the Master module instance t
11. Itis expected that the clocks of the master stations in the system may have some drift between them The masters are not synchronised between them These assumptions were applied to the simulation models of both approaches by setting the offset of the message streams and its period using probabilistic variables However the response time depends on the message stream period particularly in the Bridge based approach for that reason the simulation runs have been made independently for each master s message stream set For the master to which we want to perform the measurements the message stream periods were set to a constant value For the other masters the message streams parameters were set using a triangular distribution function In the repeater based approach the period of the message streams being measured has been set to 40 ms with no initial offset The period and the initial offset of the other message streams has been set using triang 38 40 42 and triang 0 38 40 respectively Similar rules are used in the bridge based approach the period of the message streams being measured has been set to 8 ms with no initial offset and the period and the initial offset of the other message streams has been set using triang 7 8 8 8 2 and triang 0 7 8 8 respectively In the following section a performance analysis based in simulation results is presented The results for each master message stream set have been obtained
12. S it a request frame 2 No Open a transaction in LOT Send frame Build an IDF to ComFunc Start timer Build a SC frame Remove transaction from LOT Build response frame with the response Wait Tsor End Send SC frame Wait Tsor Send response frame Figure C 16 Receive Frame from Domain Procedure A LOT entry is in the WAITING state since its creation until receiving a response At reception of the related response frame the LOT entry changes to the FINALISED It is deleted from LOT when the response frame is sent to initiator Simulation Models Implementation 147 If there is no entry in the LOT associated with this request then a new entry is opened in the LOT and a timer loaded with the bm idt timer parameter value is started After that an IDF is coded and sent to the ComFunc If there is a match with another LOT entry the frame is discarded If there is a LOT entry with same parameters and its state is FINALISED then a response frame is built and transmitted as a response to the initiator After that the associated entry is deleted from LOT Receive Frame from ComFunc Procedure Figure C 17 shows the procedure when a BM receives a frame an IDF from the ComFunc 1 e from the other BM of a bridge The BM starts by consulting its RT using the isRoute function to determine if the frame should be relayed or not Receive Frame from ComFunc Procedure Receive a frame Yes DA belongs to
13. If the received frame is an invalid frame containing bit errors the Master continues in the same state A Master in the ACTIVE IDLE state continually analyses all received frames If a token frame is received when This Station TS is skipped 1 e the address of TS lies within the address range spanned from the sender address to the receiver address in the token frame then it removes itself from the logical ring and evolves to the LISTEN TOKEN state transition 13 When the Master is in the USE_TOKEN state 1 e when it holds the token frame it behaves according to the message dispatching procedure a detailed description of this procedure is presented in Section C 1 3 A Master in the USE_TOKEN state can perform one of 4 transitions 2 3 5 and 7 The Master stays on the same state transition 2 if it transmits a frame that does not need to receive a response frame e g when using the SDN service see Section C 1 6 for more details It changes to the AWAIT DATA RESPONSE state transition 3 if it sends a message that requires a response frame e g when using the SRD service see Section C 1 6 and returns to the USE_TOKEN state when it receives a response frame or the Slot Time Tsz expires transition 4 Transition 5 occurs when the Master transmits an FDL Request Status frame and evolves to the AWAIT STATUS RESPONSE state The Master returns to the USE TOKEN state transition 6 when it receives either a response frame or th
14. The BS waits for Tip before transmitting a Beacon frame Wireless mobile stations receive these Beacon frames and changes to domain according to the location vector parameter which defines the domain location for each wireless mobile station more details in Section C 2 2 6 4 Summary This chapter presented the main architectural components used on RHW2PNetSim and some implementation details Additionally we had also described the format of the NED files required for the configuration of the modules used in this simulator Chapter 7 Bridge Based Hybrid Wired Wireless PROFIBUS Network Architecture Model This chapter describes how the bridge based approach has been implemented in the Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator It presents the main architectural modules and the main configuration parameters 7 1 Introduction As mentioned in Chapter 2 the Intermediate Systems IS of the bridge based approach operate at Data Link Layer DLL level as bridges A bridge can interconnect two or more domains Although it is assumed that a bridge only interconnects two domains Each bridge is composed by two modified PROFIBUS masters called Bridge Master BM A BM is capable of receiving all frames arriving at its physical interface and forwards them to the other BM of the bridge according to the routing information These BMs operate almost as standard PROFIBUS masters and are assigned a PROFIBUS DLL address Co
15. 0 34800 0 11 2 10 06433 3 65463 17 8 9 87200 0 13 14 0 12200 0 12 3 11 88800 0 13480 18 9 6 67867 0 14 15 0 04700 0 13 4 12 65600 0 00840 19 10 3 76400 0 15 16 0 01000 0 14 5 14 38667 0 00027 10 11 1 92033 0 116 17 0 00033 0 15 6 14 93633 0 11 12 0 87200 0 Figure 9 4 depicts a response time histogram for message stream S in both scenarios The repeater based scenario presents the MinRT 1 22 ms and MaxRT 18 49 ms values smaller than in the bridge based scenario The MinRT and MaxRT values in the bridge based scenario are 8 80 ms and 26 80 ms respectively In the repeater based scenario the histogram for S is similar to the histogram of S as it would be expected since the use of repeaters creates a broadcast network Comparative Performance Analysis in an Error Free Environment 97 09 AR O as ae etry e c y E es Mic ss 06 r NN S 05 alo SSS FSE A ea Miss Fo 02 TIAA ma o F PE 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Response Time ms Figure 9 4 Response time of the message stream S7 In the bridge based scenario the timing behaviour of message stream S is different than for S pe since S Me is an IDT Therefore such kind of transaction requires that the initiator performs at least one Application Layer AL retry before obtaining a response meanwhile stored at the BM BM M7 The period of this message stream is
16. 50 0 0 0 250 0 0 0 0 250 50 0 0 0 25 agv 1 color 0 2 0 6 0 3 1 0 agv 1 ant freq 2 3 agv 1 ant pt 1 agv 1 ant gt 1 agv 1 ant gr 1 agv 2 name S6 agv 2 vel 6 agv 2 path 80 10 0 70 20 0 40 20 0 0 125 40 20 0 0 125 70 20 0 80 10 0 80 10 0 70 20 0 40 20 0 0 125 40 20 0 0 125 70 20 0 80 10 0 agv 2 color 0 1 0 5 0 6 1 0 agv 2 ant freq 2 3 agv 2 ant pt 1 agv 2 ant gt 1 agv 2 ant gr 1 Figure 8 7 Configuration file related to the agv parameters excerpt 8 7 Output Data Files This simulator produces two kinds of output files One contains information about the timings when the network handoff procedure was active the domains to which the stations belong to and additional information these files use the extension li The other type of file contains a text file with the information required to set the location vector parameter used in the RHW2PNetSim and BHW2PNetSim these files use the extension sd Figure 8 8 presents part of the output file related to the agv module instance named M3 using the network configuration presented in Figure 8 2 The first column Time refers the timestamp when the radio signal quality was evaluated The second column SHandProc indicates the instant in time when the handoff procedure started and the third column EHandProc refers to the end of the handoff procedure The radio frequency value the BS name and domain name are in fou
17. 8 Technological Context Communication Infrastructure 2 2 2 Data Link Layer DLL Message Cycle In PROFIBUS only master stations may initiate transactions whereas slave stations do not transmit on their own initiative but only upon master requests The station that sends an Action Frame the first frame transmitted in each transaction is the initiator of the transaction while the addressed one is the responder A transaction or message cycle consists on the request or a send request frame from the initiator always a master station and the associated acknowledgement or response frame from the responder either a master station or a slave station but typically a slave station All stations monitor all the requests but will only acknowledge or respond if and only if they are the addressees in the initiator s request Moreover the acknowledgement or response frame must arrive before the expiration of the slot Time Tsz which is a master DLL parameter otherwise the initiator repeats the request a number of times defined by the max_retry limit another master s DLL parameter Token Passing The token is passed between masters in ascending address order The only exception is that in order to close the logical ring the master with the highest address must pass the token to the master with the lowest one Each master knows the address of the previous station PS Previous Station address the address of the following station
18. CP10 i S6 Domain Domain gt D4 Frame gt 03 J CP7 knows the frame s length CP7 receives the frame s first bit l R3 delays the frame during Tiom Tiom CP10 starts the frame transmission after delaying the frame to guarantee the CP10 receives the a transmission without time gaps frame s first bit D3 receives the frame s first bit R3 receives the frame s first bit Figure 6 15 Simplified timing behaviour of the module instances that model a repeater After delaying the frame CP7 passes the frame to the ComFunc module instance named R3 The frame is delayed by R3 during the time defined by a PDF configured by its parameters using the pdf delay prefix after that it passes the frame to the Connection Point module instance named CP10 CP10 computes the frame last bit reception instant and the frame last bit transmission instant in order to determine the first bit transmission frame instant in its domain In this calculation the CP10 has to respect the idle time defined by parameters that use pdf tidm prefix between two consecutive transmissions The frame is transmitted to the Domain module instance named D3 by CP10 through repeater con connection D3 delays the frame according to the frame length and then the frame is transmitted through domain con connections arriving at S6 All these delays
19. Figure 3 8 Every wireless DMM still holding the token inquires all wireless mobile stations in order to detect if 32 Error Handling Improvements for the Bridge based Architecture they are present in its domain using Discovery messages After the discovery sub phase finish a RU message is transmitted by the DMMs whenever they detect that a wireless mobile station is in their domain These messages are used by the BMs to update their RTs A Phase 3 starts IDMP ends DMM BM DMM GMM DMM BM DMM M7 M10 M9 M6 M6 M5 M8 IDMP ends de spt tspr SA A SBT SET sBT SBT _ Leer FBES spt MA Token 4 M8 ara Token SBT sBT EBF Beacon m Ee ser gt M9 Token Beacon EBF Token Phase 3 ends SST ae ug yl Token A Beacon Beacon _ EBF _ j lt l M6 D v y fay v M y D Y v D Figure 3 7 Simplified timeline of the Tree like topology mechanism for Phase 3 Note that if a wireless mobile station is not discovered during the discovery sub phase due to transmission errors or ifa RU message is corrupted by a transmission error the transactions with these stations would not possible until the next mobility procedure In order to handle with problem the DMMs must periodically send Discovery messages addressed to the wireless mobile stations if such stations are found the DMMs broadcast RU message
20. Figure 7 13 OMNeT Bm module NED definition Whenever a new transaction is open in the LOT an IDT related timer is started which is used to clear the transaction from the LOT when it expires The value of this timer is assigned by the bm idt timer parameter In the same way at the reception of a SMP message the BM IDMP Abort Timer TemiDMpAbor is loaded and started This timer is used to detect errors during the evolution of the IDMP The value of this timer is assigned by bm idmp abort timer parameter The BM operation depends on its Routing Table RT and on its List of Active Stations in Domain LASD These elements are generated by the controller module instance at simulation initialisation and are updated in run time The pum module is a simple module that models the DMM functions required by the IDMP This module is responsible for controlling Phase 3 and Phase 4 of the IDMP Figure 7 14 depicts the pum module NED definition in which there is only one parameter TS parameter The LBMD and LWMSD are generated by the controller module instance at simulation initialisation and are updated in run time There are other parameters that must be assigned to the DMM module like dmm_idmp abort timer n beacon and beacon_len parameters for instance but to simplify the configuration procedures they are assigned to the Domain module instead simple DMM parameters MS numeric gates in dil gateln out dil_gateOut endsimpl
21. If output timeout idt parameter is set equal to one it means that the BM module instance will be configured to gather information about which IDT transactions were aborted In the same way if output timeout idmp parameter is set equal to one the BMs DMMs and GMM will be configured to gather information about IDMP timers expiration simple Controller parameters _gmm string _fmob numeric _gmm_phase1_alert_timer numeric _gmm_phase1_abort_timer numeric _gmm_phase2_alert_timer numeric _gmm_phase2_abort_timer numeric _output_timeout_idt numeric _output_timeout_idmp numeric _domain string _inter_domain string endsimple Figure 7 4 Controller module NED definition of the BHW2PNetSim The parameters domain and inter domain bold lines in Figure 7 5 are strings which define the configuration of domains and bridges Both of them are written using a predefined structure based in tags Figure 7 5 presents an example of the parameter values related to the Controller module instance for the network depicted in Figure 2 12 theBHW2PNet controller _gmm M6 theBHW2PNet controller _tmob 200ms theBHW2PNet controller gmm phase1 alert timer 20ms theBHW2PNet controller gmm phase1 abort timer 30ms theBHW2PNet controller gmm phase2 alert timer 10ms theBHW2PNet controller gmm phase2 abort timer 20ms theBHW2PNet controller output timeout idt 1 theBHW2PNet controller output timeout idtmp 1 theBHW2PNet co
22. codes it according to the IDP the frame format of IDP is presented in Table 2 1 and stores internally information about the transaction in a structure called List of Open Transactions LOT Meanwhile the PROFIBUS DLL of the initiator retries transmitting the same request since the BM does not respond before the expiration of the Tsz The DLL retries are executed by the initiator a number of times specified by the max_retry limit a DLL parameter The IDreg frame is relayed by the bridges until reaching the last BM which belongs to the responder domain denoted as BM res stands for responder BM es decodes the original request frame and transmits it to the responder which can be a standard PROFIBUS station for example a wired PROFIBUS slave When decoding the frame the BM reconstructs the original frame as transmitted by the initiator it even puts the initiator address SA on the request frame Thus from the responder s perspective the initiator seems to belong to the same domain When the BM es receives the response to that request it codes that frame using the IDP and forwards it through the reverse path until reaching the BM ni where it will be decoded and properly stored Since for an IDT the response to the original request takes longer than for an IADT the initiator AL must periodically repeat the same request until receiving the related response After BM having received and stored the correspondent response frame
23. ctrl _gateln y domain_gateOut A domain_gateln Figure 5 10 OMNeT Master module composition module Master parameters TS numeric _num_streams numeric _name string _pdf_tid1_type numeric _pdf_tid1_par1 numeric _pdf_tid1_par2 numeric _pdf_tid1_par3 numeric gates in domain gateln ctrl gateln bridge gateln out domain gateOut ctrl gateOut bridge gateOut submodules phy layer Master PHY dil layer Master DLL parameters TS TS _num_streams _num_streams _pdf_tid1_type _pdf_tid1_type gatesize upper_gateOut _num_streams upper_gateln _num_streams stream Msg Stream num streams connections nocheck phy layer upper gateOut gt dll layer lower gateln phy layer upper gateln lt dll layer lower gateOut phy layer lower gateln lt domain gateln phy layer lower gateOut gt domain_gateOut phy layer ctrl gateln lt ctrl gateln phy layer ctrl gateOut gt ctrl gateOut dil layer bridge gateln lt bridge gateln dil layer bridge gateOut gt bridge gateOut for i 0 num streams 1 do dil layer upper gateOut i gt streamfi lower gateln dil_layer upper_gateln i lt stream i lower_gateOut endfor endmodule Figure 5 11 Master module NED definition 51 52 PROFIBUS Simulation Model theProfibusNet master 2 TS 3 theProfibusNet master 2 name M3 theProfibusNet master 2 pdf tidf type 3 theProfibusNet mast
24. gt BM DMM GMM TI cS bridge_gateOut lt 4 e dil gateln dil gateOut dil gateln dil gateOut dil gateln dil gateOut bridge gateOut q 4 y 4 y 4 y bm_gateOut bm_gateln upper_gateOut upper_gateln dmm_gateOut dmm_gateln gmm_gateOut gmm_gateln DLL lower_gateln lower_gateOut 4 y lower_gateln lower_gateOut Figure 7 11 OMNeT master DLL module composition Figure 7 12 presents the DLL NED definition Its NED definition is very simple and is very similar to the definition of the master DLL module presented in Chapter 5 except for the definition of the bridge gates simple DLL parameters TS numeric _pdf_tid1_type numeric _pdf_tid1_par1 numeric _pdf_tid2_type numeric _pdf_tid2_par1 numeric _pdf_tsdr_type numeric _pdf_tsdr_par1 numeric gates in upper_gateln lower_gateln bridge_gateln out upper_gateOut lower_gateOut bridge_gateOut endsimple Figure 7 12 OMNeT DLL module NED definition 74 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model The Bm module is a simple module that models the mechanisms necessary for the IDP and the IDMP related functions Figure 7 13 presents the Bm NED definition The address of the BM the same of the Master module module instance is set by the TS parameter simple BM parameters MES numeric bm idt timer numeric bm idmp abort timer numeric gates in bridge gateln dll gateln out bridge gateOut dll gateOut endsimple
25. it is ready to respond to a new repeated request from the initiator The response frame is exactly equal to the frame transmitted by the IDT responder Considering the network scenario illustrated in Figure 2 10 Figure 2 11 represents a simplified timeline regarding a transaction between master M3 and slave S6 Figure 2 11 assumes the typical behaviour of PROFIBUS DP where the slaves read their inputs periodically placing their image on the DLL by using the generic Service upd req primitive The image of the input values is placed in a buffer which is used by the protocol to build a response to a specific request An indication can be transmitted to the higher layers every time a slave receives a request This type of procedure is usually referred to as buffered operation Codes the frame using IDP and opena transaction in LOT Initiator M3 Responder s6 Bridge 1 Bridge 2 M8 M5 M6 M9 paH AL DP DLL DLL AL DP Token Des SIDE Service upd req Service req a PROFIBUS originating a request PROFIBUS frame ae Token lt _ Service con No_Data Token PROFIBUS request Token ee PROFIBUS response Token 4 PROFIBUS request Service req Service con No_Data Service req D Token PROFIBUS 0 request Codes the response Decodes the IDF and frame using IDP close the transaction in LOT y v D Di D PROFIBUS respon
26. that handles the IDMP Table 2 2 synthesizes these messages Table 2 2 Format of the IDMP messages requests Frame Frame Header Frame Data SD DA SA FC DAE SAE MC Data Start Mobility Procedure SD3 27 GMM 6 55 55 1 Ready to Start Mobility Procedure SD3 GMM Bri 6 55 55 2 Prepare for Beacon Transmission SD3 27 GMM 6 55 55 3 Ready for Beacon Transmission SD3 GMM DMM 6 55 55 4 Start Beacon Transmission SD3 27 GMM 6 55 55 5 Station Route Update SD3 127 Bri 6 55 55 6 Nis Inquiry SD3 Bri DMM 13 55 55 7 Void SDI DMM DMM 6 Since most of the messages are sent in broadcast mode thus not requiring any response the frames are coded using high priority SDN frames Therefore the FC code value of most protocol message is set to 6 The field Mobility Code MC codes the type of operation that must be performed when the destination station receives the frame For the Beacon message is used the same type of message used in the RFieldbus system which is described in Rauchhaupt 2002 This message must have a specific format which allows the wireless mobile station to evaluate the radio channel quality The Inquiry message request may have a response from the addressed BM thus it is coded as a SRD high service In that case the response to that service can only contain a mobility related message Technological Context Communication Infrastructure 25 from the output queue of the addressed BM F
27. through the ctrl_con connection a message indicating that it is starting to transmit Beacon frames and sends another message to indicate the end Beacon message transmission Wireless mobile stations at reception Of Beacon frames send to the Controller through the ctrl con connection a message indicating to which domain they want to change according to their location vector parameter see Section 6 2 3 The Controller manages this information in order to disconnect the Master and Slave from the Domain to which they are connected and connect them to the destination Domain However a Master Or a Slave can only change to a Domain where Phase 3 has taken place DLL Operation As mentioned the DLL must support additional functionalities In order to perform the mobility procedure when the DLL state machine is in the ACTIVE_IDLE or the USE_TOKEN state after the execution of any tasks it must check if this station acts as a DMM If it does succeed in which the state DMM is more precisely if it is not in INACTIVE state Figure C 32 Figure C 33 presents the DLL mobility procedure All IDMP related messages are transmitted as high priority messages Therefore whenever there are IDMP related messages into the output queues the DLL sends them Its state machine evolves according to the message s type or to the DMM state machine Therefore it has to check the message type and in the case of being an IQ REQ message the DLL evolves to the WAIT INQU
28. 1 bs 0 ant gt 1 bs 0 ant gr 1 bs 1 name BS2 bs 1 domain D3 bs 1 position 80 0 37 5 0 5 0 bs 1 color 0 0 1 0 0 0 1 0 bs 1 ant freq 2 3 bs 1 ant pt 1 bs 1 ant gt 1 bs 1 ant gr 1 Figure 8 6 Configuration file related to the bs parameters excerpt The definition of the bs module instance position defined by bs x position parameter is done by Cartesian coordinate x y z with origin at the centre of the ground In this case BS1 is located at position 50 0 0 0 5 0 and BS2 is located on the opposite side at 50 0 0 5 0 The colour of the bs module instances is defined by bs x color parameter and is set by four values all in the range between 0 0 to 1 0 The first three are the RGB values and the fourth is the transparency Each bs module instance operates at different radio channel BS1 operates on the 2 4GHz bandwidth and BS2 on the 2 3GHz bandwidth defined by bs x ant freq parameter The transmitted power of each antenna module instance is defined by bs x ant pt parameters the 88 Mobility Simulator transmitter gain is defined by the bs x ant gt parameters and the receiver gain is defined by bs x ant gr parameters which have been set equal to one in all bs module instances Figure 8 7 shows part of the configuration file related to an agv module In this example there are two agv module instances defined by sim agv_num parameter One is called M3 and the
29. 120 250 500 1000 30 60 120 250 500 1000 Delay us Delay us Figure 9 8 Influence of the IS delay on MaxRT In the repeater based scenario the internal delay of the repeater has a stronger influence on the MeanRT and MaxRT values due to the increase on the message cycle latencies Additionally the internal delay of the repeater requires a new setting of the Tp parameter of the MM and consequently the mobility procedure takes longer In the case of the bridge based scenario the internal delay of the bridge has a small influence on the response time values of message stream S an IADT since the frames exchanged in these kind of transactions are not relayed by bridges The small MaxRT value increase is mainly due to the Comparative Performance Analysis in an Error Free Environment 101 increase of the IDMP related latencies The effect on message stream S an IDT is attenuated due to several repetitions performed by the initiator until retrieving a response from the IDT BM It is also important to note that the internal delay of the ISs has a strong influence in the repeater based scenario throughput As mentioned there was the need to increase the message stream period to 80 ms which is twice the message stream period of the base configuration Consequently the number of transaction decreased for half For this reason in the bridge based scenario the number of concluded transaction is 1000 more for IADTs and 480 more for
30. 2006 Environment Standard deviation o dB Outdoor 4 to 12 Office hard partition 7 Office soft partition 9 6 Factory line of sight 3 to 6 Factory obstructed 6 8 It is important to select a free space reference distance that is appropriate for the propagation environment Usually in large coverage cellular systems a 1 km reference distance is commonly used in microcellular systems much smaller distances such as 100 m or 1 m are used The reference distance should always be in the far field of the antenna so that near field effects do not alter the reference path loss 8 4 Simulation Model This simulator models the radio signal strength using the Log normal Shadowing model Since this model takes into account not only the distance between the transmitter and receiver but also the empirical characteristics of the environment The station mobility is calculated according to the wireless mobile station velocity and its path on the plant floor In order to illustrate the simulation model Figure 8 2 presents an example with the same topology as the scenarios presented in Figure 2 3 and Figure 2 12 The network comprises four domains two wired domains D and D and two wireless domains D and D Three intermediate systems IS1 IS2 and 183 interconnect the wired and wireless domains The network also comprises seven wired stations S1 S2 S3 S4 S5 M1 and M2 three mobile wireless stations M3 M4 and S6
31. Burst Error Periodic Model dl 132 B 5 Parameterization of the Bit Error Model Parameters Used in both Simulators 133 ANNEX C IMPLEMENTATION OF THE SIMULATION MODE LG ccscssscssssssssessecesssersecssesesseees 135 Cul PROFIBUS DEL critical ano mara e ii 135 C11 Token Recovery Procedure icon orate cua E KETE S N 135 C 1 2 Token Reception Procedure ceccsseesecsesseceeceecesceseceeseseseeeeseeeeeeeeeeseeseeeaeeaeeeaeenaeenees 136 C 1 3 Message Dispatching Procedure cceccesccescessesseeseeeeceeteeeceseeeseeseecseeseecaeceaeeaeenseenees 136 C 14 Pass Token Procedure ate e tdo te e Meas a re t 138 ETS GAP Update Proce dure tn O e e 138 G 1 6 Send Frame Procedure sisters lins ieo essa a le ea e ia tio o dd 141 C 2 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulator 145 C 2 1 Send Beacon Procedure ccccsseesecseceecescesececesecsceseceseeseseseeeeeseeeeeenseesecaeeeneceeeaeenses 145 C222 Stations Mobil a ed leo e o aE iR 145 C 3 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulator 145 C 3 1 Inter Domain Protocol 0 ccececsseesecseceseceeceseceeceseeseceseeeeeseeencesseeaeeaeeeaeesaecaeceaecaeenaeenees 146 C 3 2 Inter Domain Mobility Procedure Implementation erre 148 ANNEX D TOOLS FOR SIMULATION OUTPUT ANALYSIS csssscsssssssccscssssessscseesessssssesessecssesesseees 159 Dll
32. C 1 6 of Annex C The remainder of this Section will focus on the implementation of the Master DLL and the DLL simple modules of the RHW2PNetSim and of the BHW2PNetSim respectively For that purpose in this chapter we use the term Master when referring to the PROFIBUS DLL module The operation mode of a standard master PROFIBUS DLL is supported by a state machine composed by 10 states IEC 2000 For simplification reasons in both implementations only 8 states are considered 56 PROFIBUS Simulation Model According to the PROFIBUS protocol after turning on the power of a master station it will go into the LISTEN TOKEN state in order to generate the List of Active Stations LAS and GAP List GAPL However in RHW2PNetSim and BHW2PNetSim all Master module instances start in the ACTIVE IDLE state with all configurations and parameterization performed by the controller module instance at the simulation setup In order to start the network simulation operation the Controller module instance sends a token frame to the Master module instance that acts as a Mobility Master MM in the case of RHW2PNetSim or sends a token frame to the Domain Mobility Manager DMM of each domain in the case of the BHW2PNetSim Then the state machine of the Master module instances evolves to the USE_TOKEN state Figure 5 20 presents the Master state machine diagram related to the implementation of the PROFIBUS DLL in both simulators In this state machine diagram a
33. C 26 dllReceiveToken dllholdtoken true if state WTOKEN then state INQUIRY processRBTmessage J DIANA A Figure C 25 d11ReceiveToken function pseudo code algorithm The processRBTmessage function builds a RBT message and sends it to the DLL If this Master is also acting as a BM then the DLL forwards the message to it The BM forwards the RBT message to ComFunc module instance connected to it or passes the RBT message to the DLL according to the routing information Simulation Models Implementation 153 dl processRBTmessage Les if 3 msg buildIDMPmessage RBT addr_gmm TS 4 passToDLL msg 5 Figure C 26 processRBTmessage function pseudo code algorithm When a DMM receives an IDMP related message the processIDMPmessage msg function is automatically invoked The pseudo code algorithm depicted in Figure C 27 is only related to the reception of the PBT message Depending on the DMM state the actions triggered by the PBT reception are different 1 processIDMPmessage msg Be i 3 switch state 4 case INACTIVE De switch msgKind msg de gasa IRENE 3 Ey startIDMPaborttimer hg if dllholdtoken true then 9 state INQUIRY TOs processRBTmessage TES WAS erse SA state WTOKEN 14 end LS A Se EIS end Sa case WTOKEN T9 end 208 case INQUIRY PAL switch msgKind msg 223 cage PBT PRIR processRBT
34. D M Woo et al 2005 OpenGL R Programming Guide The Official Guide to Learning OpenGL R Version 2 Addison Wesley Professional Siemens 2005 Gateway IWLAN PB Link PN IO for Industrial Ethernet Manual Part BL2 Release 7 2005 C79000 G8976 C200 02 Simon J 2002 Excel Programing Your visual blueprint for creating interactive spreadsheets New York Hungry Minds Inc Smith R 2004 Open Dynamics Engine V0 5 User Guide Available online at http www ode org References 123 Sousa P L L Ferreira et al 2006 Repeater vs Bridge Based Hybrid Wired Wireless PROFIBUS Networks a Comparative Performance Analysis In proceedings of 11th IEEE International Conference on Emerging Technologies and Factory Automation ETFA 06 Prague Czech Republic pp 1065 1072 Sousa P L L Ferreira et al 2007 Repeater vs Bridge Based Hybrid Wired Wireless PROFIBUS Networks a Comparative Performance Analysis over Error Prone Mediums To be published Stankovic J A 1989 Real Time Computing Systems the Next Generation In Tutorial Hard Real Time Systems Stankovic J and Ramamritham Editors IEEE Computer Society Press Los Alamitos USA pp 14 38 Tovar E and F Vasques 1999 Cycle Time Properties of the PROFIBUS Timed Token Protocol In Computer Communications Elsevier Science No 22 pp 1206 1216 Tovar E and F Vasques 1999 Real Time Fieldbus Communications Using PROFIBUS Networks In IEEE Tra
35. Density gt Distribution No closed form Location parameter Ld oo 00 Parameters Scale parameter O gt 0 Range Mean oo 00 Variance q Mode o 2 2 Exponential exp o B Interarrival times of customers to a system that occur at constant rate time to failure of a piece of equipment 1 B if x20 Possible Applications Fx otherwise Density Probability Distributions Functions x 127 l e f if x20 Distribution F x O otherwise Parameters Scale parameter B gt 0 Range o 00 Mean B Variance B A Mode 0 3 Triangular triang a apex b Possible applications Used as rough model in absence of data 2 x a if a lt x lt apex b a apex a 2 b x fO if apex lt x lt b b a b apex 0 otherwise Density Peni 2 b a 4 0 a apex b x 0 if x lt a x a if a lt x lt apex b aNapex a Distribution F x 2 b x gt AAA A if apex lt x lt b b a b apex 1 if b lt x a b and Apex real numbers with 4 lt apex lt b 4A isa location parameter Parameters b a isa scale parameter apex is a shape parameter Range a b a b apex Mean S 3 128 Probability Distributions Functions a b apex ab aapex bapex Variance 18 Mode apex 4 Uniform uniform a b Used as a first model for quantity that is felt to be randomly varying between a Possible applications and b but a
36. Features 5 21232 ne a e a a E a E a e a a a 7 2 2 2 Data Cink Layer DEL paa 8 2 2 3 Application Layer AL PROFIBUS DP 00 cece eeceeccescesscesceeseeseeceecaeecaecaecneenseeeenseens 12 2 3 Relevant Details on the Repeater Based Hybrid Wired Wireless PROFIBUS Architecture 12 2 3 1 Repeater Operation apa 13 2 3 2 Wired Wireless Domains Interconnection cesceescesseescesseeseeseeeseeseecaeenaecaeenseesenseens 13 2 33 Trafico A dap tation lt i e e a Ta a e ae E ATE e A a a paie iaaa 14 2 3 4 The Mobility Procedure erre reeeeeene aeee eenereeeenenaceraeneranearanenaa 16 2 4 Relevant Details on the Bridge Based Hybrid Wired Wireless PROFIBUS Architecture 17 2 4 1 Supporting Inter Domain Transactions cccceceesseesceeseeeeeseeesceseecaeeseecaeeaecseenseeeeeseens 18 2 4 2 Supporting Inter Domain Mobility cceccceceesceeceesceeceeeceseeeseeseeesecaeecaeesaeceesseeeeenseeas 21 Ds De UNI lA codes Sata AA AREAS RE UPE CARR E AR ERRORS ERES 25 CHAPTER 3 ERROR HANDLING IMPROVEMENTS FOR THE BRIDGE BASED ARCHITECTURE 27 3 ly Tatrod cti n 3 0 P RR AR UR RR RR ON IDR eo ONCE E RREO RE RREO DRE ERR DR 27 3 2 Error Handling in the TDCi ass 27 3 2 1 Possible IDP Error Situations ccccceccssssesserssonscoesenecoesscessessoeesesscessecsseessonssensensesnss 27 3 2 2 Improving Error Handling in the IDP eee ceeesecseeececeesececeseceeeseceeceeeeeeeeeeeeneeneeenes 28 3 3 Handling T
37. Ferreira 2005 approaches are both compatible with standard PROFIBUS and both extend the PROFIBUS protocol in order to support wireless communication This chapter describes the architecture of the standard PROFIBUS simulation model Section 5 2 which is used by the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim and the Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BHW2PNetSim for the simulation of repeater and bridge based approaches respectively These modules implement most of the relevant features of the PROFIBUS protocol For each module this chapter also presents its main parameters which can be configured by the use of NED files This kind of text file allows the definition of the network configuration and the setting of the module parameters The chapter continues by presenting the implementation of the PROFIBUS DLL simulation model Section 5 3 5 2 PROFIBUS Basic Architecture The PROFIBUS functions are implemented in the HW2PNet Controller Domain Master and Slave OMNeT modules The left side of Figure 5 1 shows the main OMNeT modules used in both simulation models The Hw2PNet module represents the entire network that is the system module in the context of the OMNeT framework The Controller is the module that coordinates the simulation and performs several tasks such as parameterization configuration of the other module instances and it is also responsible for the setup of t
38. Figure 20 DL Estate machine IDMP diosa nianna e A R E E E RA Figure 8 1 Wireless communication model eesssesseeeesseeceesessensseseeseessessacscsceseessensvasassceasacscnsenssesseavssssescasseneeavassesseaveese Figure 8 2 Network scenario j Figure 8 3 View of the MSim front end component 0 0 0 eee ceseesesesescescseseeescseseeecseseaescscsececsesseecseseseeessesseesasseeeeesseseeeeasatee Figure 8 4 Top view of the MSim front end component ecesessescseseeeeseseseescseseeeescsesececseaseecseseseeesseaeeesasseseeesaeseeenasate Figure 8 5 Configuration file related to global simulation parameters excerpt Figure 8 6 Configuration file related to the bs parameters excerpt Figure 8 7 Configuration file related to the agv parameters excerpt Figure 8 8 Output file of wireless mobile station M3 excerpt eeeseseesceesesseseseseeecseseseeecsesesseessesseesasseseeetseseeeeaeaeee Figure 8 9 Output location file of wireless mobile station M3 excerpt Figure 9 1 Repeater based hybrid wired wireless PROFIBUS network Figure 9 2 Bridge based hybrid wired wireless PROFIBUS network errante Figure 9 3 Response time histogram for the message stream SU oa cccecsssssssssssssesssscessecsssecssscsssecsssccssecssuccssecssucsssessseessecssueessecs Figure 9 4 Response time of the message stream S 2 Figure 9 5 Response time histog
39. IQ REQ message 154 and passes it to DLL After that it waits for the DLL notification before processing a new message Meanwhile the DLL sends an IQ REQ message and informs the DMM if it received a response or if the Ts expired During this process the DMM can receive a SBT message and it sets variable sbt rcv to true After ending the current IQ REQ message processing the IDMP Phase 2 ends and Toym 1DMPAbort is stopped If pmm s domain is a wired domain then the IDMP ends and its state machine evolves to the INACTIVE state After the message dispatching procedure presented in Section C 1 3 is performed Simulation Models Implementation Otherwise it evolves to the BEACON_TX state and the IDMP Phase 3 begins Inquiry SubPhase Procedure sbt_rcv false Select a BM from LBMD Build an IQ REQ frame Pass frame to DLL DLL notification 2 Yes y Stop Tomm omP abort Isa wired domain 2 state BEACON_TX Figure C 29 Inquiry SubPhase Procedure Phase 2 The cmm starts the commanded by the DMMs Phase 3 by issuing a SBT message but the remainder of this phase is in the wireless domains For the wired domains the IDMP ends and the message dispatching procedure presented in Section C 1 3 is performed The Beacon messages are used by the wireless mobile stations to evaluate the quality of adjacent radio channels The mM sends a pre defined number of Beacon messages Figu
40. Management FMA1 2 which is responsible for the management of the layers 1 and 2 the PhL and the DLL respectively The PROFIBUS PhL can use the RS 485 standard over twisted pair or coaxial cable for the transfer of data with bit rates up to 12 Mbit s For special applications it is also possible to use other types of physical media like optical fibre power cable or RS 485 IS for intrinsically safe applications The PROFIBUS DLL uses a token passing procedure to grant bus access to masters and a master slave procedure used by masters to communicate with slaves or other masters Slaves do not have communication initiative They are only capable of transmitting a response or an acknowledgement upon master request The token is passed between masters in ascending Medium Access Control MAC address order thus the masters organise network access in a logical ring fashion The PROFIBUS standard considers two different types of Application Layer profiles PROFIBUS FMS Fieldbus Message Specification which is being abandoned due to design complexity and cost and PROFIBUS DP Decentralised Peripherals which is being increasingly adopted for industrial automation and process control applications PROFIBUS DP is particularly suited for the cyclic exchange of data between master Programmable Controllers PC etc and slave devices valves I O devices drives etc In this dissertation DP is assumed to be used in master and slave devices
41. PHY ctrl gateOut ctrl gateln lower gateOut lower gateln sa ctrl gateOut ctrl gateln domain gateOut domain gateln Figure 5 18 OMNeT Slave module Figure 5 19 presents the slave DLL NED definition The address of This Station is set on its TS parameter The Tspr parameter is assigned according to the mode defined in Section 5 2 for these types of parameters which can receive values from PDF functions simple Slave DLL parameters NS numeric pdf tsdr type numeric pdf tsdr part numeric gates in upper gateln lower gateln out upper gateOutl lower gateOut endsimple Figure 5 19 siave DLL module NED definition 5 3 PROFIBUS DLL Basic Implementation In this Section an overview of the PROFIBUS DLL basic implementation in both simulators is provided A more detailed description can be found in Annex C In a standard PROFIBUS a slave state machine is composed by two states OFFLINE and PASSIVE_IDLE In our implementation of both simulators the slave state machine only uses one state the PASSIVE_IDLE state A slave does not have initiative it only responds to requests addressed to it The Slave DLL behaviour depends on the kind of the received frame In the implementation of both simulators a slave is only able to receive SDN SRD and FDL Request Status request frames More details about this behaviour can be found in Sections C 1 5 and
42. Probability Distribution Functions features coooccniconnncnnnnncnononononnnanonnononnnnnncnonnn crono nin on oran anne rn anni anen naar Table A 2 Probability Distributions Functions simulators parameters Table B 1 Bit Error Model simulators parameters VI Performance Analysis of Wireless enabled PROFIBUS Networks Abstract Most of the industrial community is very reluctant to integrate new technologies in their consolidated automation systems either by preconception or by the lack of matureness of these technologies When addressing communication systems for control applications these fears become even more acute Usually these communication systems are based on fieldbus networks which provide adequate levels of performance dependability timeliness and maintainability The PROFIBUS PROcess FleldBUS 1s the most widely used fieldbus with over 15 million nodes worldwide in applications ranging from discrete part automation to process control During the cellular phone and WLAN boom of the last decade soon it became evident that wireless radio based communications could leverage a whole new set of potentialities in field level and control applications Moving parts in machinery hand held equipment wearable computers transportation equipment and autonomous vehicles are just a few examples requiring wireless mobile communications However the requirements of real time systems usually served by fieldbuses impose
43. RHW2PNetSim and BHW2PNetSim through the setting of the location vector parameter The information produced by this simulator is very important for the quality of the simulation results since it provides a way to determine the domain to which a station belongs through time Chapter 9 Comparative Performance Analysis in an Error Free Environment This chapter provides a comparative performance analysis between the repeater and bridge based architectures In this comparison study we studied the influence of varying certain network parameters on the response time and throughput of message streams existing in a network scenario Additionally we also present several response time histograms which characterize the network behaviour of both architectures 9 1 Introduction This chapter provides a comparative performance analysis between the repeater and bridge based architectures For that purpose we performed a set of simulation runs varying several important network parameters on the network performance like bit rate internal delay of the Intermediate Systems ISs and the frame size This comparative performance analysis is based on the message stream response time and on the number of transactions of each message stream These comparative metrics were chosen because the response time of the message streams reflect the timing behaviour of the entire network The number of transactions gives us an idea of the throughput which can be achieved by the
44. TSDR Open Transaction Close Transaction Figure D 1 Screenshot of Timeline Visualisation Tool BHW2PNetSim s5 S4 M2 cP7 crio X 131 7 88 W 41 H 13 Line 2 Send Frame or Token M4 DA 43 SA 2 DAE 4 SAE 4 Sid 4 Mid 1 M3 Time 0 006889 Length Time 0 000330 CPs Length its 165 CP6 3 s2 st M5 Mm ig cP5 E cP8 li EE a l a E a s6 E ml gt 0 006320 us 0 013520 us MM Send Token Frame 4 Receive Token TIDITSDR Open Transaction Close Transaction Figure D 2 Screenshot of Timeline Visualisation Tool RHW2PNetSim In order to the RHW2PNetSim and BHW2PNetSim gather this information the Controller module output gant diagram parameter must be set equal to 1 This diagram is built using two kinds of files One kind contains the network configuration with extension cfg and is generated by the Controller module instance The other kind contains the module instance events with extension evt which are generated by the other modules instances D 3 Output Data Analysis Tool In order to extract information from the output data files and especially due to amount of information generated by he RHW2PNetSim and BHW2PNetSim a tool was developed which provides a fast way to decode text files containing the simulation results and present simulation statistical results in a Tools for Simulatio
45. The ComFunc module instance establishes connections between the connection Point module instances that belong to the repeater through the com func con connections The Connection Point module instances establish the connections to the Domain module instances by the repeater_con connections Repeater Connection_Point repeater_con Domain Wired i com_func_con ComFunc com_fun c_con Connection_Point Moe A A eee re repeater_con Domain Wireless Figure 6 8 Repeater s module instances and their connections ComFunc ComFunc is a simple OMNeT module The NED definition of the ComFunc module is given in Figure 6 9 The main function of this module is to model the internal relaying delay ta see Section 2 3 2 Frames relayed by the repeater are delayed by this module The delay value is assigned to the parameters with the prefix pdf delay simple ComFunc parameters _name gates in out endsimple _pdf_delay_type _pdf_delay_par1 _pdf_delay_par2 _pdf_delay_par3 numeric numeric numeric numeric string ctrl gateln ctrl gateOut Figure 6 9 comFunc module NED definition Figure 6 10 presents part of a configuration file related to a ComFunc module instance This instance is called R1 and the delay introduced is equal to 30 ps The parameter pdf delay type defines if the delay is constant 0 or random according to a PDF see Anne
46. Transadions Thousands N of Transactions Thousands a s Mt M2 M3 M3 S S S2 S Message Stream Message Stream Figure 10 17 Number of transactions using different MeanBER probability in wired and wireless domains and the MSASR rules 3 MeanBER probability of the wireless domain 10 MeanBER probability of the wireless domain 10 Concluded transactions Concluded transactions S Mi Ss e oes Sf Message Stream Message Stream Figure 10 18 Percentage of transactions that do not miss its deadlines using different MeanBER probability in wired and wireless domains and the MSASR rules 114 Comparative Performance Analysis in an Error Prone Environment These simulation results show that using the MSASR the MeanRT of message streams is slightly higher However the number of transaction and the percentage of transactions that do not miss its deadlines are higher The number of transactions increases slightly especially for message streams 9 8 and Fa using the MSASR In spite of a small decrease for message stream S rash the other messages streams increase the percentage of transactions that do not miss its deadline In Section 10 5 2 we presented a performance comparison between the repeater and bridge based approaches using a MeanBER probability of 10 in wired domains and a MeanBER probability of 10 and 10 in wireless domains The percentage of transactions that do not miss its deadline was higher fo
47. applications Moving parts in machinery hand held equipment wearable computers transportation equipment and autonomous vehicles are just a few examples requiring wireless mobilecommunications However the requirements of real time systems usually served by fieldbuses impose the use of predictable and reliable communication services which provide certain guarantees on eventual delivery of packets and delivery times Therefore running real time applications using wireless technology can be especially challenging because the real time and reliability requirements are more likely to be jeopardized than they would be over a wired channel The RFieldbus architecture driven by the European Project IST 1999 11316 consortium has provided a complete solution where multiple segments and multiple wireless cells are interconnected via Physical Layer PhL Intermediate Systems operating as repeaters This solution is compatible with standard PROFIBUS but the fact that all messages are broadcast throughout the network leads tosome problems namely no error containment between different segments and low responsiveness to failures Additionally it is also necessary to set the network parameters in a particular way which guarantees the operation of the network and leads to a lower performance These facts triggered the analysis and proposal of an alternative approach where the Intermediate Systems ISs operate at the Data Link Layer DLL level as bridges T
48. auxiliares de an lise de resultados Foi tamb m desenvolvida uma outra ferramenta que permite simular a mobilidade das esta es m veis Palavras chave Redes de comunica o h bridas cabladas e sem fios Simula o de redes de comunica o Sistemas de Tempo Real Comunica es de Tempo Real Redes Industriais Chapter 1 Overview Fieldbus communications are now the most common way of interconnecting sensors actuators and control devices in manufacturing and process control applications The widespread use of wireless communication systems in the information technology industry and the availability of mature technology has triggered the appearance of fieldbus solutions which operate based on wireless technologies This dissertation compares two of those approaches based in results extracted from simulation One approach extends PROFIBUS fieldbus technology by the use of repeaters which interconnect wired and wireless network segments and the other approach uses bridges for the same purpose 1 1 Introduction Most of the industrial community is very reluctant to integrate new technologies in their consolidated automation systems either by preconception or by the lack of matureness of these technologies When addressing communication systems for control applications these fears become even more acute That is why only a few fieldbus communication systems consolidated their market position due to their technical features and also
49. breaks into several parts so that the received signal is a multiple delayed copy of the transmitted signal This phenomenon is known as multipath propagation Radio wave quality depends on the type of antennas the frequency and the bandwidth On the other hand the strength of a radio wave decreases as the distance between transmitter and receiver increases fading and a radio wave may be altered due to electromagnetic interference and noise Another aspect is the Doppler spread Doppler spread results when radio waves transmitted to or from a moving device undergo a shift in frequency if transmit receive distance changes with time The difference in frequency of the received signal and the transmitted signal is called the Doppler shift In a multipath propagation the angles of arrival of the multipath components are different and each has a different Doppler shift Consequently the wave propagation is very difficult to characterize because it requires a very detailed representation of the objects in the environment and is computationally very complex to treat In Tranter Shanmugan et al 2003 several simulation models for radio wave propagation are presented In Rappaport 1996 the author presents the main wireless communication principles and more specifically it also presents some models for radio wave propagation According to this author the Log normal Shadowing is a more general and widely used model In this models the power at the rec
50. capabilities of the IDP we had proposed to use instead of the SDN service the SDA service which allows the retransmission of faulty frames Additionally the original IDMP had very limited error handling capabilities In an error prone environment this protocol could lead to blocking situations therefore we had proposed an error handling mechanism which permits to solve the problems detected In Chapter 10 we will analyse the behaviour of these enhancements to the original protocols in error prone environments Chapter 4 Technological Context Simulation Software This chapter gives a description of how a simulation study should be performed and it presents the main characteristics of the simulation environment framework used as a basis for the developed of the simulators used within this dissertation 4 1 Introduction A simulation is the imitation of the operation of a real world process or system over time Banks Nelson et al 2001 For the study of any system it is necessary to develop a model that represents the system Since a system is a collection of entities e g people or machines that act and interact together toward the accomplishment of some logical end the details and behaviour of these entities must be represented in that model A simulation model can be classified along three different dimensions A simulation model can be static or dynamic discrete or continuous and deterministic or stochastic A static simulation mod
51. conclude that the bridge based approach presents a better performance The response time of IADTs is much smaller in any case and for the IDTs if it is not smaller is very close of the response time obtained by the repeater based approach and less influenced by parameter settings The throughput is always higher for all message streams in same cases 2000 more Another important aspect is that we have also proposed a set of rules for address attribution to all master station in the network BM included and showed that this set of rules results in performance gains particularly on the number of transactions that do not miss their deadlines Chapter 10 presents a performance comparison between these two approaches considering communication over error prone mediums From the simulation results we can conclude that the bridge based approach presents a better performance Since globally the response time of its message streams is smaller the throughput is always higher and the percentage of transactions that do not miss their deadline is higher for most message streams 11 2 3 Software Tools The software tools developed within this thesis are also a very important contribution for further studies on hybrid wired wireless PROFIBUS networks These tools are available for free and they can be downloaded from the web site http www hurray isep ipp pt activities hw2pnetsim The following list describes a set of tools which were developed Repeat
52. domain Discard the frame Store the response Is there Change the entry LOT to fine Yes any entry in LOT Is ita request frame No Pass to DLL Figure C 17 Receive frame from ComFunc procedure If the frame must be relayed the Bm verifies using the DA frame field if the addressed station belongs to its domain If it does not belong then the BM queues the frame for transmission without changes If it succeeds then this Bm will act as either a BM or a BM es If it acts as BM it checks in LOT for an entry related to this transaction and if it succeeds then it changes the entry s state to finalised stores the response frame and stops timer associated with this LOT entry Otherwise it discards the response frame If it acts as BM the BM stores information about this transaction which will enable it to code an IDF containing the response The received IDF is decoded and queued in exactly in its original frame format as transmitted by the transaction initiator Send IDF Procedure Once the SDA service is only implemented to be used by IDP then the SDA service is presented in this section as send IDF procedure Figure C 18 This procedure is performed by DLL module instance of each Master module instance which is operating as BM As mentioned the SDA service can be modelled using the SDR service 148 Simulation Models Implementation Send IDF Procedure state AWAIT_DATA_RESPONSE retry_cou
53. first term makes sure that there is sufficient difference to the maximum possible T p between two frames recall Eq 2 4 and Eq 2 5 The second term ensures that the master stations start their token claiming procedure in different moments after an error has occurred 2 2 3 Application Layer AL PROFIBUS DP The PROFIBUS DP DP for short protocol IEC 2000 is specially suited for the exchange of data between controllers typically masters and field devices like I O drives or valves typically slaves DP provides the functionalities to configure field devices and to perform cyclic exchange of data between the controller and the field devices The main functionalities of PROFIBUS DP are related to the reading and writing of variables from to slave devices The retrieval of data is made cyclically by the DP protocol according to the timing parameters configured by the user This is an important feature of this protocol since 1t enables the communication between stations in different domains in the Bridge based solution 2 3 Relevant Details on the Repeater Based Hybrid Wired Wireless PROFIBUS Architecture A hybrid wired wireless fieldbus network is composed of wired and wireless stations A wired domain is a set of stations physically connected through a wired bus A wireless domain is composed by a set of wireless stations that intercommunicate either directly or indirectly via wireless e g radio channels Wireless communications ma
54. for both message streams is related to the fact that the initiator is a wireless mobile station which is more influenced by the transmission errors not only on message transmission but also on the IDMP MeanBER probability of the wireless domain 10 MeanBER probability of the wireless domain 10 1600 1600 1400 1400 2 1200 e 1200 Se oe 3 4 1000 S 4 1000 a amp Sa q 3 800 Eg 800 F 2 600 E 2 600 SE oe gt 400 gt 400 200 200 0 0 Si Mi SiM2 Sm S3M3 St Mi Ss M2 S2M3 SBMS Message Stream Message Stream Figure 10 11 Number of transactions using different MeanBER probability in wired and wireless domains Figure 10 12 presents the percentage of transactions that do not miss its deadline for each message stream In the repeater based scenario the percentage of transactions that do not miss the deadline is higher than in the bridge based scenario especially for message streams Sy and S MeanBER probability of the wireless domain 104 MeanBER probability of the wireless domain 10 100 100 no no 5 5 3 El O 3 98 E 98 5 amp ES 9 ES 9 7 3 3 9 3 8 oO a 3 s amp s 94 94 gmi Ss Ae sy SNS SM S M2 se Sue Message Stream Message Stream Figure 10 12 Percentage of transactions that do not miss its deadline using different MeanBER probability in wired and wireless domains Although the results seem to be more favourable to the repeater based architecture particularly in terms o
55. had been implemented according to the principles presented in Alves 2003 which guarantees that there is no need to queue a frame in the first repeater which relays a frame Nevertheless in the following repeaters the queuing of frames might occur For that reason each Connection Point module instance is provided with a queue 6 3 2 Mobility Procedure The mobility procedure is triggered in a periodical fashion The periodicity is defined by the _tmob parameter In this implementation we assume that the MM is a dedicated master 1 e it only controls the mobility procedure For that reason a mobility related timer is loaded with a value defined by the _tmob parameter which is started when it starts operating After the mobility related timer expires and when it holds the token frame it broadcasts a Beacon Trigger BT frame The BT is an unacknowledged frame therefore after sending the BT 66 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model frame it waits Typ to schedule the next action according to the message dispatching procedure presented in C 1 3 Usually it passes the token frame to its Next Station NS After sending the BT the mobility related timer is reloaded Connection Point module instances that are operating as BSs receive a BT frame and after relaying the received BT frame they start sending a pre defined number defined by n beacon parameter of Beacon frames see Section C 2 1 for more details
56. iea 71 7 4 Simulator Implementation meenen iiie e ei aE i 75 7 4 1 Bridge IDP Functionalities ccccccscsesecsscsssocssensccsesnsceceseesensesesseesscessessensesstectsesteenes 75 7 4 2 Operation of the IDMP related Agents cccccccseesseseceseeseceeceseceeceeeeeeeeeeeceseeeneeneeeaes 75 TS SUMMALY in EE E A E AE idas 79 CHAPTER 8 MOBILITY SIMULATOR e a r ss beu os clone cobced ceeteassabcecgneneedes snnsnsbecedeve cessed shes 81 SLI UCM A AE OR RE 81 8 2 OpenSceneGrap a sis amas a i E E E E EE R E O ea 82 821 Scene Graph Concept nrinn ione o R E E O E hes 82 PONDA E DATE A fam E E E NN 82 8 3 Wireless Communications Radio Signal Propagation csccsccssesesceseeeeceseeeeeeseeseeeseeneeenes 82 4 Simulation Model i n ias 84 H 8 5 Simulator Architecture tai an a tn dla lado cas 85 8 6 Simulator Configura e dd a desea thes 87 8 7 Output Data Files nan Sas Sie eect a RAS Antes 88 SOS UM A o cht uate A n 89 CHAPTER 9 COMPARATIVE PERFORMANCE ANALYSIS IN AN ERROR FREE ENVIRONMENT 91 91 tro ducho tt til tia das Co AGO esos d o 91 9 2 Network Scenario Configuration cccscescesessscesceeeceseeeeeeseeseecaseeeecaeeseecaeeaecsaeeseesenseenseees 91 9 2 1 Repeater Based Scenario cecccescesscescesseeseessecseeesecaeesaecaseaeceaeeaeceeseaeseeseeeseseeeeseneeneeenes 92 9 2 2 Bridge Based Scenario coocococnooncononononnnnononnnonnonn non nono non nron non n ron nro nr nn rra n rro rra r rara rana
57. network This chapter starts by presenting in Section 9 2 the network scenario and respective parameters to be used on the comparison study Section 9 3 presents simulation results upon variation of the some network parameters Finally in Section 9 4 we summarize our comparative study 9 2 Network Scenario Configuration The network scenarios presented in the Figure 9 1 and Figure 9 2 were used to compare the performance of the two approaches In the remainder of this chapter these scenarios are referred to as network base configuration The network comprises four domains two wired domains D and D and two wireless domains D and D These domains are interconnected by three Intermediate Systems ISs which act either as a repeater or as a bridge according to the correspondent network scenario It is assumed that the wireless domains D1 and D3 use the 802 11b Direct Sequence Spread Spectrum DSSS PhL at 2 0 Mbit s coding every character using 8 bits The frames have a head of 32 bits and no tail The reason for the use of a frame head is related to the specific requirements of the DSSS modulation schema used by 802 11b These bits are used by the receiver to acquire the incoming signal and to synchronise the demodulator The wired domains D and D use a standard PROFIBUS Physical Layer PhL operating at 1 5 Mbit s and 0 5 Mbit s for domains D and D respectively Since these domains use the RS 485 standard for the transm
58. occur during the evolution of the IDMP are analysed and the section continues by discussing two alternative mechanisms which support the error handling Section 3 4 presents the reasons which supported the error handling mechanism implemented in our simulator 3 2 Error Handling in the IDP 3 2 1 Possible IDP Error Situations As outlined in Chapter 2 the bridge based approach requires a new kind of transaction Inter Domain Transaction IDT in which the involved stations do not belong to the same domain The IDP error handling mechanism is a simple mechanism based on timers which are handled by the IDT s BM Therefore whenever a new transaction is opened at the BM LOT a timer the BM IDT Abort Timer TBm IDT4bort is loaded with the worst case time required by the BM to complete a transaction for additional details about how to obtain these values see Ferreira 2005 If the timer expires before the completion of the transaction then the entry at the LOT is deleted and the IDT can be reinitialised by the next initiator s request In order to illustrate the influence of transmission errors in the IDP consider the network example presented in Figure 2 10 Figure 3 1 presents a simplified timeline regarding a transaction 28 Error Handling Improvements for the Bridge based Architecture between master M3 and slave S6 When master M3 is holding the token frame 1t sends a PROFIBUS request addressed to slave S6 Using the informati
59. of the request After receiving the acknowledge frame the initiator prepares the next message cycle Otherwise it performs a number of retries according to the setting of the max_retry limit parameter These changes are expected to improve the error correction characteristics of the IDP since the protocol can now recover from frame errors faster than using the timer associated with each LOT entry In the remainder of this section we show a comparative analysis between both services Station Delay Responder Tspr Request Frame sc Acknowledge Frame Idle Time Tip Figure 10 3 Message cycle using the SDN or the SDA service For both cases the MeanRT the number of transactions and the percentage of transactions that do not miss the deadline are presented in Figure 10 4 Figure 10 5 and Figure 10 6 respectively In these figures the simulation results which were obtained using the SDN service are identified by the SDN tag and the SDA tag for simulation results obtained using the SDA service Using the SDA service the MeanRT is globally smaller for all message streams than using the SDN service except for message stream S an IADT The reason for these results is that there are less failed transactions therefore IDTs errors are recovered using PROFIBUS retries instead of using the timer on the BM LOT The exception is the message stream S which is affected by the increase on the network traffic Globally t
60. of this figure and on the remaining figures in this chapter a R or a B before the message stream symbol RS and BS specify that the values are related to the repeater based or to the bridge based architecture respectively In the repeater based scenario the minimum response time MinRT value is equal to 1 23 ms and the maximum response time MaxRT value to 16 09 ms 1 0 9 0 8 El 0 7 E 0 6 E 0 5 E 0 4 E 0 3 E 0 2 E 0 1 E od Tol Wa ee 2 3 4 5 6 T 8 9 10 11 12 13 14 15 16 17 Response Time ms Figure 9 3 Response time histogram for the message stream S In the bridge based scenario the MinRT value and the MaxRT value of message stream S i are 0 27 and 4 27 ms respectively Nevertheless it is important to note that 96 20 of the transactions present a response time smaller or equal to 1 ms see Table 9 7 and 98 25 of the transactions have a response time value smaller than 1 23 ms which is the MinRT of the repeater based scenario In this scenario S benefits from the smaller setting of the T p parameters as well as from the traffic segmentation resulting from the use of the bridges The first reduces the message cycle duration while the second reduces the traffic within domain D Table 9 7 Response time of the message stream S Interval RS BS Interval RS BS Interval RS BS ms ms Go ms 4 Go 0 1 0 96 20190 16 7 12 43433 0 112 13
61. only ns 2 and OMNeT were assessed has possible solutions for the implementation of the simulation models in this dissertation While ns 2 is a network simulation classic it has many drawbacks when compared with OMNeT which is a more modern and structured simulation package The following summarises a number of advantages of OMNeT over ns 2 The OMNeT simulation kernel is a class library the components are developed as any other class library and then linked with the executable library There is no need to modify OMNeT sources anywhere In contrast ns 2 tends to be a bit monolithic to add implementations to it it is necessary to download the full source and modify it in several places 40 Technological Context Simulation Software OMNeT follows a modular approach the model is assembled from self contained building blocks These components are reusable in other simulations ns 2 has some considerably detailed built in concepts about nodes agents protocols links packet representation network addresses etc This often increases the difficulty in developing models that include even slightly different concepts OMNeT is completely flexible and generic it is possible to simulate anything that can be mapped to active components that communicate by passing messages In OMNeT it is possible to fight model complexity by using hierarchical design any complex component can be implemented as one unit or built out
62. ring Error skipping A master station must remove itself from the logical ring when a token frame is transmitted in which its address is skipped i e the address of TS lies within the address range spanned by the sender and the receiver of the token frame DLL Frame Formats PROFIBUS DLL defines three types of request response frames which are the Fixed Length with no Data Field the Fixed Length with Data Field and the Variable Data Field Length as illustrated in Figure 2 1 a c and d respectively Each of these three types includes the following fields Destination Address DA Source Address SA and Frame Control FC The FC field is an octet where the frame type is specified and the function code for more details the reader is referred to Section 4 7 3 of CENELEC 1996 These frames also include the Start Delimiter SDx Frame Check Sequence FCS and the End Delimiter ED Variable data field length frames additionally contain two Data Length fields LE and LEr and they can optionally include the Destination Address Extension DAE and Source Address Extension SAE in the Data field These extension fields can be used to identify the AL service which originated the frame as well as the destination service PROFIBUS also defines the Short aCknowledgement frame SC and the Token frame illustrated in Figure 2 1 b and e respectively The first consists of a single byte frame and it is used as positive ackn
63. sIntro duce aa EE eee 159 D 2 Timeline Visualisation Tool voii iaa dde da 159 D 3 Output DataAnalysis Tool A eee ese OU ee da 160 D 3 1 Message Stream Response Time ccccesecesesseceseesecescesecescesecnsceeeeeeeeeeeaseaeeeneeaeenees 161 D322 State Machine a ee ee RR SR E 163 D 3 3 Probability Distribution Function c cccceseesecesceeecesceeeceseeeeeneeeeeesseesecaeeaeceeeaeenees 163 D3 4 Btt Error Model sho cee Re alee Miele ies cede hee RS a EIS 164 ANNEX E ACRONYMS AND SYMBOL cssssssssssssssesscsossscsassssecesassssenenasessensessossessosssesnssosssssusensesesassace 169 ESA CEOMYINS o e ca AR aca ES rp 169 IS E A E A O 171 IV List of Figures Figure 1 1 Repeater based network example minimo ARE 3 Figure 1 2 Bridge based network example 4 Figure 2 1 PROFIBUS DEL frame formats sr aii 0 Figur 2 2 Idle Time p tametet dass 11 Figure 2 3 Repeater based hybrid wired wireless PROFIBUS network example 3 Figure 2 4 Timing behaviour of a repeater x 14 Figure 2 5 Increasing queuing delay by a repeater A 15 5 6 6 7 8 9 Figure 2 6 Using additional idle time for media adaptation esesseeeeeseeeeecseseesescseeeeecsesseesscseseeesseseeecasseeeesseeeeeaeaee Figure 2 7 Slot Time Est ade a pata nt a Figure 2 8 Mobility procedure Figure 2 9 Basic components of a bridge Figure 2 10 Bridge based hybrid wired wireless PROFIBUS network
64. sesuais colada iaa diia E a sregesaanisvapesaraviaepensaeed Figure C 19 IDMP Phasel Prod A A en ce eel TE Figure C 20 IDMP Phase 2 procedure Figure C 21 handleIDMPMessage msg function pseudo code algorithm Figure C 22 processIDMPmessage msg function pseudo code algorithm Jl 51 Figure C 23 processRSMPmessage function pseudo code algorithm eee 152 Figure C 24 handleTimers msg function pseudo code algorithm cseseesescseeeeseseeeeeeseseecaeseseeeceeseseeesseaeeeeaesesseeees 152 Figure C 25 dllReceiveToken function pseudo code algorithm sa Figure C 26 processRBTmessage function pseudo code algorithm ereta 153 Figure C 27 processIDMPmessage msg function pseudo code algorithm eee 153 Figure C 28 handleTimer msg function pseudo code algorithm 2159 Figure C 29 Inquiry SubPhase Procedure Phase 2 154 Figure C 30 Send Beacon Procedure Phase 3 154 Figure C 31 Discovery SubPhase Procedure phase 4 0 ccceccesesesseseseseeeesesessescseseseesescseeseseseecscsesesesseaseesacsesseesacaeeeeassesaeeees 155 Figure C 32 Message dispatching procedure IDMP ssessseseeseseseseesesesseeeseseseeescsseecseseseeacseseecsescsseesacsesseavscaeeeasseeseeeas 156 Figure C 33 Mobility procedure sie Figure D 1 Screenshot of Timeline Visualisation Tool BHW2PNetSim
65. soft real time systems it is acceptable to miss some of them occasionally In practical engineering contexts the occasional loss of some deadline can be tolerated This is either because the consequences of the loss can be negligible i e one defectuous part per thousand or because the robustness of the involved control algorithms imply the ability to react properly at the next invocation step without serious consequences Within this context some commercial Siemens 2005 and research solutions Lee and Lee 2001 Willig 2002 Miorandi and Vitturi 2004 for providing the traditional PROFIBUS with wireless extensions have been proposed Nevertheless these solutions are quite limited either in terms of number of segments or wireless cells and in the support of mobility The RFieldbus architecture RFieldbus 2000 RFieldbus 2000a driven by the European Project IST 1999 11316 consortium has provided a complete solution where multiple segments and multiple wireless cells hereafter segments and wireless cells will be referred as domains are interconnected 2 Overview via Physical Layer PhL Intermediate Systems operating as repeaters This solution henceforth referred as repeater based is compatible with standard PROFIBUS but the fact that all messages are broadcast throughout the network leads to some problems namely no error containment between different domains and low responsiveness to failures These facts triggered the analysis
66. the BEACON_TX state When the Dmm is in the BEACON TX state it passes to the DLL a number of Beacon messages defined by n beacon parameter transition 27 At the end of the Beacon messages transmission the DMM state machine evolves to the IDENT state and the DLL state machine evolves to the DISCOVERY state transition 28 When the DMM is in the IDENT state it sends Discovery messages D_REQ addressed to the wireless mobile stations to the DLL in order to detect which stations belong to its domain When the DLL sends D_REQ message its state machine evolves to the AWAIT DISCOVERY RESPONSE state transition 29 and as soon as it receives a response or when Ts expires it returns to the DISCOVERY state transition 30 Whenever this happens the DMM is notified and in case it has received a message it passes it to the DMM When this operation ends the DMM builds and broadcasts RU messages Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 79 AWAIT_INQURY_RESPONSE BEACON_TX 28 26 27 22724 23 INQUIRY MODE AWAIT DISCOVERY RESPONSE 32 25 21 20 1 31 Figure 7 20 DLL state machine IDMP After that the pum state machine evolves to the INACTIVE state and the DLL state machine enters into the USE_TOKEN state transition 31 Therefore the IDMP ends and DLL perform another action according to the dispatching message procedure presented in Section C 1 3 In order to handle transmission e
67. the Beacon Trigger frame from the Mobility Master MM sends to the Controller through ctrl con connection a message indicating that Beacon frames will be transmitted After ending the transmission of the Beacon frames according to the procedure described in C 2 1 they sends to the controller a new message indicating that they finish the Beacon frames transmission The Master and Slave module instances which model wireless mobile stations at reception of the Beacon frames send to Controller through ctrl con connection a message indicating which domain they want to change according to their location vector parameter The Controller manages this information in order to disconnect the Master or the Slave from Domain to which they are connected and connect to the destination Domain However a Master or a Slave can only change to a domain where the Beacon frames were transmitted C 3 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulator In this section are detailed the procedures specific of the BHW2PNetSim These procedures are presented separately and are related to the state machine diagram transitions described in Section 7 4 146 Simulation Models Implementation C 3 1 Inter Domain Protocol Receive Frame from Domain Procedure Figure C 16 presents the procedure performed by a Bm when it receives a frame from its DLL module instance The BM starts by checking if the frame is addressed to a station in another domain usi
68. the GMM sends the SMP message If the GMM receives a RSMP message from all BMs in the network before the expiration of the Toumpister it stops both timers If Toum p14lerr expires i e if it did not receive a RSMP message from all BMs it sends again a SMP message and waits for the reception of a RSMP Error Handling Improvements for the Bridge based Architecture 33 message from all BMs in lack If it receives a RSMP from the BMs in lack before the expiration of the Temm prabor 1t evolves to Phase 2 Otherwise the IDMP is aborted Each BM starts its Tam iDmpAbor When it receives the first SMP message The BM will reply with a RSMP message when its LOT is empty If a BM has already sent a RSMP message and it receives again a SMP message it also replies again with a RSMP message If the IDMP takes longer than Tgm IDMPAbor the BM aborts the IDMP and starts accepting new IDTs Otherwise the Tgm IDMPAbort 1S stopped at the end of Phase 2 Considering the system scenario illustrated in Figure 2 12 Figure 3 9 represents a simplified timeline of IDMP Phase 1 For the sake of simplicity it is assumed that there are no open transactions in the LOTs of the BMs in the network When the IDMP is triggered the GMM M6 starts Tomm pister And Temm P1Abort and a SMP message is broadcasted by network As soon as the BMs receive a SMP message they start the Tgm IDMP4bort and no new IDTs are accepted In this example it is assumed that there are no open tra
69. the use of predictable and reliable communication services which provide certain guarantees on eventual delivery of packets and delivery times Therefore running real time applications using wireless technology can be especially challenging because the real time and reliability requirements are more likely to be jeopardized than they would be over a wired channel The RFieldbus architecture driven by the European Project IST 1999 11316 consortium has provided a complete solution where multiple segments and multiple wireless cells are interconnected via Physical Layer PhL Intermediate Systems operating as repeaters This solution is compatible with standard PROFIBUS but the fact that all messages are broadcast throughout the network leads to some problems namely no error containment between different segments and low responsiveness to failures Additionally it is also necessary to set the network parameters in a particular way which guarantees the operation of the network and leads to a lower performance These facts triggered the analysis and proposal of an alternative approach where the Intermediate Systems ISs operate at the Data Link Layer DLL level as bridges This approach required two new protocols one for supporting the communication between stations in different network segments the Inter Domain Protocol IDP and another to support the mobility of wireless stations between different wireless segments the Inter Domain Mobili
70. to control from which Bms the cmm received a Ready to Start Mobility Procedure RSMP message a copy of the LBMN is used the BMlist After that it builds a Start Mobility Procedure SMP message starts the GMM_Phase_1 Alert Timer TGmMm PlAlert and the GMM_Phase _1 Abort Timer TemaepiAbort timers sets the retry variable to false which is used to control the retry of the SMP message evolves to the WRSMP state and then passes the SMP message to the DLL Simulation Models Implementation 149 After that it waits for a RSMP message from the BMs in the network Whenever it receives a RSMP message from a BM it removes the corresponding address SA from the list It stays on this state until receiving a RSMP form all Bms in network i e until the list of BM address is empty Meanwhile if the Temm P14lert EXpires it sends again a SMP message and continues waiting for the RSMP from the Bus However if Temm P14bort eXpires the cum aborts the IDMP and then evolves to the INACTIVE state BMlist LBMN Start timers Tomm ptalert and Tomm PtAbort Build SMP message state WRSMP retry false Pass SMP to DLL Receive a RSMP 2 Is SA in Yes BMlist No Yes Remove SA from BMlist No Tomm PtAlert expired 2 Is BMlist empty 2 Tomm P1Abort expired 2 No Figure C 19 IDMP Phase 1 procedure Figure C 20 presents the IDMP Phase 2 procedure This procedure is similar to the previous but now the M
71. when its LOT is empty it replies with a RSMP message to its father node Therefore Phase 1 ends when the GMM has received a RSMP message from its son nodes This procedure supports errors on the transmission of the SMP message and on its reply message Figure 3 5 illustrates this mechanism considering the network example presented in Figure 2 12 GMM M6 repeatedly sends a SMP message to BMs M6 and M9 until receiving a RSMP message from them BMs M6 and M9 stop accepting new IDTs and repeatedly send a SMP message until receiving a RSMP message from BMs M5 and M10 respectively When leaf nodes BMs M8 and M7 have their LOTs empty they reply with a RSMP message to their father nodes Phase 1 ends when the GMM receives a RSMP message from all of its son nodes BMs M6 and M9 Note that the occasional transmission errors are overcome by periodic message re transmissions In Figure 3 5 a transmission error occurs when BM M10 is sending a RSMP message consequently BM M9 does not receive a RSMP message from BM M10 In order to receive a RSMP message from BM M10 BM M9 re transmits the SMP message until receiving a RSMP message The mechanism concerning to Phase 2 is very similar to the mechanism used in Phase 1 but now the tree is composed by the GMM as root and the DMMs as leafs Error Handling Improvements for the Bridge based Architecture 31 Phase 1 starts BM BM BM GMM BM BM BM M7 M10 M9 M6 M6 M5 M8 Token
72. wireless stations between different wireless domains the Inter Domain Mobility Procedure IDMP 1 3 Research Objectives The main objective of this dissertation is to compare the timing behaviour of the repeater and bridge based approaches over error free and error prone environments Additionally we also intended to show that the bridge based approach implementation is feasible and propose additional error detection and correction mechanisms which would improve its performance over error prone environments To achieve these objectives two simulation tools have been developed one to the repeater based approach and another to the bridge based approach and a set of result analysis tools Additionally we have also developed another tool to simulate the mobility of wireless stations 1 4 Contributions of this Dissertation The main research contributions of this dissertation are the following An analysis of the timing behaviour of the repeater and bridge based approaches based on simulation results The proposal of error correction and detection protocol for the bridge based approach A comparative performance analysis between repeater and bridge based approach considering communications over error free medium Sousa Ferreira et al 2006 Overview 5 A comparative performance analysis between repeater and bridge based approach considering communications over error prone medium Sousa and Ferreira 2007 The above rese
73. with different frame formats or different bit rates for instance A repeater does not perform any address filtering thus a broadcast network is created Consequently the use of repeaters implies a single MAC address space To give a better intuition of the interconnection of wired and wireless communication systems a repeater based hybrid wired wireless network example is presented in Figure 1 1 The network comprises four domains Two are wired domains D and D and the other two are wireless domains D and D Three intermediate systems operating as repeaters R1 R2 and R3 interconnect the wired and wireless domains The network comprises two wired masters M1 and M2 five wired slaves S1 S2 S3 S4 and S5 and three wireless mobile stations two masters M3 and M4 and one slave S6 Wireless communications are relayed through Base Stations BSs A BS operates using two wireless channels one to receive frames from wireless stations uplink channel and other to transmit frames to wireless stations downlink channel The network comprises two BSs BS1 and BS2 which relay the wireless communications of wireless domains D and D respectively Overview 3 nl Jigs Wireless domain domain Y Y N lt y 3 EA 1 D Radio eee l ma 1 i 4 Uplink q Downlink Figure 1 1 Repeater based network example In this approach all messages transmitted either by t
74. with no Acknowledge PROFIBUS Standard SLR Single Logical Ring SMP Start Mobility Procedure SRD Send and Request Data PROFIBUS Standard TBM IDMPAbort BM_IDMP_Abort Timer TBM IDTAbort BM IDT Abort Timer TCP IP Transport Control Protocol Internet Protocol TpMM IDMPAbort DMM_IDMP_Abort Timer Tomm P14bort GMM Phase 1 Abort Timer Temm P14lert GMM Phase 1 Alert Time TGMM P2Abort GMM Phase 2 Abort Timer T GMM P2Alert GMM Phase 2 Alert Time Tau Gap Update Time TI Transaction Identifier Tip Idle Time Tipm Minimum idle time Trr Real Rotation Time TS This Station PROFIBUS Standard Tsp Station Delay of the Initiator Time Tspr Station Delay of a Responder Time Ts Slot Time Tsm Safety Margin Time Tsyn Synchronisation Time Tip Transmission Delay Time Tra Token Holding Time Tro Time Out Time Trr Target Rotation Time WCRT Worst Case Response Time E 2 Symbols Symbol Description a Lower limit of an interval apex Mode b Upper limit of an interval c The speed of the light d The distance between transmitter and receiver in meters The close in reference distance which is determined from measurements close to the transmitter D Communication Domain i T Carrier frequency in Hertz 172 F J G G G L maxTspr min Tspr MeanBER n Nem Nem N y Pb Php Polg Pber Pr err Pg Pelb Pelg PL d PLo do Pd si tp Tem Tem Tup inj ng inj Csr Tip Tip2 ta bik To
75. 00 2 30 BS2 D3 49 17 0 00 0 00 48 80 104 40 0 400500 0 400000 0 405000 2 30 BS2 D3 48 33 0 00 0 00 49 34 120 04 0 600500 0 600000 0 605000 2 30 BS2 D3 47 50 0 00 0 00 49 87 117 47 0 800500 0 800000 0 805000 2 30 BS2 D3 46 67 0 00 0 00 50 42 127 38 Figure 8 8 Output file of wireless mobile station M3 excerpt Figure 8 9 presents part of the output file related to the agv module instance M3 which contains the domain sequence to which a wireless mobile station belongs to during the simulation runs In this file the information is organized as tuples separated by a colon The first element indicates during how many handoff procedures the agv module instance stays in the domain referred in the second element In this example the agv module instance M3 stays in domain D1 for 43 network handoff procedures and then it changes to the domain D3 where it stays for another 10 handoff procedures The information contained in this file can be assigned to the location vector parameter see Section 6 2 3 for details which determines to which domain a Master or a Slave module instance belongs to 43 D1 10 D3 29 D1 1 D3 12 D1 2 D3 3 D1 3 D3 3 D1 Figure 8 9 Output location file of wireless mobile station M3 excerpt 8 8 Summary This chapter describes the Mobility Simulator which permits to determine to which domain a wireless mobile station belongs to at a specific point in time That information is then feed into the
76. 1404965 039 97 846 2575 14 23609 0 396 96 519 2588 13 8597 0 408 97 94 2626 13 99751 0 425 97 45 2566 1469568 D 364 98 94 2590 13 95077 nao aaa EAN 14 29109 YOO wWN Figure D 6 Screenshot of spreadsheet created by Message Stream Response Time Central Limit Theorem option D 3 2 State Machine A Master module models a PROFIBUS master and additionally can model the BM DMM and GMM functionalities separately or simultaneously Each of these elements has its own state machine The information about the state machine transitions of these elements are recorded in text files with stt extension Each line of this kind of file represents a transition The transition instant Time the state name and a brief explanation about reason that causes the transition appear in first second and third column respectively Figure D 7 illustrates an example of this kind of files related to a Master module instance Time State name Description 0 019468 USE_TOKEN Received token from 6 0 019534 AWAIT_STATUS_RESPONSE Waiting for a FDL response from 3 0 019648 USE_TOKEN Slot time expired 0 019648 PASS_TOKEN Trying to pass the token to 5 0 019671 CHECK_TOKEN_PASS Waiting for activity from 5 0 019759 ACTIVE_IDLE Activity detected from 5 0 019848 USE_TOKEN Received token from 6 0 019914 AWAIT_STATUS_RESPONSE Waiting for a FDL response from 4 0 020028 USE_TOKEN Slot time expired 0 020028 PASS_TOKEN Trying to pass the
77. 2PNet controller _inter_domain lt l gt lt n gt R1 lt n gt lt cp gt CP5 CP8 lt cp gt lt pos gt 400 150 lt pos gt lt I gt lt l gt lt n gt R2 lt n gt lt cp gt CP9 CP6 lt cp gt lt pos gt 50 150 lt pos gt lt I gt lt l gt lt n gt R3 lt n gt lt cp gt CP10 CP7 lt cp gt lt pos gt 120 400 lt pos gt lt l gt Figure 6 4 Configuration file related to the controller module instance of the RHW2PNetSim excerpt The meaning of most of the tags used on the domain string has been described in Chapter 5 lt cp gt and lt cp gt tags enclose the names of the connection Point module instances that are connected to the Domain module instance the names must be separated by colon lt bs gt and lt bs gt tags enclose the name of the connection Point module instance which operates as a BS of a wireless domain In the particular case of Figure 6 4 the second domain D2 is described by lt d gt lt n gt D2 lt n gt lt m gt Ml M5 lt m gt lt s gt S1 82 83 lt 5 gt lt cp gt CP6 CP5 lt cp gt lt bs gt lt bs gt lt pos gt 200 1 50 lt pos gt lt d gt it is composed by two Master module instance M1 and M5 and three Slave module instances S1 S2 and S3 and this Domain module instance is connected to two Connection_ Point module instances CP6 and CP5 The Domain module instance is depicted in the screen at position 200 150 The parameter inter domain is a string that is similar to the domain string Thi
78. 7 B S 0 0 0 0 0 75880 0 31097 0 00858 0 00014 27 Interval R S B S ms Y 70 21 0 0 121 22 0 0 122 23 0 0 123 24 0 0 124 25 0 2 60896 125 26 0 0 64214 126 27 0 0 00028 It is important to note that if message stream S had been queued in first place instead of third the results obtained in the repeater based scenario would be equal to 1 41 ms and 16 30 ms for MinRT and MaxRT respectively In the bridge based scenario the results would be equal to 8 18 ms and 25 05 ms for MinRT and MaxRT respectively From this results we conclude that the message streams queuing order has higher influence in the repeater based scenario than in the bridge based scenario due to the fact of a single transaction in the repeater based scenario takes much more time It is also important to note that the bridge based scenario presents a much higher throughput than the repeater based scenario Figure 9 6 shows a histogram of the number of transaction for each message stream N of Transactions Thousands 1600 1400 1200 1000 M1 Si M2 S Message Stream M3 SA w o Figure 9 6 Number of message stream transactions In the bridge based scenario the number transactions for message stream S is approximately 500 more for message streams SP and SJ the ratio drops to 240 more since these are IDTs The main reason to this disparity in results is due to the traffic segmentation
79. 93 92 35 ESTAS O 94 9 3 Performance Adalid ica 95 9 3 1 Base Configuration Results c 2 cccc ccccccccccseececsesessntencecceseteetesee condecunsteeseesdaseddedectdecddestenee 96 9 3 2 Variability of the Message Stream Response Time as a Function of the Bit Rate 99 9 3 3 Variability of the Message Stream Response Time as a Function of the ISs Delays 100 9 3 4 Variability of the Message Stream Response Time as a Function of the Maximum Frame LENA trie een win ates arn eee nets ie a ees we en aT 101 D A SUMMA Y inneni aE eid lata tes E nO ETR deste nate eis ese aan te een een te 102 CHAPTER 10 COMPARATIVE PERFORMANCE ANALYSIS IN AN ERROR PRONE ENVIRONMENT NO A 103 TO LItrO UC A A A AA a i 103 10 2 Gubert Elliot Channel Models ici at di Nao EA da asa 104 10 3 Network Scenario Configuration ooooccioninnnonnnononnnonncnnnonnonn non nnnn non nronn corr on nro rra nr nora 105 10 4 Improved IDP Performance 0 cccecceeccessesseesceeeecseeesecacesaececesaecssensecescsseseseseseneeeeseneenseenes 106 10 5 Performance Comparison between Repeater and Bridge based Architectures 107 10 5 1 Comparison Using the same BER in All DomainS oononcnocnnnncnocononnononancnnnoncnnonncnnons 107 10 5 2 Comparison using different BERs in Each Domain oooonnonicnocicnnccnoccononcncnnnananncnncnnons 109 10 6 Address Assignment Rules ooooococcnccionoconocnnononnonnnonnonnconnonn con e nro AEE E E N 110 10 6 1 Compar
80. Alves Bangemann et al 1999 provides a seamless handoff for all kinds of wireless mobile stations master slave Due to the broadcast nature of the network the mobility procedure just encompasses a mechanism for radio channel assessment and switching The basics of this procedure are outlined next The Mobility Master MM ie the master that has the responsibility of triggering the mobility procedure sends periodically a special unacknowledged frame the Beacon Trigger BT This BT frame is broadcast to the entire network and causes each BS to start transmitting a pre defined number of Beacon frames in its downlink radio channel Wireless mobile stations receive these Beacon frames assess the signal quality of all downlink radio channels and switch to the radio channel set with best quality Figure 2 8 shows the simplified operation of the mobility mechanism considering the network scenario depicted in Figure 2 3 in which master MM is operating as Mobility Master Tie a bogap EE gt es Fr O O O O O E EE EH NaN E dr O EL E OSS STA Oll O Repeater Delay Beacon Trigger crx Beacon Frame Token frame Figure 2 8 Mobility procedure In Alves 2003 the author shows how to calculate the number of beacons to be transmitted in each domain which guarantees that every wireless station is capable of evaluating the signal quality of each radio channel set The setting of this parameter is different between each repea
81. IDTs 9 3 4 Variability of the Message Stream Response Time as a Function of the Maximum Frame Size The variation of the frame size impacts the duration of message transactions not only due to the increase on the message cycle time but also in the case of the repeater based approach due to the need to increase some network timing parameters To perform this comparison we have chosen to vary the frame size of message stream S This message stream is the first message stream of master M3 a wireless mobile station and the responder is slave S4 which belongs to domain D The size of the request and response frames varies between 15 and 250 bytes Figure 9 9 depicts those results Response Time ms Response Time ms 15 50 100 150 200 250 15 50 100 150 200 250 Frame Size byte Frame Size byte Figure 9 9 Influence of the maximum frame size on response time Once again in the repeater based scenario there was the need to increase the period to 160 ms and to adjust the Tsz Tip and Tpz parameters for every frame size The period for the others messages streams was set using triang 140 160 180 and the offset was set using triang 0 140 160 All message streams are affected by the increase of the maximum frame size In the bridge based scenario this influence is stronger for message streams which are routed through the same domains as S which is the case of message stream Si But for S that influence is ignora
82. IMP Errors seu sciet cece Setese sce Becca esta wedce det data 29 3 3 1 Possible IDMP Error Situations oooooncnioninonnonnnononnconnonncon non nnonn E E E 29 3 3 2 Tree like Topology Based Mechamism cccscccssesseesseesececesececeseceeceseeeeeneeesesseeeneenes 30 3 3 3 Timer Based Mecha a citas 32 3 34 The Adopted Mecha Mica eds 35 3A SUMIMALY A cosunh cubs RD RE A ND RR 36 CHAPTER 4 TECHNOLOGICAL CONTEXT SIMULATION SOFTWARE cssssssssessssssssessecescsessessees 37 AIN UC ELON wos soe A UNE OR A ERR da RR O 37 4 2 The Basic Steps of a Simulation Study ooooonoonicnnoconononnnonnnononnnonncon non nnon non nron non nn corro nnnn rra nnnnnnns 37 4 3 Simulator Implementation ccccceseescesceescesceesceseeseceseeeeecaeeeaecaeesaecasesaecaseeaeceesesecerseeeneeses 39 4 4 OMNeT Objective Modular Network Testbed in C eee 40 4 4 1 Messages Gates and Links eee ron nono non nono ron cron ron n ron n nn rra nr rn nnnnnnos 40 4 4 2 Modelling Delays Bit Error Rate and Data Rate eee 41 4 4 3 An OMNeT Example Model ccccceccesceseesscesceeseeeceeeecaeeesecaeesaecaeesaecnaeeeeeesetenseeas 41 4 4 4 Event Based Simulation cccccesccsseesecescesecesceeceseeeeeeneeeecaecesecaeesaecaeeaeeaeeaeeneseeenseeas 42 A O NA 44 CHAPTER 5 PROFIBUS SIMULATION MODEL sssssssssssssssssssssssessosssesescssvesessossnssnssossessnssossossnssesessoassess 45 Dells AMUPOCUCHON asso e
83. IRY RESPONSE state After that it waits for a valid frame or for the expiration of the Ts and changes to the INQUIRY MODE state Similarly if the message is a D REQ 156 Simulation Models Implementation the DLL evolves to the AWAIT_DISCOVERY_RESPONSE state and waits for a response or for the expiration of the Tsz timer and changes to the DISCOVERY state Message Dispatching Procedure Mobility Procedure Is a DMM and DMM state INACTIVE One high Yes priority message one_high_msg true processed GAP Update Procedure Is the high Teup lt 0 and No priority message GAP_Turn false queue empty 2 Pass Token Procedure Is the low priority message queue empty Yes Send Frame Procedure Figure C 32 Message dispatching procedure IDMP Whenever a message is received or Tsz timer expires the DMM is notified about this If a message is received it is passed to the DMM If it is neither an IQ REQ nor a D REQ messages the DLL simply sends the frame The evolution of the DLL state machine related with the INQUIRY MODE DISCOVERY and BEACONT TX states is controlled by the pum Simulation Models Implementation 157 Mobility Procedure Is high priority Yes message queue empty Build Frame 2 Isa Yes D_REQ 2 Yes state AWAIT_DISCOVERY_RESPONSE Send frame Send frame Notify DMM state AWAIT_INQUIRY_RESPONSE Send frame
84. Kind msg 6 case SMP Us startIDMPerrortimer 8 if isLOTempty then 9 processRSMPmessage 10s state WINQUIRY iba ij Las else 15 state WIDT_END 14 end oR case RU 163 updateRT msg Fiy end TBs T9 end 20 case WIDT_END ZEL if msgKind msg SMP and isLOTempty then 22 processRSMPmessage 23 state WINQUIRY 24 25 end BAG case WINQUIRY 20 switch msgKind msg 28 case SMP 295 processRSMPmessage SO end Sk case PBT BIE clearWirelessMobileAddrFromRT SiS end 34 case SBT Sy stopIDMPtimer 30 state INACTIVE Sl end 33 case IQ REQ 39 sendCommandToDLL 40 end 41 case RU 42 updateRT msg 43 end 44 45 end 46 47 Figure C 22 processIDMPmessage msg function pseudo code algorithm The behaviour of the BM operation executed by the processIDMPmessage msg function depends on the BM state and the type of the IDMP message received If the Bm is in the INACTIVE state then only SMP or RU messages are processed The Bm RT is updated when it receives a Route Update RU message independently of the Bm state If it is a SMP message BM IDMP Abort Timer Tamipmpabor is loaded with the bm idmp abort timer parameter value and is started using startIDMPerrortimer function line 7 According to its LOT the BM can evolve to either the WIDT END state if the LOT is not empty or to the WINQUIRY state if
85. M sends a Prepare for Beacon Transmission PBT message to all pms in the network and they should reply to the Gmm using a Ready for Beacon Transmission RBT message when they are holding the token frame The reception of the RBT message from a DMM means that it is operating in inquiry mode In order to start the IDMP Phase 2 the cum builds a PBT message starts the Gum Phase 2 Alert Timer TGmm P24lert and the CMM Phase 2 Abort Timer TGum P24b0r timers sets the retry variable to false which is used to control the retry of the PBT message evolves to the WRBT state and it sends the PBT message to the DLL After that it waits for a RBT message from each Dmm in the network Whenever it receives a RBT message from a DMM it removes the corresponding Source Address SA of the DMM from the list It stays on this state until receiving a RBT from all pms in the network i e until the list of DMM address is empty Meanwhile if the Temm P24lert expires the cmm retransmits a PBT message and continues waiting for RBT messages from remaining DuMs However if Temm P24bori expires the GMM aborts the IDMP and evolves to the INACTIVE state 150 Simulation Models Implementation IDMP Phase 2 Procedure DMMlist LDMMN Start timers Temv P2ater and Temu pzabor Build PBT message state WRBT retry false Pass PBT to DLL Discard frame Temm P2alert expired No Temm P2Abort expired 2 No Figure C 20 IDMP Phase 2 proc
86. MM may again receive a PBT message transition 4 and 5 respectively If a DMM is in the WTOKEN state then it again checks if the DLL is holding the token If it is it sends again a RBT message and evolves to INQUIRY state transition 3 It also sends another RBT message if it is already in the INQUIRY state transition 5 When a DMM receives a SBT message from the cmm two transitions may occur transition 6 or 13 The pmm evolves either to INACTIVE state if it is a wired domain transition 13 or to the BEACON _TX state transition 6 if is a wireless domain The Tpmm IDMP4bort 18 Stopped in both cases In the BEACON TX state it transmits transition 7 a predefined number of Beacon frames defined by the n beacon parameter It is in this Beacon frame transmission period that wireless mobile stations may change to a new domain When this period ends transition 8 the DMM evolves to the IDENT state and the Dmm tries to detect if wireless mobile stations are present in its domain by 78 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model inquiring them using Discovery request D_REQ messages transition 9 If wireless mobile stations are detected then it broadcasts a RU message and its state machine evolves to the INACTIVE state transition 10 If the Tomm 1DMPAborr timer expires its state machine evolves to the INACTIVE state from either WTOKEN or INQUIRY states This event is supported by transitions 12 and 14 respecti
87. MSASR rules The increase on the percentage of transactions that do not miss its deadline is a consequence of having a more available network since the BMs are less time out of the logical ring However the increase in the MeanRT is related to the fact that transaction initiators the system masters are more time out of the logical ring than using the MSASR since the lowest addresses are assigned to the BMs Comparative Performance Analysis in an Error Prone Environment 113 10 6 2 Comparative Performance Analysis using different BERs in Each Domain The MeanRT the number of transactions and the percentage of transactions that do not miss its deadline are presented in Figure 10 16 Figure 10 17 and Figure 10 18 respectively On the left side of these figures the results for both network scenarios in which the MeanBER probability of wired domains is equal to 10 and of the wireless domains is equal to 10 are presented On the right side of these figures the MeanBER probability of wired domains is equal to 10 and on the wireless is equal to 10 are presented MeanBER probability of the wireless domain 10 MeanBER probability of the wireless domain 10 Busasa MeanRT ms Message Stream Message Stream Figure 10 16 MeanRT using different MeanBER probability in wired and wireless domains and the MSASR rules MeanBER probability of the wireless domain 10 MeanBER probability of the wireless domain 10 1600 N of
88. NS Next Station address and obviously its own address TS This Station address If a master station receives a token addressed to itself from a station Source address SA of the token frame a frame format description is presented latter registered in the List of Active Stations LAS as its predecessor PS SA then this master becomes the token owner and may start processing message cycles On the other hand if a master receives a token frame from a station which is not its PS it assumes that an error has occurred and it will not accept the token frame However if it receives a subsequent token from the same station it accepts the token and assumes that the logical ring has changed In this case it updates the original PS value by the new one in its LAS table If after transmitting the token frame and within the Slot Time the master detects bus activity it assumes that its successor owns the token Therefore it ceases monitoring the activity on the bus In case the master does not recognise any bus activity within the Tsz it repeats the token frame and waits another Tsz If it recognises bus activity within the second Tsz it stops working as an active master assuming a correct token transmission Otherwise it repeats the token transmission to its NS for the last time If after the second retry there is no bus activity the token transmitter tries to pass the token to the next successor It continues repeating this procedure unt
89. N_TOKEN state The second token frame transmission counts as the first token rotation After the second token transmission the token claimer processes message cycles and tries to discovery the station with the address following its during its Tyr At the expiration of Tyr usually it did not found any station therefore it passes the token to itself second token rotation At this point the other stations particularly the stations whose address differ by two are ready to enter in logical ring Table 10 2 shows the master s address considering the network scenario identified by B used in Section 9 2 2 and the master address assignment according to the MSASR rules Bmsasr Table 10 2 Master s address Address Address Master Busasr B Master Busasr B M1 5 1 M6 2 6 M2 3 2 M7 1 7 M3 9 3 M8 7 8 M4 10 4 M9 6 9 M5 4 5 M10 8 10 To evaluate the impact of these rules we performed two sets of simulations First we used a MeanBER probability equal in all domains and varied the MeanBER probability between 10 and 10 112 Comparative Performance Analysis in an Error Prone Environment These results were compared with the simulation results presented in Section 10 5 1 In the second set we fixed the MeanBER probability of the wired domains to 10 and varied the MeanBER probability of wireless domains equal between 10 and 10 These results were also compared with the simulation results presented in Section 10 5 2 10 6 1 Comparativ
90. ProfibusNet domain 0 TSL 115 heProfibusNet domain 0 max_retry_limit 1 heProfibusNet domain 0 _bem_type 1 heProfibusNet domain 0 _bem_par1 0 00001 Figure 5 9 Configuration file related to Domain module instance excerpt 5 2 4 Master A Master module is a compound module that maps a master station It is composed by three modules Master PHY Master DLL and Msg Stream In each Master module instance there is one instance of Master PHY and Master DLL modules The number of the msg stream module instances can be from 1 up to 64 A Master module is connected to the Domain module through gates domain_gateIn and domain _gateOut Master PHY and Master DLL model the PhL and DLL of the PROFIBUS protocol respectively The Msg_Stream module models the operation of the Application Layer AL therefore it can be configured to periodically request services from the DLL These modules are hierarchically organized as illustrated in Figure 5 10 As mentioned compound modules are modules composed of one or more sub modules Any module type simple or compound module can be used as a sub module Further sub modules may use parameters of the compound module The compound module definition specifies how the gates of the compound module and its immediate sub modules are connected Connections that span multiple levels of the hierarchy are not allowed This restriction enforces compound modules to be self contained These concepts are pr
91. Repeater based domain parameters Domain Toi Ti Ts Trr D and D 2132 1088 2856 39712 D 1447 740 2142 27520 D 100 100 714 4858 9 2 2 Bridge Based Scenario The bridge based network scenario Figure 9 2 comprises two wired master M1 and M2 two mobile wireless master M3 and M4 five wired slaves S1 S2 S3 S4 and S5 and one mobile wireless slave S6 Domains are interconnected by three bridges B1 B2 and B3 and each of them is composed by two Bridge Masters BMs B1 M8 M5 B2 M6 M9 and B3 M10 M7 Each wired wireless domain has its own logical ring In this example four different logical rings exist D M8 gt M3 gt M5 D MI gt M5 M6 D M9 gt M4 gt M10 and D M2 gt M7 Note that the wireless mobile stations M3 and M4 can belong to wireless domains D and D Concerning the IDMP M6 assumes both the role of GMM and the DMM of wired domain D BMs M5 M9 and M7 assume the role of DMMs for wireless domain D wireless domain D and wired domain D respectively Table 9 4 presents the addresses of the master stations The slave addresses are equal to those addresses used in the repeater based scenario presented in Table 9 2 The timing parameters have been set according to the recommendation of the PROFIBUS standard IEC 2000 therefore the T p and the Ts parameter have been set to 100 and 115 bit times respectively The Trg parameter has been set according to the formulation
92. S timed token passing a bridge needs to have two network interfaces both supporting the same DLL and specifically the same MAC protocols This means that such a dual port PROFIBUS bridge would contain two master stations Figure 1 2 presents a bridge based hybrid network example similar to the repeater based hybrid network example presented in Figure 1 1 The network is composed by two structured wireless domains D and D and two wired domains D and D In the system there are two wired masters M1 and M2 two wireless mobile masters M3 and M4 five wired slaves S1 S2 S3 S4 and S5 and one wireless mobile slave S6 Three bridge devices are considered Bl M8 M5 B2 M6 M9 B3 M10 M7 Network operation is based on the Multiple Logical Ring MLR approach Ferreira and Tovar 2004 Therefore each wired wireless domain has its own logical ring In this example four different 4 Overview logical rings exist one for each domain D M3 gt M8 D M1 gt M5 gt M6 D M4 gt M9 gt M10 and D M2 gt M7 x Wireless domain domain A N N O a Mesa Base staton p 3 n 3 a dh de Radio Wy area ease BS1 1 Figure 1 2 Bridge based network example As a consequence of the MLR this approach requires two new protocols one for supporting the communication between stations in different domains the Inter Domain Protocol IDP and another to support the mobility of
93. SG Profibus msg profibus NULL 4 cMSG Self msg_self NULL 5 handle the message according to the simple module algorithm 6 if msg gt isSelfMessage usually used as timer Ve msg_self cMSG_Self s msg Ba switch msg profibus gt getAction OF case REPLY MESSAGE T0 buildResponseMessage amp msg_profibus E send msg_profibus out abs Tae 14 ise 1 Wee etset 17 msg profibus cMSG Profibus msg 18 switch msg self gt getType 15 case REQUEST_FRAME if it is a request frame 20 if msg_profibus gt getDA TS amp amp Bak DER msg_self gt setAction REPLY MESSAGE CAS js scheduleAt simTime TSDR msg self 24 BI 26 Ee break 28 20 30 sala Ss Figure 4 7 handleMessage cMessage msg function C code SlaveStation 4 5 Summary This chapter discussed the main characteristics of the OMNet simulation environment used in this dissertation It described its main characteristics and provides an example of how a PROFIBUS message cycle can be modelled in such architecture Chapter 5 PROFIBUS Simulation Model The repeater and the bridge based approaches are both compatible with standard PROFIBUS therefore their simulation software share the same standard PROFIBUS modules This chapter describes the entities which enable the simulation of a standard PROFIBUS network 5 1 Introduction The repeater Alves 2003 and bridge based
94. SP NTRANS 10 9875 0 002884 1350 D3 DI 197548 DI D3 125094 mas 000 2901 DA eso sees oaas mas essa sore DO D3 25476 DI DI 116020 Figure D 5 Screenshot of spread sheet created by Message Stream Response Time Analysis option Additionally it also provides information about the domain location of the initiator and responder during a transaction This information is particularly important for IDTs involving wireless mobile stations Since this option shows to which domain the initiator was belonging when a message was queued QUE and to which domain it belongs when the message was sent REQ Another kind of information provide by this option is related to the domain location when the initiator sends the request REQ the domain location of the responder when it replies RESP Central Limit Theorem The Central Limit Theorem is an option Figure D 6 that provides a way to compute the confidence interval of the message stream response time values according to the central limit theorem Law and Kelton 2000 The lower bound and the upper bound of this interval is computed as mean value MEAN less error ERROR value and mean value more error value respectively Tools for Simulation Output Analysis 163 STREAM 1D SA DA SAE DAE 1 1 41 1 1 CENTRAL LIMIT THEOREM MIN MAX MEAN ERROR 0 275 _ 99 994 13 91383 0 139599 RUN MIN MAX NREG MEAN 0 332 98 394 2583
95. The wireless communications are relayed by two base stations BSs which are included in the ISs Wireless mobile stations M3 M4 and S6 move in a specific path consequently the radio signal quality varies according to the propagation model Further in the mobility model of the wireless mobile station it is possible define stop points The handoff procedure is triggered in a periodic fashion and the BSs transmit a special purpose frame the Beacon frame during a pre configured amount of time which the wireless mobile stations use to assess the radio signal quality and change to the radio frequency of the BS with the strongest radio signal Mobility Simulator 85 AE PathofM3 PathofM4 CZL Path of S6 Stop points Figure 8 2 Network scenario 8 5 Simulator Architecture The following modules compose the MSim bs agv box pc antenna camera nhp and ground All these modules are implemented as C classes There are two kinds of modules the modules used for simulation and the modules used to compose the scenario The first group includes the modules antenna which models a radio antenna bs which models a BS agv which models a wireless mobile station and nhp which models the handoff procedure The second group includes box pc and camera The box can be used to model a wall or a machine the pc is used to model a personal computer or similar device The camera is used on the agv module with esthetical purpose The
96. Timer Based mechanism for Phase 2 eee Figure 3 11 Timeline example for Phase 3 and Phase 4 e Figure 4 1 Simulation study steps Source Law and Kelton 2000 ceseseeseseseeseseseeeeecsesseeseesenseesseseeesaeseeeesseseeeeaeaeee Figure 4 2 Simple and compound modules Figure 4 3 Module s gates and connections Figure 4 4 Message transmission Figure 4 5 PROFIBUS transactions events Figure 4 6 handleMessage cMessage msg function C code MasterStation INIA ACEON iaa 43 Figure 4 7 handleMessage cMessage msg function C code SlaveStation aeee 44 Figure 5 1 Modules connections and associated gates Figure 5 2 Network definitions a ny E E dado aiii Figure 5 3 HW2PNet module NED definition cerro E R E RE RARER Ei Figure 5 4 Configuration file related with HW2PNet module instance excerpt Figure 5 5 Simulator output window screenshot ocicccinninnnnnnnnnncnnnonccnnoninnnns AT Figure 5 6 Controller module NED definition Figure 5 7 Configuration file related to the Controller module instance excerpt a 49 Figure 5 8 Domain module NED definition 0 cececeesesseseesesseseseesecseseseesesccsscaesaesecseesceecaecsesaeaecaeeecsesseaeaeeaeeaeeecaeeaeeeceeeaeeaeeees Figure 5 9 Configuration file related to Domain module instance excerpt Figure 5 10 OMNe T Master module composition rii esseseseeseses
97. VS I IPP HURRAY Q e oe y y gt Technical Report Performance Analysis of Wireless enabled PROFI BUS Networks Paulo Baltarejo Sousa HURRAY TR 070607 Version 1 0 Date 10 06 2007 Technical Report HURRAY TR 070607 Performance Analysis of Wireless enabled PROFIBUS Networks Performance Analysis of Wireless enabled PROFIBUS Networks Paulo Baltarejo Sousa IPP HURRAY Polytechnic Institute of Porto ISEP IPP Rua Dr Ant nio Bernardino de Almeida 431 4200 072 Porto Portugal Tel 351 22 8340509 Fax 351 22 8340509 E mail pos dei isep ipp pt http www hurray isep ipp pt Abstract Most of the industrial community is very reluctant to integrate new technologies in their consolidated automation systems either by preconception or by the lack of matureness of these technologies When addressing communication systems for control applications these fears become even more acute Usually these communication systems are based on fieldbus networks which provide adequate levels of performance dependability timeliness and maintainability The PROFIBUS PROcess FleldBUS is the most widely used fieldbus with over 15 million nodes worldwide in applications ranging from discrete part automation to process control During the cellular phone and WLAN boom of the last decade soon it became evident that wireless radio based communications could leverage a whole new set of potentialities in field level and control
98. a ke m D y D y D v a Stop timer s Transmission error Y v Expire timer Figure 3 11 Timeline example for Phase 3 and Phase 4 Due to transmission errors a wireless mobile station may not be discovered during the discovery sub phase or a RU message may not arrive to all BMs in network A mechanism similar to the GAP Update mechanism is triggered on the DMMs in order to detect wireless mobile stations To perform this objective the DMMs periodically send Discovery messages to all wireless mobile stations and broadcast RU messages containing information about wireless mobile stations Obviously this mechanism introduces a small overhead to the network 3 3 4 The Adopted Mechanism The Tree Like Topology mechanism assures that the four phases of IDMP will always take place However this mechanism increases the network traffic in each domain i e some kind of messages are periodically transmitted Therefore there is the need to define the periodicity of each repetition And in order to minimize the IDMP latencies this period must take in attention the message s kind For example a SMP message takes longer to reach a leaf node than a SBT message because when a SMP message is transmitted the network is in normal operation i e the token frame is rotating by domain s masters and the IADTs and IDTs are both enabled When a SBT is transmitted the domains are in inquiry sub phase the DMMs are holding the tok
99. a PROFIBUS Num sistema RFieldbus a liga o entre componentes cablados e componentes sem fios feita atrav s de dispositivos de interliga o que operam ao n vel da Camada F sica como repetidores Esta solug o compat vel com o standard PROFIBUS mas o facto de todas as mensagens serem enviadas para todos os n s da rede a n o conteng o de erros entre os diferentes segmentos e a necessidade de se efectuar uma configura o especial dos par metros da rede levam a uma diminui o do seu desempenho Estes factos levaram an lise e proposta de uma nova abordagem na qual os dispositivos de interliga o operam como pontes bridges em ingl s e por isso ao n vel da Camada de Liga o de Dados Esta abordagem define dois novos protocolos um para processar transac es entre esta es pertencentes a meios de comunica o diferentes o Inter Domain Protocol e outro para processar a mobilidade das esta es m veis entre os segmentos sem fios o Inter Domain Mobility Procedure O principal objectivo desta disserta o comparar o comportamento temporal destas duas abordagens em ambientes sem e com erros Adicionalmente tamb m se quer mostrar que a abordagem baseada em bridges poss vel de ser implementado num sistema real Para tal foram desenvolvidas duas ferramentas de simula o uma para a arquitectura baseada em bridges e outras para arquitectura baseada em repetidores assim como um conjunto de ferramentas
100. action 2 Transaction 3 MMMM Di tee tst2 ters En qi 0 Es q2 0 m E q3 0 Request Frame Resp Response Frame Idle Time a Repeater Delay E Additional Idle Time Figure 2 6 Using additional idle time for media adaptation Another consequence of using ISs that act as repeaters is related to the master PROFIBUS DLL parameter the Slot Time Tsz Within the context of this approach Ts assumes a particular importance On one hand Ts must be set large enough to cope with the extra latencies introduced by the repeaters On the other hand Ts must be set as small as possible such as the system responsiveness to failures does not decrease dramatically that is a master must detect a message token loss or a station failure within a reasonably small time The timing diagram depicted in Figure 2 7 illustrates a transaction sequence that is relayed by two repeaters One interconnects domains D and D and another 16 Technological Context Communication Infrastructure interconnects domains D and D It is assumed that the PhL frame duration in D and D is half the PhL frame duration in D and that gt is constant for the sake of simplicity Request Frame Resp Response Frame Station Delay Responder E Repeater Delay Figure 2 7 slot Time Ts The setting of these parameters must be performed according to the procedures described im Alves 2003 2 3 4 The Mobility Procedure The mobility procedure
101. ad a Token L PBT RBT Do RBT Pl gy M7 Token Token RBT ue 4 M6 Phase 2 ends slaen 8 M9 1 D y y D v Y Y D M Y D Figure 3 6 Simplified timeline of the Tree like topology mechanism for Phase 2 When all DMMs are holding the token frame the GMM starts Phase 3 by sending a SBT message This procedure requires the use of a new kind message the Emitting_Beacon_Frame EBF message which signals that a son node is transmitting Beacon frames Contrarily to the IDMP outlined in Section 2 4 2 the GMM will wait for EBF message from its node sons The mechanism is similar to the previous When a leaf node receives a SBT message it sends an EBF message to its father And in order to assure that its father node receives the transmitted EBF message it waits during some time for a repeated SBT message If it does not receive a SBT message it means that its father node received correctly the transmitted EBF message After if the domain is wireless it starts emitting Beacon frames otherwise it starts processing message cycles This mechanism is illustrated in Figure 3 7 where the GMM repeatedly sends a SBT message until receiving an EBF message from its son nodes at this point its role in the IDMP ends However the IDMP continues in wireless domains controlled by DMMs Each wireless DMM emits Beacon frames during a predefined time After the end of the Beacon transmission that is when Phase 3 ends Phase 4 starts
102. and a Microsoft Excel based tool to output data analyse D 2 Timeline Visualisation Tool Figure D 1 shows a screenshot of the Timeline Visualisation Tool which provides a way to show the network events using Gant Diagrams This tool was developed using Microsoft Foundation Classes MFC Prosise 1999 and C programming language This figure depicts a diagram drew using the data files generated by BHW2PNetSim concerning the network scenario presented in Figure 9 2 In this figure it is possible to see the events accomplished by each module instance When a Master module instance operates also as BM the events related to BM module instance are separately shown On the other hand the events accomplished by a bridge are also separated by each BM module instance that composes it For example Figure D 1 shows that the bridge B3 is composed by BMs M7 and M10 thus the events of each BM module instance BM_M7 and BM_M10 that composes a bridge are individually shown To illustrate the importance of this tool in Figure D 1 three transactions are highlighted using arrows one IADT between master M1 and slave S1 and two IDTs between master M2 and slave S6 and master M1 and slave S5 The transaction between master M2 and slave S6 can be surprising since this transaction is an IDT and it finishes during the first AL request of this transaction This happen because when master M2 sends the request an open a transaction is created by BM M7 and when maste
103. and proposal of an alternative approach where the Intermediate Systems ISs operate at the Data Link Layer DLL level as bridges Ferreira Alves et al 2002 Ferreira Alves et al 2003 Ferreira Tovar et al 2003 This approach henceforth referred as bridge based required two new protocols one for supporting the communication between stations in different domains the Inter Domain Protocol IDP and another to support the mobility of wireless stations between different wireless domains the Inter Domain Mobility Procedure IDMP 1 2 Research Context PROFIBUS was standardised in 1996 as an European standard CENELEC 1996 It is based on the International Standards Organisation ISO Open System Interconnection OSI reference model however collapsed to just three layers Physical Layer PhL Data Link Layer DLL and Application Layer AL It is designed to provide different qualities of service in terms of timeliness providing intrinsic mechanisms that distinguish the way high and low priority messages are transmitted Its DLL employs a token passing mechanism to grant the medium access Moreover the PROFIBUS Medium Access Control MAC protocol being based on the measurement of the real token rotation time induces a well defined timing behaviour to the worst case message response time since the upper bound for the actual token rotation time can be know a priori Tovar and Vasques 1999 Therefore the PROFIBUS protocol i
104. ansmits two token frames addressed to itself After that every master will be joining to the logical ring using the GAP update mechanism However when a master transmits a token addressed to itself all masters that are not in the LISTEN_TOKEN state evolve to that state since they are skipped of the logical ring When a master station is in the LISTEN_TOKEN state it shall monitor the bus activity in order to identify which master stations already belong to the logical token ring For that purpose token frames are analyzed and the station addresses contained in them are used to generate the LAS After listening to two complete identical token rotations the master must remain in the LISTEN TOKEN state until it is addressed by an FDL Request Status transmitted by its predecessor to which it must respond indicating its readiness to enter into logical ring When a master station is in the LISTEN TOKEN state all frames are neither acknowledged nor answered except FDL Request Status frames In order to clarify the problem of this mechanism suppose a network situation in which the token owner is the station with the lowest address in the logical ring and it loses the token frame due to two consecutive errors in token frame transmissions Therefore it evolves to the LISTEN TOKEN state and clears its LAS as well as it NS and PS parameters As it has the lowest address its Tro timer expires first Consequently it evolves to the CLAIM TOKEN state and
105. arch contributions were based on a set of tools which have been developed within the objectives of this dissertation The Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator The Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator The Mobility Simulator Tools for simulation output analysis which have been used to validate the simulation models and to extract information from the output data files generated by the simulators Timeline Visualization Tool Output data Analysis Tool Message Stream Response Time Analysis Central Limit Theorem State Machine Statistical Analysis Frame Accounting 1 5 Structure of the Dissertation The structure of this dissertation is as follows Chapter 2 presents an overview of the PROFIBUS protocol and the most relevant aspects of the repeater and bridge based architectures In Chapter 3 some problems of the bridge based architecture related to the management of errors are identified and a new set of mechanisms to overcome these problems is proposed Chapter 4 surveys some of the existing simulation tools for the development of network simulation models and describes in more detail the adopted simulation environment The repeater and the bridge based architectures are both compatible with standard PROFIBUS therefore their simulation model implementation share the same standard PROFIBUS modules Chapter 5 describes the entities which enable the simulat
106. as the aggregate result of 100 runs each with 120 s of duration using a different seed value in order to improve the randomness of the data It is interesting to note that the results presented in this chapter are equivalent to the aggregate of the 170 hours of real operation According to the Central Limit Theorem Law and Kelton 2000 the simulation results presented in the next section presents a confidence level higher then 99 95 The largest confidence interval for a message stream response time has a range of 8 of the mean response time value 9 3 Performance Analysis In this section we present and analyse some simulation results upon variation of some network parameters bit rate ISs internal delay and maximum frame size The message streams used on the comparison between these two approaches were S Sp and S3 one IADT and two IDTs respectively where S involves mobile stations 96 Comparative Performance Analysis in an Error Free Environment In the repeater based scenario there was the need to adjust the period of the message streams because with some parameter setting the network enters into saturation since the network is used beyond its maximum throughput 9 3 1 Base Configuration Results This subsection discusses the results obtained using the base configuration described in Section 9 2 Figure 9 3 shows a histogram of the measured response time values for S in both scenarios Note that in the subtitle
107. aster changes the NS and updates the LAS and GAPL Otherwise it updates only the GAPL In any cases the Master evolves to the USE_TOKEN state 140 Simulation Models Implementation Send FDL_Request_Status Procedure state AWAIT_STATUS_RESPONSE Addr Get Next Addr from GAPL Build FDL Request Status frame Send FDL Request Status frame Receive a Ts expired SL pe response as bit errors No 2 Is a slave station 2 a NS Addr Master Yes state USE_TOKEN Discard the frame Update LAS station ready Update GAPL Figure C 6 Send FDL Request Status procedure Receive FDL Request Status Procedure Figure C 7 presents the procedure when a station receives an FDL Request Status frame The frame is discarded if it contains bit errors or if is not addressed to it Otherwise a response frame containing the state of the responder is transmitted In spite of PROFIBUS DLL defines three states in our simulator only two were implemented Slave Station if it is a slave station and Ready to Enter Logical Ring for the master stations Receive FDL Request Status Procedure Receive a FDL Request Status frame No il dle eae Yes Discard the frame Yes Build response frame Send response frame Figure C 7 Receive FDL_Request_Status procedure Simulation Models Implementation 141 C 1 6 Send Frame Procedure The PROFIBUS DLL protocol defines four data transfer services The Send Data with Ackn
108. ate of the 130 hours of real operation According to the Central Limit Theorem Law and Kelton 2000 the simulation results presented in the next sections present a confidence level higher then 99 95 The largest confidence interval for a message stream response time has a range of 5 of the mean response time value 10 2 Gilbert Elliot Channel Model It is well known that transmission errors occur in bursts Willig and Wolisz 2001 i e there is a correlation between consecutive errors The Gilbert Elliot model takes into account this correlation This model is a two state discrete time Markov chain as illustrated in Figure 10 1 One state represents good channel conditions and the other bad channel conditions To each state is assigned a steady state Bit Error Rate BER probability In this model we define ps as the probability of continuing in the good state p the probability of continuing in the bad state Each of these states has a BER probability p and p for good and bad state respectively Consequently 1 Peje is the probability of evolving from good to bad state and 1 p p is the probability of evolving from bad to good state The algorithm works by generating for each bit in a frame a random number and compares it to the respective BER A second random number is generated to determine whether the model stays in the actual state or changes into the other state for the next bit It is assumed that bit errors occur independen
109. ative Performance Analysis using the same BER in All Domains 112 10 6 2 Comparative Performance Analysis using different BERs in Each Domain 113 1027 SUMMARY ON 114 CHAPTER 11 CONCLUSIONS AND FUTURE WORK ccccsscssscsssscsccscsscssscecsessssesesessessseseesecesssessesssssesees 117 11 1 Research Context and Objectives ccccccccescessesssessessecececseceecsecseeseceseeseseeserseseseeseaeeneeeaes 117 11 2 Main Contributions A sina sliver inden 118 11 2 1 Enhancements to the Bridge Based Approach cccesceescesseesceseeeeeceeeseeneeeaeensenaees 118 11 2 2 Comparison between the Repeater and Bridge Based Approaches 118 11 2 3SoftWwate Tools mi A A a EE 118 1 32 Future work ita A AA A iaa 119 REFERENCES secem A EEE TN EETA E 121 ANNEX A PROBABILITY DISTRIBUTION FUNCTIONS eseseseseesesesorsscoessseosesesercrereososososesesererersosososesese 125 AL cSt OU Ct Oi PAET A E E 125 A 2 Parameterization of Continuous Distributions s ssssssssseeesseesesersesersseeessesessreeseesesreseseesse 125 A 3 Parameterization of Stochastic Parameters in the Simulators ooonooninnnncnnncinciocccononnconnnonoons 128 MI ANNEX B BIT ERROR MODELS istmo an ere aer eeo e ssh dei ste docs pace ides cb ses diant sai esco sbt das can 131 Ble Introduction atenas O E ott Hesse eee 131 B 2 Independent Channel Model erriren id 131 B3 Gibert Elliot Channel Model ii taa 131 BA
110. ay integrate with it like Open Dynamics Engine Smith 2004 8 2 2 OpenGL OpenGL Shreiner Woo et al 2005 is a software interface to the graphical hardware This API consists of about 150 distinct commands that developers use to specify the objects and operations needed to produce interactive 2D and 3D applications OpenGL is designed as a streamlined hardware independent interface to be implemented on many different hardware platforms In order to achieve these qualities no commands for performing windowing tasks or obtaining user input are included in OpenGL OpenGL does not provide high level commands for describing models of 2D and 3D objects With OpenGL models must be built from a small set of geometric primitives points lines and polygons 8 3 Wireless Communications Radio Signal Propagation In wireless communication the information is delivered to the transmitter and modulated into radio waves The radio wave is radiated through the air using a radio channel to the receiving antenna At the receiving antenna the radio wave is demodulated and the transmitted information is extracted Figure 8 1 In a wireless communication based on Base Station BS the coverage area is divided into small coverage areas called cells in which wireless communications are all relayed by a BS In such systems all stations transmit on one channel uplink and listen on a second channel downlink The BS functions are to receive a frame
111. b Pblg Figure B 1 Gilbert Elliot model 132 Bit Error Models One state represents a good channel conditions and the other one bad channel conditions Each state is assigned a constant Bit Error Rate BER probability p in good state and p in bad state It is assumed that the bit errors occur independently from each other Let and t the mean duration in good state and in bad state respectively The steady state probability for being in good state can be obtained as follows E B 2 Pag t 0 In same way the steady state probability for being in bad state can be obtained as follows t Pob B 3 t t The mean BER is given by MeanBER P pg Po Pop Ps B 4 The probability of a transition occurs from good to bad state is computed as Pre B 5 The probability of a transition occurs from bad to good state is computed as Pow 1 py B 6 The Gilbert Elliot model is computationally expensive since for each frame s bit two uniform experiments have to be executed The algorithm works by generating for each bit in a frame a random number and compares it to the respective BER A second random number is generated to determine whether the model stays in the actual state or changes into the other state for the next bit This will slow down the simulation performance In order to overcome this drawback a simplified Gilbert Elliot model can be used This simplification is accomplished by assuming that in good
112. ble contrarily in the repeater based scenario all message streams are severely affected It was necessary to increase the message stream period in the repeater based scenario to 160 ms consequently the number of transaction performed in the bridge based scenario is 2000 more for IADTs and 943 more for IDTs As an example the number of transactions for message streams S and S in the bridge based scenario considering a frame size of 250 Bytes is 1500000 and 102 Comparative Performance Analysis in an Error Free Environment 707809 respectively in the repeater based scenario the number of transactions is 75000 for both message streams 9 4 Summary In this chapter we have performed a performance comparison between the repeater and the bridge based architectures based on simulations results in an error free environment We have carried out experiments which showed the influence of varying certain network parameters in message response times From these experiments we noted that in the bridge based approach the variability of the response time histograms is smaller than in the repeater based approach Although in some cases the maximum response time for IDTs can be superior The bridge based approach benefits from the multiple logical ring segmentation which isolates the traffic between domains permitting lower response times for IADTs Additionally the network segmentation permits the independent setting of the network parameter
113. bout which little else is known l o if a lt x lt b f b a 0 otherwise A Ha Density 1 b a 4 E A 0 a b x 0 if x lt a x a t Distribution F x if a lt x lt b b a 1 if b lt x a and b real numbers with a lt b Parameters a isa location parameter b a isascale parameter Range a b a b Mean 2 2 b a Variance AS 12 Mode Does not uniquely exists A 3 Parameterization of Stochastic Parameters in the Simulators The RHW2PNetSim and BHW2PNetSim allow setting some parameters using PDFs The name of all these parameters uses the pdf prefix For example the parameters associated to Tspr are the following pdf_ tsdr type pdf tsdr parl pdf tsdr par2 and pdf tsdr pars Where the pdf tsdr type indicates which PDF will be used to generate the value of the Tspr and the other parameters are the arguments of the PDF Table A 2 presents how the simulator parameters must be set according to the PDF Probability Distributions Functions 129 Table A 2 Probability Distributions Functions simulators parameters Probability Distribution Functions Parameters Constant Normal Exponential Triangular Triangular _pdf_ type 0 1 2 3 4 pdf parl Value q B a a _pdf_ _par2 o apex b Epdm pan b Annex B Bit Error Models The use of bit error models in communication simulation has been widely studied In this dissertation we had used three models The Independent Channel Model the Gi
114. bution Function PDF or by a constant value The PDFs implemented require at most four parameters One of them defines which PDF is used and the other three are the arguments of the PDF The name of all these parameters uses the paf prefix For example the parameters associated to Tspr are the following pdf tsdr type pdf tsdr parl pdf tsdr par2 and _pdf_tsdr_par3 The _pdf_tsdr_type indicates which PDF will be used to generate the value of the Tspr and the other parameters are the arguments of the PDF The PDFs supported by both simulators are described in detail in Annex A Additionally the same parameters can also be used to make a configuration using constant values HW2PNet Master ctrl con Master M1 domain_gateln domain_gateOut domain_con E Domain ctrl_con Controller domain_con connection domain_con Slave ctrl con station gateOutM1 station_gatelnM1 Domain D1 y EE Output gate Input gate Figure 5 1 Modules connections and associated gates One of the most important steps of a simulation study is to analyze output data generated by the simulator Law and Kelton 2000 Our simulators enable gathering information about the response time of transactions information about state machine transitions of each module information about the PDFs in use and information about the Bit Error Model BEM in use The name of the parameters related to thes
115. chitecture of the RFieldbus Deliverable D1 3 RFieldbus project IST 1999 11316 Alves M E Tovar et al 2002 Real Time Communications over Hybrid Wired Wireless PROFIBUS based Networks 14th Euromicro Conference on Real Time Systems Vienna Austria Banks J J S Carson Il et al 2001 Discrete Event System Simulation New Jersey Prentice Hall Barros L M 2005 A Short Introduction to the Basic Principles of the Open Scene Graph Available online at http www openscenegraph org Behaeghel S K Nieuwenhuyse et al 2003 Engineering Hybrid Wired Wireless Fieldbus Networks a case study In Proceedings of the 2 International Workshop on Real Time LANs in the Internet Age RTLIA03 Porto Portugal pp 111 114 Burns A and A Wellings 2001 Real Time Systems and Programming Languages Ada 95 Real Time Java and Real Time POSIX Addison Wesley Longmain Carvalho J A Carvalho et al 2005 Assessment of PROFIBUS Networks Using a Fault Injection Framework In Proceedings 10th IEEE International Conference on Emerging Technologies and Factory Automation ETFA 05 Catania Italia pp 227 234 CENELEC 1996 General Purpose Field Communication System European Norm Chang X 1999 Network simulations with OPNET In proceedings of the Winter Simulation Conference Phoenix Az pp 307 314 Elliot E 1963 Estimates of error rates for codes on burst nise channels Bell Systems Tech Journal vol 42
116. ckages have been evaluated in this dissertation OPNET ns 2 and OMNeT OPNET is widely held as the state of art in network simulation It is a suite of products that combines predictive modelling and a comprehensive understanding of networking technologies to enable design deployment and management of network infrastructures network equipments and networked applications In particular OPNET Modeller is a development environment allowing to design and study communication networks devices protocols and applications However OPNET is a commercial product with some limited academic licensing programmes ns 2 Network Simulator 2 is a discrete event simulator targeted at networking research ns 2 provides substantial support for simulation of TCP routing and multicast protocols over wired and wireless local and satellite networks The full source code of ns 2 can be downloaded from the Internet and it can be compiled for multiple platforms including the most popular Linux flavours and Windows OMNeT Objective Modular Network Testbed in C is a public source object oriented modular discrete event simulation package that can be used for modelling communication protocols computer networks traffic modelling multiprocessors and distributed systems OMNeT also supports animation and interactive execution The OPNET simulation package was discarded because it was a commercial product at the moment when this choice was made Therefore
117. connections simple Connection_Point parameters _pdf_tidm_type numeric _pdf_tidm_par1 numeric _pdf_tidm_par2 numeric _pdf_tidm_par3 numeric _name string gates in com_func_gateln domain_gateln out com_func_gateOut domain_gateOut endsimple Figure 6 12 Connection_Point module NED definition Figure 6 13 presents part of a configuration file related to a connection Point module instance This instance is called CP8 Since the parameter pdf tidm type is set to zero then the inactivity period between the two consecutive frames is constant and equal to 100 bit times theRHW2PNet cp 3 _name CP8 theRHW2PNet cp 3 pdf tidm type 0 theRHW2PNet cp 3 pdf tidm_par1 100 Figure 6 13 Configuration file related to the Connection Point module instance excerpt 6 3 Simulator Implementation This simulator encompasses the functions concerning the PROFIBUS described in Chapter 5 related to the domains interconnection and to the mobility procedure As previously mentioned the repeater model is made up of two simple modules ComFunc and Connection Point In our simulator implementation IS and BS functions are supported by the modules that model a repeater It is assumed that the operation mode of the repeater is cut through 6 3 1 Interconnection To interconnect different domains it is necessary to convert the received frame to the format of the destination domain and transmit the frame with a pr
118. d D and two structured wireless domains D and D Three repeaters R1 R2 and R3 interconnect the domains The wireless communications are relayed by two BSs BS1 and BS2 included in the repeaters The network also comprises three wired masters M1 M2 and MM two wireless mobile masters M3 and M4 five wired slaves S1 S2 S3 S4 and S5 and one wireless mobile slave S6 In order to guarantee the operation and interoperability of the hybrid wired wireless fieldbus network there can be no more than one possible path between any two domains tree like topology Technological Context Communication Infrastructure 13 1 e no closed loops can exist The mobility procedure only makes sense when there are more than one wireless domain and when these domains are structured The mobility procedure will be detailed later Wired communication interface A ta ot Base Station pp 4 4 I I I E i v cao A 1 Y 4 Wireles communication E i a f interface 1 D Se 1 gt i 1 a E i Radio coverage area i E of BS1 k 4 i O i Figure 2 3 Repeater based hybrid wired wireless PROFIBUS network example 2 3 1 Repeater Operation As mentioned in this approach the interconnection between domains is done by ISs operating as repeaters Traditionally operating just as a signal regenerator a repeater can also interconnect two networks with different PhL protocols e g different bit rates A repeater is c
119. d defines several timing parameters some of which are relevant in the context of this document such as the Idle Time Tm the Slot Time Tsz and Time Out Time Tro parameters which are briefly explained next There are two Idle Time Tp parameters T p and Tip Tip is a period of inactivity inserted by a master station after an acknowledgment response or token frame This parameter must be set as follows Tip max T yyy Toy MINT spp gt T spy 2 2 where Tsyy Synchronisation Time is the minimum time interval for an idle bus state before a station may accept the beginning of an action or token frame Tsy Safety Margin Time is the time that elapses after the end of the Tsyy which is required by the receiver circuitry to be ready to start receiving a frame minTspr is the minimum Station Delay of a Responder Time Tsp is the Station Delay of the Initiator Time after which the initiator is ready to start receiving a frame from the responder Figure 2 2 depicts an example where the Transmission Delay Time Trp due to the network propagation delay is also illustrated Tmz is the idle time inserted by a master station after transmitting an unacknowledged request frame Tpz must be set as follows Tips max T yy T sy Max T spp 2 3 where maxTspp is the maximum Station Delay of a Responder Time The Slot Time Tsz is used by a master station to detect if a transaction with a slave or with its successor in the token pas
120. d that the number of transactions is equivalent But the percentage of transactions that do not miss its deadline increase significantly Also when comparing with the repeater based approach the percentage of transactions that do not miss its deadline is now higher for the bridge based approach Chapter 11 Conclusions and Future Work This chapter reviews the research context and objectives of this dissertation summarizes the most relevant contributions and highlights the possible directions of future work 11 1 Research Context and Objectives Nowadays industrial automation applications collect large benefits from the use of fieldbuses for the interconnection of distributed devices Usually these systems are supported by a wired infrastructure But there is now an increased pressure to extend the capabilities of fieldbuses with wireless communication functionalities in order to support mobile devices Wireless communication systems are of particular interest in supporting mobile machine parts mobile vehicles and temporary or frequently reconfigured production lines for example The integration of wireless communications in a fieldbus system creates new challenges It is expected that the level of performance of the wireless extensions to be at least similar to those existing in wired fieldbus On the other hand there is the need to interconnect different media To sum up the level of throughput reliability and real time performance of suc
121. dges as received without being coded Note that response frames are transformed into request frames by the IDP 2 4 2 Supporting Inter Domain Mobility The Inter Domain Mobility Procedure IDMP is a hierarchically managed procedure where one master in the overall network the Global Mobility Manager GMM is responsible for periodically starting the IDMP and controlling some of its phases Additionally in each domain one master controls the mobility of stations belonging to that domain the Domain Mobility Manager DMM Finally the BMs must implement specific mobility services The GMM must know the addresses of all BMs and DMMs in the system Each DMM must know the addresses of the BMs that belong to its domain as well as of the wireless mobile stations The wireless mobile stations implement specific services which enable them to evaluate the quality of the radio channels These services are assumed to be similar to the ones used in the repeater based approach For example and concerning the network scenario illustrated in Figure 2 12 M6 assumes the role of GMM and DMM for wired domain D BMs M5 M9 and M7 assume the role of DMMs for wireless domain D wireless domain D and wired domain D respectively The role of the management agents and the different phases of the proposed handoff mechanism will be described next Figure 2 12 Bridge based hybrid wired wireless PROFIBUS network example IDMP Phases of th
122. dresses between TS and NS This range of addresses in the logical ring is referred as GAP whereas the list containing the status of all stations in the GAP is called GAP List GAPL Each master in the logical ring examines its GAP periodically in the interval given by the Tup timer Its expiration indicates the moment for GAP maintenance GAP addresses are examined in ascending order except the GAP addresses which surpasses the HSA 1 e the HSA and address 0 are not used by a master station In this case the procedure is continued at address 1 after checking the HSA If a station acknowledges positively with the state Not Ready to Enter Logical Ring or Simulation Models Implementation 139 Slave Station it is accordingly marked in the GAPL and the next address is checked If a station answers with the state Ready to Enter Logical Ring the token frame holder changes its GAPL and LAS accordingly as well as its NS and passes the token frame to the new NS Pass Token Procedure Get new NS from LAS retry_counter 2 state PASS_TOKEN 2 retry_counter 0 error_frame_counter 0 Build token frame to NS Transmit the token frame state PASS_TOKEN retry_counter Token frame transmitted is equal to the received 2 error_frame_counter 2 No 2 Yes state LISTEN_TOKEN Yes Ts expired 2 No Receive a frame 2 state ACTIVE_IDLE Figure C 5 Pass Token Procedure This is accompli
123. e Figure 7 14 OMNeT Dmm module NED definition The cmm module is a simple module that models the GMM required functionalities This module is responsible for the control of Phase 1 and Phase 2 and also the beginning of Phase 3 of the IDMP For its operation the GMM must be provided with the LBMN and also with the LDMMN These lists are also generated by the controller module instance at simulation initialisation The IDMP is triggered in a periodical fashion The value for this period is assigned to the Controller module through the _tmob parameter Figure 7 15 depicts the cum module NED definition in which there is only one parameter Ts parameter simple GMM parameters TS numeric gates in dil gateln out dil gateOut endsimple Figure 7 15 OMNeT cum module NED definition Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 75 7 4 Simulator Implementation 7 4 1 Bridge IDP Functionalities When a frame is received by a BM it tests 1f the addressed station using the Destination Address DA of the frame and its RT is reached by forwarding the frame to the other Bm of the bridge If the test succeeds then the frame is forwarded to the comrunc module instance The ComFunc module relays the frame to the other Bm that composes the bridge The frame received from the ComFunc is passed by the BM to the DLL which queues the message Figure 7 16 illustrates the interconnection schema between
124. e IDMP The IDMP evolves through 4 phases as shown in Figure 2 13 The objective of these phases is to insure that the procedure will not generate errors that the inaccessibility periods are minimal especially in the case of IADTs and that the wireless mobile stations are able to evaluate all wireless radio channels and switch to the best one The proposed mechanism is synchronous in some of its phases Phase 1 and Phase 2 as well as the beginning of Phase 3 But in the case of Phase 3 the ending of it in the domains is not synchronised and Phase 4 runs asynchronously for each domain 22 Technological Context Communication Infrastructure GMM sends Start Mobility Procedure SMP message GMM sends Prepare for Beacon Transmission PBT message GMM receives all Ready for Beacon Transmission RBT Beacon transmission period ends Bridges finish all N inter domain transactions operation ur Phase 1 DMMs rec the PBT and holding the token gt Inquiry phase DMMs inquiring BMs for RBT Phase 3 Beacon transmission gt Normal 1 Mobile l stations join l operation new domains i LN 17 1 Phase 4 All Bridges replyed with Ready to Start Mobility Procedure RSMP messages DMMs start sending Ready for Beacon Transmission RBT messages GMM sends Start Beacon Transmission SBT messages Bridg
125. e Performance Analysis using the same BER in All Domains Figure 10 13 Figure 10 14 and Figure 10 15 present the MeanRT the number of transactions and the percentage of transactions that do not miss its deadline respectively These simulation results show that the MeanRT increases using the MSASR The number of transactions is very similar using both network configurations But with the MSASR the percentage of transaction that do not miss its deadline increases especially for simulation results in which the MeanBER probability is equal to 10 MeanRT ms 0 105 104 103 105 104 102 105 104 103 105 10 4 10 3 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 13 MeanRT using equal MeanBER probability in all domains and the MSASR rules 1600 sy 1400 1200 Busasa 1000 E B 3 8 SNR 38838 N of Transactions Thousands o 8 o 10 10 10 3 2 10 10 10 19 10 37 108 Mean BER probability Mean BER probability Mean BER probability 10 10 10 Mean BER probability Figure 10 14 Number of transactions using equal MeanBER probability in all domains and the MSASR rules Concluded transactions 10 105 104 10 10 10 10 10 104 10 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 15 Percentage of transactions that do not miss its deadline using equal MeanBER probability in all domains and the
126. e Ts expires in Section C 1 5 the Gap update procedure is described in detail From the USE TOKEN state it changes to the PASS TOKEN state transition 7 when its Token Holding Time Tyr expires The Tyr is computed at token frame reception according to Eq 2 1 The transitions 8 9 10 and 11 are handled by the pass token procedure a detailed description of this procedure is presented in Section C 1 4 From the PASS_TOKEN state it evolves to the CHECK_TOKEN PASS state transition 8 after transmitting a token frame to its Next Station NS In the CHECK TOKEN PASS state 2 transitions 9 and 12 can occur If it detects a valid frame in its domain then it changes to the ACTIVE _ IDLE state transition 12 Otherwise if after expiring Tsz no activity is detected it returns to the PASS TOKEN state transition 9 and stays into these two states PASS TOKEN and PROFIBUS Simulation Model 57 CHECK_TOKEN_PASS until passing the token to another station in its LAS or to itself transition 11 USE_TOKEN q 5 AWAIT STATUS RESPONSE Figure 5 20 Master state machine diagram In order to detect a defective transceiver when a Master is transmitting a token frame in the TOKEN PASS state it also receives the token frame If it detects a difference between the transmitted and received frame it waits in the CHECK TOKEN PASS state Tsz for activity from its NS If no activity is detected after expiring Tsz it again transmits th
127. e based approach increases less significantly The decrease on the number of transactions is similar in both approaches when the MeanBER probability increases The percentage of transactions that do not miss its deadline is similar for MeanBER equal to 10 and 10 but for MeanBER equal to 10 the percentage of transactions that do not miss its deadline is always higher for the bridge based approach than for the repeater based approach The results of these last experiments were a little surprising The MeanRT of the bridge based approach is more constant the number of transactions is also higher but the percentage of transactions that do not miss its deadline is higher for the repeater based approach than for the bridge based approach These results triggered a carefully analysis of the results We realise that the setting of master station addressees should be done in a way which permits achieving a higher number of transactions that do not miss its deadline Since the network availability depends strongly on the BMs we defined a set of rules to assign an address to master stations To evaluate the impact of these rules in the bridge based approach we performed another set of simulation runs using two transmission error configurations one in which the MeanBER probability was set equal to all domain and another in which the MeanBER probability is higher for wireless domains From these simulation sets we concluded that the MeanRT increases slightly an
128. e functions use the output prefix Master Slave and Domain modules are all identified by a parameter called name The value of this parameter must be unique in the overall network since it identifies a module instance To simplify the parameterization of the module instances all common parameters to the network are associated with the Controller module and all common parameters to the domain are associated to the Domain module These parameters are used by the Controller module instance to do the station parameterization This characteristic makes the simulation configuration less complex and less error prone In the next sections further details are provided concerning model architecture and implementation 5 2 1 HW2PNet In OMNeT to actually get a simulation that can be run it is necessary to write a network definition A network definition declares a simulation model as an instance of the system module in this case of the HW2PNet module A network definition is declared with keyword network followed by the network instance s name and ends with the keyword endnetwork Figure 5 2 presents the network definition in which the system module instance is called theProfibusNet No simulation parameters are assigned in the network definition since they are assigned by the configuration file named omnetpp ini network theProfibusNet HW2PNet Endnetwork Figure 5 2 Network definition PROFIBUS Simulation Model 47 HW2PNet
129. e parameters and the parameters related to the Beacon message were calculated with the help of the RFieldbus System Planning application which is described in Behaeghel Nieuwenhuyse et al 2003 Table 9 2 Station s address Master Address Slave Address M1 1 S1 41 M2 2 S2 42 M3 3 S3 43 M4 4 S4 44 MM 5 S5 45 S6 46 In this approach the setting of the Tpz parameter on the MM 2677 bit times must be made differently in relation to the remaining stations in the network This is because after transmitting the Beacon Trigger message this master enters into an inactivity period for the duration of the channel assessment period which allows the wireless mobile stations to assess the quality of the other radio channels Additionally a repeater always introduces a minimum inactivity period between two consecutive frames being forwarded This value the minimum idle time Tpm has been set to 100 bit times In order to guarantee that at the token arrival there will always be enough time to execute all pending high priority traffic the master Trr parameter has been set according to the formulations proposed in Tovar and Vasques 1999 assuming no errors However the Trg parameter has to be equal to all masters in each domain therefore the Tyg parameter was set considering the high value computed of each domain Table 9 3 presents the settings of the Tip Tip2 Ts and Trp master DLL parameters for each domain Table 9 3
130. e same rules were used to set the Temm P24lert and the Temm P2Abort equal to 4 090 ms and 8 181 ms respectively The Tam 1DmpPAbor Was set equal to 30 490 ms which is the sum of Temm PlAbort and Tomm P24bort The Tbum iDmpAbor Was set equal to Tomm pzAbor 8 181 ms because this timer is started at the begining of Phase 2 and finishes at the end of Phase 2 The relation between the IDMP timers is illustrated in Figure 10 2 DMM GMM BM Tomm P 1Alert Temm P1Abort TBmM IDMPAbort Temm r2alert TDMM IDMPAbort T GMM P2Abort Figure 10 2 IDMP timer s settings To detect and handle errors during the execution of an IDT a BM assigns to every entry in its List of Open Transaction LOT a timer called BM IDT Abort Timer TemiprAbor If an entry is still in the LOT when the associated timer expires then the entry is deleted from the LOT TamiprAbor timer must be set with a value which allows the execution of a transaction Two approaches were possible to set this timer one based in the WCRT presented in Ferreira 2005 and the other based on simulation results Obviously the first mechanism is much more pessimist Therefore we have set the Tgm 1074b0 equal to 33 022 ms which is the MaxRT of all message streams provided by the simulation results presented in Section 9 2 106 Comparative Performance Analysis in an Error Prone Environment In an error prone environment a message stream response time can potentially be unli
131. e that OMNeT does not provide any native mechanism for mobility A simple module is declared with keywords simple followed by the modules name and endsimple The parameters and the gates can be specified in the declaration of a simple module Figure 5 6 presents the NED definition of the controller simple module Parameter domain is a string which defines the configuration of the domain It is written using predefined structure based in tags The Controller module instance extracts information from these strings to perform the network configuration simple Controller parameters _output_gant_diagram numeric _output_resp_time numeric _output_states numeric _output_pdf numeric _output_bem numeric _output_period numeric output path string endsimple Figure 5 6 Controller module NED definition The parameters with the output prefix are related to the output data files If one of these parameters has a value of one it means that the information referred to by the parameter must be gathered by the module instances Table 5 1 A detailed description of the output data files is presented in Annex D Figure 5 7 presents a configuration example of the domain parameter used in BHW2PNetSim there is a slightly difference in the configuration of the parameter of the two simulators by the network showed in Figure 5 5 Table 5 1 Summary of the output data information Parameter Information output gan
132. e timing behaviour of the repeater and bridge based approaches in an error free and error prone environments Additionally we also intended to show that the bridge based approach implementation is feasible and propose additional error detection and correction mechanisms which would improve its performance over error prone environments To achieve these objectives two simulation tools have been developed for the repeater based approach and another to the bridge based approach and a set of analysis tools Additionally we have also developed another tool to simulate the mobility of wireless stations As outlined next the objectives of this dissertation were achieved 118 Conclusions and Future Work 11 2 Main Contributions 11 2 1 Enhancements to the Bridge Based Approach Error detection and correction algorithms were not previously considered in the bridge based approach proposed in Ferreira 2005 Therefore this approach has been enhanced with the necessary mechanisms to be used over error prone mediums as summarised next Originally the IDP defined that IDFs were transmitted using the SDN PROFIBUS service This service is an unacknowledged service Therefore any IDF transmission error results in a failed IDT In this dissertation we proposed that IDFs must be transmitted using the SDA PROFIBUS service The SDA service is an acknowledge service i e the frame sender receives a special frame Short Acknowledge frame confirming the recept
133. e token frame if it again detects a difference between the transmitted and received frames its state machine evolves to the LISTEN TOKEN state transition 10 This process is designated as heardback removal in Willig and Wolisz 2001 Whenever a Master evolves to the LISTEN TOKEN state the LAS is cleared and it starts listening on the medium for at least two successive identical token cycles During this time it is not allowed to send or respond to data frames or to accept the token but it builds the LAS After that its state machine evolves to the ACTIVE IDLE state transition 14 The master state machine evolves to the CLAIM TOKEN state from the LISTEN TOKEN state when the Tro timer expires transition 16 In this case there is the need to recover the token Section C 1 1 presents more details about this procedure In this procedure it transmits two token frames addressed to itself and if no difference between transmitted and received frames occur its state machine evolves to the USE TOKEN state transition 18 Otherwise the state machine evolves to the LISTEN TOKEN state transition 17 as a consequence of heardback removal 5 4 Summary This chapter provided a high level overview about the implementation of the main PROFIBUS standard functionalities available in both the Repeater and Bridge based simulators For details about the implementation of these modules the reader is referred to the Annex C The next two chapters desc
134. ecise timing which guarantees minimal delays For that purpose there is the need to know the length of the DLL frame the Baud_rate parameter value of the interconnected domains and how each frame is coded Figure 6 14 and Figure 6 15 illustrate a transaction between a Master module instance named M2 and a Slave module instance named S6 according to the network configuration presented in Figure 6 4 where M2 belongs to domain D4 and S6 belongs to domain D3 In order to simulate bit by bit reception Connection_Point module instance named CP7 delays the frame just enough time for it to know its length Note that the number of bits necessary to know Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 65 the frame length depends on the PhL frame format and the PROFIBUS DLL frame type For instance wireless PhL frames include a head tail fields and each DLL character is coded using 8 bits Wired PhL frames do not include any head or tail fields but each DLL character is coded using 11 bits Concerning the frame length PROFIBUS DLL has two kinds of frames fixed and variable length frame In order to know the length in the first case the first DLL character SD field is enough but in the second case there is the need to receive the second DLL character the LE field i Repeater ComFunc i ee me Master Connection Point Connection Point Slave M2 CP7
135. ection and detection mechanisms was proposed The enhanced version of the IDMP uses a set of timers to detect and handle errors in its messages Four timers are assigned to the GMM two timers are used to detect and handle errors during the Phase 1 GMM Phase 1 Alert Timer ToumPiAter and GMM Phase 1 Abort Timer TGy prabort and the other two are related to the Phase 2 Phase_2 Alert Timer Tomm p2Alert and GMM Phase 2 Abort Timer Tomm p2Abort Additionally each BM and each DMM also controls a timer the BM IDMP Abort Timer Tem IDMPAbort and the DMM IDMP Abort Timer TpmmiDMPAbor respectively For a detailed description of this mechanism the reader is referred to Section 3 3 3 The setting of these timers can be based on the Worst Case Response Time WCRT analysis proposed in Ferreira and Tovar 2004 but that would probably lead to a very low network performance Therefore we based our setting on simulation results According to our simulation results the maximum duration of IDMP Phase 1 and IDMP Phase 2 are 11 156 ms and 4 090 ms respectively Since these results were obtained considering an error free medium these values can be used to set the alert and the abort timers Therefore the Temm PlAlert Was set to 11 156 ms and the Tomm p14b0r Was Set to 22 311 2 Temm piAter The abort timer was set to the double of the alert timer since this setting permits the completion of Phase 1 even when the SMP message is not received by any BM Th
136. ed e e a e 109 Figure 10 10 Response time using different MeanBER probability in wired and wireless domains 109 Figure 10 11 Number of transactions using different MeanBER probability in wired and wireless domains 110 Figure 10 12 Percentage of transactions that do not miss its deadline using different MeanBER probability in wired and wireless domains Figure 10 13 MeanRT using equal MeanBER probability in all domains and the MSASR rules ooooiccncnncnononcncnononcnnnaninnnnons 112 Figure 10 14 Number of transactions using equal MeanBER probability in all domains and the MSASR rules 112 Figure 10 15 Percentage of transactions that do not miss its deadline using equal MeanBER probability in all domains and the MSASR rules 112 Figure 10 16 MeanRT using different MeanBER probability in wired and wireless domains and the MSASR rules 113 Figure 10 17 Number of transactions using different MeanBER probability in wired and wireless domains and the MSASR NA NN ON 113 VI Figure 10 18 Percentage of transactions that do not miss its deadlines using different MeanBER probability in wired and wireless domains and the MSASR TUles citada ida ted 113 Figure 10 19 Percentage of transactions that do not miss its deadline using different MeanBER probability in wired and wireless domains Figure B 1 Gilbert Elliot model Figure B 2 Simplif
137. ed the number of transactions Figure 10 8 is much higher for the bridge based scenario than for the repeater based scenario The AL deadline for each message stream was set according to the WCRT analysis considering an error free transmission In both scenarios the percentage of concluded transactions for message stream S is 100 considering the MeanBER probability equal to 10 and equal 10 With MeanBER probability equal to 10 the percentage of concluded transactions is slightly lower in the repeater based scenario than in the bridge based scenario Additionally in the bridge based scenario the number of transactions is approximately 500 more than in the repeater based scenario In both scenarios message stream S presents basically the same percentage of concluded transactions considering a MeanBER probability of 10 and 10 However with a MeanBER probability equal to 10 the percentage of concluded transaction is significantly lower in the repeater based scenario than in the bridge based scenario Further the number of transactions is approximately 250 more in the bridge based scenario than in the repeater based scenario The percentage of concluded transactions for message streams SP and S considering the MeanBER probability equal to 10 and equal 10 are slightly smaller in the bridge based scenario than in the repeater based scenario The main reason is that the initiator of these message streams is a wireless mobile master
138. edure When the cmm receives a RBT from all pms in the network it builds a SBT message and passes the Start Beacon Transmission SBT message to the DLL This message commands the DMMs to start emitting Beacon frames After the cum evolves to the INACTIVE state The remaining actions of the IDMP are controlled independently by the omms in each domain BM Operation A received message is catalogued by a BM according to the frame format described in Chapter 2 If it is an IDMP related message then it is handled by the handleIDMPMessage msg function presented in Figure C 21 A received message is passed to the DLL using the passToDLL msg function if it is received from ComFunc through bridge _gateln gate Otherwise if a message is received from the DLL it is only forwarded to ComFunc by sendToComFunc msg function if the addressed station line 6 can be reached by forwarding the message through the other BM In any case the message will be processed by processIDMPmessage msg function Figure C 22 Simulation Models Implementation 151 sendToComFunc msg processIDMPmessage msg 1 handleIDMPMessage msg Chee ast 3 if msgArrivalGate msg bridge gateIn then 4 passToDLL msg Sa Giles Ge if isRoute msg getDA then ie 8 Sha Figure C 21 handle IDMPMessage msg function pseudo code algorithm 1 processIDMPmessage msg Bo il 3 switch state 4 case INACTIVE Se switch msg
139. eir RT at the reception of the PBT message update its RT Figure 3 11 presents an example timeline of Phase 3 and Phase 4 In this example a transmission error occurs when the SBT message is forwarded by DMM M6 and consequently the BM M5 DMM M8 and BM M8 will not receive the SBT Since domain D is a wired domain no further IDMP related action will be taken in this domain Stations in this domain may start performing message cycles However BM M5 will not accept new IDTs since it did not receive a SBT message But when the Tzm 1DmPAbor eXpires it will accept new IDTs Since DMM M8 did not receive a SBT message it did not transmit Beacon messages and no wireless mobile stations entered to or left from its domain Thus to update the BMs RT DMM M8 sends a RU message containing the information about wireless mobile station that still belong to its domain when Tpmm IDMP4bort had expired TDMM IDMPAbort Tomm P2Alert TBM IDMPAbort Termm e2Abort TBM IDMPAbort DMM BM DMM GMM DMM M7 M M9 M6 M6 Transmission TBM IDMPAbort SBT ser D SBT SBT SBT e TDMM IDMPAbort TBM IDMPAbort BM DMM M5 M8 m Void Beacon Token Token yl Token Void a Void Beacon Token Discover i M3 Token Discovery Token M4 Al Void Discovery TommiDMpabor S6 TBM IDMPAbor void RU RU 4 TBM IDMPAbort K Ri
140. eity in bit rates and in PhL frame formats in a broadcast network imposes the consideration of some kind of traffic adaptation scheme The timing diagram depicted in Figure 2 5 illustrates a sequence of transactions where one repeater interconnects the two domains and it is assumed that the PhL frame duration in D is twice the PhL frame duration in D and that 1 is constant for the sake of simplicity Note that since the idle time is defined as the duration of a predefined number of idle bits separating consecutive frames in the network its duration is assumed to be different for the two domains Technological Context Communication Infrastructure 15 Figure 2 5 illustrates an increasing queuing delay q lt q2 lt q3 caused by the different physical media that will impact on the system turnaround time f for certain transactions The for a message transaction is the time elapsed since an initiator ends transmitting a request frame until it starts receiving the correspondent response frame For example in the case of the request that corresponds to transaction 3 which is addressed to a responder in domain D the system turnaround time for this transaction t 3 will be affected by the cumulative queuing delay q in the repeater Transaction 1 Transaction 2 Transaction 3 D ts tst2 tog qo E gz eee n_ Be Request Frame Resp Response Frame a Idle Time E Repeater Delay Figure 2 5 Increas
141. eiver P d can be calculated as follows P d P G PL d G 8 1 where P is the transmitted power G is the transmitter antenna gain G is the receiver antenna gain d is the distance between transmitter and receiver in meters PL d is the average path loss at distance d between transmitter and receiver and is given by PL d PL d 10n108 lt X 8 2 0 where n is the path loss exponent which indicates the rate at which the path loss increases with distance X is the shadowing term the zero mean Gaussian random variable in dB with standard deviation of 0 PLo d is the free space path loss distance dy d is the close in reference distance which is determined from measurements close to the transmitter and is given by 4 PL d 20108 8 3 where 2 is the wavelength in meters and is related to the carrier frequency by 84 Mobility Simulator A 8 4 where fis the carrier frequency in Hertz and c is the speed of the light 3 10 m s Table 8 1 and Table 8 2 list some typical path loss values exponents and the standard deviation on specific environments respectively Table 8 1 Path Loss Exponents for Different Environments Source Vignaux and Muller 2006 Environment Path Loss Exponent n dB Free space 2 Outd id Shadowed Urban area 2 7 to 5 A line of sight 1 6 to 1 8 Bin Obstructed 4to6 Table 8 2 Standard Deviation for Different Environments Source Vignaux and Muller
142. el is a representation of a system at a particular time while dynamic simulation model represents a system as it evolves over the time If a simulation model does not contain any random components it is called deterministic otherwise is called stochastic A discrete model is one for which the state variables change instantaneously at separated points in time A continuous model is one for which the state variables change continuously according to time In practice very few systems are strictly discrete or strictly continuous but since one type predominates for most systems it is usually acceptable to classify a system as either being discrete or continuous Law and Kelton 2000 Several aspects have made simulation one of the most widely used and accepted tools in operations research and systems analysis Banks Nelson et al 2001 Particularity the availability of special purpose simulation languages massive computing capabilities at a decreasing cost per operation and advances in simulation methodologies are some of these aspects This chapter starts by presenting the basic steps of a simulation study Section 4 2 Section 4 3 presents a survey of the simulation tools for the development of network simulation models as well as the reasons that sustained the choice for the OMNeT simulation package to which some further details are provided in Section 4 4 4 2 The Basic Steps of a Simulation Study The development of a simulation study i
143. element of the conceptual model and the corresponding element of the system Additionally the conceptual model must also define the model assumptions The degree of detail of the conceptual model must be as fine as possible according to the simulation study objectives Formulate problem and plan the study e a Collect data and define a model Conceptual model valid Construct a computer program and verify Make pilot runs Programmed model valid Design experiments y Make production runs y Analyze output data y Document present and use results Figure 4 1 Simulation study steps Source Law and Kelton 2000 The third step is to validate the conceptual model which must occur before programming begins In order to validate the conceptual model a structured walk through of the conceptual model and simulation objectives must be performed If the conceptual model is valid then the simulation study evolves to the next step Otherwise the simulation study must return to the previous step After validating the conceptual model it must be translated to a computer program fourth step The choice of the tools used on the development of the computer program is crucial for the next steps More details about this will be given later After the development of a computer program that implements the conceptual model there is the need to make p
144. elton 2000 Simulation Modeling and Analysis New York McGraw Hill Lee K and S Lee 2001 Integrated Network of PROFIBUS DP and IEEE 802 11 Wireless LAN with Hard Real Time Requirement In proceedings of the IEEE International Symposium on Industrial Electronics ISIE 01 Pusan Korea pp 1484 1489 Meyer R A and R Bagrodia 1998 PARSEC User Manual Release 1 1 Available online at http cl cs ucla edu projects parsec manual Miorandi D and S Vitturi 2004 A Wireless Extension of PROFIBUS DP based on the Bluetooth System In Journal of Computer Communications Vol 27 No 10 pp 946 960 Prosise J 1999 Programming Windows With MFC Second Edition Microsoft Press Rappaport T S 1996 Wireless Communications Principles and Practice New Jersey Prentice Hall Rauchhaupt L 2002 System and Device Architecture of Radio Based Fieldbus The Rfieldbus System In 4 IEEE International Workshop on Factory Communication Systems Vasteras Sweden RFieldbus 2000 RFieldbus High Performance Wireless Fieldbus In Industrial Related Multi Media Environment Available online at http www hurray isep ipp pt rfieldbus RFieldbus 2000a RFieldbus Manufacturing Field Trial Web Site Available online at http www hurray isep ipp pt rfpilot Russell E C 1999 Building Simulation Models with SIMSCRIPT II 5 Available online at http www cs sunysb edu cse529 simscript_docs simbuild pdf Shreiner
145. ely In the repeater based scenario the number of transactions is 300000 for both message streams 9 3 3 Variability of the Message Stream Response Time as a Function of the ISs Delays The ISs delay is the time required by an IS to relay a frame between the domains which it connects either a bridge or a repeater In the repeater based approach it is the time required by the repeater to convert between frame formats In the bridge based approach it is the time required for the routing decisions for the conversion of frame formats and for its queuing on the output queue of the other BM of a Bridge In order to analyze the ISs internal delay influence on the network timing behaviour we performed six simulations in which the internal delay varied between 30 and 1000 us In the repeater based scenario there was the need to increase the message streams period to 80 ms since with higher values of the internal delay 500 and 1000 us the network entered into saturation The period for the other messages streams has been set using triang 78 80 82 and the offset has been set using triang 0 78 80 Figure 9 8 presents the MinRT MeanRT and MaxRT values for message streams S and Sas a function of the ISs delays 120 RS 120 RSM 100 100 YN Ee gt 80 80 E o 60 60 17 FE Mb a 40 40 IIA PARC A 0 0 30 60 120 250 500 1000 30 60 120 250 500 1000 _ 100 100 o 80 80 E gt 60 60 2 2 40 40 f P T a 20 20 0 0 30 60
146. emanding This chapter describes the Mobility Simulator MSim which simulates the mobility of wireless mobile stations and the radio signal strength on the current wireless mobile station location With this information it is possible to generate a vector containing the wireless domains to which the wireless mobile station belongs That information is used by the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim and in the Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BHW2PNetSim in order to determine the points in time at which the stations must move to another domain The results from this simulator are feed into the other simulators on the content of the location_vector parameter The MSim has been developed using an open source high performance 3D graphics toolkit called OpenSceneGraph Barros 2005 and the C programming language This chapter is structured as follows A brief description of the simulation environment used to develop the MSim is presented in Section 8 2 A detailed description of the adopted radio signal model is given in Section 8 3 Section 8 4 presents a description of the simulation model of the MSim The architecture of the MSim and the simulator configuration are presented in Section 8 5 and Section 8 6 respectively The output data files generated by this simulator are presented and described in Section 8 7 82 Mobility Simulator 8 2 OpenSceneGraph The OpenSceneGraph i
147. en DMM M6 receives the token frame it replies with an RBT message to the GMM and forwards the PBT message to domain D As soon as the DMM M8 receives the token frame it replies with an RBT message and the GMM receives the RBT message from DMM MS i e the RBT message is forwarded by BM MS as a response to an inquiry request from DMM M6 without transmission errors At reception of the first PBT message the DMMs start Tpmm IDMP4bor and the BMs clear all RT entries related to mobile stations As DMM MY did not reply with an RBT message the Temm Pzatert expires Therefore the GMM broadcasts again a PBT message and the DMMs reply with an RBT message Assuming that at this time no transmission errors occur then the GMM M6 receives all RBT messages from all DMMs Thus Phase 2 ends and the GMM stops the Temm P24bort Toum P2Alert Toun P2Abort DMM BM DMM GMM DMM BM DMM M7 M10 M9 M6 M6 M5 M8 TDMM IDMPAbort K D A TDMM IDMPAbort C ser pat Token ol IN RBT RBT naz A Foda PET BT oa la A Token 6 A M9 h m6 ret Token y Inquiry __ H __Inquiry e M10 M5 gt RBT sae m8 m Inquiry RBT Void m10 RBT__le Mey od y 4 MB Token Touu P2Alert Inui F _ Inquiry Sy M5 gt H _Void Inquiry___4 gt e M10 Token PET PBT y e vii RBT PBT TomuomPAbor N
148. en Token PBT Token pare PBT PBT Token RBT Inquiry ae RBT Inquiry M5 y MB PBT __ lt lt M7 M10 o L RBT ma M7 Phase 2 ends D D D D v v v Figure 2 15 Simplified timeline of IDMP Phase 2 Phase 3 After collecting all Ready_for Beacon Transmission messages from all the DMMs the GMM starts the Beacon transmission sub phase by broadcasting the Start Beacon Transmission SBT message Upon reception of this message the DMMs start emitting Beacon frames The wireless mobile stations use the Beacon frames to evaluate the quality of the different radio channels and to decide if they want to handoff or not So before the end of the Beacon transmission every wireless mobile station that wants to handoff must switch to the new radio channel Note that all wired domains that evolved to the inquiry sub phase may resume IADTs after the correspondent DMM has received the SBT message since their intervention is not required on the remaining phases of the IDMP Note also that IDTs can be relayed if the neighbouring domains have already their IDTs enabled IDTs involving wireless mobile stations are only resumed when the BMs receive Route Update RU messages specifying the location of the wireless mobile stations Figure 2 16 shows a simplified timeline related to IDMP Phase 3 assuming the network scenario presented in Figure 2 12 Phase 3 is started by GMM but is finished by DMM of each wire
149. en frame IADTs and also the IDTs are both disable and therefore these messages are relayed faster The Timer Based mechanism does not assure that the four phases of the IDMP will always take place It is possible that some wireless mobile stations are not able to assess the quality of the radio channel because there is no emission Of Beacon frames in their domains As a consequence these wireless mobile stations will continue on same domain 36 Error Handling Improvements for the Bridge based Architecture However the Timer Based mechanism was adopted The reasons for the choice are outlined next First the network traffic does not increase as in the Tree Like Topology mechanism Second this mechanism is more similar to the original IDMP than the Tree Like Topology Therefore less implementation effort is required in the Timer Based solution Third the station parameterization is less labour in the Timer Based than in the Tree Like Topology mechanism Fourth and the most important the Tree Like Topology implies that the real time analyses proposed in Ferreira 2005 for the IDMP to be invalid On the other hand the formulations proposed in Ferreira 2005 can be used to set the timers according to that worst case analysis 3 4 Summary In this chapter we identified some weaknesses of the error handling capabilities of the original IDP which used the unacknowledged SDN service to transmit IDFs between bridges To improve the error handling
150. en ring For that purpose token frames are analyzed and the station addresses contained in them are used to generate the LAS After listening to two complete identical token rotations the master must remain in the LISTEN TOKEN state until it is addressed by an FDL_Request_Status transmitted by its predecessor PS If it succeed it must respond with Ready to Enter Logical Ring and waits for the token frame addressed to it in the ACTIVE IDLE state When a master station is in the LISTEN TOKEN state all frames are neither acknowledged nor answered Ring Maintenance The ring maintenance mechanism is distributed by all master stations As mentioned each PROFIBUS master maintains two tables the GAPL and the LAS Each master station when holds the token frame checks its Gap addresses every time its Gap Update Timer Tcup expires If a station acknowledges positively to the GAP request an FDL Request Status frame with the state Not Ready to Enter Logical Ring or Slave_ Station it is accordingly marked in the GAPL and the next address is checked If a station answers with the state Ready to Enter Logical Ring the token holder changes its GAPL and passes the token to the new NS This master station which has newly been admitted to the logical ring has already built up its LAS when it was in the LISTEN TOKEN state so it is able to determine its GAPL and its NS This mechanism allows masters to track changes in the logical ring due to the additi
151. equal to 8 ms and consequently the MinRT value in the bridge based scenario is greater than 8 ms It is noticeable that 94 23 which is sum of the percentage or results in the intervals 8 9 11 12 of the transactions require only one AL retry and 4 13 which is sum of the percentage in the intervals 16 17 19 20 of the transactions required two AL retries and the remaining 1 64 three AL retries The mean response time MeanRT value is equal to 10 02 ms on bridge based scenario Table 9 8 Response time histogram for the message stream S interval RS BS Interval RS BS Interval RS BS ms ms ms 1 2 10 55100 0 10 11 3 18733 8 15101 19 20 0 0 01120 22 3 13 25933 0 J11 12 1 76800 0 28249 20 21 0 0 3 4 12 34733 0 12 13 0 92533 0 21 22 0 0 J4 5 11 97600 0 113 14 0 48300 0 122 23 0 0 5 6 12 60500 0 14 15 0 09400 0 23 24 0 0 6 7 10 97600 0 15 16 0 03367 0 124 25 0 0 14035 7 8 9 55333 0 16 17 0 00767 0 18625 25 26 0 1 47371 18 9 7 21067 6 13680 17 18 0 00233 3 09011 26 27 0 0 02337 9 10 4 84000 7966181 18 19 0 0 84291 The response time histogram of message stream S3 is shown in Figure 9 5 and Table 9 9 The results are very similar to message stream S However in the repeater based scenario the MinRT 4 61 ms and MaxRT 19 85 ms values are higher than the MinRT and MaxRT of the message streams Sand S The main reason for these r
152. er 2 pdf tidf part 11 theProfibusNet master 2 pdf tidf par2 70 theProfibusNet master 2 pdf tid1_par3 100 Figure 5 12 Configuration file related to Master module instance excerpt The following sections describes the implementation of each module that composes a Master module instance and their interactions Master_PHY The Master_PHY module models the PhL of the PROFIBUS protocol It represents the network interface of the Master module it receives messages from a Domain or from a Controller module instance and passes the messages to the Master DLL module and vice versa For that reason this module is connected to the Master compound module through four gates see Figure 5 10 domain gateIn domain_gateOut ctrl gateIn and ctrl gateout the first two are related to domain_con connections and the last two are related ctr1_con connections Figure 5 13 shows the Master PHY NED definition simple Master_PHY gates in lower_gateln ctrl_gateln upper_gateln out lower_gateOut ctrl_gateOut upper_gateOut endsimple Figure 5 13 Master_PHY module NED definition Master_DLL The Master_DLL module is structurally different in RHW2PNetSim and BHW2PNetSim On the RHW2PNetSim it is a simple module while on the BHW2PNetSim it is a compound module However the behaviour of the PROFIBUS DLL is the same as well as the implementation The Master DLL module is directly connected to the Master PHy and the Msg S
153. er Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BHW2PNetSim Conclusions and Future Work 119 Mobility Simulator is a tool which is used to simulate the mobility of wireless mobile stations the radio signal strength at a certain point in space and the wireless domain to which the station belong through time Timeline Visualization provides a way to show the network events like frame transmissions using Gant Diagrams This tool can also be used with any other kind of network simulator Output Data Analysis this tool provides a set of options to extract information from the output data files generated by the RHW2PNetSim and BHW2PNetSim 11 3 Future work The work performed during this thesis permitted to adequately characterise and compare the timing behaviour of the repeater and bridge based hybrid wired wireless PROFIBUS networks However we envision some improvements mainly in relation to the bridge based architecture which might considerably improve their performance The operation of the bridge based architecture relies completely on bridge devices which are new components in a PROFIBUS network infrastructure The BMs defined in this architecture do not implement AL functionalities An obvious add on would be to include AL functionalities in the bridges as in a standard PROFIBUS DP master This would permit to improve the performa
154. es a Ready for Beacon Transmission RBT message from every DMM in the network If Temm P24tert expires before the cum receives a RBT from every DMM in network then it re sends a PBT message and waits in WRBT state for the reception of RBT messages from the pws in lack If Temm P24borti expires before the cmm receives a RBT from the remainder DMMs then it aborts the IDMP and evolves to INACTIVE state transition 5 Otherwise it sends the Start _ Beacon_Transmission SBT message and evolves to INACTIVE state transition 5 a T a wo WRBT Figure 7 17 cum state machine A detailed description of the procedures related to the cmm operation can be found in Section E32 Bridge Master BM The role of a BM in the IDMP is based on the state machine presented in Figure 7 18 The state machine of the BM is composed of the 3 states INACTIVE WIDT_END Wait Inter Domain Transactions End and WINQUIRY wait Inquiry The Bm role during the evolution of the IDMP is essential in ensuring the finalization of pending IDTs and in the relaying of IDMP related messages When the IDMP is not active the BM is in the INACTIVE state In this state a BM can update its RT according to the information contained into Route_Update RU messages transition 1 When a BM receives a SMP message the timer BM IDMP Abort Timer TamiDMpAbor is loaded with the value of the bm idmp abort timer parameter and is started It evolves to either WIDT_END s
155. es is summarized FRAME TRANSMISSION PROFIBUS BEACON IDP IDMP NO ERROR 123767334 1870962 1515670 ERROR 8407600 342954 221932 PROFIBUS E FRAME TOKEN REQUEST DL RESPONSE REQUEST RESPONSE 4382639 3226415 20741 423497 354308 IDP REQUEST RESPONSE 178863 164091 IDMP SMP RSMP PBT RBT SBT 1Q_REQ BCN D_REQ D RESP RU VOID 15627 24213 2753 2292 1628 41277 15524 47641 18916 52061 Figure D 13 Screenshot of spreadsheet created by the Bit Error Model Frame Accounting option IDP Timeout The IDP has a error recovery mechanism which deletes an entry from a BM LOT if the timeout timer associated with that transaction expires This behaviour allows the BM to initialise a new LOT entry related to the same message stream The information about deleted IDTs in a BM LOT is recorded in a file with the tidt extension Figure D 14 shows an example of this kind of file The information gathered in this file is the DA SA message ID and the timestamp when the deletion occurred Based on the information contained in this file a spreadsheet tool allows the analysis of the results Figure D 15 depicts a screenshot of the spreadsheet where the information concerning IDTs deleted by BM M8 is shown The information is organi
156. es of connections but these messages are sent and received by simple modules Such series of connections which go from simple module to simple module are called routes Compound modules act as cardboard boxes in the model transparently relaying messages between their inside and their outside world i e they are used to aggregate other modules relaying messages from inside to the outside and vice versa without any processing Compound module Simple module Simple module Hl gt gt gt Gate Connection Figure 4 3 Module s gates and connections 4 4 2 Modelling Delays Bit Error Rate and Data Rate Connections can be assigned three parameters which facilitate the modelling of communication networks propagation delay sec bit error rate errors bit and data rate bits sec Each of these parameters is optional One can specify link parameters individually for each connection or define link types also called channel types once and use them throughout the whole model The propagation delay is the amount of time the arrival of a message is delayed when it travels through a communication channel The bit error rate has influence on the transmission of messages through the channel The bit error rate is the probability that a bit is incorrectly transmitted The data rate is specified in bits second and it is used for trans
157. es send Route Update RU messages Figure 2 13 Phases of the Inter Domain Mobility Procedure Phase 1 Phase 1 starts with a Start Mobility Procedure SMP message sent by the GMM This message is sent periodically according to the mobility requirements of the wireless mobile stations involved in the application When the BMs receive a SMP message they stop processing new IDTs from the masters belonging to their domains Nonetheless they keep handling pending IDTs still present in their LOTs and importantly they keep relaying IDF originated in other domains After completing all pending IDTs those from their LOT the BMs transmit a Ready to Start Mobility Procedure RSMP message to the GMM When the GMM has received RSMP messages from all BMs in network the Phase 1 ends Figure 2 14 shows a simplified timeline related to IDMP Phase 1 assuming the network scenario presented in Figure 2 12 For the sake of simplicity it is assumed that when a BM receives a SMP message there is no open transaction in its LOT Phase 1 starts BM M5 BM M7 M8 Token Token SMP _ SMP SMP TS MP Token Token Tok RSMP RSMP oken M7 RSMP Token l M8 SMP 110 _RSMP nee SMP Token M5 RSMP M7 gt _RSMP Token RSMP Phase 1 ends M8 D D Di D v v Figure 2 14 Simplified timeline of IDMP Phase 1 Phase 2 Phase 2 is triggered by the GMM broadcast
158. esenscessessesceesesseessesceseessesseavacseeseesssuavesssecensaeseeavessesseaseese Figute S 11 Master module NED dsfimtiOn s sencssasesastisessensta stores esta Etra ento S pipa h E cusses EOE de ir Facas RU ssa ca dee IE Figure 5 12 Configuration file related to Master module instance excerpt E Figure 5 13 Master PHY module NED definition sosen orse ns dec staves oeseoseeve aj asso dando ade bala pan a sede ESA aE ETit Figure 5 14 OMNeT Master DLL module NED definition cece eeseeeeecseseesescseeeeseseseeesseseeecasseseeesasseseeesseetenaeaeee Figure 5 15 Msg Stream NED definition Figure 5 16 Deadline missing examples Figure 5 17 Msg Stream configuration parameters of a Master Figure 5 18 OMNe TE Slave A ANN Figure 5 19 Slave DLL module NED definition i s EEr A ER AEE RAA RAE E Figure 5 20 Master state machine diagram Figure 6 1 Modules and connections of the RHW2PNetSim Figure 6 2 Screenshot of the output window of the RHW2PNetSim eecseseeseseseeeesescseeecaesesecasseseeesasseseeesseseeeeaeaees Figure 6 3 Controller module NED definition of the RHW2PNetSim Figure 6 4 Configuration file related to the Controller module instance of the RHW2PNetSim excerpt Figure 6 5 Domain module NED definition of the RHW2PNetSim Figure 6 6 Configuration file related to the Domain module instance
159. esented in the Master module NED definition depicted in Figure 5 11 In this definition some parameters are omitted since its definition is very long The address of the Master module instance is set using the TS parameter The number of message streams is defined by the num_streams parameter This parameter is used to define the number of Msg Stream module instances and also to specify the number of gates between the Master DLL module and Msg Stream module instances The parameters with pdf tidl prefix are related to T PROFIBUS DLL parameter Figure 5 12 shows part of the configuration file related to a Master module instance The parameter pdf tidl type is set to three meaning that the T p duration evolves according to a Triangular PDF A Triangular PDF requires three parameters In this case the value of T p will be between 11 bit times pdf tidl parl parameter and 100 bit times pdf _tidl par3 parameter and the mode is 70 bit times _pdf_tid1l_par2 parameter PROFIBUS Simulation Model Msg_Stream 0 lower_gateln lower_gateOut lower_gateln lower gateOut Master Msg Stream N 4 y lower_gateln lupper_gateOut 0 upper gateln 0 upper_gateOut N upper gateln N Master DLL lower gateOut upper gateOut ctrl gateOut ctrl gateln lower gateOut lower gateln upper gateln Master PHY e M ctrl_gateOut
160. esults is due to the simulation model in which message stream S is always queued in third place on M3 output queue Therefore frames related to message stream S have to wait for the transmission of frames related to the other two message streams in which the initiator is M3 This operation mode is similar to the typical behaviour of a Programmable Logical Controller PLC running PROFIBUS In the bridge based scenario the MinRT and MaxRT values for message stream Se are 8 66 ms and 26 08 ms respectively These results are similar to the results presented by message stream S as would be expected since both are IDT 98 Events 0 5 6 7 Interval ms 14 5 15 6 16 7 17 8 8 9 9 10 10 11 11 12 Figure 9 5 Response time histogram for the message stream S7 Comparative Performance Analysis in an Error Free Environment 1ALRety 8 9 10 1 12 13 m 2ALRetries 14 15 16 18 19 Response Time ms 20 21 22 O L RSI BS 3ALRetries 24 25 Table 9 9 Response time of the message stream S3 R S 1 28067 14 19700 14 40467 13 35167 13 15400 12 97300 11 11733 8 15633 B sie 0 0 0 0 75 89062 19 69454 0 08483 0 00014 Interval ms 2 13 13 14 4 15 15 16 6 17 17 18 8 19 19 20 Rs 5 47100 3 19200 1 65500 0 68300 0 26800 0 07233 0 02233 0 0016
161. et Rotation Time Trp of the token and TSL is the Slot Time Ts The number of retries is defined by the max retry limit parameter The parameters with bem prefix are used to define the channel Bit Error Model BEM in use In Annex B we describe the BEM supported by these simulators Figure 5 9 presents part of the configuration file related to one instance of the Domain module This Domain module instance is called D1 and maps a wireless domain operating at 2 MBits s 50 PROFIBUS Simulation Model Baud_rate parameter Each frame has a head of 32 bits no frame tail and each character is coded using 8 bits The G is set to one therefore the GAP Update mechanism is always active The HSA can be set differently for each domain according to the highest address of the stations that can belong to its domain The TTR and TSL parameters are set in bit times and represent the 77 and Ts PROFIBUS parameters respectively Transmission errors are modelled using the Independent Channel Model Willig and Wolisz 2001 bem type parameter equal to 1 with a bit error probability of 10 0 00001 _bem pari parameter heProfibusNet domain heProfibusNet domain heProfibusNet domain name D1 _medium 0 Baud_rate 2000000 frameHeadLen 32 frameTailLen 0 0 0 0 heProfibusNet domain 0 heProfibusNet domain 0 heProfibusNet domain 0 bitsPerChar 8 heProfibusNet domain 0 G 1 heProfibusNet domain 0 HSA 5 heProfibusNet domain 0 TTR 300 he
162. example IDP a Figure 2 11 Example timeline for an Inter Domain Transaction IDT between master M3 and slave S6 Figure 2 12 Bridge based hybrid wired wireless PROFIBUS network example IDMP Figure 2 13 Phases of the Inter Domain Mobility Procedure A Figure 2 14 Simplified timeline of IDMP Phase 1 0 0 ee eeessssesessessscsessesesessssescsecsessscsssaecaceavaesensvsssessesvessseccaseneesessesseaseese Figure 2 15 Simplified timeline of IDMP Phase 2 0 0 0 eessssssssesessesssesecsesenessccscseceesessssacscuceasaesensenssessesvassceseevseassevessesseavaese Figure 2 16 Simplified timeline of IDMP Phase 3 and Phase 4 Figure 3 1 Example timeline for an IDT between M3 and S6 considering transmission errors Figure 3 2 Example timeline for an IDT between M3 and S6 considering transmission errors and using SDA service 29 Figure 3 3 A simplified timeline for the Phase 1 Figure 3 4 Tree data structure example Figure 3 5 Simplified timeline of the Tree like topology mechanism for Phase 1 Figure 3 6 Simplified timeline of the Tree like topology mechanism for Phase 2 Figure 3 7 Simplified timeline of the Tree like topology mechanism for Phase 3 Figure 3 8 Simplified timeline of the Tree like topology mechanism for Phase 4 Figure 3 9 Simplified timeline of the Timer Based mechanism for Phase 1 eee Figure 3 10 Simplified timeline of the
163. f PROFIBUS is supported by two tables the Gap List GAPL and the LAS It may also optionally maintain a Live List LL table The GAPL consists on the address range from address TS until NS This includes all possible addresses except the address range between Highest Station Address HSA which cannot be a master s address and 127 which does not belong to the Gap Initialization is primarily a special case of updating the LAS and the GAPL If after power on of a master station in the LISTEN TOKEN state a time out is encountered i e no bus activity within Time Out Time Tro it shall claim the token in the CLAIM TOKEN state and it starts initializing the logical ring The master station with the lowest station address starts initialization by transmitting two token frames addressed to itself Destination Address DA SA TS it informs any other master stations entering a NS into the LAS that it is now the only station in the logical token ring Then it transmits an FDL Request Status frame to each station in an incrementing address sequence in order to register other stations The first master station to answer with Ready to Enter Logical Ring is registered as NS in the LAS and thus closes the Gap range of the token holder Then the token holder passes the token to its NS When a master station is in the LISTEN _TOKEN state it shall monitor the bus activity in order to identify those master stations which are already in the logical tok
164. f from the logical ring Figure C 5 depicts the token pass procedure The first step is to evolve the Master from the USE TOKEN state to the PASS TOKEN state and sets the retry counter variable to zero After that it builds the token frame addressed to its successor NS and transmits the token frame If the token frame received is equal to the token frame transmitted then it evolves to the CHECK_TOKEN PASS state If after transmitting the token frame and before the expiration of the Ts the Master receives a frame it assumes that its NS owns the token and that it is executing message cycles and evolves to the ACTIVE IDLE state If the Master does not receive a frame within the Tsz it returns to the PASS TOKEN state and it repeats the transmission of the token frame its state machine evolves again to the CHECK TOKEN PASS state for the last time retry counter 2 and waits another Tsz If it receives a frame within the second Tsz it assumes a correct token frame transmission Otherwise it continues repeating this procedure until it has found a successor from its LAS getNs If it does not succeed it transmits the token frame to itself At the first time that the received token frame is different from the transmitted frame it transmits again the token frame At the second time it evolves to the LISTEN TOKEN state C 1 5 GAP Update Procedure Each master in the logical ring is responsible for the addition and removal of masters that have ad
165. f the BHW2PNetSim PROFIBUS 8954 57 BEACON IDP 246 6 IDMP 131 2 Figure D 11 Output frame accounting file excerpt Detailed information about corrupted frames is also recorded to file In such kind of file each line is composed by several fields separated by colons see Figure D 12 The first field is the timestamp at which an invalid frame was detected the second and the third fields are the DA and SA contained in the frame respectively The frame s type appears in the fourth field The remaining fields contain the remaining frame parameters Start Delimiter SD Frame Control FC and the Mobility Code MC 0 209914 4 6 IDMP IQ_REQ REQUEST_OR_SEND_REQUEST_FRAME SEND_DATA_WITH_NO_ACKNOWLEDGE_HIGH 0 214466 44 3 IDF REQUEST_OR_SEND_REQUEST_FRAME SEND_AND_REQUEST_DATA_HIGH 0 220326 5 4 PROFIBUS FDL_REQUEST_STATUS REQUEST_OR_SEND_REQUEST_FRAME REQUEST_FDL_STATUS_WITH_REPLY 0 222138 42 1 PROFIBUS REQUEST_OR_SEND_REQUEST_FRAME SEND_AND_REQUEST_DATA_HIGH 0 226439 5 4 PROFIBUS FDL_REQUEST_STATUS REQUEST_OR_SEND_REQUEST_FRAME REQUEST_FDL_STATUS_WITH_REPLY 0 226918 45 1 PROFIBUS REQUEST_OR_SEND_REQUEST_FRAME SEND_AND_REQUEST_DATA_HIGH 0 227865 4 1 PROFIBUS TOKEN Figure D 12 Information about invalid frames relayed by a Domain module instance Figure D 13 depicts a screenshot of the spreadsheet generated by Bit Error Model Frame Accounting option where the information contained on the referred kind of fil
166. f transaction that do not miss its deadline we will show in the next section that using a careful setting of master station addresses the performance of the bridge based architecture can be improved 10 6 Address Assignment Rules In order to handle token loss due to an error on the current token owner every master listens permanently on the medium Every time the medium goes idle each master starts the Tro timer which is reset when the medium goes busy If the Tro timer expires no activity on the medium for some time a master claims the token always the master with the lowest address i e it starts behaving as if it is the current token owner and performs some frame transmissions it sends data frames or passes the token to its current Next Station NS Ifa master was not in the LISTEN TOKEN state then when Tro expires no changes occur on its parameters specifically to its List of Active Station LAS NS and Previous Station PS parameters If it is in the LISTEN TOKEN state when the Tro expires then it evolves to the CLAIM TOKEN state and assumes that it is the only station in the Comparative Performance Analysis in an Error Prone Environment 111 logical ring note that when a master evolves to the LISTEN_TOKEN state it clears all entries from its LAS and the NS and the PS parameters are set equal to This Station TS In order to recover the token and reinitialize the logical ring the master which is in the CLAIM_TOKEN state tr
167. from one station through its uplink channel and retransmit the frame on its downlink channel Several aspects have influence in the quality of wireless communications The transmission path between transmitter and receiver can vary from a simple line of sight to one severely obstructed by objects like buildings mountains and other surrounding objects These objects can cause reflection diffraction and scattering of the radio waves Reflection of the radio waves occurs when the radio wave impinges upon an object which has very large dimensions compared to the wavelength Diffraction occurs when a radio wave encounters obstructions and propagates around the edges corners and behind the obstruction causing secondary Mobility Simulator 83 radio waves to form behind the obstruction Scattering results from rough surfaces whose dimensions are of the order of the wavelength which causes the reflected energy to scatter in all directions Information source Information destination 1 Channel 1 Transmitter gt Receiver Figure 8 1 Wireless communication model Additionally the objects could be perfect dielectric perfect conductor or in between A dielectric object absorbs some of the radio wave energy while the remainder is reflected back to the medium A perfect dielectric object absorbs all radio wave energy without reflection A perfect conductor reflects all radio wave energy As a consequence a radio wave
168. g to domain D and M3 belongs to domain D Figure 8 3 View of the MSim front end component E _ 7 BS2 operating at 3 ate 2 che Ens M3 is M4 and S6 transmitting are transmitting by BS2 by BS1 2 40GHz 2 30GHz 4 Ee 2 e Figure 8 4 Top view of the MSim front end component It is also possible to interact with this simulator using the keyboard The Esc key finishes the simulation run The f key allows increasing the simulation clock and in opposite the s key decreases the clock The goal of this component is not to run a simulation but to validate the simulation environment The r key allows running simulation runs in a background process which was referred as the simulation engine module Since this component does not have any graphics output then the simulation runs can be performed in much less time Mobility Simulator 87 8 6 Simulator Configuration The simulation configuration is done through two text files One of them specifies the parameter values of all components and the other specifies the seed number for the generation of random values used in the Log normal Shadowing model For each seed number a simulation run is done Therefore the number of simulation runs depends on the number of text lines Figure 8 5 presents part of configuration file The duration of each simulation run defined by the sim duration parameter is equal to two minute
169. h getAction msg 4 case TTO TIMEOUT Se switch state Ge case ACTIVE IDLE T messageDispacthing 8 end oF case LISTEN TOKEN AKO state CLAIM TOKEN abil tokenRecovery Ue oe end 14 OB end LAY Se ELS dista 1 Figure C 1 handleSelfMessage msg function pseudo code algorithm 136 Simulation Models Implementation In order to recover the token and reinitialize the logical ring the Master which is in the CLAIM_TOKEN state transmits two token frames addressed to itself Figure C 2 The pass token procedure will be described in Section C 1 4 In this way the token frame is recovered After that every Master will be joining to the logical ring using the GAP update procedure described in Section C 1 3 Note that when a Master transmits a token addressed to itself all Masters that are not in the LISTEN_TOKEN state evolves to that state since they are skipped of the logical ring tokenRecovery 1 passToken TS passToken TS state USE_TOKEN messageDispacthing YHDOBWNE Figure C 2 tokenRecovery function pseudo code algorithm C 1 2 Token Reception Procedure The token frame is passed between masters in ascending Medium Access Control MAC address order The only exception is that to close the logical ring the master with the Highest Station Address HSA must pass the token frame to the master with the lowest one Each master knows the address of
170. h hybrid wired wireless network must fulfil the requirements of industrial automation applications In this dissertation we analysed the performance of two approaches that extend the PROFIBUS standard to support wireless communications One based on repeaters and another based on bridges In the repeater based approach the interconnection between wired and wireless segments called as domains is supported by Intermediate Systems ISs operating as repeaters 1 e at Physical Layer PhL level The support of repeaters requires a specific settings of some PROFIBUS timing parameters Slot Time and Idle Time which results in a lower responsiveness to errors and on an increased latency of the message cycles Additionally the use of repeaters creates a broadcast network The bridge based approach triggered an alternative approach where the ISs behave as bridges i e at Data Link Layer DLL level This approach the bridge based approach solves some of the problems of the repeater based approach The bridge based approach creates a Multiple Logical Ring MLR network and as consequence there is the need of two more protocols One for supporting the communication between stations that belong to different domains the Inter Domain Protocol IDP Another to support the mobility of wireless mobile stations between different wireless domains the Inter Domain Mobility Procedure IDMP The main objective of this dissertation is to compare th
171. h the new one and updates its LAS In the same way it evolves to the USE TOKEN state calculates the new Try sets to false the GAP_Turn variable and then the token reception procedure ends According to the PROFIBUS DLL only one GAP Update procedure can be performed per token visit the GAP_Turn variable is used to avoid that more than one GAP Update procedure is performed when a Master is holding the token frame The execution of the token reception procedure forces the execution of the message dispatching procedure described in the Section C 1 3 C 1 3 Message Dispatching Procedure Figure C 4 presents the message dispatching procedure when a Master holds the token frame This procedure is repeatedly performed until the Master expires Try and the pass token procedure is executed At token frame reception the period during which the Master is allowed to perform messages cycles Try is computed according to the Eq 2 1 Simulation Models Implementation 137 Token Reception Procedure Receive a token frame as bit errors 2 Yes Discard the token Yes No frame s this the 2 token from the same SA No state USE_TOKEN PS SA Tr Trr Ter Update LAS GAP_Turn false Figure C 3 Token Reception procedure Message Dispatching Procedure One high priority message processed 2 Yes one_high_msg true GAP Update Procedure Is the high priority message queue empty Teu
172. he masters e g the token or request messages or by the slaves e g responses to masters requests are broadcasted throughout the overall network Moreover all masters in the network belong to the same logical ring For this particular example the token rotation can have the following sequence gt M1 gt M2 gt M3 gt M4 gt M1 In the repeater based approach inter domain mobility is supported and is implemented in a very simple and efficient way Periodically one specific master in the system denoted as Mobility Master emits a special non acknowledged request the Beacon Trigger This message is received by all BSs in the system which in turn start to transmit Beacon frames in their respective radio channels When the wireless stations receive the Beacon frames they start assessing the quality of the different radio channels operating in the network At the end of this assessment phase wireless stations switch to the channel with the best quality Due to the broadcast nature of the network other timing parameters must also be properly set for the system to work correctly Bridges operate at DLL level Assuming a two port bridge interconnecting two different network segments frames arriving to one bridge port are only relayed to the other port if the destination address embedded in the frame corresponds to the MAC address of a station physically reachable through that other port With a MAC protocol as the one used in PROFIBU
173. he no gaps instant is the earliest instant to start relaying the PhL frame from D to D in a way that guarantees that the transmission in D is continuous Consider the example depicted in Figure 2 4 The first time instant is the data ready f4 followed by the time instant when the length of the frame is known fw The last instant thus the maximum of the three is the time instant that guarantees a continuous retransmission of the PhL frame Ta This situation usually happens when the duration of the PhL frame in D is smaller than in D Figure 2 4 Timing behaviour of a repeater 2 3 3 Traffic Adaptation Network interconnection often brings up the problem of network congestion Generally if for any time interval the total sum of demands on a resource is more than its available capacity the resource is said to be congested for that interval In the case of computer networks resources include buffer space and processing capacity in the ISs and for example if during a short interval the buffer space of an IS is smaller than the one required for the arriving traffic frame loss may occur dropped frames and the IS is said to be congested It is also true that the congestion problems depend dramatically on the type of IS used in the interconnection Particularly if the ISs act as repeater traffic congestion may occur as a result of the heterogeneous characteristics of the interconnected physical media The heterogen
174. he number of transactions is higher using the SDA service than using the SDN service The exception is message stream S3 when the MeanBER probability is equal to 10 Concerning the percentage of transaction that do not miss the deadline using the SDA service the percentage is higher than using the SDN service once again the exception is message stream S which has a value somewhat worst when considering the MeanBER probability equal to 10 and 10 Although the results for message stream S were not very interesting globally the results using the SDA service are better than using the SDN service Nevertheless we show in Section 10 6 1 that these results can be improved by the proper setting of master stations MAC addresses Comparative Performance Analysis in an Error Prone Environment 107 MeanRT ms 10 10 10 10 10 10 10 10 10 10 10 10 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 4 MeanRT using the SDN and the SDA services 1600 SDN E so 1000 N of Transactions Thous ands o s 10 10 10 10 101 10 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 5 Number of transactions using the SDN and the SDA services SDN E SDA Concluded transactions 105 104 10 i 105 10 102 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 6 Percen
175. he simulation run The Domain module models a network domain and interconnects all components in a single network domain The Master and Slave modules model a master or slave standard PROFIBUS network device On the left side Figure 5 1 shows how the main modules are interconnected There are 2 kinds of the connections ctrl con and domain con connections The ctrl con connections are used to establish the connections between the Controller module instance and all module instances in the overall system This kind of connection has no delay and is used for simulation control and configuration purposes The domain_con connections are used to establish the connections among all domain components between Master and Slave module instances and the Domain module instance On the right side Figure 5 1 there is a connection example between a Master and a Domain module instances The Master instance is called M1 and Domain instance D1 Each of these connections is composed of four gates two for each connected module instance One gate is for input and it is connected to the output gate of the other module instance and vice versa The messages are sent to D1 from M1 through a gate called domain gateout Consequently D1 receives messages from M1 through a gate called station gateInM1 D1 sends messages to M1 through a gate called station_gate0utM1 46 PROFIBUS Simulation Model Some station parameters like Tspr or T p are modelled either by Probability Distri
176. his approach required two new protocols one for supporting the communication between stations in different network segments the Inter Domain Protocol IDP and another to support the mobility of wireless stations between different wireless segments the Inter Domain Mobility Procedure IDMP The main objective of this dissertation is to compare the timing behaviour of the bridge and repeater based approaches over error free and error prone environments Additionally we also intended to show that the bridge based approach implementation is feasible and propose additional error detection and correction mechanisms which would improve its performance over error proneenvironments To achieve these objectives two simulation tools have been developed one for the repeater based approach and another to the bridge based approach and a set of result analysis tools Additionally we have also developed another tool to simulate the mobility of wireless stations O IPP Hurray Research Group 1 www hurray isep ipp pt UNIVERSIDADE TECNICA DE LISBOA INSTITUTO SUPERIOR INSTITUTO SUPERIOR TECNICO TECNICO Performance Analysis of Wireless enabled PROFIBUS Networks Paulo Manuel Baltarejo de Sousa Licenciado Dissertac o para obtenc o do Grau de Mestre em Engenharia Electrot cnica e de Computadores Orientador Doutor Lu s Miguel Moreira Lino Ferreira Co Orientador Doutor Carlos Manuel Ribeiro Almeida J ri Presidente Dout
177. iator Responder M3 s6 Bridge 1 Bridge 2 M8 M5 M6 M9 DLL AL DP AL DP DLL Codes the frame using IDP and open a transaction in LOT Decodes the IDF originating a Token Service_upd req Service req PROFIBUS o ESTA PROFIBUS frame _ Service con No_Data Token Token PROFIBUS Service req e PROFIBUS pa request 3 Service con PROFIBUS Transmission response No_Data keni Token Service req AT Codes the response PROFIBUS q PROFIBUS ack x Fame usina IDF request Decodes the IDF and Ts gt PROFIBUS close the transaction in expiration Service con response LOT Data p D D O Open transaction Close transaction th Transmission error Figure 3 2 Example timeline for an IDT between M3 and S6 considering transmission errors and using SDA service 3 3 Handling IDMP Errors The original IDMP version is also prone to errors similarly to the IDP case In this section the vulnerabilities of the IDMP are identified and solutions to those problems are outlined 3 3 1 Possible IDMP Error Situations Figure 3 3 presents an error scenario for possible IDMP errors This scenario assumes the network configuration presented in Figure 2 12 where BM M6 assumes both roles of GMM and DMM of wired domain D BMs M8 M9 and M7 assume the role of DMMs for wire
178. ich are named M1 M2 M3 and M4 and six Slave module instances S1 S2 S3 S4 S5 and S6 Its PROFIBUS Simulation Model 49 DMM is Master module instance named M1 The Domain module instance is depicted in the screen at position 400 300 This information enables the controller to set the LAS of all Master module instances as well as PS NS and GAPL parameters It also assigns the token frame to the master defined as DMM thus avoiding the need to perform the standard PROFIBUS network initialization procedure theProfibusNet controller domain lt d gt lt n gt D1 lt n gt lt m gt M1 M2 M3 M4 lt m gt lt s gt S1 S2 S3 S4 S5 S6 lt s gt lt dmm gt M1 lt dmm gt lt pos gt 100 200 lt pos gt lt d gt Figure 5 7 Configuration file related to the controller module instance excerpt 5 2 3 Domain In spite of the OMNeT capacities only one to one connections are supported One to many and many to one connections can only be achieved using special purpose simple modules Therefore it was necessary to develop a simple module the Domain module which is able to connect all stations in a domain and simulate a broadcast network The connections are created and assigned dynamically enabling the support of mobility In our model we assume that the propagation delay is ignorable The transmission delay is simulated by the Domain module as a function of the Baud_rate parameter and the message length The parameters that a
179. ied timeline of Burst Error Periodic Model oo ceeeseseesesesceeeeeseseeeceeseeeeecsesceecsesesseeseeseseessseaeeeeaeseseeeeas 132 Figure C 1 handleSelfMessage msg function pseudo code algorithm Figure C 2 tokenRecovery function pseudo code algorithm Figure C 3 Token Reception procedure cece a Figure C 4 Message dispatching proc UE soii noto transita Figure C 5 Pass Token POC Us Figure C 6 Send FDL_Request_Status procedure s Figure C7 Receive FDL Request Status procedure mii siso sor Ei EREE vase EREEREER NEEE A 140 Figure 8 Send Frame Procedure em 22 202 catteseticsh ey nia 141 Figure C 9 SDN transaction schema between master and slave 142 Figure C 10 Send SDN Procedure occocicicncncnnncnnnnnnnncnncnnncncnos 142 Figure C 11 Receive SDN Procedure 143 Figure C 12 SRD transaction schema between Master and Slave c oooocicncnccociconononnnncnnnnnnononnno nono rn cnc cnn rn oran rana cnn rn onnnnan aran 143 Figure C 13 Send SRD Proc dlte int E ncaa sas AA ca ad salves Gas POCO aed lec SET GNR que T mri eee e Figure C 14 Receive SDR Procedure Figure C 15 Send Beacon Procedure Figure C 16 Receive Frame from Domain Procedure s csscsssesseessvssnvsesconenessucssssessotscesensencsesnsevesuensesscsresnsassaetenvesesers Figure C 17 Receive frame from ComFunc procedure ee Figure C 18 Senid IDE Procedure
180. il it finds a successor from its List of Active Stations LAS Token Cycle After receiving the token a master station is allowed to execute message cycles during Token Holding Time Try that is computed as follows Tra Tx Tre 2 1 Try is equal to the difference if positive between the Target Rotation Time Trp and the Real Rotation Time Trg of the token Trg is a parameter common to all masters in the network which must be set to the expected time for the token cycle Trp is the time measured between two consecutive token receptions the token cycle PROFIBUS defines two main categories of messages high priority and low priority each using a different transmission queue that is handled differently by the DLL At the arrival of the token the Try timer is loaded with the value corresponding to the difference between Trp and Tre If the token is delayed then Try is set to zero and the master is only allowed to perform at most one high priority Technological Context Communication Infrastructure 9 message transaction Otherwise the master is allowed to perform high priority message transactions until the value of the Try timer becomes negative Low priority messages are only transmitted when the high priority queue is empty and Tr is still positive Note that once a message cycle is started it is always completed including any retries even if in meanwhile Try expires Re Initializing the Logical Ring The logical ring o
181. ility Simulator This chapter describes this tool 8 1 Introduction Every radio technology has a limited physical coverage area within which radio communications can be performed in acceptable conditions The area characteristics depend on several aspects such as the dimensions and layout of obstacles the existence of electromagnetic interference and the radio technology including its antenna type in use Therefore to cover wider areas it is necessary to divide the application area into several radio cells each one operating at its own radio channel While moving wireless mobile stations will always try to use the best radio cell signal to communicate therefore these stations will belong to different radio cells also called domains within the context of this dissertation The domain to which a wireless mobile station belongs depends on its location over time and on the signal strength in that location In a simplified model the signal strength depends on the distance to the wireless cell base station There are several models to estimate the radio signal strength Rappaport 1996 Simple models estimate radio signal strength between a transmitter and a receiver based solely on the distance between them More sophisticated models use environment information such as buildings mountains wall materials and location of obstacles These models require a very detailed representation of the objects in the environment and are computationally very d
182. ilot simulation runs with the purpose of validating the implemented model The sixth step is to verify if the computer program is valid Validation ensures that no significant difference exists between the programmed conceptual model and the real system If it is an invalid program the simulation study must return to step 2 otherwise the simulation study evolves to step 7 After the implemented model has been validated it must be specified for each system configuration the length of each simulation run and the number of independent simulation runs each run must use different seed numbers for random number generation step 7 The following step is to make simulation runs step 8 and analyze the output data produced by the simulation runs In the last step documentation must be produced about the simulation runs as well as about the simulation system and simulation implementation Technological Context Simulation Software 39 4 3 Simulator Implementation One of the most important decisions in a simulation study concerns to the choice of the simulation software Simulation software can be divided into three categories The first category includes all general purpose programming languages such as C C and Java just to mention some The second category includes simulation programming languages such as PARSEC Meyer and Bagrodia 1998 SIMSCRIPT II 5 Russell 1999 and SimPy Vignaux and Muller 2006 The third category is related to simulatio
183. inally when a wireless domain has only one BM and the bridge is also the DMM it must transmit a Void message to maintain the network activity 2 5 Summary This chapter presented an overview of the most relevant features of the PROFIBUS protocol necessary to tackle the remainder of the dissertation it also highlighted the main architectural features of the repeater and bridge based approaches The objective was to provide the reader with the necessary background and intuition for tackling the remainder chapters of this document In this chapter the mobility procedure for the repeater based architecture was described It is very simple and errors occurring during its execution can be tackled by native PROFIBUS error handling mechanisms But the mobility mechanism used on the bridge based architecture is more complex and involves the exchange of many messages between the intervening stations Errors during the execution of the IDMP must be tackled by specific mechanisms implemented in the bridges The next chapter will analyse those error situations and propose solutions for the error handling problem of the IDMP Chapter 3 Error Handling Improvements for the Bridge based Architecture The original IDMP protocol did not define any error handling mechanisms regardless of the mechanism used in PROFIBUS The IDP also relies on a simple timeout timer to control the success of an IDT As a consequence these error handling mechanisms could lead
184. ing and the purpose of each message is also the same 3 3 2 Tree like Topology Based Mechanism Due to the routing mechanism a bridge based network can be represented by a tree since there is no close path between any two stations Usually a tree data structure places one node called the root then proceeds down connecting one or more nodes beneath each node on the previous level until n nodes have been placed The nodes below each node are called its sons the node above each node is called its father The sons are themselves the roots of trees called the sub tree A leaf node is one which is not a root of any tree or sub tree Figure 3 4 shows the above concepts Node 0 is the root of the tree and has two sons nodes 1 and 2 consequently it is the father of these nodes Since node 1 is not a leaf node then it is a root of a sub tree Dr Ome Dn Figure 3 4 Tree data structure example This error handling proposal requires some adaptations to the original IDMP For Phase 1 the tree is composed by the GMM as root and the network BMs Phase is started by the GMM when it sends a SMP message and finishes when all BMs reply with a RSMP message To handle transmission errors each root node of every sub tree periodically sends SMP messages until receiving a RSMP message from its son nodes When all son nodes have replied with a RSMP message which means that their LOTs are empty then the root node stops sending SMP message After
185. ing queuing delay by a repeater Traffic congestion in a repeater can be avoided through the insertion of additional inactivity idle intervals before issuing a message transactions Alves Tovar et al 2002 Obviously the insertion of this additional idle time reduces the number of transactions per time unit when the responder is not in the same domain as the initiator The PROFIBUS MAC mechanism allows only one station master or slave to transmit at a given moment in time Every master in PROFIBUS has two different Idle Time parameters Tip and Tpz As mentioned in Section 2 2 2 a master always waits Tp after receiving a response acknowledgement or a token frame before transmitting another frame It must also wait Tmz after transmitting an unacknowledged request frame and before transmitting another frame request or token For a traditional wired network all masters may set their T p and Tpz parameters to the minimum default value which is usually adequate to cope with bit synchronisation requirements In this approach the traffic adaptation is based on the computation of the additional idle time that must be inserted by each master in order to properly encompass the interconnection of heterogeneous physical media The timing diagram depicted in Figure 2 6 illustrates a sequence of transactions where queuing delay is zero for all transactions on the first repeater This is due to the additional extra idle time Transaction 1 Trans
186. ing the Prepare for Beacon Transmission PBT message After receiving the PBT message a DMM retains the token after token reception obviously starting the inquiry sub phase When receiving a PBT message all BMs in the network clear their RT entries related to wireless mobile stations On the inquiry sub phase the DMMs start by transmitting a Ready for Beacon Transmission RBT message to the GMM signalling that they are on the inquiry sub phase ready for Beacon transmission Then every DMM sequentially sends Inquiry frames addressed to the BMs belonging to its domain The BMs use the response message to transmit any mobility related message that they require to transmit Wireless terminating domains i e wireless domains connecting to only one bridge emit Void frames in order to maintain network activity Note that in this kind of domains a DMM does not have to retrieve any mobility related message from the other bridges This procedure allows a faster communication between the GMM and the DMMs while at the same time the inaccessibility period of the wired stations is kept small Phase 2 ends when all Ready for Beacon Transmission RBT Technological Context Communication Infrastructure 23 messages are received by the GMM Figure 2 15 shows a simplified timeline related to IDMP Phase 2 assuming the network scenario presented in Figure 2 12 Phase 2 starts DMM BM DMM DMM BM DMM M7 M10 M9 M6 M5 M8 Tok
187. ion by the responder The simulation results presented in Section 10 4 showed that using the SDA service would lead to a better performance than using the SDN service The original IDMP only had very limited error handling capabilities In an error prone environment this protocol could lead to blocking situations therefore we have proposed an error handling mechanism which permits to solve the detected problems This mechanism is based on timers which control the IDMP phases These enhancements provide the bridge based approach with the necessary mechanisms to be used in an error prone environment 11 2 2 Comparison between the Repeater and Bridge Based Approaches In Chapter 9 a performance comparison between the repeater and the bridge based approaches was performed considering an error free medium and by varying some important network parameters From that comparison it has been shown that the performance of the bridge based approach is less influenced by changes on the network parameters Additionally the bridge based approach is not dependent on the network parameters or configuration i e the impact of changes on the network parameters or configuration is minimum in relation to the repeater based approach in which any change on the network parameters or configuration implies changes to the parameters settings in all stations This performance comparison was based on response time and throughput results From these results it was possible
188. ion of a standard PROFIBUS network which are common to both architectures The simulation model implementation by the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator is detailed in Chapter 6 and Chapter 7 describes the simulation model implemented by the Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator In the approaches analysed in this dissertation wireless stations are able to move between different domains Therefore in order to achieve more realistic simulation results it is necessary to know at which points in time a wireless mobile station moves between different wireless domains To obtain this information a tool was developed which simulates the mobility of wireless stations over a factory floor and the radio signal quality of different wireless domains at station location the Mobility Simulator This tool is described in Chapter 8 A comparative performance analysis between the repeater and bridge based architectures is performed in Chapter 9 which evaluates the influence of varying some network parameters in the timing behaviour of each approach In order to evaluate the performance of both architectures considering transmission over an error prone medium Chapter 10 presents simulation results in which the transmission errors are modelled according to the Gilbert Elliot Channel Model Additionally the changes proposed in Chapter 3 for the bridge based architecture of this dissertation are evaluated Fi
189. ion of the PhL frame in domain D The instant for a specific repeater depends on its operation mode either store and forward or cut through For a cut through repeater the following is assumed 14 Technological Context Communication Infrastructure When relaying a frame from D to D it cannot start being relayed while the first char of the DLL frame of D is not completely received by the repeater The PhL frame cannot start being relayed while the length of the DLL frame is not known by the repeater When relaying a frame from D to D the instant for start relaying the PhL frame must take into account that the repeater cannot run out of bits to relay from D to D i e the transmission of a PhL frame in D must be continuous without time gaps Taking these assumptions into account which are illustrated in Figure 2 4 the start relaying instant for a cut through operation mode repeater is defined as ke max ar tn t ne 2 7 where fp the data ready instant is the instant at which a predefined amount of DLL data has been received from D ready to be relayed counted since the beginning of the PhL frame in D For the cut through behaviour it is considered that it is the instant at which the first DLL character is completely received tk the length known instant is the instant at which the length of the DLL frame in D is known counted since the beginning of the PhL frame in D Pa t
190. ission of the PhL frames each character is coded using 11 bits The three additional bits are related to one start one stop and one parity check bit In wired domains the PhL frames do not have a head or a tail sequence of bits Table 9 1 shows for each domain the bit rate the length in bits of the frame head and of the frame tail as well as the number of bits used to code a character 92 Comparative Performance Analysis in an Error Free Environment We have assumed that the time required by a slave to answer a request frame Tspx can be modelled stochastically using a triangular distribution function with apex at 70 bit times and extremes at 11 and 100 bit times triang 11 70 100 This distribution has been chosen since the triangular distribution function is a rough model when there is no data available about the real distribution function Law and Kelton 2000 Henceforth the following notation for the triangular distribution function triang minimum apex maximum will be used Table 9 1 Physical media parameters Domain Parameters D andD 2000000 32 0 8 D 1500000 0 0 11 D 500000 0 0 11 The internal delay of the ISs is equal to 30 us and the maximum number of master DLL retries max retry limit parameter has been set to one The mobility procedure is triggered every 200 ms and it is assumed that the wireless mobile stations move in the simulated environment according to pre defined path In this path the
191. ist of Bridge Masters in the Network List of Domain Mobility Managers in the Network Frame Length PROFIBUS Standard Frame Length repeated PROFIBUS Standard Live List PROFIBUS Standard List of Open Transactions List of Wireless Mobile Stations in the Network Medium Access Control maximum response time Mobility Code Mean Bit Error Rate Mean response time Minimum response time Multiple Logical Ring Mobility Master Mobility Simulator Next Station PROFIBUS Standard Objective Modular Network Testbed in C Open System Interconnection Prepare for Beacon Transmission Personal Computer Point Coordinator Function IEEE 802 11 Personal Digital Assistant Probability Distribution Function Protocol Data Unit Physical Layer Programmable Logical Controller PROcess Fleld BUS PROFIBUS Decentralised Peripherals PROFIBUS Fieldbus Message Specification Previous Station PROFIBUS Standard Ready for Beacon Transmission High Performance Wireless Fieldbus in Industrial Multimedia Related Environment Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator Ready to Start Mobility Procedure Routing Table Route Update Acronyms and Symbols 171 SA Source Address PROFIBUS standard SAE Source Address Extension PROFIBUS standard SBT Start Beacon Transmission SC Short Acknowledge PROFIBUS standard SD Start Delimiter PROFIBUS standard SDA Send Data with Acknowledge PROFIBUS Standard SDN Send Data
192. its Previous Station PS the address of the Next Station NS and its own address This Station TS When a master receives a token frame addressed to itself from a master registered in the LAS as its PS then this master is said to be the token owner On the other hand if a master receives a token frame from a master which is not its PS it shall assume an error and will not accept the token frame However if it receives a second subsequent token frame from the same master it shall accept the token frame and assumes that the logical ring has changed In this case it updates the original PS value by the new one and updates the LAS and the Live List LL Figure C 3 illustrates the token reception procedure A token frame is discarded when it is an erroneous frame or if it is not addressed to the Master If the token frame is addressed to the Master and it does not contain bit errors then the Master behaviour depends on the token frame transmitter i e if the token frame Source Address SA is registered as its PS If the token frame transmitter is registered as its PS it evolves to the USE TOKEN state calculates the Token Holding Time Tyr according to the Eq 2 1 sets to false GAP_Turn variable and then the token reception procedure ends Otherwise the received token frame is discarded when a Master token frame sender is not its PS However if it receives a second token frame from the same Master then it updates the original PS value wit
193. l stripes means that the field is not available to the IDF in this specific case fixed length frames with no data are mapped into frames of fixed length with data field The equal symbol means that the field must the equal to the original embedded frame field Table 2 1 Mapping between standard PROFIBUS frames and IDFs Frame Header PROFIBUS Frame Data Original Type of defined IDP defined Frame Data LE SD DA SA FC DAE SAE TI EFT EFC Unit A Req ES SD3 10 TI 1 EFC Fixed length a no data Ack srs SD3 10 M 2 Bee q e Short sosesesese ack BBB SD3 sa DA 10 cc TI 3 HC Data y 3 Z Fixed length ES O 4 EFC w data Res Data SD2 _ o _ E E R _ len Ral na TI 6 EC len Var length Dat E 2 E 10 T TI 7 EFC In the conversion the IDFs preserve the same DA and SA except in the case of the SC frame which does not have DA or SA In this case the IDF includes the DA and SA obtained from the request frame To distinguish IDFs from other frame types the Function code of the FC field must be equal to 10 note that this feature also imposes a non standard behaviour by the BMs DLL And its remaining sub fields should be filled with the appropriate values for a PROFIBUS frame Note that all frames defined in Table 2 1 are transmitted as individual requests Finally SDN frames do not need Technological Context Communication Infrastructure 21 any conversion so they can be relayed by the bri
194. lassified according to its relaying behaviour store and forward when a PhL frame must be completely received from one port before being transmitted to the other port cut through when a repeater starts relaying a PhL frame which has not been completely received yet A repeater does not perform any address filtering This result in a broadcast network i e every station listens to every frame transmitted by any other station in the network The use of repeaters implies a single MAC address space and that only one logical ring exists in the network For that the network operation is based on the Single Logical Ring SLR 2 3 2 Wired Wireless Domains Interconnection A repeater may need to implement more than a bit by bit repeating functionality This is the case when it interconnects communication media with different PhL frame formats In order to encompass the functionalities referred each repeater has an associated internal relaying delay time tg It is assumed that the repeaters always introduce a minimum inactivity period minimum idle time Tipm between any consecutive PhL frames When a repeater receives a PhL frame from one port it must start the transmission in an instant the start relaying instant f that guarantees that the retransmission is done without time gaps t is defined as the earliest time instant for start relaying a specific PhL frame from domain D to domain D counted since the beginning of the recept
195. lbert Elliot Model and the Burst Error Periodic Model which are described in detail in this annex B 1 Introduction The purpose of this annex is to describe the Bit Error Model BEM supported by the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim and Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BHW2PNetSim Independent Channel Model Willig and Wolisz 2001 the Gilbert Elliot Channel Model Gilbert 1960 Elliot 1963 and the Burst Error Periodic Model Hazim 2006 This Annex describes the implemented BEMs B 2 Independent Channel Model This model is very simple and determines if a frame is correct or wrong that is there is a bit error in a frame As there is no correlation between two consecutive errors this model is called Independent Channel Model The result of this model is obtained using Bernoulli function Law and Kelton 2000 with parameter P err This parameter is computed as follows L Ema 1 1 Pre B 1 where L is the length in bits frame and p is the Bit Error Rate BER probability associated to the channel B 3 Gilbert Elliot Channel Model It is well known that transmission errors occur in bursts that is there is correlation between consecutive errors The Gilbert Elliot model Gilbert 1960 Elliot 1963 takes into account this correlation This model is a two state discrete time Markov chain as shown Figure B 1 Polg Pglb Pbjb Good Bad Pg P
196. le SMP SMP_ Transmission Token 4 error sup Token je SMP smp RSMP le SMP smp Token rsmp_ Token sme T 7 gt o me pe RSMP MP M10 RSMP_le M5 Token SMP 4 M6 e Token Token SMP Token gt Token Token le e SMP Phase 1 ends RSMP EA RSMP M10 gt m9 Dt D gt D D v v v v v v v Figure 3 5 Simplified timeline of the Tree like topology mechanism for Phase 1 Phase 2 starts when the GMM sends a PBT message to its son nodes In this schema the GMM periodically sends a PBT message until receiving a RBT message from all its son nodes The same mechanism is used by the sub tree root nodes They periodically send a PBT message until receiving a RBT message from its son nodes When all son nodes have replied with a RBT message then the root node of each sub tree stops sending PBT messages and if it is holding the token frame replies with a RBT message to its father node When a DMM replies with a RBT it starts the inquiry sub phase The BMs at reception of PBT message clear all entries of their RT concerning wireless mobile stations Figure 3 6 illustrates this mechanism considering the network example presented in Figure 2 12 Phase 2 starts DMM BM DMM GMM DMM BM DMM M7 M10 M9 M6 M6 M5 M8 Token 81 PBT Token Jl PBT _ gt PBT le Token PET PBT 4 p8r Token PET Token le Per AE RBT e BT uy PT Token
197. le stations As mentioned in Chapter 3 all LOT entries have an associated timers BM IDT Abort Timer TBm IDT4bort that is used to avoid endless IDTs When either of the timer Tgm IDTAbori Or the Tgm IDMPAbort CXpires the handleTimers msg function see Figure C 23 is automatically invoked If it is a timer related to the LOT entry IDT TIMEOUT action line 6 its correspondent entry is deleted If the LOT is empty and the Bm is in the WIDT_END state then the processRSMPmessage function is called and the BM evolves to the WINQUIRY state In order to recover from IDMP errors the BMs are provided with the Tgm IDMPAbore When this timer expires IDMP TIMEOUT action line 9 a BM evolves to the INACTIVE state and the IDMP ends il handleTimers msg 2 1 3 switch getAction msg 4 case IDT_TIMEOUT Be removeEntryFromLOT msg 6 if state WIDT END and isLOTempty then a processRSMPmessage 8 end OF case IDMP_TIMEOUT 10 state INACTIVE bil end Figure C 24 handleTimers msg function pseudo code algorithm DMM Operation Figure C 25 presents the pseudo code algorithm of the d11ReceiveToken function which is called by the DLL whenever its Master acts as a DMM at the token reception The dllholdtoken variable is set to true and if the DMM is in the WTOKEN state that means it has already received a SMP message then the DMM evolves to the INQUIRY state and the processRBTmessage is called Figure
198. less domain The mobility procedure is finished at reception of the SBT message for wired domains Phase 4 After the end of the Beacon transmission every wireless DMM still holding the token inquires all wireless mobile stations in order to detect if they are present in its domain using Discovery messages This period can also be referred to as the discovery sub phase From this instant onwards wireless mobile slaves are already capable of answering requests but wireless mobile masters must still enter the new logical ring using the standard PROFIBUS ring management mechanisms Since the RT entries related to wireless mobile stations have been cleared only when the BMs receive updated routing information embedded on RU messages at the end of the IDMP they may restart routing IDTs related to wireless mobile stations The RU messages are transmitted by the DMMs whenever they detect that a wireless mobile station is ready to start operating that is after the entry of a master into the logical ring or after the detection of wireless mobile slave using Discovery messages When a wireless mobile station continues in the same domain its presence is detected by a Discovery message and a RU message is transmitted by the DMM before releasing the token frame When a wireless mobile slave changes to another domain the detection in the new domain is also made by a Discovery message When a wireless mobile master changes to another domain its detec
199. less domain D wireless domain D and wired domain D respectively The GMM M6 starts the IDMP by broadcasting a SMP message After that it waits for RSMP messages from all BMs in the network Assuming that a transmission error occurs when BM M9 is sending a SMP message to BM M10 then BM M10 and BM M7 will not receive the SMP message If they do not receive a SMP message they will not reply with a RSMP message Consequently the IDMP will not evolve because the GMM M6 is expecting RSMP messages from all BMs Further the BMs that have replied stop accepting new IDTs therefore blocking the evolution of the IDMP Phase 1 starts BM BM BM GMM BM BM BM M7 M10 M9 M6 M6 M5 M8 Token le SMP SMP y Token Transmission So Ersmp RsMPZ error SMP M9 ae M6 PA smP TAS MP Token Token Toki RSMP_ oes Sa e Token le M8 Token al PSMP RSMP y RSMP le M5 4 M5 Token ek oken yl a RSMP y M8 RSMP_l e 4 M5 D D D D Y v v v v v v Figure 3 3 A simplified timeline for the Phase 1 30 Error Handling Improvements for the Bridge based Architecture To overcome such problems we propose two mechanisms the Tree like Topology Based and the Timer Based These mechanisms respect the main features of the IDMP they simply add error control and correction detection capabilities to the protocol The IDMP agents are the same and maintain the same functionalities The mean
200. main Protocol IDP The frames involved in IDTs both the standard PROFIBUS frames and the frames exchanged between the BMs are referred to as Inter Domain Frames IDFs IDFs conveying the request are called Inter Domain Request IDreq frames and equivalently the frames which convey the response are called Inter Domain Response IDres frames An Intra Inter Domain Transaction IIDT is a transaction that can be either an IADT or an IDT depending if the involved stations are either in the same domain or in different domains The stations involved in this kind of transactions are the wireless mobile stations Bridges perform routing based on the MAC addresses contained in the frames and on the RT entries of the incoming side The Inter Domain Protocol The IDP explores some PROFIBUS protocol features at the DLL and AL level which enable a master to repeat the same request until receiving a response from the responder station It defines the behaviour of the bridges and the codification of the frames exchanged between them related to a specific IDT Technological Context Communication Infrastructure 19 When a master starts a transaction with a station belonging to another domain an IDT it starts by transmitting a request frame addressed to the responder station an IDreq frame This frame is then relayed by only one of the BMs denoted as BM ini stands for initiator belonging to the initiator domain BM receives the IDreq frame
201. main are strings which define the configuration of the network domains and repeaters Both of them are written using predefined structure based in tags The Controller module instance extracts information from these strings to perform the network configuration simple Controller parameters _mm string _tmob numeric _domain string _inter_domain String endsimple Figure 6 3 Controller module NED definition of the RHW2PNetSim Figure 6 4 presents the parameter values related to the controller module instance assuming the network depicted in Figure 2 3 This file also includes information about MM Master module instance named MS and about the periodicity _tmob theRHW2PNet controller _mm M5 theRHW2PNet controller _tmob 200ms theRHW2PNet controller _domain lt d gt lt n gt D1 lt n gt lt m gt lt m gt lt s gt S6 lt s gt lt cp gt CP8 lt cp gt lt bs gt CP8 lt bs gt lt pos gt 400 300 lt pos gt lt d gt lt d gt lt n gt D2 lt n gt lt m gt M1 M5 lt m gt lt s gt S1 S2 S3 lt s gt lt cp gt CP6 CP5 lt cp gt lt bs gt lt bs gt lt pos gt 200 150 lt pos gt lt d gt lt d gt lt n gt D3 lt n gt lt m gt M3 M4 lt m gt lt s gt lt s gt lt cp gt CP9 CP 10 lt cp gt lt bs gt CP9 lt bs gt lt pos gt 50 300 lt pos gt lt d gt lt d gt lt n gt D4 lt n gt lt m gt M2 lt m gt lt s gt S4 S5 lt s gt lt cp gt CP7 lt cp gt lt bs gt lt bs gt lt pos gt 250 450 lt pos gt lt d gt theRHW
202. main parameters or attributes in the context of object oriented programming of the antenna module are frequency Pt Gt and Gr The frequency parameter contains the main radio frequency in use Pt Gt and Gr parameters are the transmitted power the transmitter gain and the receiver gain of the antenna respectively The bs module includes an instance of the antenna module Each bs module instances is characterized by the parameters name domain_name position and color The parameter name identifies a BS instance in the overall network its value should be equal to the domain name defined by domain_name parameter in which it is operating The position parameter specifies the bs module instance position and is used to calculate the distance between bs and agv module instances and the correspondent signal quality at the wireless mobile station location The color parameter permits the assignment of a colour for easy visual identification of the wireless domain The agv module also includes an instance of the antenna module The main parameters of the agv module are name path velocity and color The parameters name and color have the same purpose as in the bs module The parameter path is used to assign a path to the agv module instance using a list of points in Cartesian notation Additionally it is possible to define stop points by defining the stop duration together with path point coordinates To network handoff procedure is modelled by the nhp
203. mber between 0 and 5 and one ComFunc module instance labeled repeater x where x can be 0 1 or 2 Note that the connection Point module instance which is symbolized by a large ball also models the BS functions 60 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model HW2PNet Repeater com_func_con com_func_con Connection_Point Connection_Point repeater_con ctrl con Domain domain con domain con Master ctrl con ctrl con Slave repeater con ctrl con Controller Domain ctri_con Figure 6 1 Modules and connections of the RH W2PNetSim slave 5 cp 5 cpl2 slavel3 slave 4 Figure 6 2 Screenshot of the output window of the RHW2PNetSim 6 2 1 Controller The configuration mode used by the Controller assumes that all Master Slave Connection Point and ComFunc module instances have a unique identifier in the overall network Therefore each module has a parameter called name Figure 6 3 presents part of the NED definition of the Controller module related to the configuration of repeater based networks To define the Mobility Master MM of the network it is Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 61 necessary to assign the name of the Master module instance to the _mm parameter The mobility procedure period is defined by the _tmob parameter The parameters domain and inter do
204. message 24 end DO E SS BI end Bes ES PAS sea Sal Figure C 27 processIDMPmessage msg function pseudo code algorithm If it is in the INACTIVE state the DMM IDMP Abort Timer Tpum 1DMPabor is loaded with _dmm_ idmp abort timer parameter value and is started by the invocation of the startIDMPaborttimer function Then if its DLL is holding the token it evolves to the INQUIRY state and the processRBTmessage function is called Otherwise the DMM evolves to the WTOKEN state When another PBT message is received and if the DMM is in the INQUIRY state the processRBTmessage function is called again i e a new RBT is sent If the pmm is in the WTOKEN state then no action is taken When Tbum iDMPAbor expires the handleTimer msg function see Figure C 28 is automatically invoked and it evolves to the INACTIVE state from any other state handleTimer msg switch getAction msg case IDMP TIMEOUT dllholdtoken false state INACTIVE HOoIS Us wNH 0 Figure C 28 handleTimer msg function pseudo code algorithm When a DMM is in the INQUIRY state the relaying of the IDMP related messages is ensured by transmitting IQ REQ messages to the Bms belonging to its domain Figure C 29 The goal of this message is to allow the Bms to perform the relaying of IDMP related messages to the cmm or broadcast by cmm For that purpose the bm selects a BM which belongs to its domain builds an
205. meter is used to define if a Master or a Slave module instance models a mobile station assigned with one or not assigned with zero Figure 6 7 depicts part of the configuration file related to a Master module instance which models a wireless mobile station called M3 This station stays in domain D1 for five mobility procedures and then it changes to domain D3 where it will stay for another 10 mobility procedures This sequence of events repeats itself until the end of the simulation theRHW2PNet master 2 _name M3 theRHW2PNet master 2 _is_mobile_station 1 theRHW2PNet master 2 _ vector location 5 D1 10 D3 Figure 6 7 Configuration file related to the Master module instance of the RHW2PNetSim excerpt In order to set the location vector parameter according to the radio channel quality and the mobility of wireless mobile station the Mobility Simulator MSim has been developed This simulator models the radio wave propagation according to the Log normal Shadowing model Rappaport 1996 and the mobility of wireless mobile stations A detailed description of this simulator is found in Chapter 8 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 6 2 4 Repeater Architecture There is no module called repeater the repeater is in fact an abstraction since its operation is supported by three module instances one instance of the ComFunc module and two of Connection Point module Figure 6 8
206. mission delay calculation The sending time of a message normally corresponds to the transmission of its first bit and the arrival time of the message corresponds to the reception of the last bit Figure 4 4 A B Message transmission ta te Transmission delay Propagation delay Message arrival Figure 4 4 Message transmission In OMNeT the length of a message does not depend of its data structure composition but on the length attribute This attribute is used to compute the transmission delay when the message travels through a connection with an assigned data rate 4 4 3 An OMNeT Example Model An OMNeT model consists of the following parts NED language topology description s which describes the module structure and respective parameters gates etc These are files with ned suffix NED files can be written with any text editor or using the GNED graphical editor Simple modules are C sources files with h cc cpp suffix The simulation system provides the following components Simulation kernel which contains the code that manages the simulation and the simulation class library It is written in C compiled and put together to form a library 42 Technological Context Simulation Software User interfaces OMNeT user interfaces are used with simulation execution to facilitate debugging demonstration or batch execution of simulatio
207. mited since errors can occurs in all frames related to a transaction Therefore the Application Layer AL handles these events by assigning a timer to each message stream This timer is loaded using the deadline parameter of the Msg_Stream module and it is reloaded every time a variable related to a transaction is updated The simulations in this chapter were performed using a deadline parameter of 100 ms which is higher than the WCRT of all message streams calculated according to the formulations presented in Ferreira 2005 In this chapter we assume the same parameter setting and the same message streams as used in Chapter 9 10 4 IDP Performance using SDA Frames In Section 3 2 2 we proposed changes to the IDP protocol which can improve its performance in error prone environments In the original IDP IDF frames were transmitted using the SDN unconfirmed service One possible improvement is to use the SDA service which permits for a BM receiving an IDF to transmit a confirmation and in case of error the initiator can repeat the request The left side of Figure 10 3 shows a SDN message cycle example The initiator I sends a request frame to the responder R which receives and processes the frame after waiting Tip a new message cycle can be initiated The right side of Figure 10 3 shows a SDA message cycle example The initiator sends a request frame and waits for the acknowledge frame from the responder confirming the correct reception
208. module This module has the following parameters period duration exponent std dev and do The period parameter is used to define the periodicity of the mobility procedure and the parameter duration the respective duration The exponent std_dev and do are related to the Log normal Shadowing model used in this simulator 86 Mobility Simulator The values of path loss exponent standard deviation and close in reference distance are assigned to the parameters exponent std dev and do respectively Each simulation run has a predefined duration which is set in global simulation parameter called sim duration The MSim is composed by two components one is the front end and the other one is the simulator engine The front end component can show two graphical 3D views of the simulation It is useful to validate the simulation configuration Figure 8 3 and Figure 8 4 present a screenshot of the simulator views In the view presented in Figure 8 3 the user is able to zoom and rotate of the simulation environment The other view Figure 8 4 shows a top level view of the simulator execution This component allows checking the position and the radio frequency of bs instances the path and the stop points of the agv instances When these screenshot were taken the agv instances M4 and S6 were operating at the same frequency as bs instance BS2 at 2 3GHz and agv instance M3 was operating at the same frequency of BS1 at 2 4GHz This means that M4 and S6 belon
209. module instances composing a bridge separated by colon lt pos gt and lt pos gt indicate the position of the ComFunc module instance for graphical representation purposes The first bridge presented in Figure 7 5 lt b gt lt n gt Bl lt n gt lt m gt MS5 M8 lt m gt lt pos gt 350 150 lt pos gt lt b gt is referred to as Bl and it is composed by two Master module instances M5 and M8 which are depicted at position 350 150 This bridge interconnects two domains D1 and D2 The Controller module instance use the domain and inter domain parameters information to perform the parameterization of the module instances such as the RT of each BM LBMD of each DMM and LDMMN of a GMM just to mention some parameters It also stores in internal variables the structure of the network Using this information the network configuration can be changed when a Master or Slave module instance moves between wireless domains Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 71 7 3 2 Domain Figure 7 6 presents the Domain module NED definition The dmm_idmp abort timer parameter is used to set the duration of the DMM DMM IDMP Abort Timer Tpyacipmpabort Of this domain The number of Beacon frames that are transmitted by DMM during the IDMP and its length are defined in the n_beacon and beacon_len parameters respectively simple Domain Parameters _dmm_idmp_abort_timer numeric _n_beacon numeric _beacon_len nume
210. module is a compound module that contains all other module instances An OMNeT network has at least one instance of each module The number of module instances is specified in the HW2PNet module instance Figure 5 3 presents part of the OMNet NED definition of the Hw2PNet module This kind of compound module definition must be contained between the keywords module and endmodule It is composed by the module parameters and by its sub modules Additionally in the declaration of the compound modules elements like gates and connections can be specified Parameters are mainly used to define the module behaviours These parameters can be strings or numeric values as well as random values from different PDFs Within a compound module parameters can define the number of sub modules as well as the number of gates In this case the gates and connections are assigned dynamically at run time module HW2PNet parameters _num_domains numeric _num_masters numeric _num_slaves numeric Submodules master Master _num_masters slave Slave _num_slaves domain Domain _num_domains controller Controller endmodule Figure 5 3 Hw2PNet module NED definition Figure 5 4 depicts parts of the configuration file with the settings related to the Hw2PNet module instance which is the simulation network This is a typical PROFIBUS network composed by only one domain num domains parameter This network is composed by four masters num ma
211. more adequately the timing behaviour of the network In this set of figures it is also shown the message stream response times obtained in an error free environment the simulation results presented in Section 9 3 1 these results are identified by the tag EF Using the MeanBER probability equal to 10 or to 10 the influence of transmission errors on the MeanRT is smaller since these results are very similar to the results provided from the simulation runs considering an error free medium But the effects of errors are clearly visible on the MaxRT for all message streams As expected the simulation results in the repeater based scenario are very similar for all message streams The simulation results in the bridge based scenario show that the MeanRT values appear very close to the MinRT values and the influence of the transmission errors on the MeanRT is smaller than in the repeater based scenario Further it is noticeable that MeanRT increases much more in the repeater based scenario than in the bridge based scenario when the MeanBER probability is equal to 10 120 s fo 2 Response Time ms Oo s Response Time m 0 0 EF 10 10 10 EF 10 10 10 EF 10 10 10 EF 10 10 10 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 7 Message stream response time considering the MeanBER probability equal for all domains As it would be expect
212. mulator records information about it Figure D 16 depicts part of this file tidmp extension which contains information about the expiration of the IDMP timers concerning the GMM The first column contains the time when the timer expired and in second contains the identification of the expired timer timestamp Timer 108 211156 Phase 1 alert 109 211156 Phase 1 alert 109 222311 Phase 1 abort 109 412431 Phase 2 alert 109 611156 Phase 1 alert 109 622311 Phase 1 abort 111 211156 Phase 1 alert 111 216033 Phase 2 alert 111 413714 Phase 2 alert 113 011156 Phase 1 alert 114 211156 Phase 1 alert 114 222311 Phase 1 abort 108 211156 Phase 1 alert 109 211156 Phase 1 alert 109 222311 Phase 1 abort Figure D 16 Output IDMP alerts and aborts file excerpt To analyse this results a spreadsheet based tool has also been developed Figure D 17 depicts a screenshot of these results which contains the number of timer that the IDMP was triggered and the IDMP timers that expired Tools for Simulation Output Analysis 167 GMM_M6 IDMP TIMEOUT IDMP 59900 PHASE 1 ALERT PHASE 1 ABORT PHASE 2 ALERT PHASE 2 ABORT Figure D 17 Screenshot of spreadsheet created by the Bit Error Model IDMP Timeout option Channel State Quality In order to model burst sensitive models like the Gilbert Elliot bit error model there is the need to compute the state of the channel during time This information can also be reco
213. n PROFIBUS System turnaround time Synchronization period of at least 33 idle bit periods Transmission Delay is the propagation delay in the bus Token Holding Time Time out time Target Token Rotation time The transmission error period The shadowing term the zero mean Gaussian random variable in dB with standard deviation of Constant rate The wavelength in meters Standard deviation Mean value
214. n Output Analysis 161 convenient format Output Data Analysis Tool was developed using Microsoft Excel and Visual Basic for Applications VBA Simon 2002 Figure D 3 depicts a screenshot of this tool This tool permits the analysis of the message stream response time stations state machine evolution over time probability distribution functions and data related to bit error models Output Data Analysis Tool Message Stream Response Time State Machine Analysis Analysis Probability Distribution Functi Central Limit Theorem Analysis Bit Error Model Frames Accounting IDP Timeout IDMP Timeout Channel State Quality Figure D 3 Screenshot of the Output Data Analysis Tool D 3 1 Message Stream Response Time In order to compute the message stream response time Master and slave module instance are able to gather information about transactions This information is stored in text files which use the srt extension and contain information depicted in Figure D 4 In the first column with the header ID is the identifier of the stream The follows column contains the Destination Address DA the Source Address SA the Destination Address Extension DAE and the Source Address Extension SAE The sixth column contains the time when the stream is queued on the DLL module instance output queue The first transmission of the request frame appears in the seventh column named FTxReq the first transmission of the resp
215. n different domains the Inter Domain Protocol IDP and another to support the mobility of wireless stations between different wireless domains the Inter Domain Mobility Procedure IDMP 18 Technological Context Communication Infrastructure Figure 2 10 Bridge based hybrid wired wireless PROFIBUS network example IDP 2 4 1 Supporting Inter Domain Transactions Definitions and Concepts In PROFIBUS a message transaction involves a request by the initiator and an immediate response by the responder station In a bridge based network when a transaction involves stations in two different domains which is a consequence of the MLR that sequence of events is not possible since the request frame must be relayed by the bridge s until reaching the responder Similarly the response must be relayed by the bridge s until reaching the initiator Thus three types of transactions must be considered Intra Domain Inter Domain and Intra Inter Domain transactions An IntrA Domain Transaction IADT is a transaction that involves stations in the same domain In this case the initiator and responder stations operate according to the rules defined by the standard PROFIBUS protocol An Inter Domain Transaction IDT is a transaction which involves stations in different domains In such type of transaction the request and response frames are relayed by the bridge s and their respective BMs using a specific protocol the Inter Do
216. n environments such as OPNET Chang 1999 ns 2 Fall and Varadhan 2006 and OMNeT Varga 2004 Nowadays the use of the general purpose languages is considered not appropriate for the development of simulation models with some level of complexity However understanding how to develop a simulation model in a general purpose language helps to understand the basic concepts of the simulation On the other hand if the execution speed of the simulation is an important feature then general purpose languages can be a good choice Simulation languages provide more flexibility for the simulation developer than the general purpose programming languages The simulation developer has greater flexibility in designing and implementing the simulation model since much work has been done at the simulation language level usually as function libraries Network simulation packages can provide a more comprehensive support than simulation languages They include the basic constructs for the development of network simulations typically require less programming effort and have a smoother learning curve when compared to simulation languages Many network simulation packages include some type of pre built and reusable models of networking protocols devices and applications Additionally they also provide means for using and creating user interfaces to the simulation models facilitating the development debugging and understanding of the code Three simulation pa
217. n of the Master_PHY and Msg_Stream modules can be found Master Msg Stream 0 Msg Stream N lower gateln lower gateOut lower gateln lower gateOut bridge gateln bridge gateln upper gateOut 0 upper gateln 0 upper_gateOut N upper gateln N Master DLL bridge_gateOutq lower_gateln lower_gateOut bridge gateOut A upper gateOut upper gateln Master PHY ctrl gateOut ctrl gateln lower gateOut lower gateln y y 4 ctrl gateOut ctrl gateln domain gateOut domain gateln Figure 7 8 OMNeT Master module composition of the BHW2PNetSim module Master parameters TS numeric _station_mobile string vector_location string bm_idt_abort_timer numeric bmid mp abort timer Numeric _name string submodules dil layer Master DLL parameters TS TS _vector_location_location_vector bm idt abort timer bm idt abort timer bm idmp abort timer bm idmp abort timer gatesize upper gateOut num streams upper gateln num streams endmodule Figure 7 9 Master module NED definition of the BHW2PNetSim Master DLL As mentioned a Master DLL module models the PROFIBUS DLL and the necessary functions to support part of the IDP and IDMP functionalities Consequently in the BHW2PNetSim a Master module besides modelling a simple PROFIBUS DLL master can also operate as a BM and o
218. n oval shape represents a state and an arrow a transition For better identification within of the oval shape the state identification is written and each transition is identified by a number In the description of the Master state machine diagram it is assumed that the Time Out Time Tro and the Slot Time Tsz related timers are automatically started and stopped in the following situations The PROFIBUS protocol defines that when a master frame s last bit is transmitted or when a master frame s last bit is received a timer is loaded with the Tro value and is started hereafter referred to as Tro timer The Tro timer is stopped after receiving the first bit of the following frame The value of this timer is set according to the Eq 2 2 Whenever a request frame s last bit is transmitted by an initiator that requires either a response or an acknowledgement a timer is loaded with Ts parameter value and is started hereafter referred to as Tsz timer When a Master is in the ACTIVE IDLE state 3 transitions 1 19 and 13 are possible Transition is triggered either by the reception of a valid token frame from its Previous Station PS see Section C 1 2 for more details or when Tro expires Transition 19 occurs when a Master receives either a valid frame without bit errors not addressed to it or if the frame is valid and addressed to it but is not a token frame for example an FDL Request Status frame more details are given in Section C 1 5
219. n this dissertation where it is possible to use PDFs in some of its parameters like the message stream period and T p parameters just to mention two This annex describes the PDFs which have been implemented to model the behaviours of certain parameters of the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim and Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BHW2PNetSim A 2 Parameterization of Continuous Distributions For a given family of continuous distributions e g normal or gamma there are usually several alternative ways to define or parameterize the probability density function However if the parameters are defined correctly they can be classified on the basis of their physical or geometric interpretation as being on of three basic types location scale or shape parameters A location parameter y specifies an abscissa x axis location point of a distribution s range of values usually y is the midpoint or the lower endpoint of the distribution s range in the latter case location parameters are sometimes called shift parameters As y changes the associated distribution merely shifts left or right without otherwise changing Also if the distribution of a random variable X has a location parameter of 0 then the distribution of the random variable Y X y has location parameter of y A scale parameter B determines the scale or unit of measurement of the values in the ranges
220. nally Chapter 11 summarises the contributions of this dissertation provides conclusions and describes some lines of work that can potentially be explored as a natural sequence of the work described in this dissertation Chapter 2 Technological Context Communication Infrastructure This chapter presents an overview of the PROFIBUS protocol as also the most relevant aspects of the repeater and bridge based architectures 2 1 Introduction The repeater and bridge based architectures extend the PROFIBUS protocol to support wireless communications and mobility of the stations between wireless domains Both approaches are compatible with PROFIBUS protocol In this chapter we describe the most relevant characteristics of PROFIBUS protocol Section 2 2 The chapter continues by presenting the most relevant issues of the repeater Section 2 3 and bridge based architectures Section 2 4 The objective is to provide the reader with the necessary background and intuition for tackling the remaining chapters of this dissertation 2 2 Relevant Details on PROFIBUS 2 2 1 General Features PROFIBUS was standardised in 1996 as an European standard CENELEC 1996 It is based on the International Standards Organisation ISO Open System Interconnection OSI reference model however collapsed to just three layers Physical Layer PhL Data Link Layer DLL and Application Layer AL There is also a transversal management functionality called Fieldbus
221. nce analysis of both alternatives Section 10 5 presents a comparative performance analysis between the repeater and bridge based architectures considering transmission over error prone mediums Additionally in Section 10 6 we also propose a set of rules for MAC address distribution which tries to reduce the impact of the PROFIBUS token recovery mechanism in the bridge based approach Finally in Section 10 7 we summarize our comparative study The metrics used in these comparisons were the following response time number of transactions and percentage of concluded transactions The response time of a message streams reflect the timing behaviour of the entire network The number of transactions gives us an idea of the throughput of the network and the percentage of concluded transactions shows how the network recovers from errors In both the simulation models a transaction can miss the deadline in two situations when the response 104 Comparative Performance Analysis in an Error Prone Environment time of a transaction is higher than its deadline and when due to queuing delays the action frame the first request frame of a transaction is sent after its deadline see Section 5 2 4 for details The simulation results presented in this chapter were obtained as the aggregate result of 100 runs each using a different seed value for random value generation It is interesting to note that the results presented in this chapter are equivalent to the aggreg
222. nce of the network and at the same time reduce the number of devices required In our opinion the token recovery mechanism of the PROFIBUS is not very efficient therefore we think that if the BMs were provided with better mechanisms then the network performance can be improve particularly over error prone environments The Mobility Simulator used in this thesis assumed that wireless mobile stations always change to a new radio channel whenever the quality of the radio channel in another domain is better than the actual this can possibly result in performing more handoffs than necessary Consequently a possible enhancement to the handoff mechanism is to implement a histeresis mechanism Additionally our mobility simulator should also provide results considering the influence of obstacles in the environment and noise sources like welding machines It is also clear from our results that the IDMP has a high impact on message streams response time therefore a potential improvement would be to develop an enhanced IDMP with a better timing performance Most of the work and methodologies developed within context of this thesis can also be applied to evaluate other wireless technologies like IEEE 802 11 or ZigBee network in industrial scenario References Alves M 2003 Real Time Communications over Hybrid Wired Wireless PROFIBUS Based Networks Porto University of Porto PhD thesis Alves M T Bangemann et al 1999 General System Ar
223. nd slaves belong to its domain Therefore the LASD is a list of all masters and slaves that belong to the domain The other two components GMM and DMM are optional and their functions are related to the IDMP The DMM functionalities require two data structures the List of Bridge Masters in the Domain LBMD and the List of Wireless Mobile Stations in the Network LWMSN The LBMD is a list that contains the domain s BMs addresses and is used in the IDMP inquiry sub phase The LWMSN is list of address of all wireless mobile stations present in the network and is used in the IDMP discovery sub phase 68 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model The GMM must also be provided with two data structures the List of Bridge Masters in the Network LBMN and the List of Domain Mobility Managers in the Network LDMMN The LBMN is a list of address of all BMs present in the network and is used to control the received RSMP messages during the IDMP Phase 1 The LDMMN contains all network DMM addresses and is used to control the received RBT messages during the IDMP Phase 2 Adapted MAC promiscuous mode IDP and IDMP support BM DMM GMM r O Mandatory 1 i Optional ls as Bridge Master Bridge Master Figure 7 1 Bridge architecture Figure 7 1 also shows the Common Functionalities box which is supported by a shared memory area and is responsible for the communication between the two BMs of a bridge
224. ng the isRoute function which consults its Routing Table RT If no match is found the frame is discarded Otherwise the frame is processed Next the Bm verifies if the initiator of the transaction through the SA of the frame belongs to its domain If it does not belong it checks if it is a duplicate frame and if it is the frame is discarded Otherwise a SC frame is sent and the frame is forwarded to the other BM using the ComFunc If the frame sender belongs to the BM domain there is the need to check the type of frame If the kind of frame is a response frame it means that this BM was the last BM in the transaction path BM es an Inter Domain Frame IDF is coded using the received frame and is forwarded to ComFunc If the frame is a request frame it means that this is the first BM in the transaction path BM then it is necessary to check the service s type If it is not a SRD service then it must be a SDN service the frame is forwarded to ComFunc Otherwise if it is a SRD service it matches the received frame with all entries in the List of Open Transactions LOT Each entry in LOT can be in one of two states WAITING and FINALISED Receive Frame from Domain Procedure Receive a frame Discard the frame No ERGUIDA Yes No Match transaction in the LOT s ita duplicate frame 2 SA belongs to domain 2 No Build an IDF with the response ls ita SRD frame 2
225. ng the master with the HSA must pass the token to the master with the lowest one Each master knows the address of the PS the address of the NS and its address TS as well When Try expires or when no more messages are available on the queues low and high priority a master passes the token frame If after transmitting the token frame and after the expiration of Slot Time Tsz timer the token transmitter detects bus activity it assumes that its NS owns the token and is performing message cycles Otherwise if the token transmitter does not recognize any bus activity within the Tsz it re sends the token frame and waits another Tsz It assumes that its NS owns the token frame thereafter if it recognizes bus activity within the second Tsz If after the second retry there is no bus activity the token transmitter tries to pass the token to the next master on its LAS It continues repeating this procedure until it has found a successor from its LAS If it does not succeed the token transmitter assumes that it is the only one left in the logical ring and transmits the token frame to itself In order to detect a defective transceiver when a master is transmitting a token frame it reads back from medium all transmitted bits If it detects a difference between the transmitted and received bits it waits Tsz for any activity from NS If no activity is detected after expiring Tsz it transmits again the token frame if an error occurs then it removes itsel
226. nowledged service therefore when an initiator sends a SDN frame it will not receive a response or acknowledge from the responder Figure C 9 illustrates a SDN transaction between a Master and a Slave In a Master the Msg_Stream emulates the behaviour of an AL protocol periodically producing messages which are passed to the Master DLL The Master DLL stores the message in its output message queues high or low according to the message priority In a Slave a Msg_Stream has a similar behaviour but in this case the Slave_DLL only refreshes the content of the variables modelled by the msg stream module When a Master holds the token frame it processes messages cycles according to message dispatching procedure described in Section C 1 3 If a Master has messages in its output queues it pops a message builds a frame passes to the Master_PHY and then sends to the Domain module to which it is connected The Domain broadcasts the frame to every Master and Slave connected to it The Slave PHY receives the frame from Domain and passes to the Slave DLL The Slave DLL matches the frame information Destination Address DA SA Destination Address Extension DAE and Source Address Extension SAE with its configured message streams If it succeeds then it registers the information about this transaction for output analyses and discards the frame Otherwise the Slave DLL discards the frame 142 Simulation Models Implementation Master Slave Msg_S
227. ns There are several user interfaces written in C Simulation programs are built from the above components First the NED files are compiled into C source code using the NEDC compiler which is part of OMNeT Then all C sources are compiled and linked with the simulation kernel and a user interface to form a simulation executable The simulation executable is a standalone program which can be run in any machine When the program is started it reads from a configuration file usually omnetpp ini settings that control how the simulation is run and values for model parameters The configuration file can also specify several simulation runs in the simplest case they will be executed by the simulation program one after another or executed on a parallel environment 4 4 4 Event Based Simulation As mentioned an OMNeT model consists of hierarchically nested modules which communicate between them using messages Each message can be exchanged directly between simple modules or via a series of gates and connections The local simulation time advances when a module receives messages from another module or from itself Self messages are used by a module to schedule events at a later time In the initialization step OMNeT builds the network it creates the necessary simple and compound modules and connects them according to the NED definitions OMNeT calls the initialize functions of all simple modules which is usually used to ini
228. nsactions in their LOTs therefore the BMs reply with a RSMP message However a transmission error occurs when a RSMP message from BM M7 is transmitted by BM M10 As no retries are performed by BM M10 the GMM will not receive all RSMP messages Consequently the Tgyi praterr expires Then the GMM broadcasts again a SMP message All BMs reply with a RSMP message even the BMs that had already replied Assuming that no transmission errors occurred the GMM receives all RSMP messages from all BMs in the network and then it stops the Toum p14borr Tomm PtAlert TGMM P1Abort BM BM BM GMM BM BM BM M7 M10 M9 M6 M6 M5 M8 H TBM IDMPAbort K O A Te loMPAbort acia E CD sue a TU H ora Eesme sme Token TBM IDMPAbort K M9 gt 4 M6 ja A TBM IDMPAbort Q sme E csm ur Token Token Tok RSMP RSMP oxen M7 gt Token l me a SMP RSME bsp lt M10 gt RSMP RSMP AA M10 or Token RSMP_ 4 M5 47 M5 rswe Transmission M7 error Token rsme Al RSMP_le M8 47 Me Tomm Ptalert Pte Token le SMP SMP Token Sy RSMP RSMP M9 M6 SMP SMP Token g SMP Token SMP E RSMP RSMP_ Token S m e R Token le M8 SP mo RSMP MP M10 RSMP Token RSMP_ M5 e M5 RSMP M7 A AER Token 5 RSMP A RSMP Je
229. nsactions on Industrial Electronics Vol 46 No 6 pp 1241 1251 Tranter W H K S Shanmugan et al 2003 Principles of Communication Systems Simulation with Wireless Applications Prentice Hall Varga A 2005 OMNet User Manual version 3 1 Available online at http www omnetpp org doc manual usman html Vignaux T and K Muller 2006 SimPy Manual Available online at http simpy soucreforge net SimPyDocs manual html Weber B 2006 PROFIBUS continues to climb as the world s most popular fieldbus Press release PROFIBUS International Support Center Available online at http www profibus com Willig A 1999 Analysis of the PROFIBUS Token Passing Protocol over Error Prone Links In Proceedings of the 25 Annual Conference of the IEEE Industrial electronics Society IECON 99 San Jose USA pp 1246 1252 Willig A Wolisz A 2001 Ring Stability of the PROFIBUS Token Passing Protocol over Error Prone Links In IEEE Transactions on Industrial Electronics Vol 48 No5 pp 1025 1033 Annex A Probability Distribution Functions The simulators developed within the aim of this dissertation permit the setting of some its parameters using probabilistic distribution functions PDFs This annex describes the characteristics of the PDFs which can be used A 1 Introduction Most simulation studies have variables which exhibit a stochastic behaviour This is also the case of the simulators developed withi
230. nsequently they take part on their domain logical ring This chapter presents the main architectural components of the Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator BHW2PNetSim in Section 7 2 In Section 7 3 some relevant details of the implementation of the BHW2PNetSim are described 7 2 Bridge Architecture Figure 7 1 illustrates the main building blocks of a two port bridge In order to support the required functions there must be a set of mechanisms related to the Inter Domain Protocol IDP and to the Inter Domain Mobility Procedure IDMP These mechanisms operate at DLL level and consequently the existing PROFIBUS DLL must be adapted The operation of the IDP and IDMP are managed by three components Bridge Master BM Global Mobility Manager GMM and Domain Mobility Manager DMM The BM component which gives the device its name and is mandatory contains the routing and the IDF handling functions which are crucial to the IDP and to the IDMP These two functions are associated with three data structures the Routing Table RT the List of Open Transactions LOT and the List of Active Stations in Domain LASD Frames are relayed by a BM according to the information contained in its RT The LOT is used to store information about on going IDT in which the BM assumes the role of BM see Chapter 2 for details For the BM which operates as BM es i e the last BM in the transaction path it has to know which masters a
231. nter 0 Build frame Send frame Ts expired Receive a S No SL BP No fetry_counter as bit errors Yes 2 No Discard frame state USE TOKEN Figure C 18 Send IDF Procedure max retry limit The first step is to evolve the Master state machine from the USE TOKEN to the AWAIT DATA RESPONSE state and set the retry counter variable to zero After that the Master pops a message from its message output queue builds a frame and transmits it After it waits for the reception of a SC frame If an invalid SC frame is received with bit errors then it is discarded and the retry counter variable is increased The retry counter variable is also increased if no frame is received within the Tsz When the retry counter variable reaches the max retry limit a Master parameter limit then the master state machine evolves to the USE TOKEN state C 3 2 Inter Domain Mobility Procedure Implementation GMM Operation In this Section we explain the operation of the cmm with the error handling mechanism described in Chapter 3 Figure C 19 presents the procedure that corresponds to Phase 1 of the IDMP We assume that the cmm is configured prior to runtime with the List of Bridge Masters in the Network LBMN and the List of Domain Mobility Managers in the Network LDMMN which are the list of all Bms and a list of all bums present in the network respectively See Section 7 4 2 for more details In order
232. ntroller domain lt d gt lt n gt D1 lt n gt lt m gt M8 M3 lt m gt lt s gt lt s gt lt dmm gt M8 lt dmm gt lt pos gt 400 300 lt pos gt lt d gt lt d gt lt n gt D2 lt n gt lt m gt M6 M1 M5 lt m gt lt s gt S1 S2 S3 lt s gt lt dmm gt M6 lt dmm gt lt pos gt 200 150 lt pos gt lt d gt lt d gt lt n gt D3 lt n gt lt m gt M9 M10 M4 lt m gt lt s gt S6 lt s gt lt dmm gt M9 lt dmm gt lt pos gt 50 300 lt pos gt lt d gt lt d gt lt n gt D4 lt n gt lt m gt M7 M2 lt m gt lt s gt S4 S5 lt s gt lt dmm gt M7 lt dmm gt lt pos gt 250 450 lt pos gt lt d gt theBHW2PNet controller _inter_domain lt b gt lt n gt B1 lt n gt lt m gt M8 M5 lt m gt lt pos gt 350 150 lt pos gt lt b gt lt b gt lt n gt B2 lt n gt lt m gt M6 M9 lt m gt lt pos gt 50 150 lt pos gt lt b gt lt b gt lt n gt B3 lt n gt lt m gt M10 M7 lt m gt lt pos gt 120 400 lt pos gt lt b gt Figure 7 5 Configuration file related to the controller module instance of the BHW2PNetSim excerpt The meaning of the tags used in the domain parameter was already defined in Chapter 5 The parameter inter domain defines the configuration of a bridge The meaning of the tags is similar to that described in Chapter 6 for the repeaters Tags lt b gt and lt b gt define a bridge lt n gt and lt n gt are used to set the name of the ComFunc module instance between tags lt m gt and lt m gt enclose the names of the Master
233. nvolves several steps Usually a simulation study is not a simple sequential process but often there is the need to go back to a previous step In Law and Kelton 2000 the authors divide the simulation study in ten steps However in Banks Nelson et al 2001 the development of a simulation study evolves through 12 steps Figure 4 1 shows the steps that compose a typical simulation study based on the methodology proposed by Law and Kelton 2000 The first step involves defining the goals of the study and determining what needs to be solved The problem is further defined through objective observations of the process to be studied Care should be taken to determine if simulation is the appropriate tool for the problem under investigation If simulation is the appropriate tool for the problem then the simulation study evolves to a second step The goal of this step is to collect data about the system under study and delineates the conceptual simulation model Information must be collect about system layout operating procedures 38 Technological Context Simulation Software model parameters input probability functions and performance measures to be analyzed These tasks must be carefully done because sometimes the information is inaccurate and the operating procedures are not formalized After collecting information of the system under study the conceptual model can be developed It is not necessary to do one to one correspondence between each
234. o the _gmm parameter The parameter _tmob is used to define the period of the IDMP The duration of the GmMmM_Phase_1 Alert Timer Tommrlaler GMM Phase 1 Abort Timer TGMM Pl4bort CMM Phase 2 Alert Timer TGya p2Alert s Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 69 GMM Phase 2 Abort Timer TGum p24bor described in Chapter 3 are set by gmm phasel_ alert timer gmm phasel abort timer gmm phase2 alert timer and gmm phase2 abort timer parameters respectively These parameters are only configured on the Master module instance that acts as GMM HW2PNet A AA a RAR RA domain_con ctrl_con Domain domain_con ctrl_con Slave Figure 7 2 Modules and connections of the BHW2PNetSim domain_con Slave ctrl con E theBHw2PNet masteftg ri ster 2 domain 0 slavel3 slavel4 Figure 7 3 Screenshot of the output window of the BHW2PNetSim In order to gather the output data files information about aborted IDT and IDMP the Controller module is provided with two more parameters output timeout idt and _output_timeout_idmp apart from the parameters described in Chapter 5 A detailed description of the output data files generated by both simulators the RHW2PNetSim and BHW2PNetSim is presented in Annex D as well as some tools used to extract information from them 70 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model
235. of several smaller components In ns 2 models are flat OMNeT has a powerful interactive graphical environment where it is possible to examine nearly everything during execution ns 2 only includes Network AniMator NAM which is a playback tool For the reasons presented above OMNeT has been chosen 4 4 OMNeT Objective Modular Network Testbed in C OMNeT Varga 2004 is an object oriented modular discrete event simulator which provides a reusable component framework where the system components can be independently built characterized and assembled into larger components and models The basic system components are built using the C programming language and then assembled into larger components and models using a high level language named NED an OMNeT specific scripting language An OMNeT model consists of hierarchically nested modules which communicate between them using messages OMNeT models are often referred to as networks The top level module is the system module The system module contains sub modules which can also contain sub modules themselves The depth of module nesting is not limited consequently providing a useful way to reflect the logical structure of the system in the model structure Figure 4 2 System module Compound module Simple module Simple module Simple module Figure 4 2 Simple and compound modules Modules that contain sub modules are termed c
236. of the distribution The standard deviation o is a scale parameter for the normal distribution A change in B compresses or expands the associated distribution without altering its basic form Also if the distribution of a random variable X has a scale parameter of 1 then the distribution of the random variable Y BX has a scale parameter of B A shape parameter a determines distinct from location and scale the basic form or shape of a distribution within the general family of distributions of interest A change in a generally alters a distribution s properties e g skewness more fundamentally than a change in location or scale Some distributions e g exponential and normal do not have shape parameter while others e g beta may have two The Table A 1 gives the information relevant to the PDFs implemented in both simulators The range indicates the interval where the associated random variable can take on values Also listed are the mean expected value variance and mode i e the value at which the density function is maximized 126 Probability Distributions Functions Table A 1 Probability Distribution Functions features 1 Normal N u o Errors of various types e g in the impact point of a bomb quantities that are the sum of a large number of the other quantities by virtue of central limit theorems 1 y f x Raro so A Possible applications 2 e 20 for all real numbers X 0 57
237. of the RHW2PNetSim excerpt 62 Figure 6 7 Configuration file related to the Master module instance of the RHW2PNetSim excerpt 62 Figure 6 8 Repeater s module instances and their connections e eeeeeeereeaeaeeeeraeeeererarerararatanaa Figure 6 9 ComFunc module NED deftones riiai oeii ira nri e i EERE VE E NEEE NEER EE EES EEEE EAEN Figure 6 10 Configuration file related to one ComFunc module instance excerpt Figure 6 11 Connection Point module connections seine a a R E RRR Figure 6 12 Connection Point module NED definition ss sssssesssssssrersssessssrrtstsrsstsrsrtstertsrsttrteterenrsresterestsrertntarasrsrentnteterre ret Figure 6 13 Configuration file related to the Connection_Point module instance excerpt Figure 6 14 Wired wireless interconnection example Figure 6 15 Simplified timing behaviour of the module instances that model a repeater Figufe 7 1 Bridge arChitecture sis neern aroei eo aaoi EA u E r A PESEE pa doce Ere r TEA TERA TET SEER SER R apa sitas Figure 7 2 Modules and connections of the BAW2PNetSim essesssessesssssssisrrtetsrssrsrtstsrsstsrrtsterertnrertstereresrrintaresrsrentnterente ret Figure 7 3 Screenshot of the output window of the BHW2PNetSim Figure 7 4 Controller module NED definition of the BHW2PNetSim
238. of the response frame transmission event ez at the instant t Event e corresponds to the reception of the response frame s last bit by master M1 Technological Context Simulation Software 43 In order to model the system described above two simple modules can be used one to model the master station hereafter called MasterStation module and another to model the slave station hereafter called slavestation module Figure 4 6 shows part of the handleMessage cMessage msg function or method of the MasterStation module This function is automatically invoked at reception of every message Therefore the arriving of the token frame to the MasterStation module instance called M1 is handled by this function This message is a PROFIBUS message where the DA and SA are equal to TS and PS line 20 respectively M1 computes the Try line 22 and in order to wait Tip before sending a frame it schedules a self message to arrive at instant T p counted from the current instant the current simulation time is given by simTime function line 24 When M1 receives a self message line 6 that marks the end of the inactivity time Tp it removes a message from its message output queue line 10 and starts transmitting the request frame addressed to the SlaveStation module instance called S1 line 11 AL void MasterStation handleMessage cMessage msg 2 1 he cMSG Profibus msg profibus NULL 4 cMSG Self msg self NULL 5 handle the me
239. ompound modules in opposite to simple modules which are at the lowest level of the module hierarchy The simple modules of a model contain the algorithms coded using C programming language The full flexibility and power of the C programming language can be used in conjunction with the OMNeT simulation class library Further OMNeT has a consistent object oriented design Thus Object Oriented Programming concepts like inheritance and polymorphism can be used to extend the basic functionality of the simulator 4 4 1 Messages Gates and Links Modules communicate by exchanging messages which represent frames or packets in a computer network These messages can contain arbitrarily complex data structures Simple modules send messages through gates or directly based on their unique identifier Messages can arrive from another module or from the same module self messages are used to implement timers Gates are the input and output interfaces of modules Messages are sent out through output gates and arrive through input gates Each connection also called a link is created within a single level of the module hierarchy and is composed by two gates Within a compound module one can connect the corresponding gates of two sub modules or a gate of one sub module and a gate of the compound module Figure 4 3 Technological Context Simulation Software 41 Due to the hierarchical structure of the model messages typically travel through a seri
240. on joining and removal leaving of stations This is accomplished by examining at most one Gap address per token visit using an FDL Request Status frame after the execution of all high priority transactions and if the value of Typ is still positive Error Handling Additionally in order to enhance the communication system s reliability PROFIBUS handles some operational or error states concerning logical ring management In Carvalho Carvalho et al 2005 and Willig and Wolisz 2001 is presented which fault tolerant mechanisms are activated and their 10 Technological Context Communication Infrastructure effects on the network behaviour The most important error situations within the context of this document are the token lost heardback removal and error skipping Token lost This abnormal situation is clearly recovered by means of a continuous monitoring activity performed by each master in the logical ring If a period of inactivity longer than the Tro is detected then the token is claimed by the master with the lowest address in the logical ring and the logical ring is re initialised Heardback removal Whenever a master is sending a token frame it must hear from the medium all transmitted bits in order to detect a defective transceiver If the token frame sender detects differences between the transmitted and the received token frame in two consecutive transmissions then it must remove itself from the logical
241. on contained in its RT and List of Active Stations in Domain LASD BM M8 RT which operates as BM for this transaction opens a new entry in its LOT related to this transaction After coding an IDF BM M8 transmits the frame to BM M5 When BM MS has the right to access the medium it transmits the IDF but assuming that a transmission error occurs in domain D the frame is discarded by all stations belonging to domain D Since IDF frames are coded using standard non acknowledge PROFIBUS frames there is no retransmission by BM M5 and consequently the frame is lost and the IDT can only be recovered when the BM IDT Abort Timer expires Codes the frame using IDP and open a transaction in LOT Initiator Responder M3 s6 Bridge 1 Bridge 2 M8 M5 M6 M9 ETE AL DP DLL DLL AL DP Token Service_upd req Service req 4 PROFIBUS pes Token Transmission error Service con MT DF No_Data Token vf Token Service req a PROFIBUS request Token Service con No_Data Token T Service req 4 PROFIBUS request Service con 3 3 No_Data D D O Open transaction i Transmission error Figure 3 1 Example timeline for an IDT between M3 and S6 considering transmission errors 3 2 2 Improving Error Handling in the IDP The main reason for the original protocol was mainly related to provide simple functionalities capable of being implemented in resource constrained devices since the receiving BM wo
242. onally the Master DLL also implements the IDP and IDMP functions namely the ones related to the BM the DMM and the GMM as well as the necessary extension to the PROFIBUS DLL These modules are hierarchically organized as illustrated in Figure 7 8 Figure 7 9 presents part of a Master module NED definition the omitted parameters were defined in Chapter 5 Concerning the parameters presented some were described in Chapter 5 like Ts name and num streams parameters while is mobile station and location vector parameters were detailed in Section 6 2 3 of the Chapter 6 When a Master module instance is a BM there is the need to assign two timers One related to the entries in its LOT bm iat abort timer parameter and the other one related to the duration of the IDMP bm idmp abort timer parameter See Section 3 2 1 and Section 3 3 3 for details Figure 7 10 shows part of the configuration file related to one master module instance named M10 If this Master module instance acts as a BM all IDTs opened in its LOT must be finalized within 20 ms after being created bm idt abort timer parameter and the IDMP must end 40 ms after a Start Mobility Procedure SMP message has been received defined by the bm idmp abort timer parameter 72 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model In the following section a description of the Master DLL module and its interactions are presented In Chapter 5 a detailed descriptio
243. onse frame appear in the eighth column named FTxResp and the last reception 1 e when the transaction is finished appears in the ninth column named LRecep The last column displays path related information the first item is the initiator s domain when message is queued the second item is the domain name to which the initiator belongs when the first request is transmitted on the third item appears the domain name to which the responder belongs when it replies on the fourth item contains the domain name to which the station belongs when the transaction finishes ID DA SA DAE SAE Queued FTxReq FTxResp LRecep Domains 7 46 3 7 7 0 000000 0 000560 0 001120 0 005619 D1 D1 D3 D1 7 46 3 7 7 0 010000 0 010588 0 011265 0 015621 D1 D1 D3 D1 7 46 3 7 Z 0 020000 0 020505 0 021207 0 025593 D1 D1 D3 D1 7 46 3 7 7 0 030000 0 030675 0 032323 0 035648 D1 D1 D3 D1 7 46 4 7 A 0 040000 0 040580 0 042337 0 045570 D1 D1 D3 D1 7 46 3 Y 7 0 050000 0 050675 0 051481 0 055802 D1 D1 D3 D1 7 46 3 7 7 0 060000 0 060522 0 061403 0 065519 D1 D1 D3 D1 Figure D 4 Output response time file excerpt 162 Tools for Simulation Output Analysis Note that if the transaction is a SDR the transaction then it only finishes when the initiator receives the response frame But if it is a SDN then the transaction finishes when the responder receives the request frame In the last case no contain any information The number of transaction that missed i
244. onsidering two version of the IDP Originally the IDP defined that IDFs were transmitted using the unacknowledged SDN service Therefore any IDF transmission error leads to an error on an IDT which can only be solved using the LOT timer associated with that IDT In this dissertation we proposed that an IDF should be transmitted using the acknowledge SDA service Our simulation results show that it is possible to achieve a higher performance using the SDA service Globally there is a reduction on the MeanRT and an increase on the number of transactions which do not loose its deadline In the second comparative analysis we compared the performance of the repeater based approach and the bridge based approach considering communications over error prone mediums Two transmission error configurations were used In the first we performed simulations in which the MeanBER probability was set equal in all domains either wired or wireless In the second configuration we used a fixed MeanBER probability for the wired domains and varied the MeanBER probabilities for wireless domains From the simulation results provided by the first set of simulation runs we conclude that the repeater based approach is much more influenced by transmission errors than the bridge based Comparative Performance Analysis in an Error Prone Environment 115 approach since the MeanRT increases considerably with an increase in the MeanBER probability Contrarily the MeanRT of the bridg
245. or Mario Serafim dos Santos Nunes Vogais Doutor Manuel Alberto Pereira Ricardo Doutor Lu s Miguel Moreira Lino Ferreira Doutor Carlos Manuel Ribeiro Almeida Junho de 2007 UNIVERSIDADE TECNICA DE LISBOA INSTITUTO SUPERIOR INSTITUTO SUPERIOR TECNICO TECNICO Performance Analysis of Wireless enabled PROFIBUS Networks Paulo Manuel Baltarejo de Sousa Licenciado Dissertac o para obtenc o do Grau de Mestre em Engenharia Electrot cnica e de Computadores Orientador Doutor Lu s Miguel Moreira Lino Ferreira Co Orientador Doutor Carlos Manuel Ribeiro Almeida J ri Presidente Doutor Mario Serafim dos Santos Nunes Vogais Doutor Manuel Alberto Pereira Ricardo Doutor Lu s Miguel Moreira Lino Ferreira Doutor Carlos Manuel Ribeiro Almeida Junho de 2007 Table of Contents CHAPTER 1 SAA O OO 1 LLO UC e s5e65 2 35 e este isctsegs e e Pegtve sph E eee vin besos sone ced ima Sena et ins 1 1 2 Research Context sarado A sio 2 13 Research ODjECtH Ves 0 ra a E E R eats a eee ig ee ne eee 4 1 4 Contributions of this Dissertationen a a S 4 1 5 structure ofthe Dissertation spears rice ee chew ite da iowa acc ad ed me 5 CHAPTER 2 TECHNOLOGICAL CONTEXT COMMUNICATION INFRASTRUCTURE 7 Dl Introduction st rsss E E lense eh setts levees lac senate EE EA 7 2 2 Relevant Details on PROFIBUS 0 ccceeccecseessesseesceseecsecseecseceaecsecesesecsseseecuseseeeseseeeeseeeaeeaeesaes 7 2 21 General
246. or station since that type of frame does not have a DA field Therefore the first problem can be solved by using a Transaction Identifier TI which enables matching the request and the respective response while to solve the second problem it is required that every IDF must have a destination address field The TI is a sequence number assigned by the BM which must also be included in the response frame similar to a TCP IP sequence number This field is used by the BM to distinguish between response frames related to different pending transactions The IDF is a new frame that embeds the original frame Therefore to reconstruct the original frame one of the fields that must be stored in the IDF frame is the original frame function code which is stored as Embedded Function Code EFC and an identifier which enables BM and BM es to identify the type of the embedded frame the Embedded Frame Type EFT Considering the three types of data frames defined in PROFIBUS the IDP converts Frames of Fixed Length with no Data Field to Frames of Fixed Length with Data Field and both Frames of Fixed Length with Data Field and Frames with Variable Length Data Field to Frames with Variable Length Data Field Table 2 1 illustrates the proposed mapping between standard PROFIBUS frames and the IDFs In the table a rectangle with a dash means that the field is not used in the IDF because it is not present in the original frame A rectangle with diagona
247. or tie a AA E RBT a Token MB PBT RBT Inquiry __ Inquiry lt _ _PBTI 754 M10 Ms O ae Lo RBT RBT i aaa e rd GD ReT_1 m8 fo Sl Void MB void Toum p2abort H y 3 va p D y 4 D y y y D Y M O Start timer s O Expire timer a Stop timer s E Transmission error Figure 3 10 Simplified timeline of the Timer Based mechanism for Phase 2 After collecting all RBT messages from the DMMs the GMM broadcasts a SBT message and its role in the IDMP ends For the wireless mobile stations to assess the quality of the radio channel in all domains Phase 3 must occur almost simultaneously in all wireless domains If a transmission error occurs at the transmission of the SBT message some DMMs will not transmit Beacon messages If a DMM of a wireless domain does not transmit Beacon messages the wireless mobile stations present in its wireless domain are not able to perform radio channel assessment and consequently they stay in the same domain On the other hand other wireless mobile stations are not able to evaluate the radio quality of the domain where an error occurred The IDMP ends when the Tpmm IDMP4bort expires But at the expiration of the Tpum 1DmPAbor the DMMs send RU messages related to the wireless mobile Error Handling Improvements for the Bridge based Architecture 33 stations present in its domain In this way the BMs which have cleared all entries related to the wireless mobile stations from th
248. other S6 defined by the agv x name parameter Their velocity is set to 8 m s and 6 m s for M3 and S6 respectively The path and the stop points are defined by the agv x path parameter This parameter is a string that is written using a pre defined structure as follows Each point of the path is defined using the coordinates x y and z The path is a set of points separated by colons When a point is also a stop point the stop time is defined after character For example M3 starts at point 90 10 0 it follows in the direction of point 80 20 0 then goes to point 70 30 0 As soon as it arrives to the last point 1t follows to the next point in the path 40 30 0 where it stops for 0 125 seconds After that 1t follows to the next point 40 30 0 and it again stops for 0 125 seconds The path definition is cyclical therefore when it reaches the last point it continues to the first The parameter freg of the agv antenna module instance changes during the simulation run its value depends on the domain to which the agv module belongs to sim agv_num 3 agv 0 anme M3 agv 0 vel 8 agv 0 path 90 10 0 80 20 0 70 30 0 40 30 0 0 125 40 30 0 0 125 70 30 0 80 20 0 90 10 0 90 10 0 80 20 0 70 30 0 40 30 0 0 125 40 30 0 0 125 70 30 0 80 20 0 90 10 0 agv 0 color 0 6 0 5 0 1 1 0 agv 0 ant freq 2 4 agv 0 ant pt 1 agv 0 ant gt 1 agv 0 ant gr 1 agv 1 name M4 agv 1 vel 6 agv 1 path
249. output color red output color green and output color blue The length in bits the period and offset of the first activation of the message streams can be assigned either using random or constant values 54 PROFIBUS Simulation Model Initiator Responder Initiator Responder AL DP DLL DLL AL DP AL DP DLL DLL AL DP Token Transmission error Token Transmission error Service req PROFIBUS Service req PROFIBUS Start transaction gt i reg lest Service_upd req adele A pei aes request Servica_upd 1eg A E ly 4 e Service con PROFIBUS Service con PROFIBUS T di Deadline No_Data request retry No Data request tretry A ransmission error ty Transmission error Deadline Token vt Service req N Service req L Transmission error Service upd req PROFIBUS Service_upd req lt _ _ewuesgt Service con Token e Service con No Data No Data PROFIBUS Transmission error AA Socera PROFIBUS Y Service req at E gt Response time IX request Token Service_upd req ea es e PROFIBUS Service upd req Finish transaction L Y je tangaction request e Service con PROFIBUS Y Data response D x v D v Figure 5 16 Deadline missing examples Figure 5 17 depicts an example of an Msg_Stream module instance configuration This message stream involves a Mas
250. owledge SDA service which allows an initiator to send a message and immediately receive the confirmation The Send Data with No Acknowledge SDN is an unacknowledged service The Send and Request Data with Reply SRD is based on a reciprocal connection between an initiator and a responder and requires either an acknowledgement or a response The Send and Request Data with Reply CSRD is a cyclic service based on the acyclic SRD In this simulator only the SDN and SRD services were implemented The SDA service was implemented based on SDR service just by adequate the setting of the request and response frame sizes The Msg_Stream module emulates the behaviour of an AL protocol Periodically it produces messages which are passed to the Master_DLL module in case of the RHW2PNetSim or DLL module in case of the BHW2PNetSim which stores the message in its output message queues as a function of the message priority When a Master has the right to access the medium i e when it holds the token frame it pops messages from its output message queues and it checks which service will be used If it is a SDN then after transmitting the message it can schedule a new action according to the message dispatching procedure Otherwise it has to wait for the response or the T s expiration Figure C 8 Send Frame Procedure No e Yes Send SDN Procedure Send SRD Procedure Figure C 8 Send Frame Procedure Send SDN Procedure A SDN is an unack
251. owledgement to a request 5 a Fixed length frame w no data field b Short acknowledgement frame sD3 DA sa FC Data 8 Bytes Fos E c Fixed length frame w data field sD2 LE Ler sD2 Data max 246 Bytes Fos ED d Variable data field length frame sD1 DA sa FC Fes SD4 DA SA e Token frame Figure 2 1 PROFIBUS DLL frame formats Data Link Layer Services PROFIBUS defines 4 types of data transfer services Send Data with Acknowledge SDA Send Data with No acknowledge SDN Send and Request Data SRD and Cyclic Send and Request Data CSRD The SDA service allows a user to transmit data to another station and receive a Short Acknowledge confirming its reception by the responder station The SDN service permits to transfer data to a single station to a group of stations multicast or to all stations broadcast The SRD service Technological Context Communication Infrastructure 11 allows the transmission of a message to another station and the retrieval of a response This service can be used for example to send the output settings for an I O device and retrieve the state of the device s input ports The CSRD builds upon the SRD service adding the capability of transferring data periodically according to the user requirements The CSRD service is usually not implemented in current commercial hardware platforms Timing Parameters The PROFIBUS standar
252. p lt 0 and GAP_Turn false 2 No Is the low priority message queue empty Pass Token Procedure Yes Send Frame Procedure Figure C 4 Message dispatching procedure Independently of the Try value a Master is allowed to transmit at least one high priority message After high priority message will be processed while 7 gt 0 When there are no more high priority messages to dispatch then low priority messages and GAP update related messages can be transmitted 138 Simulation Models Implementation The GAP update procedure is triggered when the Gap Update Timer Teup expires and the Try gt 0 If Try lt 0 the GAP update procedure is postponed for the next token holding period However only one GAP update procedure is performed per token visit It should be pointed out that once a high or low priority message cycle or GAP update procedure is started it is always completed it is not pre empted including any retry or retries even if Try lt 0 When Try lt 0 or when the output message queues are empties the Master passes the token frame to another station In this section a high level message dispatching mechanism was described The pass token send frame and GAP Update procedures are detailed in the following sections C 1 4 Pass Token Procedure The PROFIBUS protocol defines that token frames are passed between masters in ascending MAC address order The only exception is that to close the logical ri
253. parameter value is set to 1 it means that it models a SDN service In the case of being 3 it is a SRD service The PROFIBUS SDA service can be modelled by using a SRD with a one byte response frame PROFIBUS Simulation Model 53 simple Master_DLL parameters TS numeric _pdf_tid1_type numeric _pdf_tid1_par1 numeric pdf_tid2_type numeric _pdf_tid2_par1 numeric _pdf_tsdr_type numeric _pdf_tsdr_par1 numeric gates in upper_gateln lower_gateln bridge_gateln out upper_gateOut lower_gateOut bridge_gateOut endsimple Figure 5 14 OMNeT master DLL module NED definition The message generation can be active or inactive If the value of the active parameter is 0 then no messages are generated otherwise messages are periodically generated simple Msg_Stream parameters DA numeric DAE numeric SA numeric SAE numeric DATA_UNIT numeric Serv_class numeric service numeric _active numeric _deadline numeric _output_color_red numeric _output_color_green numeric _output_color_blue numeric _pdf_length_type _pdf_period_type numeric pdf of set type numeric gates in lower_gateln out lower_gateOut endsimple Figure 5 15 Msg Stream NED definition Typically a transaction or message cycle consists of the request or a send request frame from the initiator and the associated response frame from the responder especially for SRD The respon
254. pp 1977 1997 Fall K and K Varadhan 2006 The ns Manual Available online at http www isi edu nsnam ns doc ns_doc pdf Ferreira L 2005 4 Multiple Logical Ring Approach to Real time Wireless enabled PROFIBUS Networks Porto Portugal University of Porto PhD Thesis Ferreira L M Alves et al 2002 Hybrid Wired Wireless PROFIBUS Networks Supported by Bridges Routers In proceedings of 4th IEEE International Workshop on Factory Communication Systems Vasteras Sweden pp 193 202 Ferreira L M Alves et al 2003 PROFIBUS Protocol Extensions for Enabling Inter Cell Mobility in Bridge Based Hybrid Wired Wireless Networks In proceedings of 5th IFAC International Conference on Fieldbus Systems and their Applications Aveiro Portugal pp 283 290 122 References Ferreira L E Tovar et al 2003 Enabling Inter Domain Transactions in Bridge Based Hybrid Wired Wireless PROFIBUS Networks In proceedings of 9th IEEE International Workshop on Emerging Technologies and Factory Automation Lisboa Portugal pp 15 22 Gilbert E 1960 Capacity of a burst noise channel Bell Systems Tech Journal vol 39 pp 1253 1266 Hazim S 2006 Inaccessibility in PROFIBUS Due To Transient Faults Baghdad Electronics amp Communications Dept College of Engineering University of Baghdad Technical Report 3 2 IEC 2000 IEC61158 Fieldbus Standard for use in Industrial Systems European Norm Law A M and W D K
255. provided by the MLR Comparative Performance Analysis in an Error Free Environment 99 approach the lower overhead caused by the IDTs smaller settings of Tp parameter and additionally the message stream period is much lower 8 ms and 40 ms for the bridge and repeater based scenarios respectively In the following subsections we will analyse the network timing behaviour when certain network parameters are varied 9 3 2 Variability of the Message Stream Response Time as a Function of the Bit Rate This subsection analyses how the setting of different bit rates in some network domains affects the timing behaviour of the two approaches For this purpose the results presented were obtained by varying the bit rate in domain D Figure 9 7 compares the MinRT MeanRT and MaxRT values of the two scenarios for messages streams S and S assuming the base configuration described in Section 9 2 and by varying the bit rate in domain D from 0 5 Mbit s to 5 Mbit s In these conditions parameters Tsz Tip and Typ must be recalculated for every bit rate in the repeater based approach and these changes are applied to all domains In the bridge based scenario the parameter changes only affect domain D In Figure 9 7 and in the following figures of this section the MinRT MeanRT and MaxRT values are identified by a dash The MinRT and MaxRT values are placed on the lower and upper extremes of the line and the MeanRT is placed between MinRT and Ma
256. r M2 finishes the retry BM M7 already has received the response frame from the responder One of the factors that cause this situation is related to the difference between the bit rates of domain D 0 5 Mbits s to which master M2 belongs and domain D 2 Mbits s to which slave S6 belongs This tool was also of paramount importance for debugging and validating of the both simulation models since it provides a temporal overview of the network events Further it is possible to check the characteristics of all events by a double click on the event object Figure D 2 shows a screenshot of this 160 Tools for Simulation Output Analysis feature using output data files generated by RHW2PNetSim concerning the network scenario presented in Figure D 1 In this figure a message box shows the information related to the indicated event a S5 sa M2 N f NN A ENS M7 N N NN N N oe ESO A ss E E BM_M10 MO MI 81 TI M1 MI Eh fll MI AI ol Mi MI MI TMI M4 1 81 lt 81 1 41 fll AMAI Al 81 81 MI mt s6 mm E M9 1 4 sl q 1 4 sa 4 A 4 q 4 1 1 si 4 BM_M9 BM_M6 A M6 8 9 49 fl A fl A A A 01 AA A AIAN s3 s2 si a EV ee fli SIA AL ISA S IA M5 T NNA A A 39 9 9 9 AN AA aa BM_M5 BM_M8 0 Me E mi Mi Mi 81 GI NL MI MI NL MI MI MI ML MI MI MI MI MI MI 4 M3 Hl TI MI 01 01 1 ft 1 1 81 fl 1 1 81 01 fl 1 AO fl MODS 0 014800 us 0 022000 us MM Send Token Frame 4 Receive Token TID
257. r all message streams in the repeater based scenario than in the bridge based scenario Figure 10 19 shows the percentage of transactions that do not miss its deadline for repeater based scenario from the simulation presented in Section 10 5 2 which are identified by a R and the percentage of transaction that do not miss its deadline for the bridge based scenario using the MSASR schema proposed in this section MeanBER probability of the wireless domain 10 MeanBER probability of the wireless domain 10 100 O R O R a Busasr a Busasa Concluded transactions Concluded transactions o q S M1 S M2 sm S 3 S Mt S M2 SM SI Message Stream Message Stream Figure 10 19 Percentage of transactions that do not miss its deadline using different MeanBER probability in wired and wireless domains From these results we can conclude that using the MSASR the percentage of transactions that do not miss its deadlines increases significantly when compared with the results of the repeater based scenario it is shown that for some message streams the bridge based scenario presents a higher percentage of transactions that do not miss its deadline 10 7 Summary In this chapter we presented set of simulation results where transmission errors were modelled according to the Gilbert Elliot Channel Model These results were used to make three comparative performance analyses First we made a comparative analysis of the bridge based approach c
258. r as a DMM and or as a GMM For that reason the Master DLL module is a compound module composed by 4 simple modules DLL BM DMM and GMM as shown in Figure 7 11 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 73 The DLL module models the PROFIBUS DLL as well as the required adaptations in order to support the IDP and the IDMP BM DMM and GMM modules model IDP and IDMP agents theBHW2PNet master 9 TS 10 theBHW2PNet master 9 _name M10 theBHW2PNet master 9 idt error timer 20ms theBHW2PNet master 9 idmp error timer 40ms Figure 7 10 Part of configuration file related to Master module instance The DLL module Figure 7 8 and Figure 7 11 is directly connected to every module that composes a Master DLL module and also to the master PHY and the msg Stream modules through the gates of the Master DLL It is connected to the Master PHY module instance through lower gateIn and lower gateOut gates and it is connected to N msg Stream module instances through gates upper gateIn x and upper gateOut x Similarly The Bm module is connected to ComFunc module instance through gates bridge gateIn and bridge _gate0ut upper_gateln upper gateOut Master DLL bridge gateln bridge gateln gt
259. ram for the message stream 3 ccccsssssssssssseessssessecsssecssscsssecssecesseessuceasecssucsasecssseeasecssseeasecs Figure 9 6 Number of message stream transactions ii nooo on nan nn onnn non on nana an ona nen an ona nan on anna anan nan en ancianos Figure 9 7 Influence of D4 bit rate on the message stream response time values eee 99 Figure 9 8 Influence of the IS delay on MaxRT Figure 9 9 Influence of the maximum frame size on response time A Figure 10 1 State machine for the Gilbert Elliot error model eee eereeeeeeereeareereraeerereeanaa Figur 10 2 IDMP timer SEUS iaa Figure 10 3 Message cycle using the SDN or the SDA service sea Figure 10 4 MeanRT using the SDN and the SDA services eeeeeeereeeenerareeeaaeeraearaeereearertnanaa Figure 10 5 Number of transactions using the SDN and the SDA services ir eeeereeeeerereeererenanaa 107 Figure 10 6 Percentage of transactions that do not miss its deadline using the SDN and the SDA services Figure 10 7 Message stream response time considering the MeanBER probability equal for all domains Figure 10 8 Number of transactions considering the MeanBER probability equal for all domains 109 Figure 10 9 Percentage of transactions that do no miss its deadline considering the MeanBER probability equal for all dOMAINS uses oco ta ds nt e de e
260. rates the transaction schema between a Master and a Slave When a Master holds the token frame its Master DLL pops a message from one of its message output queues builds a frame and passes it to the Master PHY which sends to the Domain to which it is connected The Domain broadcasts the frame to every Master and Slave connected to it The Slave DLL receives the frame from Slave PHY and tries to determine if there is a match between the frame and a message stream configuration If it finds a match then it builds a response frame and passes it to the Slave _PHY that sends it to the initiator through the Domain otherwise the frame is discarded The Domain broadcasts the frame to all Master and Slave module instances connected to it The Master PHY passes the response frame to the Master_DLL If it is the addressed to Simulation Models Implementation 143 the Master it stores the information about this transaction and then discards the frame If is not addressed to it then the frame is discarded Receive SDN Procedure Receive a SDN frame Yes Yes No Discard the frame Figure C 11 Receive SDN Procedure Finish transaction Store transaction information Master Slave Msg_Stream Msg_Stream Message output queues Master_DLL Slave DLLZ DA SA Message streams DAE e configuration SAE DATA O Frame Message
261. rded to output data files by each Domain module instance Figure D 18 shows an example of this kind of file which uses the est extension The identification of the BEM used and its parameters are written in the first line The following lines show in the first column the timestamp when state change occurred and the second column show when the new state The main objective of this feature is to validate the error model in use by displaying statistical data regarding its operation GILBERT_ELLIOT 0 327037 0 672963 0 000082 0 002889 0 0000000000 GOOD 0 0000005000 BAD 0 0000010000 GOOD 0 0000020000 BAD 0 0000025000 GOOD 0 0000040000 BAD 0 0000050000 GOOD 0 0000060000 BAD 0 0000065000 GOOD 0 0000085000 BAD 0 0000095000 GOOD Figure D 18 Output channel state quality file excerpt Figure D 19 shows a screenshot of the spreadsheet related to channel state quality of one domain This tool summarizes information regarding the periods in time during which the channel in one domain has been in the good or in the bad state of the Gilbert Elliot bit error model The tool provides some statistical parameters like minimum MIM maximum MAX mean MEAN standard deviation STD DVT and the number of times that this domain was in this state N REG Additionally it also constructs a histogram of these timings D1 CHANNEL STATE QUALITY
262. re C 30 presents the send Beacon procedure state IDENT n_beacon_counter Figure C 30 Send Beacon Procedure Phase 3 Simulation Models Implementation 155 During Phase 4 the DMMs of the wireless domains try to detect which wireless mobile stations are present in their domains Every DMM knows the addresses of all the wireless mobile stations belonging to the network LWMSN And for each of them transmits a Discovery message D_REQ using the DLL services After sending the D REQ message the DLL informs the DMM with the result which can be the reception of a response or the Ts expiration The pmm collects this information and after all D_REQ messages have been processed it broadcasts a RU message containing the addresses of the wireless mobile stations in its domain The IDMP ends and the bum evolves to the INACTIVE state Figure C 31 presents the Send Discovery Procedure Discovery SubPhase Procedure WMSlist LWMSN state INACTIVE Build RU message No Pass RU message to DLL Remove an address from WMSlist Build discovery D REQ frame Pass discovery frame to DLL DLL notification Store information Yes Figure C 31 Discovery SubPhase Procedure phase 4 Station Mobility In order to model the mobility of station between domains in the BHW2PNetSim the pum of the wireless domains also operates as a BS At reception of the SBT message from cmm sends to the Controller
263. re are several stop points at specific locations see Figure 9 1 and Figure 9 2 The domain location of cach wireless mobile station was set according to the results provided by the Mobility Simulator MSim described in Chapter 8 Another important detail concerns the cap Update factor G which is set to 1 in all domains in order to have the GAP Update mechanism always active This feature effectively increases the network load but since the FDL Request Status frames used by the GAP Update mechanism have low priority the response time of the high priority message streams is only minimally affected 9 2 1 Repeater Based Scenario The repeater based network scenario Figure 9 1 is comprised of three wired masters M1 M2 and MM two mobile wireless master M3 and M4 five wired slaves S1 S2 S3 S4 and S5 and one mobile wireless slave S6 Master MM is the Mobility Master MM Figure 9 1 Repeater based hybrid wired wireless PROFIBUS network Comparative Performance Analysis in an Error Free Environment 93 Table 9 2 presents the station s address used in this network scenario The Highest Station Address HSA master parameter was set equal to five in all masters The repeater based approach requires the specific setting of the T p and Ts parameters which depend on the maximum size of the frames relayed by the repeaters the number of repeaters in cascade the bit rate in each medium and the delays in each repeater Thes
264. re common to all modules connected to a domain are assigned to the Domain module and the Controller module instance performs the domain configuration and other module instance parameterization using this information Figure 5 8 presents the Domain simple module NED definition The parameter medium defines if the Domain module instance maps into wired different to zero or wireless equal to zero domain Due to the use of different media in the network the format of the wired and wireless frames is different As an example each DLL character can be coded using 8 or 11 bit for wireless and wired frames respectively The wireless frames can also include additional preamble and header fields The parameters bitsPerChar frameHeadLen and frameTailLen are the number of bits per character the number on the bits of the frame head and the number of the bit on the frame tail respectively simple Domain parameters Baud_rate numeric bitsPerChar numeric frameHeadLen numeric frameTailLen numeric G numeric HSA numeric TTR numeric TSL numeric max_retry_limit numeric _bem_type numeric _bem_par1 numeric _bem_par2 numeric _bem_par3 numeric _bem_par4 numeric _name string gates in ctrl_gateln out ctrl_gateOut endsimple Figure 5 8 Domain module NED definition The parameter G is the Gap Update factor The HSA parameter defines the Highest Station Address in the domain The TTR parameter is the Targ
265. red Wireless Network Simulator RHW2PNetSim and the Bridge Based Hybrid Wired Wireless Network Simulator BHW2PNetSim In Chapter 5 Chapter 6 and Chapter 7 the behaviour of each module was described using the description of its state machine The particular specification of each transition is detailed in this annex Therefore this annex is a complement to those chapters C 1 1 Token Recovery Procedure In the PROFIBUS protocol a token lost is detected when a master does not detect any network activity for a time defined by its Time Out Time Tro parameter which is set by Eq 2 2 A timer is loaded with Tro parameter value and is started in two situations First when the frame transmitter transmits the frame s last bit Second when a master receives frame s last bit The timer is stopped when the first bit of the following frame is received When Tro timer expires and the Master module instance is in the ACTIVE_IDLE state it starts performing message cycles according to the message dispatching procedure described in Section C 1 3 But if it is in the LISTEN TOKEN state then it evolves to the CLAIM TOKEN state and the token recovery procedure starts This procedure has two objectives First recovering the token frame and second to reinitialize the logical ring Note that when a Master evolves to LISTEN TOKEN state all List of Active Stations LAS entries are deleted Figure C 1 1 handleSelfMessage msg A 3 switc
266. redes baseadas na tecnologia PROFIBUS acr nimo de PROcess Fleld BUS s o o tipo fieldbus mais utilizado em todo o mundo em aplica es de automa o e controlo Durante a ltima d cada assistiu se ao aumento exponencial da utiliza o de sistemas de comunica o sem fios wireless communications quer atrav s do telefone celular vulgo telem vel quer atrav s das redes locais sem fios wireless local area network cedo ficou evidente que a tecnologia de comunica o sem fios poderia impulsionar um novo conjunto de potencialidades nas aplica es de automa o e controlo Ve culos auto guiados sensores em partes m veis de maquinaria e terminais port teis s o alguns dos exemplos que requerem comunica o sem fios Todavia os requisitos dos sistemas de tempo real geralmente servidos por fieldbuses imp em a utiliza o de servi os de comunica o previs veis e confi veis que proporcionem certas garantias relativas entrega de mensagens e do respectivo tempo em que s o entregues Consequentemente o uso de comunica es sem fios em aplica es de tempo real pode ser um desafio dado que a probabilidade dos requisitos de tempo real n o serem cumpridos maior do que usando sistemas de comunica o cablados O projecto Europeu RFieldbus executado entre os anos de 2000 e 2002 foi uma importante iniciativa no sentido da concep o de um sistema de comunica o industrial do tipo fieldbus h brido suportado pela tecnologi
267. ribe the implementation of the specific functions required by the repeater and bridge based approaches Chapter 6 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model This chapter describes how the repeater based approach has been implemented in the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator by presenting the main architectural modules and its configuration parameters 6 1 Introduction The Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim is based on the network model developed within the aims of the RFieldbus project RFieldbus 2000a and particularly on the architecture detailed in Alves 2003 It has been developed using the OMNeT discrete event simulation platform described in the Chapter 4 The RHW2PNetSim is composed by 7 main components modules HW2PNet Controller Master Slave Domain ComFunc and Connection Point The modules Hw2PNet Controller Master Slave and Domain control the main PROFIBUS functionalities and they were already detailed in Chapter 5 Therefore in this chapter only the modules which implement the repeater functionalities are described the ComFunc and Connection Point modules For the sake of simplicity the functionalities of the Base Stations BSs are incorporated into the repeaters and the repeaters are only able to operate in cut through mode This chapter starts by describing the main architectural components of the RHW2PNe
268. ribution Functions Analysis option D 3 4 Bit Error Model The information about the Bit Error Model BEM used in simulation runs are recorded in several output data files First includes information about the number of correct and corrupted transmitted frames Second includes detailed information about corrupted frames transmitted Third includes information about IDT deleted and fourth includes information about IDMP aborted The last one includes information about channel state quality and the frame transmitted by each Domain module instance The third and fourth are only generated by the BHW2PNetSim and fifth is generated only 1f the BEM used is either Gilbert Elliot Model simplified or not or Burst Error Periodic Model Frame Accounting The number of valid and invalid frames is also recorded to files with cfr and efr extensions as well as the information about the invalid frames relayed by each Domain module instance The information is grouped into four groups PROFIBUS BEACON IDP and IDMP related frames where Tools for Simulation Output Analysis 165 the first is related to standard PROFIBUS frames the second are the beacon frames the third are the IDF used by the IDP protocol and the last group represents the frames related to the IDMP For each group is presented the number of valid and invalid frames relayed by a Domain module instance Figure D 11 presents the information generated by a Domain module instance o
269. ric _hame string endsimple Figure 7 6 Domain module NED definition of the BHW2PNetSim Figure 7 7 shows a configuration example of a domain in which the IDMP must end after 40 ms since the DMM receives a PBT message from the GMM and the pmm has to transmit 14 Beacon frames during Phase 3 of the IDMP The length of each of Beacon frame is 10 bytes theBHW2PNet domain 0 _dmm_idmp_abort_timer 40ms theBHW2PNet domain 0 _n_beacon 14 theBHW2PNet domain 0 _beacon_len 10 theBHW2PNet domain 0 name D1 Figure 7 7 Configuration file related to the Domain module instance of the BHW2PNetSim excerpt 7 3 3 Master On the Bridge based Hybrid Wired Wireless PROFIBUS simulator a master module can be used to simulate a PROFIBUS master wired or wireless or a BM It is composed by three modules Two of them are simple modules Master PHY and Msg Stream and the other is a compound module Master DLL As described in Section 5 2 4 each master module instance is composed by one instance of Master PHY and Master DLL modules and at most 64 msg Stream module instances This kind of master module Figure 7 2 and Figure 7 8 is connected to the Domain module through gates domain gateIn and domain_gateOut and it can be connected to the comrunc module through gates bridge gateIn and bridge gateout when it operates as a BM Master PHY and Master DLL model the PhL and DLL of the PROFIBUS protocol respectively Additi
270. rrors of the RU message when the IDMP is not active the DMM sends in a periodical fashion D REQ messages to wireless mobile stations and if the wireless mobile stations acknowledge positively the DMM sends a RU message containing information about the new stations in a domain When the DLL sends the D REQ message its state machine evolves to the AWAIT DISCOVERY RESPONSE state transition 32 and returns to USE TOKEN state transition 33 when it receives a response or Ts expires Section C 3 2 presents a detailed description of the DLL procedures that these transitions 7 5 Summary This chapter presented the main architectural components of a bridge and the implementation of the Bridge Based Hybrid Wired Wireless PROFIBUS Network Simulator Additionally we have also described the format of the NED files required for the configuration of the modules used in this simulator The following chapters will present the mobility simulator used to simulate the physical displacement of the wireless mobile station and results obtained by the simulators Chapter 8 Mobility Simulator In order to achieve more realistic simulation results it is necessary to know at which points in time a wireless mobile station moves between different wireless domains To obtain this information a tool was developed which simulates the mobility of wireless stations over a factory floor and the signal quality of the different wireless domains at station location the Mob
271. rs are used by the Controller module instance to parameterize the Connection Point module instance which is operating as a BS simple Domain parameters _beacon_len numeric n_beacon numeric beacon_gap numeric _name string endsimple Figure 6 5 Domain module NED definition of the RHW2PNetSim During the mobility procedure a BS will transmit 14 Beacon frames with an interval of 25 us between them and the length of each Beacon frame is equal to 10 bytes Figure 6 6 theRHW2PNet domain 0 beacon len 10 theRHW2PNet domain 0 n beacon 14 theRHW2PNet domain 0 beacon gap 25us theRHW2PNet domain 0 _name D1 Figure 6 6 Configuration file related to the Domain module instance of the RHW2PNetSim excerpt 6 2 3 Parameters location vector and is mobile station Besides the parameters referred in Chapter 5 Master and Slave modules have two more parameters One called location vector and the other called is mobile station these parameters are highlighted in Figure 6 7 The location vector is a string which defines the location of each Master and Slave module instance during time In order to limit the size of the configuration files used the location vector parameter is written in a compact format Each location is represented by a tuple n_mob Dx where n_mob represents the number of mobility procedures during which the Master Or Slave module instance will stay on domain Dx The is mobile station para
272. rth Freq fifth BS and sixth D columns respectively The seventh column Position x y z contains data related to the agv position The distance between the Mobility Simulator 89 agv instance and the bs instance with the best radio quality is presented in the eighth column Dist The signal power at the receiver location appears in the last column Pr Time SHandProc EHandProc Freq BS D Position x y z Dist Pr 0 200000 0 200000 0 205000 2 30 BS2 D3 49 17 0 00 0 00 48 80 104 40 0 400500 0 400000 0 405000 2 30 BS2 D3 48 33 0 00 0 00 49 34 120 04 0 600500 0 600000 0 605000 2 30 BS2 D3 47 50 0 00 0 00 49 87 117 47 0 800500 0 800000 0 805000 2 30 BS2 D3 46 67 0 00 0 00 50 42 127 38 1 000 000 1 000 000 1 005 000 2 30 BS2 D3 45 84 0 00 0 00 50 97 120 68 1 200 500 1 200 000 1 205 000 2 30 BS2 D3 45 00 0 00 0 00 51 54 120 72 1 400 500 1 400 000 1 405 000 2 30 BS2 D3 44 17 0 00 0 00 52 11 117 13 1 600 500 1 600 000 1 605 000 2 30 BS2 D3 43 33 0 00 0 00 52 68 128 74 1 800 500 1 800 000 1 805 000 2 30 BS2 D3 42 50 0 00 0 00 53 27 128 30 2 000 500 2 000 000 2 005 000 2 30 BS2 D3 41 67 0 00 0 00 53 86 115 78 2 200 500 2 200 000 2 205 000 2 30 BS2 D3 40 84 0 00 0 00 54 45 124 12 2 400 500 2 400 000 2 405 000 2 30 BS2 D3 40 00 0 00 0 00 55 06 123 01 2 600 500 2 600 000 2 605 000 2 30 BS2 D3 39 17 0 00 0 00 55 66 126 10 2 800 500 2 800 000 2 805 000 2 30 BS2 D3 38 34 0 00 0 00 56 28 118 07 0 200000 0 200000 0 2050
273. s Consequently they take part of the domain s logical ring to which they are connected For the sake of simplicity it is assumed that BMs do not support any AL functionalities Figure 2 9 presents a bridge B3 with two BMs M7 and M10 Wireless domains in the bridge based architecture are structured Therefore in each wireless domain there is the need of a BS which can be included in the bridge wireless front end Bridge B2 depicted in Figure 2 9 is such kind of bridge A Wired communication interface E m6 Wired Bridge Master Wireless Bridge Master O e Base Station N Wireless communication interface Figure 2 9 Basic components of a bridge Figure 2 10 illustrates an example network that comprises two structured wireless domains D and D and two wired domains D and D In the system there are two wired masters M1 and M2 two wireless mobile masters M3 and M4 five wired slaves S1 S2 S3 S4 and S5 and one wireless mobile slave S6 Three bridge devices are considered B1 M8 M5 B2 M6 M9 B3 M10 M7 Network operation is based on the Multiple Logical Ring MLR approach described in Ferreira Alves et al 2002 Therefore each wired wireless domain has its own logical ring In this example four different logical rings exist D M3 M D M1 MS M6 D M4 gt M9 gt M10 and D M2 gt M7 This approach requires two new protocols one for supporting the communication between stations i
274. s The network mobility procedure is triggered every 0 2 s defined by the nhp period parameter The duration of each network handoff procedure is equal to 0 005 s defined by the nhp duration parameter The path loss exponent defined by nhp exponent parameter is set to five and the standard deviation defined by nhp std_dev parameter is set to 6 8 The close in reference distance defined by nhp do parameter is set equal to one meter The simulated factory floor dimensions are 100 m per 200 m defined by ground width and ground length parameters The output directory for the simulation output files is called output files defined by the output dir parameter sim duration 2 nhp period 0 2 nhp duration 0 005 nhp exponent 5 bnp std_dev 6 8 bmp d0 1 ground width 100 ground length 200 output dir output_files Figure 8 5 Configuration file related to global simulation parameters excerpt Figure 8 6 presents part of a configuration file related to the bs module instance parameters The system is composed by two bs instances defined by the sim bs num parameter One is called BS1 and the other is called BS2 defined by the bs x name parameter Wireless communications in domains D1 and D3 are done through BS1 and BS2 respectively bs x domain parameter sim bs_num 2 bs 0 name BS1 bs 0 domain D1 bs 0 position 80 0 37 5 0 5 0 bs 0 color 1 0 0 0 0 0 1 0 bs 0 ant freq 2 4 bs 0 ant pt
275. s e g Tip and Tsz in every domain Contrarily in the repeater based approach the parameter setting depends on the network parameters and configuration resulting on higher duration for a message cycle The segmentation also permits a better responsiveness to errors transmission and token loss in the bridge based approach since Tsz can be set to smaller values Additionally the segmentation operated by the bridges permits a higher throughput of the overall network which can be confirmed in our experiments since in the bridge based case the number of message transaction performed is in all cases much higher It is also noticeable that the messages queuing order has practically no influence in the maximum response time of a message stream in the bridge based approach contrarily to the repeater based approach From the experiments in which the network parameters have been varied it can be concluded that the repeater based approach is more influenced by these changes especially when the maximum frame size in the network is increased Nevertheless the use of repeaters leads to a simpler solution since the repeater devices only operate at the PhL level contrarily to the bridge based approach which requires a more complex set of protocols the IDP and the IDMP implemented at the DLL level Additionally the mobility procedure used in the bridge based approach leads to higher inaccessibility times for the wireless mobile stations since these s
276. s able to support guaranteed real time traffic There are two kinds of stations master and slaves Only masters have access to the medium and slaves only have access to the medium when they reply to a master request The benefits of using wireless technologies are manifold The ease of equipment installation the systems configuration flexibility the ability to evolve and the cuts in cabling and maintenance costs just to mention some However wireless channels are prone to more transmission errors caused by either channel outages which occur when the received signal strength drops below a critical threshold and or interference Therefore the use of wireless technologies in a real time system can jeopardize the timing requirements and reliability of this kind of system To integrate wired and wireless technologies in the same network system there is the need of a special purpose device that is able to interconnect the different mediums systems called Intermediate System IS The behaviour of the ISs is very important in this context Within the context of this dissertation only IS operating as repeaters and as bridges are considered Traditionally a repeater act as signal regenerator but it can also interconnect communication systems with different Physical Layer PhL protocols For that purpose a repeater may require the implementation of more than bit by bit repeating functionality This is the case when it interconnects communication systems
277. s also constant since the pdf length type parameter is equal to zero at 15 bytes pdf length pari parameter The first message does not have any initial offset The colour of this message stream on the Gant diagram is 100 255 0 in RGB notation 5 2 5 Slave A slave is a compound module which maps into a standard PROFIBUS slave station It is structured similarly to the Master module The Slave PHY module is equal to the master PHy module The Msg_Stream module is the same module used by the Master module and is used for the same purpose although in the case of a Slave module it generates periodical response messages The Slave DLL module is a simple module which maps the PROFIBUS DLL of a slave In each Slave module instance there is one instance Of Slave PHY and Slave DLL modules The number of the Msg_Stream module instances can be from 1 up to 64 As shown in Figure 5 18 the Slave module structure is very similar to the Master module structure presented in Figure 5 10 Since the Master_PHY and Msg_Stream modules were already described in Section 5 2 4 in this sub section only the Slave DLL module will be described PROFIBUS Simulation Model 55 Slave Msg Stream 0 Msg Stream N lower gateln lower gateOut lower gateln lower gateOut upper gateOut 0 upper gateln 0 upper gateOut N upper gateln N Slave DLL lower gateln lower gateOut upper gateOut upper gateln Slave
278. s an open source cross platform graphical toolkit for the development of high performance graphical applications such as visual simulation flight simulation games virtual reality scientific visualization and modelling Based around the concept of scene graph detailed below it provides an object oriented framework on top of OpenGL freeing the developer from implementing and optimizing low level graphics calls and provides many additional utilities for rapid development of graphics applications Written entirely in standard C and OpenGL it makes full use of the Standard Template Library and it runs on all Windows platforms OSX GNU Linux IRIX Solaris and FreeBSD operating systems 8 2 1 Scene Graph Concept The scene graph concept is tree like It starts with a top most root node which encompasses the whole virtual world be it 2D or 3D The world is then broken down into a hierarchy of nodes representing spatial groupings of objects object positions object animations or definitions of logical relationships between objects The leaves of the graph represent the physical objects themselves the drawable geometry and their material properties A scene graph is not a complete game or simulation engine although it may be one of the main components of such an engine Its primary focus is the representation of 3D worlds and efficient rendering Physics models collision detection and audio are left to other development libraries that a user m
279. s between 10 and 10 Figure 10 10 Figure 10 11 and Figure 10 12 present the response time values the number of transactions and the percentage of transactions that do not miss its deadline On the left side of these figures the MeanBER probability used was set to 10 for wireless domains and on the right side the wireless domains BER was modelled using a MeanBER probability equal to 10 The results presented in Figure 10 10 show that in the repeater based scenario the MeanRT is more affected by the increase of the BER than in the bridge based scenario MeanBER probability of the wireless domain 10 MeanBER probability of the wireless domain 10 Response Time ms Response Time ms sim sim sms sms sm sm sms Sam sii sm ss Sam sm sm sm sam Message Stream Message Stream Message Stream Message Stream Figure 10 10 Response time using different MeanBER probability in wired and wireless domains The number of transactions for each message stream is depicted in Figure 10 11 In the bridge based scenario the number of transaction is much higher for all message streams As would be 110 Comparative Performance Analysis in an Error Prone Environment expected the number of transactions is smaller when the MeanBER probability is equal to 10 on both scenarios especially for IDTs In the bridge based scenario the decrease on the number of transactions is especially higher for message streams S and S3 The larger decrease
280. s proposed in Tovar and Vasques 1999 in order to guarantee that at token arrival there will be enough time to execute all 94 Comparative Performance Analysis in an Error Free Environment pending high priority messages The HSA was set differently for each domain according to the highest address for all stations belonging to that domain This setting reduces the impact of the GAP Update mechanism since in this way the master with the highest address has a minimum number of station addresses to inquiry from 0 to the station with the lowest address in the logical ring Table 9 5 presents the settings of the Tips Tpz Tst Trr and HSA master DLL parameters for each domain Table 9 4 Master s address Master Address Master Address M1 1 M6 6 M2 2 M7 il M3 3 M8 8 M4 4 M9 9 M5 E M10 10 a Ee i PathofM3 PathofM4 TIT Path of S6 Stop petits Figure 9 2 Bridge based hybrid wired wireless PROFIBUS network Table 9 5 Bridge based domain parameters Domain Tip1 Tmz Ts Trr has D and D 100 100 115 2076 8 10 D 100 100 115 2046 7 D 100 100 115 1306 6 9 2 3 Message Streams A message stream is a periodic sequence of message cycles related for instance to the reading of a sensor Each message stream associates an initiator a master with a responder usually a slave The notation S is used to identify a message stream i from an initiator station x e g S is the first message stream of ma
281. s rita evs desen a ral de ate cued eden WH ie de e A O 45 5 2 PROFIBUS Basic Architecture ee non nono ron rr nn ron nr nn non n ron rro rra nr narran 45 DD ds A A A E ARO CO 46 AR Caa KO IS NEA ce pee ere re tee eee 48 5 2 3 Doar oras ep AT 49 5 24 A AS A 50 A ON 54 5 3 PROFIBUS DLL Basic Implementation re ereceeereceraeereeenererererennaes 55 DA SUMMARY contain eres ea ADE EA DITOS ESTAS A Toa ia ansias 57 CHAPTER 6 REPEATER BASED HYBRID WIRED WIRELESS PROFIBUS ARCHITECTURE SIMULATION MODEL ieccdvcsncsvcdsssssccescccnc ccosiesenssseetedasvetesonetiverseas tedestesensoseseodesuatvecess ioo atiae oispa Soie 59 Os Introd uctiOn GAR REAR ADO ek es E E ee RR ed ee eee BE Es os 59 6 2 Simulators Architect ua id 59 6 2 L Controll ai A ai 60 62 2 DMa Secor keh awk Su eee aie Bsa Re ee ek Re Ra 62 6 2 3 Parameters location vector and is mobile station ea 62 6 2 4 Repeater Architecture Leucas te econ eer A Shae eos 63 6 3 Simulator Implementation siseses possue asus E non penis ent pasig en ene 64 6 3 1 Interconnection in 64 6 3 2 Mobility Procedure tens 65 GA SUMMARY Iia eors ee A A acia 66 CHAPTER 7 BRIDGE BASED HYBRID WIRED WIRELESS PROFIBUS NETWORK ARCHITECTURE MODEL BASANE E EE OEE A A AE E E E E 67 Polis MMO LUTEA TON a RREAN ENE AE E A E E E EN tat dd toed 67 7 De Bridge ATOM AE E EE A A des 67 7 3 simulator ATCRI ECtULS 2 EN A a aaa eRe Sees 68 TIL Coti ad 68 ESA O 71 Ted Des O E
282. s string defines the repeater configuration and the meaning of the tags is the following lt r gt and lt r gt define a repeater lt n gt and lt n gt enclose the name of the ComFunc module instance between tags lt cp gt and lt cp gt and separated by a colon appear the names of the Connection Point module instances lt pos gt and lt pos gt are used to define the location of the repeater The first repeater presented in Figure 6 4 lt r gt lt n gt RI lt n gt lt cp gt CP5 CP8 lt cp gt lt pos gt 400 150 lt pos gt lt r gt is referred to as R1 it is composed by two Connection_Point module instances CP5 and CP8 and is positioned at 400 150 Note that this repeater interconnects domains D1 and D2 The Controller module instance stores into internal variables the structure of the network By manipulating this information it changes the network configuration when wireless mobile stations Master and Slave module instances move between domains 62 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 6 2 2 Domain Figure 6 5 presents the Domain simple module NED definition The length of the Beacon frame is set by beacon_len parameter n beacon parameter is used to define the number of the Beacon frames relayed in the mobility procedure The time interval between two consecutive Beacon transmissions is set on the beacon gap parameter thean parameter referred to in Section 2 3 3 These paramete
283. se Service con Data O Open transaction Close transaction Figure 2 11 Example timeline for an Inter Domain Transaction IDT between master M3 and slave S6 The initiator also uses a buffered communication mode where the user and the initiator s protocol stack interface with the PROFIBUS DP through a memory area which allows reading and writing variables that represent the state of local or remote variables It is the responsibility of PROFIBUS DP to periodically update the variables using primitives of the type Service req 20 Technological Context Communication Infrastructure Inter Domain Frame Formats IDFs are used by the IDP for proper transmission of frames between bridges These frames must contain information that enables decoding the embedded original request response and matching the information stored in the BM LOT and the respective response The PROFIBUS protocol allows a request using a variable data field length frame with DAE to be answered by a fixed length response frame without data field thus not supporting DAE Therefore BM would not be capable of matching two different requests from the same initiator addressed to the same responder but with different DAE The PROFIBUS DLL protocol also defines that requests using variable data field length frames can be replied with a SC frame Obviously if no special IDF format was used the bridges would be unable to route the SC frame back to the initiat
284. se time of each transaction is computed from the time in which the request frame is queued on the DLL output message queue until it receives the response frame However the response time of a transaction can be theoretically unlimited in error prone environments In order to deal with transmission medium characteristics real time systems must be provided with mechanisms to detect and handle these error situations Burns and Wellings 2001 In a communication system these mechanisms are implemented at all levels of the communication stack At the AL level the Msg Stream module is provided with a parameter the deadline parameter which is used to detect if a transaction is not concluded before the expiration of the deadline In our simulation model we consider that a transaction misses its deadline in two situations First when the response time of a transaction is higher than its deadline even when a valid response is obtained from the IDT BM this is illustrated on the left side of Figure 5 16 The second case is the most common case in which a deadline is considered missed when the response frame is not received within its deadline This case is illustrated on the right side of Figure 5 16 As mentioned these simulators produce information enabling the display of the Gant diagrams concerning message transactions To distinguish between the different message streams it is possible to assign a colour to each message stream using the parameters with
285. shed by examining at most one address per token cycle by means of sending an FDL Request Status once Tgyp has expired After a complete GAP check which may last several token rotations the timer Tgup is loaded with the value resulting from the multiplication of the Tyr by the GAP Update Factor G G represents the number of tokens rounds between GAP maintenance Upon receiving the token a GAP maintenance cycle starts immediately after all queued high priority message cycles have been processed and if there is still Tr available Otherwise the GAP maintenance is postponed to the next token reception Note that FDL Request Status messages have higher priority than low priority messages used in PROFIBUS but lower than high priority messages Send FDL Request Status Procedure Figure C 6 presents the send FDL Request Status procedure The first step is to evolve the Master state machine from the USE TOKEN to the AWAIT STATUS RESPONSE state After that it builds the FDL Request Status frame addressed to the selected station and transmits it Thereafter it waits for a valid without bit errors response frame from the addressed station within the Tsz If either Tsz expires or the received response frame is an invalid the Master evolves to the USE_TOKEN state and then the procedure ends If a valid response frame was received from the addressed station the message is parsed If the responder is a master and its state is Ready to Enter Logical Ring the M
286. sing has failed A timer is loaded with Ts at the end of the transmission of a request frame Upon its expiration the master station may execute another retry for the same request if the number of retries executed is smaller than the max retry limit parameter or it may inform the upper layers of a transmission failure A timer is also loaded with Ts after transmitting the token frame If it expires before a master has detected any activity in the bus then it signals the MAC layer in order to take the appropriate actions according to the token passing procedure Tro Tro gt gt Tor Initiator Req t Ack Resp token frame Req Token Frame Idle Time Figure 2 2 Idle Time parameter T p The Slot Time parameter Tsz must be set to the maximum between two values Tsz and Ts gt Ts can be calculated as follows To 2xT p max T opp 110it Tou 2 4 where bit is the time duration of a bit Ts can be calculated as follows Tao 2X Typ maxT p 11bit Toy 2 5 12 Technological Context Communication Infrastructure Note that all masters in the network must hold the same Tsz value due to the token passing mechanism The Time Out Time Tro controls the token passage in PROFIBUS a token lost is detected when a master does not detect any network activity for a time period defined by its Tro which is set to as follows Tr 6 T 2 n Ty 2 6 where n is the master address In Eq 2 6 the
287. ssage according to the simple module algorithm 6 if msg gt isSelfMessage usually used as timer Ki msg_self cMSG_Self s msg 8 switch msg_self gt getAction 9 case SEND MESSAGE OR dequeueMessage msg profibus DES send msg_profibus out oe HL Shes 14 1Sa 16 else LO msg_profibus cMSG_Profibus msg ise switch msg_profibus gt getType alero case TOKEN FRAME if it is a token frame 20 if msg profibus gt getDA TS amp amp msg_profibus gt getSA PS 21 AR VEA computeTHT DBF msg self gt setAction SEND MESSAGE 24 scheduleAt simTime TID msg self Sra 26 break Ph case RESPONSE FRAME if it is a response frame 28 if msg profibus gt getDA TS amp amp ES scheduleAt simTime TID msg self 30 aty break SA ada See 34 Sa BCN Figure 4 6 handleMessage cMessage msg function C code MasterStation Figure 4 7 presents handleMessage cMessage msg function of the SlaveStation module In same way this function is automatically invoked as soon as a SlaveStation module instance receives a message At reception of the message slave S1 checks if the received frame is addressed to it line 20 It schedules the sending of the response line 23 if it is Otherwise no action is taken 44 Technological Context Simulation Software W void SlaveStation handleMessage cMessage msg Ze ak 3 cM
288. st and error skipping A master leaves the logical ring i e its state machine evolves to the LISTEN TOKEN state if it transmits two consecutive corrupted token frames heardback removal A consequence of the heardback removal is the token loss When the token is lost there is the need to reinitialize the logical ring Error skipping occurs when a master in the logical ring receives a token frame in which its address is skipped after detecting such event it removes itself from the logical ring Note that in PROFIBUS networks every station in the logical ring receives all transmitted messages In this chapter we present simulation results considering transmission over error prone mediums These simulation results were obtained using the Gilbert Elliot Channel Model Gilbert 1960 Elliot 1963 to model the transmission errors a brief description of this model is presented in Section 10 2 Error detection and correction algorithms were not previously considered in the bridge based approach proposed in Ferreira 2005 Therefore this approach has been enhanced with the mechanisms proposed in Chapter 3 enabling operation over error prone mediums The proposed mechanisms are based on timers Section 10 3 shows how to set these timers We have also proposed an enhancement to the IDP in which the IDFs are transmitted using the SDA PROFIBUS service instead of using the SDN PROFIBUS service In Section 10 4 we present a comparative performa
289. starts the token recovery mechanism sending a token frame addressed to itself At this instant the others masters which were in the ACTIVE IDLE state evolve to the LISTEN TOKEN state and will stay in this state until two identical token rotations were performed The availability of the bridge based network can be degraded due to this mechanism since when a BM is out of the logical ring then it is not able to process IDTs To decrease the impact of the PROFIBUS token recovery mechanism we defined a set of rules for the attribution of master station addresses This schema is called Master Station Address Setting Rules MSASR The lowest address in each logical ring must be given to a BM The following addresses should be separated by two or more addresses between them For instance 1f the lowest address in a logical ring is 2 the second one must be 4 or more The goal of the first rule is for the token recovery mechanism to be always performed by a BM The second rule is used to reduce the time that a station is out of the logical ring As mentioned a station which address is skipped evolves to the LISTEN_TOKEN state and waits in this state for two identical token frame rotations After that it waits for an FDL Request Status message to enter into the logical ring On the other hand when a master claims the token it sends two token frames addressed to itself Therefore the first transmission leads that the other station evolves to the LISTE
290. state all frame s bit are correctly transmitted Therefore in good state there is the need to compute if state transition occurs In context of this dissertation this model is called Simplified Gilbert Elliot Channel Model B 4 Burst Error Periodic Model The Burst Error Periodic Model assumes that the transmission errors occur in a periodic way In this model it is assumed that there are a lower Tem and a higher T m period threshold The burst length is also bounded by a minimum Nem and maximum N v number of bits The Tem and Tem parameters are set in milliseconds and the Nem and N y are set in bits Figure B 2 shows a simplified timeline using this model The transmission error period is computed using the Eq B 7 and burst length is computed using the Eq B 8 T T3 N bit time N2 bit time N3 bit time A gt Ei Good channel conditions no error Bad channel conditions burst error Figure B 2 Simplified timeline of Burst Error Periodic Model Bit Error Models 133 T uniform T T B 7 N uniform N Nou B 8 To compute the transmission error period and the burst length a uniform probability distribution function Law and Kelton 2000 is used Eq B 8 This function was chosen because this model imposes thresholds i e the transmission error period has to be enclosed within of the range Tem Tem and burst length has to be enclosed within of the range Nen Nem On the other hand is as
291. ster M1 The set of message streams presented in Table 9 6 tries to illustrate some probable transaction scenarios in the network The message streams are specified as tuples destination address request frame length in bytes response frame length in bytes and priority Comparative Performance Analysis in an Error Free Environment 95 As an example S and S are IADTS between master M1 and slaves S1 and S2 respectively S is an IDT between master M1 and slave S5 S is a transaction between two wireless mobile stations master M3 and slave S6 When both are in same domain this transaction is an IADT but if they are in different domains the system handles it as an IDT To simplify we will refer to this transaction as an Intra Inter Domain Transaction IDT Table 9 6 Message streams Stream Parameters Stream Parameters Stream Parameters SM S2 15 20 high See S6 15 20 high sie S3 15 20 high sia S5 15 20 high sp S4 15 20 high spt S6 15 20 high A PROFIBUS standard master is usually a dedicated device composed by a communication module mostly in hardware and a CPU module running the control software Therefore master stations used in our simulation have been modelled according to the following operational characteristic assumptions The variability of the master timing parameters is usually reduced as confirmed by some experimental measurements Behaeghel Nieuwenhuyse et al 2003
292. sters parameter and six slaves num slaves parameter theProfibusNet num domains 1 theProfibusNet num masters 4 theProfibusNet num slaves 6 Figure 5 4 Configuration file related with Hw2PNet module instance excerpt Figure 5 5 shows a screenshot of the simulator output window for the network configuration referred to above In this figure it is clear that the controller instance labelled controller is able to communicate with all module instances Master and Slave module instances are connected to the Domain module instance symbolized by a rectangle controller Figure 5 5 Simulator output window screenshot 48 PROFIBUS Simulation Model 5 2 2 Controller The Controller is a simple module that coordinates the simulation and performs several managing tasks acting as the simulation supervisor Parameters that are specific of one module instance or common to all module instances in the network are assigned to the Controller module instance On simulation setup the Controller module instance makes the parameter setting of the all other module instances Additionally due to memory limitations the controller module instance is responsible for periodically sending commands to other module instances commanding them to dump the information gathered to data files Finally whenever a Master or Slave module instance changes between domains this module updates the network configuration and the corresponding connections Not
293. sumed that either period or burst lengths are uniformly distributed in their defined ranges B 5 Parameterization of the Bit Error Model Parameters Used in both Simulators The RHW2PNetSim and BHW2PNetSim define a set of parameters to specify which BEM to use The name of all these parameters uses the bem prefix The bem type is used to define the BEM and the other parameters _bem_parl bem par2 bem parl bem par3 and bem par4 are used to set the BEM parameter Table B 1 presents how the simulator parameters must be set according to the BEM to be used Table B 1 Bit Error Model simulators parameters Bit Error Model Parameters No errors Independent Gilbert Elliot Bamna t Periodic Channel Model Channel Model Burst Model Channel Model _bem_type 0 1 2 3 4 as josueil Pber Pep Pew Vez _bem_par2 Pog Pos Tem bem par3 Pg DP Nem _bem_par4 P Nem Annex C Implementation of the Simulation Models In Chapter 5 Chapter 6 and Chapter 7 the main architecture of the Repeater Based Hybrid Wired Wireless Network Simulator and the Bridge Based Hybrid Wired Wireless Network Simulator were described In this annex we describe the implementation details of the Repeater Based Hybrid Wired Wireless Network Simulator and the Bridge Based Hybrid Wired Wireless Network Simulator C 1 PROFIBUS DLL The goal of this annex is to provide a more detailed description of the implementation of the Repeater Based Hybrid Wi
294. t diagram Information necessary to build event Gant Diagrams Output resp time Information about the response time of each transaction output states Information about module instances state machine and their transitions output pdf Information about the probability distribution functions used output bem Information about the bit error model used output period Period for dumping the gathered information on to data files output path The path of the directory output files The meaning of the tags used in domain parameter are the following lt d gt and lt d gt specify a domain the tags lt n gt and lt n gt enclose the name of the Domain module instance lt m gt and lt m gt enclose the name of the masters belonging to the domain which are separated by a colon lt s gt and lt s gt tags are similar to the previous case but associated with slaves lt dmm gt and lt dmm gt define the Master module instance that is the DMM of the domain lt pos gt and lt pos gt indicate the position of the domain for graphical representation purposes Note that coordinate 0 0 represent the top left corner of the window as shown in Figure 5 5 In this particular case the first domain lt d gt lt n gt D1 lt n gt lt m gt M1 M2 M3 M4 lt m gt lt s gt S1 S2 S3 S4 S5 S6 lt s gt lt dmm gt M1 lt dmm gt lt pos gt 400 300 lt pos gt lt d gt is referred to as D1 and is composed by four Master module instances wh
295. tSim and its configuration parameters Section 6 2 Details regarding the implementation of the RHW2PNetSim are discussed in Section 6 3 6 2 Simulator Architecture In order to model a repeater two simple modules were developed the Connection Point and ComFunc modules Figure 6 1 illustrates these modules The Connection Point is a simple module that establishes the connection with the Domain module A repeater must include at least two of these module instances The ComFunc is also a simple module that links Connection Point module instances of the same repeater In addition to the ctr1_con and domain_con connections already referred to in Chapter 5 there is another kind of connection the repeater_con This kind of connection is used to connect a repeater to a Domain module instance For that purpose the Domain module is provided with a set of input and output gates whose names use the repeater_gate prefix Figure 6 2 depicts a graphical representation of the network presented in Figure 2 3 In this figure it is clear that the Controller instance labeled controller is able to communicate with all module instances for parameterization and control purposes Master and Slave module instances are connected to their correspondent Domain module instances symbolized by a rectangle or a cloud for the case of wired or wireless domains respectively Each repeater is composed by two Connection_ Point module instances labeled cp x where x is a nu
296. tage of transactions that do not miss its deadline using the SDN and the SDA services 10 5 Performance Comparison between Repeater and Bridge based Architectures In this section we use the network scenarios described in Section 9 2 to compare both approaches For that purpose we performed two sets of simulation runs On the first set the MeanBER probability varies from 10 to 10 in all domains independently of its type On the second set the MeanBER probability of the wired domains was set to 10 while the MeanBER probability of wireless domains varies from 10 to 10 This second set emulates a scenario in which the BER in a wireless domain is different from the BER in a wired domain 10 5 1 Comparison Using the same BER in All Domains In this section we present the simulation results considering the MeanBER probability equal for all domains Figure 10 7 Figure 10 8 and Figure 10 9 show the response time the number of transactions and the percentage of transaction that do not miss the deadline for the message streams set used in this study In the legend of these figures an R or a B specifies that the values are related to the repeater or to the bridge based architecture respectively 108 Comparative Performance Analysis in an Error Prone Environment Figure 10 7 shows the response time graphics where the MinRT MeanRT and MaxRT values are signalled However we focus on the MeanRT to do the comparative analysis Since it reflects
297. tate transition 2 or WINQUIRY state transition 3 depending if its LOT is empty or not respectively In the WIDT END state a BM waits until all open IDTs contained in its LOT are finalized transition 4 In this state 1f it receives a duplicated SMP message then it replies with a RSMP message After this the Bm will not accept new IDTs When all IDTs have been completed it sends a RSMP message to the cvm and enters into the WINQUIRY state transition 5 In the WINQUIRY state Bm only communicates with its domain DMM using the Inquiry service In this state when a BM receives an Inquiry request IQ REQ message transition 6 and if there is a IDMP related message on the DLL high priority output message queue then it commands the DLL to transmit that message as a response otherwise no response is transmitted In this state if it receives another SMP message then it will reply with another RSMP message transition 6 When a BM receives a PBT message it clears all wireless mobile station related entries in its RT and stays in the same state When it receives a SBT message it changes into INACTIVE state transition 7 and the Tam ipmpAbor 18 Stopped If the TamiDmPabor EXpires the BM evolves to the INACTIVE state from either the WIDT_END state transition 8 or WINQUIRY state transition 7 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model 77 A more detailed description of the procedures that support this transi
298. tations must deregister from the original domain and register in the destination domain The results in this chapter were obtained considering the inexistence of errors in message transactions in the following chapter we will present some results which characterise the behaviour of both networks in the error prone environments Chapter 10 Comparative Performance Analysis in an Error Prone Environment This chapter provides a comparative performance analysis between the repeater and bridge based architectures considering its operation over error prone mediums In this comparison study we analysed the influence of transmission errors on the network performance of both architectures Additionally a set of rules are proposed which try to reduce the impact of the PROFIBUS token recovery mechanism in the bridge based approach 10 1 Introduction When considering communications over an error prone and lossy medium performance degradations mainly stem from two sources one is the loss of data making retransmissions necessary the other is the station outage 1 e when a master station is out of the logical ring as a consequence of the fault recovery mechanisms used by the PROFIBUS protocol to handle and detect errors The simulation models used in this study considers that a frame is corrupted 1f 1t contains a bit error independently of which bit or bits are wrong Three stations outage situations can occur heardback removal token lo
299. ter and it must be set prior to runtime as well as the mobility period i e the time interval between the queuing of each BT by the MM Technological Context Communication Infrastructure 17 2 4 Relevant Details on the Bridge Based Hybrid Wired Wireless PROFIBUS Architecture A Bridge Based Hybrid Wired Wireless PROFIBUS Network is composed by wired and wireless stations The wireless stations have a wireless interface as defined in RFieldbus Alves Tovar et al 2002 Rauchhaupt 2002 The communication between stations is based on the PROFIBUS protocol with specific extensions to support wireless communications and mobility In this approach the IS operate as bridges A bridge is a network device capable of relaying PROFIBUS DLL frames between the domains to which the bridge is connected Although a bridge can interconnect more than two different domains for the sake of simplicity it is assumed that a bridge only interconnects two different domains The bridge relaying decision is based on a Routing Table RT which determines whether an incoming frame is to be relayed to the other port or not A bridge is constituted by two Bridge Masters BMs A BM is a modified PROFIBUS master capable of receiving all frames arriving to its physical interface and forwarding them to the other BM of the bridge according to the routing information contained on RT These BMs operate almost as standard PROFIBUS masters and are assigned a PROFIBUS DLL addres
300. ter module instance with address 3 SA and Slave module instance with address 46 DA The SAE and DAE have the same value which is equal to 7 heProfibusNe heProfibusNei heProfibusNei heProfibusNe heProfibusNe heProfibusNei heProfibusNe heProfibusNe heProfibusNe heProfibusNe heProfibusNei heProfibusNe heProfibusNe heProfibusNe heProfibusNe heProfibusNei heProfibusNe heProfibusNe master 2 stream 1 DA 46 master 2 stream 1 SA 3 master 2 stream 1 DAE 7 master 2 stream 1 SAE 7 master 2 stream 1 DATA_UNIT 0 master 2 stream 1 Serv_class 1 master 2 stream 1 service 3 master 2 stream 1 _active 1 master 2 stream 1 _deadline 100ms master 2 stream 1 _pdf_length_type 0 master 2 stream 1 _pdf_length_par1 15 master 2 stream 1 _pdf_period_type 0 master 2 stream 1 _pdf_period_par1 0 005 master 2 stream 1 _pdf_offset_type 0 master 2 stream 1 _pdf_offset_par1 0 0 master 2 stream 1 _output_color_red 100 master 2 stream 1 _output_color_green 255 master 2 stream 1 _output_color_blue 0 Figure 5 17 Msg_Stream configuration parameters of a Master As the Serv_class parameter is equal to 1 this is a high priority message stream using the SRD service since the service parameter is equal to 3 The period of this message stream is constant since the pdf period type parameter is equal to zero at 0 005 s pdf period pari parameter The length of the message i
301. than other messages even the PROFIBUS high priority messages After the pmm state machine has left the INACTIVE state the DLL enters into the INQUIRY MODE state at the reception of the token transition 20 or after completing any task when it is holding the token transition 21 In the INQUIRY MODE state four transitions may occur transition 22 23 25 and 26 If it sends an Inquiry request IQ REQ message then the DLL evolves to the WAIT INQUIRY RESPONSE state transition 19 and when it detects a valid frame or the Tsz expires its state machine evolves to the INQUIRY MODE state transition 24 At this moment the DLL notifies the DMM about what happened In the INQUIRY state the pum commands its domain Bms if it is the case to send messages For that purpose it sends IQ REQ messages to its domain BMs in sequence allowing them to transmit any mobility related messages on their output message queue transition 22 From the INQUIRY MODE state the DLL state machine can evolve to the USE TOKEN state transition 25 or to the BEACON_TX state transition 26 If the domain s type is wired the DLL state machine evolves to the USE TOKEN state and the pum state machine evolves to the INACTIVE state Therefore the IDMP ends and the DLL performs another action according to the message dispatching procedure presented in Section C 1 3 If the domain is wireless then the DLL evolves to the BEACON TX state and the DMM state machine also evolves to
302. the LOT is empty It stays in the WIDT END state until all pending transaction are finalised When the LOT is empty the BM evolves to the WINQUIRY state The processRSMPmessage function Figure C 23 is called whenever a SMP message is received and the LOT is empty This function builds a RSMP message which can be forward to the other BM of the bridge or queued in its own DIL When a BM is in the WINQUIRY state the DMM controls the message cycles of its domain and a BM only responds to Inquiry request IQ REQ message from its pum When a BM receives the IQ REQ message from its DMM it commands its DLL to send IDMP related messages If there is any IDMP 152 Simulation Models Implementation related message in the output queue then these messages otherwise it does not respond In this way all IDMP related message are relayed during inquiry sub phase processRSMPmessage msg buildIDMPmessage RSMP addr_gmm TS if isRoute msg then sendToComFunc msg else sendToDLL msg INS wNH Figure C 23 processRSMPmessage function pseudo code algorithm When a BM receives the PBT message it clears all entries related to wireless mobile stations using the clearWirelessMobileAddrFromRT function from the RT At the reception of the PBT message a BM stops the Tgm IDMPAbort and enters into the INACTIVE state From this point forward the BM is capable of relaying IDTs except for the ones related to wireless mobi
303. tialize the data members The handleMessage function is called during event processing This means that the behaviour of each module is coded in this function The finish function is called when the simulation terminates successfully it is usually used to write statistics at the end of a simulation run In order to clarify these concepts Figure 4 5 presents a typical PROFIBUS network transaction which consists on the request frame from the initiator master M1 and the associated response frame by the responder slave S1 The initiator has to wait an Idle Time Tp before sending a request frame and the responder has to wait Station Delay Responder Time Tspr before sending a response frame e e2 es e4 es Request Frame Do Response Frame 1 Token Reception i Station Delay Responder Idle Time Frame Reception Figure 4 5 PROFIBUS transactions events Assuming that master M1 initiator has just received the token event e at the instant to then it will schedule a self message for instant which marks the beginning of a request frame transmission and the end of the T event e Event e at instant t represents the reception of the request frame s last bit by slave S1 This instant is calculated as function of the request frame s length and the data rate of the connection As soon as the slave S1 receives the request frame s last bit it schedules a self message to simulate the end of the Tspr and begin
304. time inserted by a master ES after an acknowledged request PDU PROFIBUS The data ready instant The length known instant Transmitter fall time The relaying delay time Time within which a master station shall be ready to receive an acknowledgement or response after transmitting a request Real Rotation Time Acronyms and Symbols 173 T SDI T SDI T SDR Tser Ts Tsu T SL2 Tsm Tsyn Trp Tra Tro Trr gt 02 Station delay of the initiator which is measured with respect to the receipt of the last frame last bit until an initiator is ready to transmit again Station Delay of the Initiator Station Delay of a Responder Set up time which expires from the occurrence of an event e g interrupt last octet sent or Synchronous Time expired until the necessary reaction is performed e g to start Synchronous Time or to enable the receiver The Slot Time is a parameter set in every master that defines the timeout for listening for activity in the bus after having transmitted an acknowledged request or token Maximum time the initiator waits for the complete reception of the first frame character of the acknowledgement response frame after transmitting the last bit of the request frame Maximum time the initiator waits after having transmitted the last bit of the token PDU until it detects the first bit of a PDU either a request or the token transmitted by the station that received the token Safety margi
305. tion is made by the update of the LAS and or GAPL of the DMM of the new domain After detecting the presence of wireless mobile stations the DMM broadcasts a RU message When a RU message is received by a BM it updates its RT according to the information contained in the message 24 Technological Context Communication Infrastructure In Figure 2 16 is shown a simplified timeline related to IDMP Phase 4 assuming the network scenario presented in Figure 2 12 Phase 4 is started independently by each DMM and ends also independently after the discovery sub phase IDMP ends Phase 3 starts DMM BM DMM GMM DMM DMM M7 M10 M9 M6 M6 M5 M8 IDMP ends SBT SBT SBT spt e SBT T SBT SBT lt lt Token Beacon Token i Token RBT Beacon___ m e Phase 3 ends and Token Discovery Phase staris Token 1 M Discovery Phase 3 ends and Ta M4 Phase 4 starts iscovery Teken eso S60 A A Phase 4 ends Token RU a E MA a Phase 4 ends e RU i 4 3 z D y D 4 D Figure 2 16 Simplified timeline of IDMP Phase 3 and Phase 4 Inter Domain Mobility Procedure Messages To reduce the cost and complexity of implementing the IDMP this procedure is based on standard features offered by PROFIBUS Therefore all mobility related messages use standard Frames of Fixed Length with Data Field addressed to a specific Source Address Extension SAE e g 55
306. tion is presented in Section C 3 2 Figure 7 18 Bm state machine Domain Mobility Manager DMM The operation mode of the DMM is based on the state machine diagram presented in Figure 7 19 The state machine of the DMM is composed of five states INACTIVE WTOKEN Wait Token INQUIRY BEACON_TX Beacon Transmission and IDENT Identification The DMM goes into the INACTIVE state after power on When the Dmm receives a PBT message the DMM IDMP Abort Timer TbwmmIDmPabor is loaded with dmm idmp abort timer parameter value and is started Then it evolves to either WTOKEN state transition 2 or INQUIRY state transition 3 if the DLL is holding the token or not respectively If the DLL is holding the token at reception of a PBT message it sends a RBT to the cmm and its state machine evolves to the INQUIRY state transition 2 otherwise it evolves to the WTOKEN state transition 1 and waits for the token frame reception As soon as its DLL receives the token frame it sends a RBT message to the cum and evolves to the INQUIRY state transition 3 BEACON TX Figure 7 19 pMM state machine Thereafter the DMM uses the Inquiry service in order to exchange IDMP related messages with the Bms present in its domain transition 5 If a bum does not have any other Bm belonging to its domain then it transmits Void frames in order to maintain the network activity When it is in the WTOKEN state or in the INQUIRY state a D
307. tly from each other A detailed description of this model is found in Annex B Poig 1 Pgig Pbjb Good Bad Pg Pb 1 Pojo Figure 10 1 State machine for the Gilbert Elliot error model Table 10 1 summarizes the Gilbert Elliot Channel Model parameters according to each mean BER MeanBER probability used in the comparative analysis presented in this chapter see Section B 3 for more details Table 10 1 Parameters for Gilbert Elliot Channel Model MeanBER Pog Pop Pp Po 10 0 925074102 0 074925898 0 00000082 0 00012334105 104 0 925074102 0 074925898 0 0000082 0 0012334105 107 0 925074102 0 074925898 0 000082 0 012334105 The bad and good steady state probabilities are set to 7 5 and 92 5 respectively The BER probability for each state varies according to the MeanBER probability For instance the bad state BER probability p is 1 23 10 and the BER probability in good state pg is 8 2 107 for a MeanBER probability equal to 10 and for the MeanBER probability equal to 10 p increases to 1 23 10 and Pg increases to 8 2 10 These parameters are set according to the work of Willig and Wolisz 2001 Comparative Performance Analysis in an Error Prone Environment 105 10 3 Network Scenario Configuration Error detection and correction algorithms were not previously considered for the IDMP in the bridge based approach proposed in Ferreira 2005 Therefore in Chapter 3 an enhanced version of the IDMP with error corr
308. to big enterprise lobbies From these PROFIBUS PROcess FleldBUS IEC 2000 is the most widely used with over 15 million nodes worldwide Weber 2006 in applications ranging from discrete part automation to process control During the cellular phone and WLAN boom of the last decade soon it became evident that wireless radio based communications could leverage a whole new set of potentialities in the field level and control applications Moving parts in machinery hand held equipment wearable computers transportation equipment and autonomous vehicles are just a few examples requiring wireless mobile communications However the requirements of real time systems usually served by fieldbuses impose the use of predictable and reliable communication services which provide certain guarantees on eventual delivery of packets and delivery times According to Stankovic 1989 a real time computing system is a system in which its correctness depends not only on the logical results of computation but also on the time at which results are produced Therefore running real time applications using wireless technology can be especially challenging because the real time and reliability requirements are more likely to be jeopardized than they would be over a wired channel Traditionally real time systems are classified as being either hard or soft Burns and Wellings 2001 For hard real time systems it is imperative that no deadline is missed whilst in
309. to fault situations and to situations in which the protocol state machines can be blocked In this chapter the error situations which can occur during the evolution of the IDP and the IDMP are analysed and solutions to the problems are proposed 3 1 Introduction Wired fieldbus networks usually exhibit a low Bit Error Rate BER however the integration of wireless communications radio based in such kind of network may increase the BER since a wireless transmission medium can not be shielded from the influence of noise sources Nevertheless this increase can be somewhat reduced by the use of more robust modulation technologies like spread spectrum and frequency hopping and by the use of more sophisticated error detection and correction mechanisms The IDP relies on a timeout timer to control the success of an IDT This timer is started by the BM when the IDT is initialised and cleared if it receives a response The IDMP has been originally developed without error handling mechanisms but in this protocol if a frame is lost or corrupted due to transmission errors the evolution of the mobility procedure could be blocked Consequently one of the main objectives of this chapter is to propose improved error handling mechanism for the IDP and for IDMP This chapter is organised as follows In Section 3 2 we analyse the error handling mechanism used by the IDP protocol and propose some improvements Then in Section 3 3 the error situations which can
310. token to 5 0 020051 CHECK_TOKEN_PASS Waiting for activity from 5 0 020139 ACTIVE_IDLE Activity detected from 5 0 020228 USE_TOKEN Received token from 6 0 020294 PASS_TOKEN Trying to pass the token to 5 0 020317 CHECK_TOKEN_PASS Waiting for activity from 5 Figure D 7 Output state machine file excerpt State Machine Statistical Analysis The State Machine Analysis option Figure D 8 provides a fast way to summarise the information This option builds histogram related to each transition computing the number of times N REG that a Master module instance was in each state Additionally it computes the minimum MIN and maximum MAX time spending in each state as well as the mean MEAN and the standard deviation STD DVT D 3 3 Probability Distribution Function The information about the random values generated by the probability distribution function PDF are recorded into several text files different extension are used for example the files related with the Typ and with the Tspx has tid and tsdr extensions respectively Figure D 9 presents an example of this kind of files The first line is used to identify the PDF and its parameters In this case the PDF is a triangular distribution function with apex at 50 bit times and extremes at 11 and 70 bit times 164 Tools for Simulation Output Analysis T3 WRSMP WRBT MIN MAX MEAN _ STD DVT N REG 1823 11045 5225 2 786 59900
311. tream Msg_Stream Message output queues Slave_DLL DA SA Message streams DAE configuration SAE DATA Frame Message Master_PHY Slave_PHY Figure C 9 SDN transaction schema between master and slave Master_DLL Figure C 10 presents the send SDN procedure The Master pops a message from message output queues builds a frame and then transmits it Thereafter Master waits Tpz before performing another action according to the message dispatching procedure Note that it stays on the same state USE_TOKEN Build frame Send frame Figure C 10 Send SDN Procedure Receive SDN Procedure Figure C 11 presents the receive SDN procedure At frame reception the Slave DLL starts by checking if it is a valid frame or not The frame is discarded if it is an invalid frame Otherwise it checks if it is addressed to it or not DA TS If not the frame is discarded If it is the information about this transaction is stored for output analysis and the frame is discarded Send SRD Procedure The SRD is based on a reciprocal connection between an initiator and a responder and requires either an acknowledgement or a response from the responder Using this service the initiator is able to send data in the request frame and receive data from the addressed station in the response frame Figure C 12 illust
312. tream modules It is connected to the Master PHY module instance through lower gateIn and lower gateOut gates and it is connected to M Msg Stream module instances through 64 gates upper_gateIn x and upper _gateOut x where x is a number between 1 and 64 Figure 5 14 presents part of the master DLL NED definition used in the RHW2PNetSim Its NED definition is very simple since most of its parameters are dynamically configured by the Controller module instance Msg Stream The msg Stream module models the typical behaviour of the AL It can be configured to periodically request services from the Master DLL module instance through the lower gateout gate Each Msg_Stream module instance must be configured with the parameters necessary to build PROFIBUS messages Figure 5 15 shows the Msg_Stream NED definition The parameters DA and sa refer to Destination Address and Source Address respectively The local access address to the AL is defined in the SAE Source Address Extension and the remote access address to the AL is defined in the DAE Destination Address Extension The parameter DATA_UNIT maps the content of a frame data field For simplification reasons this parameter is a numeric data field Serv_class parameter defines the priority high or low for the data transfer and the service parameter defines if a message maps into a Send Data with No Acknowledge SDN or a Send and Request Data with Reply SRD PROFIBUS service If the service
313. try_counter Send frame Receive a response 2 Ts expired as bit errors 2 No Finish transaction Discard frame Store transaction information state USE_TOKEN Figure C 13 Send SRD Procedure Receive SRD Procedure Receive a SRD frame Yes Yes No Yes Build response frame Discard the frame Wait Tsor Send response frame Figure C 14 Receive SDR Procedure Simulation Models Implementation 145 C 2 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulator Following are described the procedures that are specific of the Repeater Based Hybrid Wired Wireless Network Simulator This section is a complement of the Section 6 3 C 2 1 Send Beacon Procedure Figure C 15 presents the send Beacon procedure The first step is to set the n_beacon_counter variable equal to n_beacon a BS parameter After that while n beacon_counter variable is higher than zero the BS waits T7p builds a Beacon frame and then transmits the Beacon frame Send Beacon Procedure n_beacon_counter n_beacon n_beacon_counter gt O No Yes n_beacon_counter Figure C 15 Send Beacon Procedure Build beacon frame Wait Tiom Send beacon frame C 2 2 Stations Mobility In order to model the mobility of a station between domains in the Repeater Based Hybrid Wired Wireless PROFIBUS Network Simulator RHW2PNetSim the Connection Point which is operating as a BS at reception of
314. ts deadline is also stored in a file This information is recorded in text files using the sdm extension by each Master module instance Message Stream Response Time Statistical Analysis The tool is capable of decode the text files described in the last Section and retrieve statistical results The response time is computed as a difference between the timestamp of the last reception ninth column of the text file and of the timestamp when the message was queued sixth column of the text file Figure D 5 shows a screenshot of a spreadsheet created by this option It provides information about message streams characteristics like minimum MIN maximum MAX mean MEAN standard deviation STD DVT number of transaction N TRANS and number of transaction that missed the deadline N TRANS DM Further this option builds a histogram of the message stream response time values STREAM ID SA DA SAE DAE 7 3 46 7 Z RESPONSE TIME MIN MAX MEAN STD DVT N TRANS N TRANS DM 0411 9875 11832 15293 468138 26059 1 10 245 0799457 374256 DOMAINS ANITIATOR 2 20 079 0069099 32348 QUE REO N TRANS 329913 0029120 13635 op D3 D3 222931 4 39747 0 019586 9189 D1 D1 244942 5 49581 0050103 23455 os H D3 DI 172 6 59414 0011465 5367 D1 D3 93 7 69 248 0005425 3008 8 79082 000446 2088 DOMAINS 9 88916 0007395 3462 REQ RE
315. two BMs In Section C 3 1 a detailed description of the IDP procedures can be found namely the receive frame from ComFunc procedure the receive frame from Domain procedure and the send IDF procedure i Bridge i i Master Master i i Master DLL Master DLL f BM o ComFunc O gt BM Message a i DLL output queues go Master_PHY Master_PHY 1 A LN Domain O gt wired O O Frame wireless O Figure 7 16 Interconnection schema between two BMs 7 4 2 Operation of the IDMP related Agents Global Mobility Manager GMM The operation mode of the cmm is based on the state machine diagram illustrated in Figure 7 17 The state machine of the cum is composed by three states INACTIVE WRSMP Wait Ready to Start Mobility Procedure and WRBT Wait Ready for Beacon Transmission The state machine diagrams shown in this chapter use an oval shape representing a state and an arrow for a transition For better identification the name of each state is written within the oval shape and each transition is identified by a number The IDMP is triggered in a periodic a fashion At power on the cmm enters into the INACTIVE state and the IDMP related timer is loaded with a time defined by the _tmob parameter When the IDMP related timer reaches zero the cum sends a Start Mobility Proced
316. ty Procedure IDMP The main objective of this dissertation is to compare the timing behaviour of the bridge and repeater based approaches over error free and error prone environments Additionally we also intended to show that the bridge based approach implementation is feasible and propose additional error detection and correction mechanisms which would improve its performance over error prone environments To achieve these objectives two simulation tools have been developed one for the repeater based approach and another to the bridge based approach and a set of result analysis tools Additionally we have also developed another tool to simulate the mobility of wireless stations Keywords Hybrid wired wireless networks Network simulation Real time systems Real time communications Fieldbus networks An lise de Performance para Redes PROFIBUS Sem Fios Resumo A maioria da comunidade industrial apenas integra novas tecnologias nos seus sistemas de automag o ap s estas terem sido extensivamente testadas e amadurecidas No caso dos sistemas de comunica o usados por aplica es de controlo esta tend ncia ainda mais exacerbada devido sua criticalidade para o funcionamento de qualquer planta fabril Geralmente estes sistemas de comunica o s o baseados em redes de campo fieldbus dado que estas disponibilizam n veis adequados de desempenho confian a de funcionamento comportamento temporal e capacidade de manuten o As
317. uld only process the received IDF when it had available resources An obvious improvement to this protocol would be to use the SDA service instead of SDN Nevertheless this change requires that the BM must receive the frame decode its content consult its RT and send a confirmation to the initiator station These operations must be done within the time allowed for the transmission of a confirmation frame which is defined by the Tspr parameter Using the SDA service all IDFs either embedding a request or a response are acknowledged Figure 3 2 presents a simplified timeline of an IDT where IDFs are transmitted using the SDA service In this case a transmission error occurs when the IDF embedding the response is transmitted between BM M6 and BM MS Since the frame has not been acknowledge by BM MS the BM M6 retransmits the frame after the expiration of Tsz Although this mechanism adds some overhead to the network it improves the error handling in error prone environments as it will be shown in Chapter 10 As illustrated in Figure 3 2 if a transmission error occurs the IDF will be retransmitted The number of retransmission is defined by max retry limit DLL parameter This mechanism can cause duplication of IDFs when the acknowledge frame is corrupted Therefore the BM has to check for IDFs duplication and it has to assure that only one IDF of each IDT is relayed by the BMs Error Handling Improvements for the Bridge based Architecture 29 Init
318. ur tra Troy Trr Acronyms and Symbols Probability function Density function Gap Update Factor The receiver antenna gain the transmitter antenna gain Number of characters in a frame Maximum delay before a responder starts transmitting a response to a request Minimum delay before a responder starts transmitting a response to a request Mean Bit Error Rate The path loss exponent which indicates the rate at which the path loss increases with distance The maximum burst length The minimum burst length The transmission error burst length The BER probability in bad state The steady state probability for being in bad state The probability of a transition occurs from bad to good state Bit Error Rate BER probability The frame error probability The BER probability in good state The probability of a transition occurs from good to bad state The steady state probability for being in good state The average path loss at distance d between transmitter and receiver The free space path loss distance d0 The power at received radio signal The transmitted power Message stream i from an initiator station x The mean duration of bad state Lower period threshold Higher period threshold The mean duration of good state Gap Update time defines the periodicity of the GAP update mechanism The no gaps instant The start relaying instant Idle time inserted by a master station after an acknowledgement response or token PDU Idle
319. ure SMP message starting IDMP Phase 1 and the state machine evolves to the WRSMP state transition 1 In order to detect and handle IDMP errors two timers are started Toum p1ater and Toum P14borr The duration of each timer is defined by the _gmm phasel alert timer and gmm phasel abort timer parameters respectively It stays in the WRSMP state until it receives a Ready to Start Mobility Procedure RSMP message from every BM transition 2 in the network However if the Tomm p141er expires before the GMM receives a RSMP from every BM in the network then it re sends a SMP message and waits in the WRSMP state until it receives RSMP messages from BMs which had not replied If it receives the remaining RSMP messages before the expiration of the Temm Pl4bort then it sends a Prepare for 76 Bridge Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model Beacon Transmission PBT message and evolves to the WRBT state transition 3 Otherwise it aborts the IDMP and returns to the INACTIVE state transition 6 When the cmm sends a PBT message Phase 2 starts and it evolves to the WRBT state In the same way to detect and handle errors during Phase 2 two timers are loaded and started Temm pz4tem and TGmmM P24bor The timer Tomm p2Ater is loaded with a time defined by the gmm phase2 alert timer parameter while the Temm P24bort 18 loaded with a time defined by the omm phase2 abort timer parameter It stays in this state until it receiv
320. vely When the Toum pmpabor Xpires it means that no wireless mobile stations have entered or left of domain and in order to update the RT of the BMs a RU message is broadcasted by the DMM with information about which wireless mobile stations continue in its domain Section C 3 2 presents a detailed description of the DMM procedures that support it state machine transitions PROFIBUS master DLL The IDMP has been designed in order to keep the number of protocol modifications low therefore most of the functions can be implemented as an independent module above the DLL or at PhL level like channel assessment functionalities required by the mobile wireless stations However some functionality related to the DMM must be implemented at the DLL level Figure 7 20 depicts the changes required in the state machine of the DLL of a Master to support the functions required by a DMM The operation of the DLL is based on the state machine diagram present in Figure 7 20 There is the need for five more states INQUIRY MODE WAIT INQUIRY RESPONSE BEACON TX Beacon Transmission DISCOVERY and AWAIT DISCOVERY RESPONSE After beginning the IDMP the DLL state machine of a BM which also operates as a DMM evolves according to the IDMP message received and the current state of the DMM state machine In this case 1 e when a Master is also acting as a DMM then it repeatedly sends messages During the mobility procedure IDMP related messages have more priority
321. which must use the IDMP service also affected by errors to move between Comparative Performance Analysis in an Error Prone Environment 109 wireless domains However with MeanBER probability equal to 10 the roles are inverted and the bridge based approach has a higher number of successful transactions 1600 1600 1600 1600 1400 1400 1400 1400 1200 1200 1200 1200 1000 1000 1000 1000 Hs 800 800 800 600 800 600 600 400 400 400 400 200 200 200 200 o o o EF 105 104 10 3 105 10 4 103 EF 105 apre 102 EF 107 10 10 Mean BER probability Mean BER probability Mean BER probability Mean BER probability N of Transactions Thousands 8 8 o Figure 10 8 Number of transactions considering the MeanBER probability equal for all domains ES 1 3 3 3 8 2 Or 2 S 9 95 95 E E 3 B gt ER 9 so 90 90 3 a 2 85 85 85 85 8 80 n 80 80 80 EF 10 104 10 EF 105 104 10 EF 105 10 10 EF 105 104 10 Mean BER probability Mean BER probability Mean BER probability Mean BER probability Figure 10 9 Percentage of transactions that do no miss its deadline considering the MeanBER probability equal for all domains 10 5 2 Comparison using different BERs in Each Domain Usually a wired medium has a lower BER than a wireless medium Therefore we have also performed a comparative analysis where we fixed the MeanBER probability of wired domains equal to 10 and varied the MeanBER probability of wireless domain
322. with Data PROFIBUS standard DA Destination Address PROFIBUS standard DAE Destination Address Extension PROFIBUS standard DCCS Distributed Computer Controlled System DLL Data Link Layer DMM Domain Mobility Manager DSSS Direct Sequence Spread Spectrum EBF Emitting Beacon_Frame ED End Delimiter EFC Embedded frame Function Code EFT Embedded Frame Type FC Frame Control PROFIBUS standard FCS Frame Check Sequence PROFIBUS standard FMA1 2 Management for PROFIBUS networks layers and 2 GAPL Gap List GMM Global Mobility Manager HSA Highest Station Address Vo Input Output IADT Intra Domain Transaction ICM Independent Channel Model IDF Inter Domain Frame IDMP Inter Domain Mobility Procedure IDP Inter Domain Protocol 170 IDreq IDres IDT IEC TDT IS ISO LAN LAS LASD LBMD LBMN LDMMN LE LEr LL LOT LWMSN MAC MaxRT MC MeanBER MeanRT MinRT MLR MM MSim NS OMNeT OSI PBT PC PCF PDA PDF PDU PhL PLC PROFIBUS PROFIBUS DP PROFIBUS FMS PS RBT RFieldbus RHW2PNetSim RSMP RT RU Acronyms and Symbols Inter Domain Request frame Inter Domain Response frame Inter Domain Transaction International Electrotechnical Commission Intra Inter Domain Transaction Intermediate System International Organization for Standardisation Local Area Network List of Active Stations PROFIBUS Standard List of Active Stations in Domain List of Bridge Masters in the Domain L
323. x A for details theRHW2PNet repeater 0 name R1 theRHW2PNet repeater 0 pdf delay type 0 theRHW2PNet repeater 0 pdf delay par1 0 00003 Figure 6 10 Configuration file related to one Comrunc module instance excerpt Connection Point Connection Point is a simple OMNeT module This module establishes the connections between the Domain module instances and assures that a frame is transmitted without time gaps Additionally 64 Repeater Based Hybrid Wired Wireless PROFIBUS Architecture Simulation Model the BS functions are also modelled in this module like for instance sending Beacon frames during the mobility procedure Figure 6 11 shows the connection Point connections Figure 6 12 presents the NED configuration of the Connection Point module The inactivity time between two consecutive frames T pm described in Section 2 3 4 can be assigned in a stochastic way For this reason the timing delay can be assigned by four parameters The parameters related to the Tipm have the pdf_tidm prefix Repeater i i Repeater amp com func con ComFunc _ com func con com func con ComFunc Le com func con 34 Connection_Point Connection_Point Connection_Point repeater_con repeater_con repeater_con repeater_con Domain Domain Domain Wired Wireless Wired Figure 6 11 Connection_Point module
324. xRT using a wider dash 30 Rs 25 Response Time ms np qa np e Response Time ms 2 a a 0 5 1 15 2 5 0 5 1 15 2 5 Bit Rate Mbit s Bit Rate Mbit s Figure 9 7 Influence of D bit rate on the message stream response time values From the observation of Figure 9 7 we can conclude that in the repeater based scenario the variability of the bit rate in domain D has a strong influence on response time of these message streams In this scenario the lower MaxRT occurs when D is operating at 1 5 MBit s but it keeps increasing afterwards The main reason for this behaviour is due to the need of inserting an additional idle time to compensate the dissimilarities of the bit rates In the bridge based scenario the bit rate variation in domain D has a small influence on the response time values of message streams Sand S since these message streams are not relayed by domain D The decrease verified in the MaxRTs value when the bit rate increases is mainly due to a reduction of the IDMP related latencies 100 Comparative Performance Analysis in an Error Free Environment Again in this case in the bridge based scenario the number of concluded transaction is almost more 500 for IADTs and 240 for IDTs than in the repeater based scenario For instance the number of transactions for message streams Sand S in the bridge based scenario considering a bit rate in domain D of 5 MBit s is 1500000 and 722900 respectiv
325. y be achieved by two ways in a direct way or via Base Station BS Ifwireless stations are able to intercommunicate directly the wireless domain is called an ad hoc domain Otherwise 1f messages are relayed by a BS the wireless domain is usually called a structured domain In this dissertation we always assume that we are using the structured approach since this is the only one that permits the mobility of wireless stations A BS operates as a wireless repeater using two radio channels one to receive frames from the wireless stations the uplink channel and another to transmit frames to wireless stations the downlink chamnel The interconnection between wireless and wired domains is done through a special device designed as Intermediate Systems IS This device has to be provided with two communication interfaces one to connect to the wired domain wired communication interface and another to connect to the wireless domain wireless communication interface In the Repeater Based Hybrid Wired Wireless PROFIBUS approach Alves 2003 the ISs operate essentially as repeaters that is they receive frames from one communication interface and retransmit those frames using the other communication interface Figure 2 3 depicts a wired wireless fieldbus network scenario The ISs operating as repeaters may include the BS functionalities in their wireless communication interfaces This network scenario comprises four domains two wired domains D an
326. zed by message stream 166 Tools for Simulation Output Analysis DA SA DAE SAE Timestamp 43 2 4 4 0 080469 46 2 5 5 0 156737 46 2 5 5 0 237376 43 2 4 4 0 250335 46 2 5 5 0 260427 46 2 5 5 0 293211 46 2 5 5 0 332400 46 2 5 5 0 445243 46 2 5 5 0 518283 46 2 5 5 0 565210 46 2 E 5 0 668706 46 2 5 5 0 748602 46 2 5 5 0 773215 43 2 4 4 0 080469 46 2 5 5 0 156737 Figure D 14 Output deleted IDTs file excerpt DA sa DAE SAE NREG ao a ef el 207629 kay af el ales Lo a a af aj 38270 Figure D 15 Screenshot of spreadsheet created by the Bit Error Model IDP Timeout option IDMP Timeout Timers Concerning IDMP four timers are assigned to the GMM and one to each BM Tam impAbor and another to each DMM Dmm IDMP Abort Timer Tpmm IDMP4bort presents in the network Two of the timers associated to the GMM are used to detect and handle the errors during the Phase 1 GMM_Phase_ 1 Alert Timer Temmpiaten and TGmm P14bort While the others two are related to the Phase 2 cum Phase_2 Alert Timer T GMM P2Atert and GMM Phase 2 Abort Timer TGmMM P24bort In Chapter 3 a mechanism was proposed to provide the IDMP with capabilities to operate in error prone environments This proposed mechanism is based on the timers BM_IDMP_Abort_Timer DMM IDMP Abort Timer GMM Phase 1 Alert Timer GMM Phase 1 Abort Timer GMM Phase 2 Alert Timer and GMM Phase 2 Abort Timer Whenever a timer expires the si

Performance Analysis of Wireless-enabled PROFIBUS

Contents

Download Pdf Manuals

Related Search

Related Contents