Home

A Hardware Platform for Communication and Localization

Contents

1. gt O 0000 Zarlink AIM100 2 System Status and Control Session Control 400 MHz Link Status Base Station Status Control BSM Company ID _ 400 MHz TX Data _ 400 MHz RX Data Start Session C Auto CCA 2 45 GHz TX Dat IMD Transceiver ID Seen eater C Ary Company TX Modulation Search ren C Any Implant RX Modulation Start Listening C Auto Listen Channel Z2L70101 for Emergency Registers Bytes Block Status Poling V Data Gather Max Blocks Packet Wake up Mode Operational 22 2450 MHz 400 MHz State Wake up Responses or Emergency Calls IMDID_ Count Company System Messages Description Remote Implant Status Control AIM Figure 48 Base Station Module BSM100 Main Form 76 As shown in Figure 47 Base Station Module BSM100 Main Form has following function tags e Link Setup Setup connection and display link properties e CCA amp Cal Perform Clear Channel Assessment CCA and calibration CAL e Data Send and receive data e Test Enable and disable carrier wave set transmitter level e Remote Implant Not applicable now The BSM100 main form for the ADK is divided into two main sections The upper section is comprised of a tabular form that allows access to the different configuration settings of the ZL7010X as well as providing for control of operational modes of the device The lower section if comprised of a static display that allows for basic system status and control for the main operati
2. d 1 where Lp is the path loss in decibels Lo is the path loss in first meter a is the path loss exponent d is the distance between the transmitter and the receiver usually measured in meters and is a constant which accounts for system losses Radio and antenna engineers use the following simplified formula also known as the Friis transmission equation for the path loss between two isotropic antennas in free space Path loss in dB Ard L 20 logyo 2 where L is the path loss in decibels A is the wavelength and d is the transmitter receiver distance in the same units as the wavelength 2 2 2 Effects of Multipath in Wideband Characterizing 14 The models discussed generally characterize the path loss of BAN devices taking into account possible shadowing due to the human body or obstacles near the human body and postures of human body Therefore we need to take fading effects into consideration here Multipath fading is caused by reflection off or loss through walls and objects in and around the transmitter and the receiver which ultimately change the distance power gradient Multipath fading is also called fast fading in contrast with shadow fading which is a consideration for long range outdoor radio links because the rapid instantaneous changes in the received signal power caused by fast changes in the phase of the received signal from different paths due to small movements Thus multipath has strong impac
3. Channel number Sample density Continuous model Settings Mobile speed CIA update rate Model file name Surto0h_SOrom tap Browse Figure 47 Channel Settings for Generate Simulation 75 Appendix B Zarlink Tutorial This section describes in detail the operation of the main components of the ZL70101 ADK the BSM100 and the AIM100 These features are important in Zarlink operation The ZL7010X Application Development Kit main form controls the launch of the ADK application and its components It provides revision of the software and also controls system level timing intervals for various functions described below The BSM100 and AIM100 button are used to launch their own interface System setting control the interval of timers used for various operations of the system The range of these settings is from 0 5 Hz to 10 Hz B 1 Base Station Module BSM100 HE 770101 Base Station Module BSM100 Main Form Link Setup CCA amp Cal Data Test Remote Implant HK Link Setup 400 MHz Normal Operation 400 MHz Emergency Operation 2450 MHz Wake up Company ID hex ofi TX Modulation v Country 0 USA v IMD Transceiver ID hex 0 o fo lo fo 1 RX Modulation v User Data v TX Modulation v Channel v Implant Type RX Modulation v Bytes Block v Channel v Max Blocks Packet v Bytes Block v Max Blocks Packet v Implant Information Company ID IMD ID Company Ta Sii Load hex hex Nann Implant Description
4. in IEEE 22nd International Symposium on Personal Indoor and Mobile Radio Communications 2011 Jie He Shen Li Qin Wang Kaveh Pahlavan A Testbed for Evaluation of the Effects of Multipath on Performance of TOA based Indoor Geolocation 88
5. 140 Data Figure 38 Comparison of CDF Plot between Cramer Rao lower bound and Localization Algorithm with 8 base stations The X label is localization error in mm Y label is cumulative probability 62 The above Figure 38 compared the result of 8 base stations between our localization algorithm and what of Cramer Rao lower bound Observation shows the result from CRLB is much more static with a small variation range from 80mm to 110mm However the algorithm result from 30mm to 150mm Although they are so different in ranging their average result at 0 6 cumulative probabilities confluence Therefore we explore and discuss what could be the possible reason of causing this effect The localization errors of total 1926 coordinates are evaluated point by point firstly in Figure 39 CRLB vs Algorithm for All Points 160 Algorithm CRLB 140 40 20 200 400 E00 sd 1000 1200 1400 1600 1500 2000 Figure 39 Point by point localization error of small intestine coordinates X label is the point number Y label is the localization error Blue lines are errors of our algorithm red is that of CRLB 63 In consistent with Figure 38 CRLB result has a gentle wave while the algorithm one has severe fluctuations To have a better understanding we picked two points from Figure 39 one with maximum localization error whose point number is 316 the other with minimum localization error of point number 68 Repeating
6. 35 355 Packets and RSol COMCCEON 2 wsseceici codes sdesxaatannncd a ninco en E seas 36 5341 PACKEURECCDUOM nas wees siandissraceasiudath A uteieues 37 3 3 2 Receive Signal Strength Indicator vis Arsncastvassicduebiensecsemledtawecs oat an A iiA 38 Chapter 4 Performance Evaluation of Communication ccccsecececcccsccccscecceces 39 4 1 Hardware Platform Channel Model Validation csssssssseeeccececeesseeeesseeeeceeeeeeeeeaas 39 411 Path LossAcc uracy Verification aiscccaadssccssacvins seasonervedentetanuatassactaaeeataaane eossedeeress 39 4AL2 FACING Accuracy Verlicat Oiaoi A TAAN 40 4 2 Communication Performance Evaluation Scenario eesssssssssssesseresssesssssrsrrrrereesssssessse 42 4 3 Performance Evaluation of BANS Communication s eeessssssssssseeerressssssssssererrereesssssesene 43 43 1 lPaPlant to Body SUP ACE nonse a a e NO 44 A D ZAP COND AIA aa a Site tstedaenes 45 A CONIC STON viss ceecniseealteasrasaduciasannace A aya pee eua A A O AR 46 Chapter 5 Performance Evaluation of Localization Technique csccsccscsseees 48 5 1 OVeErnview of RSS Based LOCalIZatiOn scnccisienccs tiie a E O T E hanes 48 SLLMaxm mMm LIK CHM OOO iis sick dsevasoessyocssterastanseaceuguas cau ends a AA 48 5 1 2 Cram r Rao lower DOUNOS sssesssssrsessreessrrersrsenrerresreeeserossrreeseeerreressreesseeesseresseeee 51 5 2 Empirical Indoor Localization Study csscccccssssscccccessccecceesceeseeseeceeseeseceeea
7. Generator Zarlink PCB Boards Figure 17 Assembled Hardware Platform Zarlink boards are at the bottom left PROPSim is in the middle 34 The assembled Hardware Platform is shown in Figure 17 PROPSim C8 and a monitor connected to it are in the upper part of this figure Zarlink Base Station Module and Application Implant Module are in the lower part of this figure Output from signal generator is connected to RF LO port in PROPSim lower port Zarlink Application Implant Module functions as input routing through PROPSim connected to Zarlink Base Station Module which is output A PC laptop is used to control Zarlink boards 3 2 4 Configuration of Channel Models PROPSim has provided various options of channel fading such as Nakagami Rayleigh and Gaussian fading In the submenu users are able to define Doppler types including classic round and flat According to NIST channel models Gaussian fading is applied in this hardware platform Beam Center frequency shift Standard deviation Beam B Center frequency shift Standard deviation The power ratio of the beams 4 B BO Custom Figure 18 Gaussian fading parameter configuration 35 However in PROPSim Channel Model Editor Gaussian fading configuration consisting 2 beams which are called beam A and beam B shown in Figure 19 Each of them can be customized in two aspects Beam Shift and Beam standard deviation Another parameter which is user definable i
8. Networks Section 2 2 presents the channel model characteristics and IEEE task group models for BAN Section 2 3 describes earlier work related to this thesis work performed in WPI s CWINS laboratory and in other organizations as well 2 1 Characteristics of BANs The development of wireless BAN technology started around 1995 when wireless personal area network WPAN technologies were being considered for communications on near and around the human body Later around 2001 this application of WPAN came to be identified as Body Area Network to emphasize the focus on communications on in and near the body only Body area network is expected to be the next breakthrough invention with great potential due to the rapid growth in physiological sensors low power integrated circuits and wireless communication A number of intelligent physiological sensors can be integrated into a wearable wireless body area network under computer assistance to serve for rehabilitation or early detection of medical conditions Unlike indoor and outdoor wireless communication these are many challenges for body area networks 24 such as Interoperability implementing a plug and play device with easy interaction System devices designing medical device and system System and device level security security of patient s data and resistance to tampering Invasion of privacy impact upon patients freedom will affect the acceptance of BAN
9. T 30 13 Where tr is trace For example let w n be a sample of N independent observations with unknown mean 9 and known variance o w n Ny 61 071 14 the Fisher information is a scalar given by 0 2 c 1 MD H La as And so the Cram r Rao bound is var 6 gt 16 Suppose 8 is an unknown detereministic parameter which is to be estimated from measurements x distributed according to some probability density function f x 8 The variance of any unbiased estimator of 8 is then bounded by the inverse of the Fisher information 8 1 TO 17 var gt where the Fisher information I O is defined by 52 al x 0 2 312 x 0 1 E a E aa 18 and I x 0 logf x 0 is the natural logarithm of the likelihood function and E denotes the expected value The efficiency of an unbiased estimator 0 measures how close this estimator s variance comes to this lower bound estimator efficiency is defined as 8 a 29 or the minimum possible variance for an unbiased estimator divided by its actual variance The Cram r Rao lower bound thus gives e 0 lt 1 20 A more general form of the bound can be obtained by considering an unbiased estimator T X of a function 0 of the parameter 8 Here unbiasedness is understood as starting that E T X wW in this case the bound is given by OK I 0 var T gt 21 where W is the deri
10. di di y yi _ 0 11 which are non linear equations Hence x can be solved in closed form from the above two equations in Equation 11 using a least squares LS algorithm 5 1 2 Cram r Rao lower bounds In estimation theory and statistics the Cram r Rao bound CRB or Cram r Rao lower bound CRLB named in honor of the person who were among the first to derive it Harald Cramer and Calyampudi Radhakrishna Rao 38 expresses a lower bound on the variance of estimators of a deterministic parameter The bound is also known as the Cram r Rao inequality or the information inequality In its simplest form the bound states that the variance of any unbiased estimator is at least as high as the inverse of the Fisher information An unbiased estimator which achieves this lower bound is said to be efficient Such a solution achieves the lowest possible mean squared error among all unbiased methods and is therefore the minimum variance unbiased MVU estimator However in some cases no unbiased technique exists which achieves the bound The Cram r Rao bound can also be used to bound the variance of biased estimators of given bias In some cases a biased approach can result in both a variance and a mean squared error that are below the unbiased Cram r Rao lower bound see estimator bias 51 For multivariate normal distribution x Na u 9 CC4 12 the Fisher information matrix has elements OM Oi G r Imk zol T C
11. enabling the streaming of data from the sensors to thecellphone IN real time carro E AE 10 Figure 4 View the small intestine with capsule endoscopy for patients with history of obscure gastrointestinal bleeding or abnormal small intestine found on small bowel series 10 Figure 5 NIST Path Loss Model Simulated Scenario ccccccssssecceeeeseeeceeesescceseeesceeeseuaeecessaganees 18 Figure 6 Functional Block Diagram of Hardware Platform for Performance ssscceeesseeeeees 22 Figure 7 Functional Block Diagram of Hardware Platform for Localization ccsseeceeeeeeeeeees 22 Figure 8 PROPSim C8 Wideband Multipath Simulator ssssenesssseesssseressseerssssserssssrressssrressseeress 23 Figure 9 Zarlink Advanced Development Kit cssccccsssssececcensessccuesscccusassessesseusssceseescscesscaners 24 Figure 10 Configuring Model Taps Parameters cccccccsssecceccsssececeeeecceceeeesccessuuseeessaeeceeeaganees 26 Figure 11 Simulation Editor User Graphic INterface ccccccssseeccccneseeeceeeseseceseensceeeseueeecessuaeeees 27 Figure 12 Simulator Control User Graphic INterface ccccccsssseccccessecceeeeuseceesseesceesseuasecessuaeeees 28 Figure 13 Base Station Module BSM100 with Dual Band Helical Antenna And ADP board is mounted upon BSM as power provider and controller with power switch on the lower Hent CONE re aa otsaceawaca tue svaneu son advevasientosoeatenanssal teases e
12. is always a shadow fading S influencing the accuracy of channel model that we take a separate look at random normal distribution which we generated 2000 sets of these rand numbers Then we graphed the CDF plot these normal random variables As shown in Figure 25 the x label is path loss variation rang in dB the Y label is the probability We limit the path loss variation from 13 dB to 13 dB here to make sure a 90 measurement successful rate in other words from 5 up to 95 By including such fluctuations in path loss model every TX RX distance pair will result in theoretical minimum distance Dmin and a maximum distance Dmax In 2D environment the predicted Dmax and Dmin Should looks like Figure 35 59 Base Station Figure 35 Maximum and minimum predicted distance in 2D environment Due to the fading effects even if we only have one receive signal strength we can have a range of possible transmitter and receiver distances As shown in Figure 36 in three dimensional environment the ranges looks like a ring between the maximum coverage Dmax and minimum coverage Dyin Thus this algorithm divided the localization space into three parts by these sphere R1 which is ranging inner the ring area R2 is on and inside ring ranging R3 is ranging outer the ring as shown in Figure 36 60 Location Area ee g R1 R2 Base Station Figure 36 Maximum likelihood ranging area division The orange dot stands for Base Station
13. place the localization performance is even better than bound Possible reasons were discussed and explained After combining with localization algorithm our hardware platform presents a Satisfactory performance of in body 3D RSS based localization with multiple commercial medical sensors Thus our hardware platform is helpful and can be used extensively for future works 69 Chapter 6 Conclusion and Future Work In this thesis we present a unique hardware platform for body area performance evaluation of communication and localization Typical medical implantable devices from the Zarlink Company are tested using the channel model simulated in the Elektrobit PROPSim C8 which provides a real time interference controllable and repeatable environment Base Station and implants serve respectively as transmitter and receiver The thesis utilizes the path loss model from National Institute of Standards and Technology NIST for the Medical Implant Communication Service MICS band Our hardware platform is first validated with respect of both the path loss and fading distribution simulation ensure an accurate simulation result Three modulations 2FSK high data rate 2FSK high sensitivity and 4FSK are compared in Chapter 4 Four different scenarios are categorized deep tissue implant to body surface near surface implant to body surface deep tissue implant to implant near surface implant to implant Received packets number is counted in the re
14. various MICS operations anc E BuildBsmapp101 Thi file ntains functions for arious MI perations an 4 Binaries eS Includes G Debug This copyrighted work constitutes an unpublished work create Release Gm TOP AppDevPlat Sw AnyBoard L Gm TOP ZL7010X Adk Sw AnyMezz L Gm TOP 2L7010XAdk Sw BsmMez2 H Gm TOP ZL7010X Adk Sw BsmMezz L Debug tgt T MSP430F1611 ccxml Active Ink_msp430f1611 cmd BuildImApp101 Active Debut Binaries A Includes Debug Gm TOP AppDevPlat Sw AnyBoard L Gm TOP 2L7010 Adk Sw AnyMezz L Gi TOP 2ZL7010XAdk Sw ImMez2 Ar Debug tat T MSP430F1611 ccxml Active nk_msp430F1611 cmd Copyright Zarlink Semiconductor U S Inc 2007 All rights The use of the copyright notice is intended to provide notic Semiconductor Inc owns a copyright in this unpublished work notice is not an admission that publication has occurred TE confidential proprietary information and trade secrets of z t Semiconductor Inc it may not be used reproduced or transmi or in part in any form or by any means without the prior wr of Zarlink Semiconductor Inc This work is provided on a ric subject to additional restrictions set out in the applicable other agreement include Adp General h UINT gt include Adp Build Build h UD16 9 include Adp AnyBoard SysTimerLib h StElapsedMs 20 ineclude Adk AnyMezz MicsHw h MAC CTRL inclu
15. widely concerned major applications as follows e A BAN network in place on a patient can alert the hospital about the impending onset of a heart attack by measuring changes in the patient s vital signs Figure 3 25 Figure 3 IMEC demonstrated new techniques based in Eindhoven incorporate a dongle that plugs into the SD card slot of a cellphone enabling the streaming of data from the sensors to the cellphone in real time e A BAN network capsule endoscopy on a diabetic patient can auto inject insulin though a pump as soon as the patient s insulin level declines Figure 4 26 Figure 4 View the small intestine with capsule endoscopy for patients with history of obscure gastrointestinal bleeding or abnormal small intestine found on small bowel series 10 2 2 Channel Model for BANs Many efforts has been put in consider of Body Area Channel Models 27 28 The IEEE 802 15 Task Group 6 BAN is developing a communication standard optimized for low power devices and operation on in or around the human body but not limited to humans to serve a variety of applications including medical consumer electronics personal entertainment and others IEEE 802 15 TG6 was formed in November 2007 and began operations as TG6 in January 2008 in Taipei It had received 34 proposals which were merged into a single candidate proposal for the 802 15 6 Standard To evaluate the performance of different physical layer proposals TG
16. 100 or AIM100 boards 31 Figure 15 Programmer Cable Adapter 3 1 2 4 MSP430 USB Debug Interface wis TEXAS INSTRUMENTS MSP430 USB Debug Interface MSP FET430UIF Figure 16 MSPFET430 USB Debug Interface The MSP FET430UIF is a powerful flash emulation tool to quickly begin application development on the MSP430 MCU It includes USB debugging interface used to program and debug the MSP430 in system through the JTAG interface or the pin saving Spy Bi Wire 2 wire JTAG protocol The flash memory can be erased and programmed in seconds with only a few keystrokes and since the MSP430 flash is ultra low power no external power supply is required Note the Zarlink board is not compatible with other USB devices in other words they cannot connect to the same USB network at the same time Therefore if we want to use this USB Debug Interface we must unplug Zarlink boards first and vice versa if we want to use Zarlink boards TI USB Debug Interface must be disconnected 32 3 2 Body Area Performance Evaluation and Localization Hardware Platform Design After giving a brief introduction and description of each component s function this section discusses how to setup the complete hardware platform for performance evaluation and localization purpose of Body Area Network 3 2 1 Design Consideration These are several matters have to be taken into consideration when we design this platform e First the unidirecti
17. 1500 b a Figure 43 Maximum error point intersection of location area The possible reason why localization scatter points looks larger than the red circle is because these maximum and minimum ranging rings are projection on to the YZ plane However they do have three dimensional ranges in X direction If we view it in 3D environment they will intersect similar with what was described in Figure 39 Then we take a look at the maximum error point in Figure 43 Similar to minimum error the left plot in Figure 43 shows all the eight ranges of different base stations the blue point part is the intersection The right plot Figure 43 is a closer look of the left one similar to the relationship of Figure 42 We can observe from Figure 43 that the intersection part is much but also the green circles are larger than that in Figure 42 larger because not only the red circles According to the above mentioned several reasons we presume the minimum error point is very close to one base station whereas maximum error point has almost equal distances 66 To verify our assumption we locate these points in human body The left plot from Figure 44 shows the implant location when error is very small The right plot in Figure 44 shows the implant location when error is large Simulating it inside human body in MATLAB or combining this figure with Figure 31 and Figure 32 we can tell the position in the left of Figure 44 is v
18. 6 proposed a list of frequency band and a number of available measurements on which the model can be based shown here in Table 1 Table 1 List of Frequency Band TG6 defined three types of BAN nodes for detailed study 1 Implant node A node that is placed inside the human body This could be positioned immediately below the skin or deeper inside the body tissue 11 2 Body Surface node A node that is placed on the surface of the human skin or at most 2 centimeters away from the surface 3 External node A node that is not in contact with human skin positioned from a few centimeters up to 5 meters away from the body Based on the frequency band listed on Table 1 and location of the communicating nodes IEEE 802 15 6 has summarized seven scenarios for Wireless Body Area Networks propagation model Different frequency band and channel models are evaluated widely 29 30 31 In this thesis we focus on the channel model for Medical Implant Communication Service MICS band 402 405 MHz which is CM1 and CM2 MICS band has a frequency band between 401 and 406 MHz in communication with medical implants Applications using such bandwidth are a pacemaker or other electronic implants In order to reduce the risk of interfering with other users of the same band the maximum transmit power is as low as 25 microwatt The maximum used bandwidth at any one time is 300 kHz which makes it a low bit rate system compared with WiFi or Blueto
19. 8 The development of wireless systems imposes heavy requirements on product development methods and testing Understanding the radio channel behavior is a key factor in developing wireless products that will operate properly in their intended application environments Elektrobit PROPSim C8 is a technology independent radio channel emulator supporting all major wireless standards and signal types in a broad frequency range covering established and future technologies It allows users to perform realistic and accurate radio channel emulation supporting the development of most demanding wireless applications such as beam forming 8x8 unidirectional MIMO emulation supported with single unit network level testing 220 MHz 6000 MHz frequency range and 70 MHz bandwidth People take benefits of PROPSim s own characteristics such as 24 e Controllable multi channel fading environment e Repeatable lab testing facility e Beamforming and multi user MIMO testing capability The PROPSim C8 plays a critical role in our platform performing repeatable NIST channel emulation at the 403 5 MHz center frequency It simultaneously supports simulation of eight independent RF Input and RF Output channels Figure 3 is the front panel of PROPSim C8 from which we can observe the eight unidirectional channels and 8 RFLO RF Local Inputs along with LED status indicators EB PROPSim contains a built in PC with numbers of software for channel modeling what ar
20. A Hardware Platform for Communication and Localization Performance Evaluation of Devices inside the Human Body by Shen Li A Thesis Submitted to the Faculty of the WORCESTER POLYTECHNIC INSTITUTE in partial fulfillment of the requirements for the Degree of Master of Science in Electrical and Computer Engineering May 2012 APPROVED Prof Kaveh Pahlavan Thesis Advisor Prof Fred J Looft Department Head Prof Allen Levesque Thesis Committee Prof Xinming Huang Thesis Committee Abstract Body area networks BAN is a technology gaining widespread attention for application in medical examination monitoring and emergency therapy The basic concept of BAN is monitoring a set of sensors on or inside the human body which enable transfer of vital parameters between the patient s location and the physician in charge As body area network has certain characteristics which impose new demands on performance evaluation of systems for communication and localization for medical sensors However real time performance evaluation and localization in wireless body area networks is extremely challenging due to the unfeasibility of experimenting with actual devices inside the human body Thus we see a need for a real time hardware platform and this thesis addressed this need In this thesis we introduced a unique hardware platform for performance evaluation of body area communication and in body localization This hardware platform utilizes a w
21. CC A EE NE E E EENE E E daa AT NEEE S EE AA 6 Chapter 2 Wireless Access for BAN Applications ssesesesesesesesesssssoeseoesesesssseseoe 7 PRE EEA EaSI BANS E elects O T E A E ESE 7 22 Channel WMIOGEIMTOFr BANS A A R 11 22t PAU VOSS MOEI acsussipecniazssctacanezaeins EEE masons 13 2 2 2 Effects of Multipath in Wideband Characterizing ccssccccsssssececeeeeeceeeeeeeeeees 14 Z2 SMACOW a CUI E sess niet stati doce E ATO TE E OA 15 22 A MICS band Pati LOSS IVIOGE seccaisuseetcc a ek ove adic ates aa a Gedeceeeer 16 223 PREVIOUS WW OMS sicecchics es RAs a hcunsalteahdueosoue dow e aa eaheed 19 Chapter 3 Design and Implementation of the Hardware Platform scsseee 21 3 1 Elements of Hardware Platform ccccccccccccsecceessssseeeeccccccceeueueaeesseeeceeecceeesseanaaegsees 21 3 1 1 Elektrobt PRORSINY GS yiesisdastecmestassata sat deaee torte aden acca iret acre 24 3 1 2 Zarlink ZL70101 Advanced Development Kit cccccccssssccceceessececeeeeeeeeeeeeeeeeees 29 3 2 Body Area Performance Evaluation and Localization Hardware Platform Design 33 52 Desin CONSIGOK ALON eea E new aeataeasbedeloniadtataeainesieas 33 3 2 2 Hardware Platform Implementation cccccssscccccsseceeeeeseececeeeececeeeneeeeeseeneeeeeas 33 E PASE Nic 6 E ene RE eee ern eee eer eee ee eee ere eee 34 3 2 4 Configuration of Channel Models cccccccsssseeccceessececeeeseceeeeessecesseeseeeeeauaeeeeeas
22. EGBGBOSB O CG CO GO et amp oe a tO a O 0000000000000 6666666566666 666 680 oo0000 00000 i Ua e ESO eg 660666 06 6 Y Axis Y Axis Figure 40 Minimum error point scatter plot of possible locations 64 OO0000 0 20 40 60 aQ 100 120 140 160 180 200 100 200 300 400 500 600 700 400 300 200 100 0 100 200 300 400 400 300 200 100 0 100 200 300 Figure 41 Maximum error point scatter plot of possible locations We speculate that the result could be related with target locations since the point with large errors always have bad localization performance Therefore the next step is to analyze these point location Figure 42 shows the intersection of location area of Minimum error point The red circle indicated the maximum range green circle indicate the minimum range of each base stations Eight base stations are in use in this figure The right plot of Figure 42 is a closer look of the right bottom part of the left one in Figure 42 We can see there is an very small red circle limited the predicted location area Minimum Intersetion Minimum Intersetion 1000 500 500 OO0000000 0000000 1000 1500 1500 1000 500 D 500 1000 1500 65 Figure 42 Minimum error point intersection of location area Maximum Intersetion Maximum Intersetion 1000 500 A ee iL T WTI DEDITI 500 1000 1500 200 300 100 100 200 100 500 1000 500
23. Environment 2005 M A Assad A REAL TIME LABORATORY TESTBED FOR EVALUATING LOCALIZATION PERFORMANCE OF WIFI RFID TECHNOLOGIES ECE Department WPI Worcester MA USA May 4 2007 Jie He Shen Li Qin Wang Kaveh Pahlavan 6th International Symposium on Medical Information and Communication Technology ISMICT2012 in IEEE International Conference on Communications ICC Ottawa Canada 2012 E A Lee Cyber Physical Systems Design Challenges in 11th IEEE Symposium on Object Oriented Real Time Distributed Computing ISORC Gill Radhakisan Baheti and Helen Cyber physical Systems in The Impact of Control Technology T Samad and A M Annaswamy eds 2011 NSF Workshop On Cyber Physical Systems 06 09 2008 NIST Foundations for Innovation in Cyber Physical Systems Workshop 02 08 2012 NSF Cyber Physical Systems Summit 08 01 2008 Shen Li Jie He Ruijun Fu Kaveh Pahlavan A Hardware Platform for Performance 85 23 24 25 26 27 28 29 Evaluation of In body Sensors in 6th International Symposium on Medical Information and Communication Technology ISMICT2012 La Jolla 2012 Jie He Kaveh Pahlavan Shen Li Qin Wang A Testbed for Evaluation of the Effects of Multipath on Performance of TOA based Indoor Geolocation IEEE transactions on instrumentation and measurement 2012 Body Area Networks Online Available http en wikipedia org wiki Body_area_
24. Firmware Elprotronic FET Pro430 This software Figure 50 helps to program the firmware on the boards included with the ZL70101 Application Development Kit To use the Elprotronic FET Pro430 follow these steps e Connect the MSP FET430UIF to the PC via USB e Connect the ADP board to the PC via USB 79 e Turn on the power switch on the ADP board or if the power switch is already on press the reset button on the ADP board e Under Microcontroller Type select the MSP430F1611 e Click Open Code File and browse to the firmware file for the target board a BsmApp101 txt for the ZL70101 base station mezzanine board b BsmApp102 txt for the ZL70102 base station mezzanine board c ImApp101 txt for the ZL70101 implant mezzanine board d ImApp102 txt for the ZL70102 implant mezzanine board e AdpApp txt for the ADP board e Under Device Action check Reload Code File click AUTO PROG This software tool is very useful when we need to reload the original Zarlink source code especially when users modified code of Zarlink board in a wrong way and lose control of it FET Pro430 helps to reset all the default settings of Zarlink When performing AUTO PROG auto progress it will check all the functions displays under that button i e Verify Security Fuse etc If progress successful a green indicator will appear otherwise a red fork will appear At the bottom of this software is the progress ba
25. ND 500 00 550 AND _ 1000 650 1200 700 1400 750 100 50 0 50 100 50 100 1600 e Base Station mo woo o woo mM 9 Implant Figure 31 Simulated Body Meshes Left is human body muscles with base station indicated by circles right is small intestine with dot indicate where the implant is 56 Figure 32 shows the overview of our scenario by putting small intestine into human body We can observe that the small intestine is in the lower part of abdomen The implant location is varying all along with every point of small intestine The coordinates of these sensors are got from the MATLAB cursor function Aiming at the place we want to attach these patch antennas read the coordinates and record it for localization The method for placing implant in small intestine is similar 200 200 400 600 500 1000 1200 1400 1600 af 3a Ff FF f F F 200 0 200 200 100 0 100 200 300 Figure 32 The human body meshes with small intestine in abdomen The green part is where the small intestine locates the black meshes are human body Red circles are sensor locations 5 4 RSS fitting for BANs For RSS based localization one of the most important thing is getting the accurate receive signal strength from device under test In this thesis Zarlink ADK includes a function reading RSSI from Implant boards The RSSI is defined by Zarlink Company Here we analyze the 57 RSS versus RSSI on their fitting aspect Figure 33 give
26. RSS is derived from Zarlink user manual in dashed line 4 1 2 Fading Accuracy Verification Further verification of fading properties is done in this part By connect Network Analyzer Agilent Technologies E8363B to PROPSim as transmitter and receiver respectively simulated channel propagation through platform is observed Gaussian fading defined in 40 Table 3 was added in our hardware platform using PROPSim C8 Due to fading effects path loss varies dramatically at any fixed range The implant to on body surface channel model is presented as an example After collecting more than 500 path loss values at each TX RX distance a curve fitting plot of path loss was generated as shown in Figure 23 is the scatter plot from the NIST results 35 120 w r E O ON E DOA IA ON VAE R 110 100 wal i l ee ce cee ce oe S Oe ae A T T 1 oe O A O a Swe E O P ETT n a ee ee ee o 7 ae a ee uj a na e e e e E a AEE E Peete ee oli i Within 20mm of the skin o sie ce aed an ee ernie poor l I See eee eee ee i i i L 0 50 100 150 200 250 300 380 450 SN range mm 5 ee i NEAR Beneat PENT VAA l ane EEE T seven a ae a 0 50 100 150 200 250 300 350 400 460 4500 Range mm Figure 23 Comparison of Path Loss vs Distance Scatter Plot Between Our Platform Simulation and NIST Statistical Model Left is our result from Network Analyzer right is NIST result used for generating statistical model T
27. Red and blue ring intersecting with each other are the possible ranging area of two base stations respectively The maximum distance Dmax and minimum distance Dmin are derived from the path loss model and RSSI value If we have these two base stations merely then the location of transmitter is predicted to be in the insect part of R2 Figure 37 give a quick glance of the predicted location White meshes in this figure indicate the intersection of two rings If we add another sensor as base station very likely the third ring will only intersect with one of the white mesh By averaging the possible location coordinates as predicted location an extra base staion could increase the localization accuracy in a great deal In this thesis we will use up to eight base stations for simulation to pursue a better result Then the centroid result i e the average of all intersected coordinates are predicted as the transmit source 61 Figure 37 White mesh are the bound of where the possible source reside 5 6 Performance Evaluation of RSS Based localization Algorithm In this part we evaluate the performance and results of our localization algorithm First we compare it with Cram r Rao lower bound since the latter one is an important metric for valuation algorithm s accuracy and applicability CDF Plot _ ta CRLB Algorithm Cumulative probability Se B 2 io S oe tS a to pan oO oo F 20 40 BO o0 100 120
28. SK high sensitivity is 2FSK high rate and 4FSK also attain good performance Moreover approaching 100 at 50 mm which is significantly better than seen for the implant to even 4FSK can reach up to 200 mm body surface case discussed previously These results indicate that the implant to implant body area communication has better link quality than implant to body surface channel 45 WLLL foxy D oo 8 RSs E WM BHH aaaeeeaa RESESSIE SEEE EEE ESEESE Ema MMMM TOMA PRA a A A A A A A A A A A A A A A A A A A A A A A A A A A A A A S T POPPA SRR RY ERREA EE KRKRKKKKRKKKKKKKKKKKKKKKKKKKKKKKKKKKKKN CI BSS II I II MOI LI WLLL LLL eed 2FSK high rate S899 2F SK high sensitivity E i J4FSK Deep Tissue Implant to Body Surface Modulation vs Link Quality 29 Oo FY OO HM QM N a 9 aJeY UoNdaday jayse4 MMMM og 2FSK high rate biti J4FSk Ray PHP Metatetetetetetetetete s Bn WLM OC ETETETT ETETETT ETET ETETETT ETETETT ETS ERA B LAI 5 d w MMMM TERE WEEE EEEE EEEE EEEE AEAEE EAIA ATAATA ATATA ATAATA TAA G O EI IO OOO PSIE Oo Distance mm BOO OOOO OOO OOO OOO OOOO OOOO OOOO CI EKER OJ LI Near Surface Implant to Body Surface Modulation vs Link Quality co Li wo Ww t mo oO q 5 5 5 5 o a 5 5 9 aey uondasay Jay3ed Distance mm Figure 27 Implant to Implant Modulation vs Link Quality On the left is Near Surface Implant on the rig
29. a into three dimensional rings the intersection of multiple rings provides the estimated location of the implant By comparing the observed localization error with Cram r Rao Lower Bound we find that the algorithm yields accurate localization performance Te PROPSim C8 L de oe 3 5 Ra EEr prms aad mae 2 res po Ses se arlink Boards _ 2 Figure 1 Major Components of Hardware Platform The eight channel wirdeband multipath simulator with an monitor is Elektrobit PROPSim C8 Two boards are Zarlink base station and implant boards 1 4 Thesis Outline This remainder thesis is organized as follows Chapter 2 defines Body Area Networks and provides background information on BANs communication and localization Chapter 3 provides a detailed introduction of our hardware platform and our study undertaken to prove its validity Chapter 4 presents link quality and comparisons for different scenarios to evaluate the communication performance of Body Area Networks Chapter 5 discusses localization scenario geolocation algorithm and evaluates localization error as a function of in body sensor location Chapter 6 summarizes our conclusions Appendices A and B demonstrated technical information on the PROPSim and Zarlink tools respectively Chapter 2 Wireless Access for BAN Applications In this chapter we discuss wireless access for BAN applications Section 2 1 provides a brief overview of Body Area
30. ading because the variations are much slower with distance than another fading phenomenon caused by multipath It is also found that shadow fading has less dependence on the frequency of operation than multipath fading or fast fading as discussed later The path loss Eq 1 will have to be modified to include this effect by adding a random component as follows Lp Lo 10a logio d X 3 Here X is a random variable with a distribution that depends on the fading component Several measurements and simulations indicate that this variation can be expressed as a log normally distributed random variable A log normal absolute fading component ends up as a zero mean Gaussian fading component when expressed in decibels In Body Area Network much empirical measurements were undertaken regarding the fading effects and characteristics 34 2 2 4 MICS band Path Loss Model An accurate verified propagation model is essential for BANs studies Sayrafian et al 35 at the National Institute of Standards and Technology NIST introduced a path loss model as shown in Equation 2 with center frequency 403 5 MHz modeling the human body using different dielectric parameters for different organs in simulating a 3D full wave electromagnetic field in HFSS PL represents path loss d is the distance between the transmitter and receiver S is 16 shadow fading which subjects to Gaussian distribution here The simulation modeled four near surface i
31. adio wave front is obstructed by an opaque obstacle and losses caused by other phenomena The signal radiated by a transmitter may also travel simultaneously along many and different paths to a receiver this effect is called multipath Multipath waves combine at the receiver antenna resulting in a received signal that may vary widely depending on the distribution of the intensity and relative propagation time of the waves and bandwidth of the transmitted signal The total power of interfering waves in a Rayleigh fading scenario can vary quickly as a function of distance which is known as small scale fading Small scale fading refers to the rapid changes in radio signal amplitude in a short period of time or propagation distance In the study of wireless communications path is usually characterized by the path loss exponent whose value is normally in the range of 2 to 4 where 2 corresponds to propagation in 13 free space 4 is for relatively lossy environments and for the case of full specular reflection from the earth s surface the so called flat earth model In some environments such as buildings stadiums and other indoor environments the path loss exponent can reach values in the range of 4 to 6 On the other hand a tunnel may act as a waveguide resulting in a path loss exponent less than 2 Path loss is usually expressed in dB In its simplest form the path loss can be calculated using the formula Lp Lo 10alog
32. algorithm several times we can observe that on small intestine point with minimum error always has small localization errors point with maximum error always has large localization errors According to Figure 39 the intersecting principle we projected the possible locations onto the YZ plane to look for clues of probable reasons Figure 40 is the minimum error point scatter plot of possible locations Compared to Figure 44 which is maximum error point scatter plots we can observe that Figure 40 has sparse distribution and smaller location areas which ranges about 200 mm in Y direction and 250 mm in Z direction However in Figure 41 points distributed very closely ranging about 700 mm in Y direction and 550 mm in Y directions Minimum Error on YZ Plane Minimum Error Localtion in Z Plane 520 450 6606066060006 006 540 oo ty oe AP O O O O a O Cre dh 0G eo amp 66 o BO 500 To Coe 2 oe amp 8 ot FOGD aat ene aah une yes at ag RESIS SER ERIE Sg e SOkrooo000000 0000 000 SeePeeaeaerrueree Ses PEECECLECOSLLELDSD mC SUS ESEER ERR SER GES el epee Peek papane a BEESSESERSeesee Eee se cB Be Sh Bik Bee Be ie te ee Be te a x SRESSSRERSRISSSR ES SS q A AURA AR SE HORE SA ER Ee SAA OSE V S a i S amp OG oTa OG OO OGG OG amp oO a amp M690 0000060000000 000 e9999999999999999908 pi mbedek pe f pata bege bet bene Aa Girit il r r E 12S eee ee Oe ee T a Oo6O056565656 6 HHKSEHKEHSKGESBS 660 5555 0 0 6 66 06000 7006 OG COCOSOGGHOGGHOS
33. ar there is no widely accepted model for wireless propagation in the human body though various studies have been made for localization and communication applications such as 4 5 Several simulation software tools were developed for BANs communication purposes Although these tools e g HFSS from Ansoft Corporation 6 and FDTD simulated in MATLAB 7 are very useful for accurately implementing channel models for signal propagation inside the human body these software tools limitations is they cannot be put to use for connecting real world medical devices which are necessary prerequisites to real world field measurement For such purposes we have developed a real time emulation hardware platform For the device under test DUT we chose the Zarlink ZL70101 Application Development Kit ADK and for multipath channel emulation we chose the Elektrobit PROPSim C8 system 8 which provides a repeatable controllable body area propagation environment Previous work on indoor communication and localization performance testbed using the PROPSim C8 is discussed in 9 10 11 and 12 In Leon T Metreaud s master thesis 13 a real time performance evaluation testbed for wireless local area networks WLAN is introduced Azimuth Systems 801W was used to provide isolated environment and capture packets by a software protocol analyzer Wild Packets Airopeek NX Elektrobit PROPSim C8 multipath channel simulator was used to provid
34. ceeceseeseeeeeeeseeeeees 49 Figure 29 RITEM based location area GiVISION cccccccccssssecceceesueeesececeesauaesseeeeesssaaaseeseeeeuanens 55 Figure 30 Maximum likelihood centroid algorithm with 4 estimated distances cccsseeeeeees 55 Figure 31 Simulated Body Meshes Left is human body muscles with base station indicated by circles right is small intestine with dot indicate where the implant is cceceeeeees 56 Figure 32 The human body meshes with small intestine in abdomen The green part is where the small intestine locates the black meshes are human body Red circles are sensor OCA ON aean hed varedilaneetiowles aan clocunie E TE O E 57 Figure 33 Relationship between Receive Signal Strength and Receive Signal Strength Indicator 58 Figure 34 Cumulative Probability Plot CDF of random Gaussian fading added in our path loss model X axis is the varying range of path loss in dB Y axis is corresponding probability lin this figure the probability 1S 90 meiss Snin a A E AN 59 Figure 35 Maximum and minimum predicted distance in 2D environment sssssssssersssseressseeee 60 Figure 36 Maximum likelihood ranging area division The orange dot stands for Base Station Red and blue ring intersecting with each other are the possible ranging area of two base STALIONS ES DECLIVELY cironi unana aa awed A aeaueewaaiestee 61 Figure 37 White mesh are the bound of where the possible Source reside sssseeceeeseeee
35. ceiver side to analyze packet reception rate under these different propagation scenarios and different modulations By varying the transmitter receiver distance under each condition we can observe from Chapter 4 that modulation with lower data rate has higher packet reception rate in other words better communication performance Moreover implant to implant channels exhibit better link quality than implant to body surface channels These results take multiple considerations into account which are intuitive Therefore our hardware platform can help provide valuable insight into how BANs will perform and how to implement devices in wireless BANs incorporating medical implants Analysis of body area localization is based on receive signal strength RSS The RSS simulation validation is conducted by fitting simulated RSS and theoretical RSS values We can 70 observe our hardware platform provides accurate performing RSS simulation Then we introduced an improved algorithm based on maximum likelihood This algorithm was first simulated in MATLAB along the propagation path of the small intestine Then the observed error was analyzed and compared with the Cram r Rao lower bound Possible causes of the observed error performance are discussed in the thesis Finally a number of simulations are performed with our hardware platform and results are analyzed by our algorithm These results show that our hardware platform and algorithm present satisfac
36. computation communication and control is crucial for future technology developments Starting in late 2006 the US National Science Foundation NSF and other United States federal agencies sponsored several workshops on cyber physical systems 19 20 21 The NSF has identified cyber physical systems as a key area of research In BANs link quality can be affected by many factors including transmission power packet size maximum retransmission times signal modulation scheme and so on In this thesis by using the NIST path loss models we created a hardware platform is that can be repeated and expanded on by others We focus on the impact of the choice of modulation scheme on link quality by calculating packet reception rate PRR observed as a function of modulation choice and path distance RSS based body area localization results are also simulated and analyzed in this thesis 1 2 Motivation This work was motivated by the need for accurate performance evaluation of communication and localization in Body Area Networks Understanding how the body area channel propagates is very important and challenging As we know the IEEE 802 15 6 is focusing on wireless BANs body area networks However to today there is no widely accepted model for wireless propagation in body area network Various studies have been made upon this problem but in body measurement and experiment are extremely difficult 1 3 Contributions of the Thesis hav
37. d inside the human body and a body mounted sensor We show how link quality is analyzed by observing the packet reception rate under three different transmission alternatives 8 Body Area Network BAN wireless body area network WBAN and body sensor network BSN are terms used to describe the application of wearable computing devices In the view of IEEE802 15 6 standardization group Body Area Networks should ee Provide communication links in and around the body Allow communications between sensors actuators and processing element Employ a hub allowing nodes to be simpler have a longer life and be less costly EEG VHON a HEARING POSITIONING A o oe i l T GLUCOSE LL BLOOD PRESSURE eae s A e a O J DNA Ir i He PROTEIN 7 CELLULAR r ee LL E z amp TSSEA _J WLAN IMPLANTS Figure 2 Body area network applications include the management of chronic disease medical diagnostic home monitoring biometrics and sports and fitness tracking This will enable wireless communication between several miniaturized body sensor units BSUs and a single body central unit BCU worn at the human body The applications of Body Area Networks include many typical applications such as EEG ECG blood pressure and positioning shown in Figure 2 provide services regarding medical sports and entertainment function Here we stressed two
38. dB simulator will be damaged 7 Constantly check the Status LEDs all the time during the experiment The number 10 LED is POWER indication system status If it is RED there is ERROR We will need to shut down and restart system The number 18 LED is STATUS If it is RED it means the input signal is too high We must stop simulating and disconnect circuits as soon as possible The number 19 LED is also called STATUS it shows RFU Radio Frequency Units status This kind of LED on the top of 72 eight RF IN port works alike LED 18 If the input signal is too high it becomes RED we must prevent this happening If it happens disconnect and stop simulating immediately After checking over the above seven rules we can now start using PROPSim in a safe condition In this thesis because NIST path loss model need only one channel we select Uni Channel Model Then click Next New Model Generation Wizard Step 1 Select Model Type om Uni Channel Model Oection of arrival Corelative MIMO Model Instant Help The selection of Uni Channel Model will create model with only one channel The selection of Direction of ariwal will force this wizard into the mode where multichannel model is calculated trom actual geometnc constellation Nest step is to define antenna array geometry The selection of Correlate will start the actual editing of the model immediately Corelatve selection will start th
39. de Adk AnyMezz MicsLib h public include for Mic 5 t E console 23 CS 7 0 82 Problems 3 C Build BuilldBsmApp 101 0 errors 0 warnings 0 infos Description lt Read Only Smart Insert Figure 51 Code Composer Studio version 4 User Graphic 3 1 2 5 Other Subsidiary Software Software named HyperSnap 7 and Button Wizard is also used in this thesis HyperSnap supports snapshot screen contents from user defined region and convert captured 81 images into word file This is useful for recording Receive Signal Strength Indicator RSSI variation Button Wizard is used to assist HyperSnap by writing and running a Script imitating mouse movement and clicking to repeat snapshot action 3 1 2 5 Other Subsidiary Hardware Other hardware components such as Signal Generator Spectrum Analyzer Network Analyzer Variable Attenuation Power Splitter Circulator PC connecting with Zarlink Advanced Development Kit and several cables are used in this thesis Signal generator is used to support RF local input Soectrum Analyzer is used to read receive signal strength network analyzer is used to collect received signal and analyze fading properties Variable attenuation is used to protect PROPSim and simulate distance change Power splitter and circulator help to fulfill circuit functions 82 References 1 2 3 4 5 6 Sofia Najwa Ramli Rabiah Ahmad Surveying t
40. e a controllable repeatable WLAN propagation environment Statistical data characterizing performance such as data rate and Received Signal Strength RSS were collected Metreaud analyzed average throughput and instantaneous throughput variation using different channels models developed under IEEE 802 11 b and 802 11 standardization activities Mohammad Heidari 14 provided a testbed for performance evaluation of indoor geolocation systems Heidari focused on the Received Signal Strength RSS based localization method along with fingerprinting for indoor environments The Ekahau indoor positioning engine was used for performance evaluation while the Elektrobit PROPSim C8 served as channel model simulator Comparison between the simulated results of the performance evaluation of the positioning engine and the real time performance evaluation of the positioning system is analyzed in this thesis Primitive error in terms of distance error is also performed Another Master s thesis from Muhammad Ali Assad 15 discussed a testbed for evaluating localization performance of WiFi RFID technology That testbed consisted of Elektrobit PROPSim C8 RF channel simulator for multipath characteristics several WiFi 802 11 access points and commercial RFID tags Assad compared the performance of the modified IEEE 802 11 channel model and the Ray Tracing channel model Ray tracing software was also been used to evaluate the performance of tw
41. e contributed in two accepted conference papers and one impending journal article This thesis is based on the work in 22 16 and 23 This thesis introduces an interference controllable repeatable real time hardware platform consisting detailed implementation validation and result analysis This hardware platform using the Elektrobit PROPSim C8 channel emulator and the Zarlink ZL70101 Advanced Development Kit ADK simulating a body area network using a statistical path loss model developed by the National Institute of Standards and Technology NIST evaluating performance of a typical in body sensor chipset Link quality is evaluated by observing the packet reception rate PRR under three different transmission alternatives and localization accuracy is evaluated using a 3 dimensional localization algorithm The relationship between packet reception rate and modulation under different path loss models and distances are observed The modulations evaluated are binary frequency shift keying BFSK and two quadrature frequency shift keying QFSK differing in raw data rate Packet reception rate shows BFSK rate has better performance and can attain good performance in longer distance This provides a metric for how to choose modulation when performing applications for body area networks RSS based localization is performed with multiple base stations on the surface of human body and one deep tissue implant By dividing localization are
42. e editor in the mode where spatial characteristics of the channel are defined by determining correlation ELUA The selection of MIMO Model will create a MIMO Model Figure 45 Channel Model Types Selection Users are able to customize center frequency channel power delay profile parameters including mean amplitude level delay and phases of multipath components fading types and other parameters in Chanel Model Editor 73 Model Taps Carrier Wave OF Center frequency Wavelength 4045 AIE Sampling 1 Sample density samples half way Sample distance Simulation Length Mobile speed Estimated simulation time 11760906 355 4 EE Number of impulse responses O0000 Number of wavelengths 1562 5 Model Generation Minimize hardware usage Balance delay accuracy and hardware usage OF Maximize delay accuracy Id Continuous model Advanced Parameters Figure 46 Configuring Channel Model In accordance with channel model editor described previously in Chapter 3 1 1 the center frequency is 403 5 MHz Average input level is set to 1 dBm The tap file we just generated in Channel Model Editor is loaded into Simulation Editor as channel model connecting with input and output When the Simulation Editor has finished configuration a smu file along with sim and ir files should be generated These files are used for performing simulation 74 Information
43. e mainly used in this thesis will be described in 3 1 1 2 3 1 1 1 PROPSim Channel Model Editor PROPSim provides a menu of model types If the channel model includes only one channel i e one tap then it is the so called Uni Channel Model If the model has two or more channels taps the model is multi channel model which has Direction of Arrival Correlative and MIMO Channel as subclasses Because we are testing MICS bands the center frequency is set to 403 5 MHz In the Model Generation we must choose maximize delay accuracy and select the Continuous simulation model Mobile speed is set at a very low value and Doppler is negligible 25 Channel Impulse Response Model Parameters Model Taps Delay ng Mean Amplitude dE Fading 0 00 0 00 Gaussian Tap 1 Properties oF Delay Description Sten E f Random hopping Sinusoidal Linear Doppler Spread Phase Shift FI Mean amplitude level a Fading Nakagami Pure Doppler Lognormal O Suzuki gt Gaussian O Custom Figure 10 Configuring Model Taps Parameters In Figure 10 Taps means channel impulse response If a channel model contains multiple taps then it behaves multipath effects For each tap in PROPSim users are able to define delay types mean amplitude level fading types and parameters For NIST path loss model one zero mean amplitude tap without delay is created The fading type should be Gaussian in accorda
44. e solid line fitting from these dots RSS is derived from Zarlink user manual in CAS SONI stossaris Noss pasta anaa lela uva yay cdot naan sat ebivaa tna eausion acanes aes eg tease ees Gaeaanes 40 Figure 23 Comparison of Path Loss vs Distance Scatter Plot Between Our Platform Simulation and NIST Statistical Model Left is our result from Network Analyzer right is NIST result used for generating Statistical model cc ceseccccesseececeeeeseeceeeeesceesseeseececseeeeceesaueeeeeeas 41 Figure 24 Comparison of Probability Distribution Function Plot Between Our Platform Measurements and NIST Statistical Model Left is our result from Network Analyzer right is NIST result used for generating statistical MOCEl ccccessseeecceeeeeceeeeeseeeeees 42 VII Figure 25 Simulation Scenarios of Different Sensor Nodes Locations Include two body surface implant on arm and chest and two deep tissue implant located lower and upper the STOMAC PRR nee ne Pree ee RS AT Se eee er 43 Figure 26 Implant to Body Surface Modulation vs Link Quality On the left is Near Surface Implant on the right is Deep Tissue IMplant cc ceeccccccccssseescccessseeseeceeesseeeeeeeeeeeeeas 45 Figure 27 Implant to Implant Modulation vs Link Quality On the left is Near Surface Implant on the FINS Deep WSSUS TIMP ANU ensis anna ee Belton ata awaits eese ees 46 Figure 28 Illustration of a simple scenario for wireless localization csesece
45. eal time performance for in body and on body wireless networks The most important feature of our platform is that we use hardware components which we are able to connect to real medical devices thus largely reproducing medical device performance as in an actual in body application The remainder of this chapter comprises three main sections Section 3 1 provides an overview of the block diagram and major elements of our hardware platform Section 3 2 presents detailed consideration of our hardware platform design Section 3 3 discusses our method of gathering data packets in simulation experiments 3 1 Elements of Hardware Platform Here we describe the major components of our hardware separately We begin by introducing the working principle of our hardware platform 3 1 1 Block Diagram 21 Variable Attenuator Count Packets Number PROPSim C8 Simulator Zarlink Base Staion Zarlink Implant Transmit Packets Receive Packets Simulate Channel Models Figure 6 Functional Block Diagram of Hardware Platform for Performance Figure 6 is the block diagram of our hardware platform Zarlink Implant Board AIM100 is served as transmitter Zarlink Base Station Board BSM100 is served as receiver As the channel is unidirectional a PROPSim C8 Wideband Multipath Channel Simulator is connected between the implant board and the base station simulating body area channel models To simulate distance changes
46. eeeeeeeseees 54 5 3 Body Area Localization Scenario seere sive ssgats nega sand E anu sioudentsgiadanioeaaceenosaoanes 56 SARS HUNE TOR BA ING scstes once cules acon annes sana se asocansneanmca A A A A 57 5 5 Body Area RSS Based Localization Algorithm cccccssecccceesseeceeeessecceeaeeeceeeseeneeeeseees 58 5 6 Performance Evaluation of RSS Based localization Algorithm ceeccccsssseeeeeeeeeees 62 5 7 Performance Evaluation of RSS Based Localization for Hardware Platform 68 Be CONC VUS ION ss 2ccetesesais Silo we ccncoeivocda E A bose aoe 69 Chapter 6 Conclusion and Future WTK csccccsccscsccscsccscsccccscccceccccecsccssesceces 70 Appendix A PROPSIM Tutorial ccscscsscsccscsccsceccccecccceccsceccsceccececcsceccscescssesens 72 Appendix B Zarlink TUtorial ccscsscscsscsccscscccceccccecccceccscescsceccscecceccccscescusecens 76 RETEFENCES iinan E EE 83 VI List of Figures Figure 1 Major Components of Hardware Platform The eight channel wirdeband multipath simulator with an monitor is Elektrobit PROPSim C8 Two boards are Zarlink base SEALIOM ANG IMDIANE DOAN Sras E A 6 Figure 2 Body area network applications include the management of chronic disease medical diagnostic home monitoring biometrics and sports and fitness tracking 0060 9 Figure 3 IMEC demonstrated new techniques based in Eindhoven incorporate a dongle that plugs into the SD card slot of a cellphone
47. ees 62 Figure 38 Comparison of CDF Plot between Cramer Rao lower bound and Localization Algorithm with 8 base stations The X label is localization error in mm Y label is cumulative Probabili eTa E wala aati el weeie eee 62 Figure 39 Point by point localization error of small intestine coordinates X label is the point number Y label is the localization error Blue lines are errors of our algorithm red is Mator CRUB xt kccseatennncsatacohautecemciena ie devapancnbass N 63 Figure 40 Minimum error point scatter plot of possible locations ccceeeeecccecsseeseeeeeesseeeeees 64 Figure 41 Maximum error point scatter plot of possible lOCAtIONS ccsseseccceesseseeseeeeesseeeeees 65 Figure 42 Minimum error point intersection Of location area eeccecccccsssesseceeeeceeesseeeeeeessenees 66 Figure 43 Maximum error point intersection Of location area ceeccecccecssseseseeeeecaaeseeeeeeeseenees 66 Figure 44 Maximum and minimum error point location on small intestine The green mesh is small intestine of typical human body red round is the implant location 006 67 Figure 45 Channel Model Types Selection cccccsssecccecssececeeesecceceesececseaeeceesueneeceeeenseceeseaggess 73 Figure 46 Configuring Channel Model secstecieisdivdssS cciue Sa hasntiileseited iessassacdssaaisneeseausabavaiaeebives 74 Figure 47 Channel Settings for Generate Simulation cccccccccccsesseeeecceceaaeseecee
48. er comes out of the output A users is able to vary attenuation at any time being sure to select the Apply button whenever making a change When finished configuring the user selects the Run Simulation button at the menu bar while being aware of any warning or errors appearing in the status window If an abnormal state occurs the user stops or pauses the simulation immediately and refers to the troubleshooting part of the Manual Please see Appendix for more manipulation MASurtoOnOnePath Simulator Control Input Output Information Channel number Model gain Total channel gair Source file name Control file name Sample density Continuous model Settings Mobile speed CIR update rate Initializing hardware Control bus test Passec Memory te T Digital c nd routing test Failed ABB loop test Passed Connected Figure 12 Simulator Control User Graphic Interface 28 I0 x 20 1 fils 59 2 As Surtoln_5 mm tap Surtoln_B mm_ sim 64 00 3 1 2 Zarlink ZL70101 Advanced Development Kit The ZL70101 Advanced Development Kit is the medical wireless product medical implantable RF transceivers from Zarlink Company It is a typical high reliable ultra low power solution for implanted ingested and sensor applications and external monitoring with programming equipment The applications includes pacemakers implantable cardioverter defibrillators ICDs neurostimulators Implan
49. ery close to a base station which locates on the bottom right of human inner back however the position in the right of Figure 44 is in the middle of human body Minimum Location Error Point Maximum Location Error Point Nese fae J aay he 100 50 0 50 100 50 0 a b Figure 44 Maximum and minimum error point location on small intestine The green mesh is small intestine of typical human body red round is the implant location As aresult we can conclude in our algorithm implant positions have strong impact on our algorithm performance When the implant is close to base station we can get even better localization performance than Cramer Rao lower bound The possible reason could be that location area is limited for a small ranging However Cramer Rao lower bound takes whole human body into consideration to derive the bounds It is intuitive that reducing sample areas will decrease the error 67 Another reason could be that we only take part of Gaussian fading into account when calculating path loss As shown in Error Reference source not found apart from all Gaussian fading only the part which has over 90 possibility is studied This will sacrifice the probability of successful calculation but will increase the localization performance as the shadow fading effects are restricted 5 7 Performance Evaluation of RSS Based Localization for Hardware Platform This part we will evaluate the localization per
50. fading model Therefore we begin by assessing the RSSI path loss accuracy of our platform and then simulated fading model is compared with NIST channel model 4 1 1 Path Loss Accuracy Verification Here we analyze the accuracy of path loss simulating A static channel model was under test in our platform because we don t want fading effects influence the result After varying path loss using platform they are verified through spectrum analyzer and converted to RSSI according to Zarlink user manual 36 Recording every corresponding measured discrete RSSI values from Zarlink graphic user interface a scatter plot of RSSI can be drawn and fitted to a line The relationships between derived receive signal strength RSS and RSSI line fitting from real collected data are graphed in Figure 22 Zarlink User Manual shows each step between every two continuous RSSI values differs in 2 5 dB Since static model is used we can observe RSSI has 39 small variation Although there are still slight differences we judge that they nearly coincide Thus we concluded that our platform is sufficiently accurate in path loss aspect 32 a 20 RSSI 4 4 Receme Signal Strength vs RSSl P o AA A an er r i 25 20 15 10 Receive Signal Strength dB Figure 22 Relationship between RSS and RSSI RSSI are real received power in discrete dots RSSI line is the solid line fitting from these dots
51. formance from real time hardware platform using RSSI values read from Zarlink user graphic interface Four base stations are used in this case A one set of sample RSSI values looks like the followings Table 5 Sample RSSI localization values from hardware platform Every time we pick one of the RSSI values from our observation convert them to receive signal strength applying our algorithm to do a RSS based localization For the sample location we get the standard deviation of error from our hardware platform 68 44 7863 mm which is a satisfactory performance result for in body localization 5 8 Conclusion This chapter described a real time hardware platform developed for three dimensional RSS based localization performance evaluations of BANs Our unique hardware platform includes multiple typical medical sensors of Zarlink base stations as receivers and a Zarlink Implant as transmit source First of all we validate the received signal strength accuracy after simulation through our hardware platform Then we introduce our improved maximum likelihood localization algorithm for the 3D ring and intersection area The centroid data from our algorithm predicts the source position With that our software and hardware simulation scenarios are presented We compare the results from our localization algorithm with the Cram r Rao lower bound The result indicates our algorithm presents a satisfactory performance however in some
52. gated over a large area We use maximum likelihood localization algorithm MLE based on the concentration readings at the sensor nodes Direct Triangulation algorithm is used to estimate the location of source The effect of the estimation error with different sensor number and different back ground noise is researched by simulations The direct triangulation algorithm is simple and intuitionistic the MLE algorithm is robust to the much noise compared to the Direct Triangulation algorithm The simulation results show the performance of the two algorithms that we can get accurate position of the contaminant source using the two algorithms if the sensor nodes reach to appropriate numbers in the field In the absence of NLOS bias i e b 0 for all i the conditional probability density function PDF of d in can be expressed as follows 37 P Tiki exp SY 8 eee ee ee JCM exp 3 9 where j a o a 10 Then the ML solution for x is the one that maximized P d x i e arg max P d x Note that solving for x requires a traverse over all possible maximum likelihood locations which requires intensive computationally work 50 For the special case of gf g for all i the maximum likelihood solution in 2 arg max P d x is equivalent to minimizing J In order to find the minimum value of J the gradient of J with respect to x is equated to zero yielding N i di x xi _ 0 aon 11 N
53. h power switch on the lower right corner 3 1 2 2 Base Station Mezzanine BSM100 board The Base Station Module BSM100 shown in Figure 11 is composed of the Base Station Mezzanine board mated to the Application Development Platform board ADP100 Also included is the Dual Band Helical Antenna for operation in the 2 45 GHz ISM band and the 400 MHz MICS band On the Base Station Mezzanine socket can connect to the MICS Test Adapter MTA100 for viewing key ZL7010X analog and digital signals 30 3 1 2 3 Applications Implant Mezzanine AIM100 board The Application Implant Module AIM100 is comprised of the Application Implant Mezzanine board mated to the Application Development Platform board ADP100 Also included is the Dual Band Printed Loop Antenna for operation in the 2 45 GHz ISM band and the 400 MHz MICS band and a MICS Test Adapter MTA100 Mating Connector Figure 14 Base Station Module BSM100 with Dual Band Printed Loop Antenna An ADP board is mounted in the other side upon AIM board as power provider and controller with power switch on the lower right corner 3 1 2 3 Programmer Cable Adapter PCA100 The Programmer Cable Adapter PCA100 connects between all ZL7010X ADK boards that have the MSP430 micro controller on board and the Texas Instruments MSP430 USB Debug Interface MSP FET430UIF 10 pin ribbon cable The PLA100 allows users to download code and run the debugger to implement and test new features on the BSM
54. hankfulness to Professor Kaveh Pahlavan for all the knowledge he has taught me as well as the wisdom of life he endowed on me over the two years Without his enlightening instruction impressive kindness and patience could not have accomplished my thesis shall extend my thanks to my committee members Professor Fred Looft Professor Allen Levesque and Professor Xinming Huang for their invaluable assistance valuable comments and reviewing of this thesis would like to express my gratitude to all colleagues in CWINS lab especially Jie He for his patience and valuable guidance in every stage of the writing of this thesis He has been more than a colleague but a co advisor to me owe much of my thesis to him Last but not least must thank my beloved parents for supporting and providing me encouragement during my study abroad Their love was the ultimate motivation in my life Table of Contents PDS ACC a E E cides evs notes aa nw ee cant amen teas cane aneemeaaeateaneas Acknowledgements sctiscucereiein incessant esd cacsonsiceceaibeulicedas couuceecea sade edesssteusvosueutacsieeesieess III LIST OF FIGUNES minasa a R VII BIS UO aS ace fecins ch satcecececacncussbcaiswoasen sous cessw ester AE Xl GIOS Sil Vessia poosscdaceswt E EE E a srcsestadeeidess XII Chapter L Mtrod ctioneainiesen a eee 1 LE BACK Bl OU d rareori TEE EAA E A T OA T 1 EZ MOUV O A aa a a A T A A 4 t3 GCONTIDUTIONS ofthe INES Seea A A T A N 4 a VPA STS
55. he Probability Distribution Function PDF plot of RSS bias is also presented in this thesis Figure 24 is the PDF plot of our practical measured path loss We can observe from this figure that with normal fitting the peak is at O dB representing zero mean standard deviation ranges from 25 dB to 25 dB which coincide with NIST s PDF plot in 35 Thus our fading model is verified Given that the path loss and fading models are both accurate we concluded that we are able to reproduce NIST statistical channel model with our real time hardware platform for BANs Consequently our platform configuration was adopted for the further results introduced in Section 4 3 41 Probability Distribution Function RSS Bias Normal Distribution i i i I L I I i I 1 i i i i I i t i i i _ _ Figure 24 Comparison of Probability Distribution Function Plot Between Our Platform Measurements and NIST Statistical Model Left is our result from Network Analyzer right is NIST result used for generating statistical model 4 2 Communication Performance Evaluation Scenario Now that the platform is validated in accuracy aspect we can prepare for starting doing measurement Simulation scenario for communication performance evaluation is discussed in this section As shown in Figure 25 the x axis is aligned to the human facing direction Two white rings represent body implantable sensors o
56. he Wireless Body Area Network in the realm of Wireless Communication in Information Assurance and Security IAS 2011 7th International Conference 5 8 Dec 2011 S Drude Requirements and Application Scenarios for Body Area Networks in obile and Wireless Communications Summit 2007 16th IST 1 5 July 2007 Fabio Di Franco Christos Tachtatzis Ben Graham Marek Bykowski David C Tracey Nick F Timmons et al The effect of body shape and gender on Wireless Body Area Network on body channels in IEEE APS Middle East Conference on Antennas and Propagation MECAP Cairo Egypt 2010 Emil Jovanov Aleksandar Milenkovic Chris Otto and Piet C de Groen A wireless body area network of intelligent motion sensors for computer assisted physical rehabilitation Journal of NeuroEngineering and Rehabilitation vol 2 no 6 2005 Marina Sukor Sharifah Ariffin and Norsheila Fisal and S K Syed Yusof Adel Performance Study of Wireless Body Area Network in Medical Environment in Modeling amp Simulation 2008 AICMS 08 Second Asia International Conference 13 15 May 2008 Ansoft HFSS 3D full wave electromagnetic field software datasheet Online Available 83 7 8 9 10 11 12 http www ansoft com products hf hfss datasheet cfm f HFSS_Flysheet pdf Sergey N Makarov Umair I Khan Md Monirul Islam Reinhold Ludwig Kaveh Pahlavan On Accuracy of Simple FDTD Models for the S
57. ht is Deep Tissue Implant 4 4 Conclusion This chapter described a real time hardware platform for communication performance evaluation of BANs The ability of PROPSim C8 to precisely emulate in body wireless propagation environments has been verified in path loss and fading aspects Zarlink boards functioned as transmitter and receiver communicating through the PROPSim C8 emulator providing valuable insight into how BANs will perform in communications with medical implants This emulation platform approach can be very estimable in facilitating BANs link performance assessments prior to implanting devices inside a patient s body Hardware platform explored the relationship between modulation selection and link performance It shows 2FSK high sensitivity modulation has the highest PRR i e better implant to implant channels exhibit better link performance for the MICS band Moreover quality than do implant to body surface channels The results discussed multiple aspects which multipath and real time properties are satisfactory and desirable Along with the repeatable our hardware platform can be used extensively for various body area network scenario channel 46 propagation models evaluation It also helps test and understand practical performance of medical devices implemented inside human body With broad application foreground our hardware platform can be utilized on other objectives such as in body l
58. ical yet repeatable evaluation of the effects of the wireless channel on WLAN performance In 9 research as already discussed in the Introduction a testbed was developed to enable real time laboratory performance evaluation of WLANs This testbed utilize an RF isolated test system Azimuth Systems 801W for isolation from external interfering sources such as cordless phones and microwave ovens and a real time multipath channel simulator Elektrobit PROPSim C8 for wireless channel emulation A software protocol analyzer Wild 19 Packets Airopeek NX is used to capture data packets in the testbed from which statistical data characterizing performance such as data rate and Received Signal Strength RSS are collected The relationship between the wireless channel and WLAN performance under controlled propagation and interference conditions is analyzed using this RF isolated multipath testbed 20 Chapter 3 Design and Implementation of the Hardware Platform In this chapter we describe a hardware platform which we developed for Real Time Body Area Networks Performance Evaluation and in Body Localization While research and development in body area networks continues to attract increasing amounts of attention it is recognized by researchers in this field that it is very difficult if possible at all to make RF measurements directly inside the human body Therefore simulation methods are seen as the best way to evaluate the r
59. ideband multipath channel simulator the Elektrobit PROPSim C8 and a typical medical implantable device the Zarlink ZL70101 Advanced Development Kit For simulation of BAN channels we adopt the channel model defined for the Medical Implant Communication Service MICS band Packet Reception Rate PRR is analyzed as the criteria to evaluate the performance of communication Several body area propagation scenarios simulated using this hardware platform are validated compared and analyzed We show that among three modulations two forms of 2FSK and 4FSK The one with lowest raw data rate achieves best PRR in other word best communication performance We also show that the channel model inside the human body predicts better communication performance than through the human body For in body localization we focus on a Received Signal Strength RSS based localization algorithm An improved maximum likelihood algorithm is introduced and applied A number of points along the propagation path in the small intestine are studied and compared Localization error is analyzed for different sensor positions We also compared our error result with the Cram r Rao lower bound CRLB shows that our localization algorithm has acceptable performance We evaluate multiple medical sensors as device under test with our hardware platform yielding satisfactory localization performance Acknowledgements First and foremost would like to express my deepest t
60. imulation of Human Body Path Loss in Sensors Applications Symposium SAS 2011 IEEE San Antonio TX February 22 24 2011 PROPSim C8 Wideband Multichannel Simulator Operational Manual August 2002 Edition Elektrobit Group Ltd August 2002 Metreaud L T Pahlavan K RF Isolated Real Time Multipath Testbed for Performance Analysis of WLANs in Information Sciences and Systems 2006 40th Annual Conference Princeton NJ 22 24 March 2006 M P K Heidari Performance evaluation of indoor geolocation systems using PROPSim hardware and ray tracing software in Wireless Ad Hoc Networks 2004 International Workshop IWWAN Oulu Finland June 2004 M Hassan Ali and K Pahlavan Site Specific Wideband and Narrowband Modeling of Indoor Radio Channel Using Ray Tracing in IEEE PIMRC 98 Boston MA September 8 11 1998 Aung Aung Phyo Wai Wei Ni Robert Hsieh and Yu Ge Development of Visualization and Performance Evaluation Testbed for Wireless Body Area Network in e Health Networking Applications and Services Healthcom 2011 13th IEEE International Conference 11 13 June 2011 84 13 14 15 16 17 18 19 20 21 22 L T Metreaud An RF Isolated Real Time Multipath Testbed for Performance Analysis of WLANs ECE Department WPI Worcester MA 2006 M Heidari A Test bed for Real time Performance Evaluation of Indoor Geolocation Systems in Laboratory
61. is not a good result Link Setup CCA amp Cal Data Test Remote Implant HK Link Setup RSSI amp CAL Data Test le lele fo fo fo fo fo fs Implant Information Company ID IMD ID Company ta hex hex Nana Implant Descriptio gt SE co000 Zarlink AIM100 System Status and Control System Status and Control Session Control 400 MHz Link Status Session Control 400 MHz Link Status IMD Status Compary ID 0x01 Company ID 0x01 C 400 MHz IMD Transceiver ID 0000001 Emergency IMD Transceiver ID Qx000001 S Stop Listening Channel 0 Channel 0 ZL701 for Emergency Tea Bytes Block 3 Bytes Block 14 2 Status Poling Data Gather ie ek Paka 1 Status Polling Data Gather Mae Blocks Packet 1 aa onal Listen for Emergency Se Sending Emergency State State Emergency Calls IMD ID Count Company Description 000001 583 Zarlink AIM100 System Messages System Messages Figure 20 User Interface of Receiving Packets for Determine Packet Reception Left is BsM100 GUI receiving and counting received packets numbers Right is AIM100 GUI transmitting packets 37 3 3 2 Receive Signal Strength Indicator Receive signal strength indicator RSSI is also measured under communication status However the method of setting up wireless access is a little bit different from section 3 3 1 In this part as shown in Figure 21 transmitting is started by the 400 MHz carrier wave under Test tag by configuring which channel to be used and transmit powe
62. lations were performed at 5 specific TX RX distances 50 mm 100 mm 150 mm 200 mm and 250 mm 20 transmissions were made at each specific TX RX distance 1000 packets at a time Therefore a total number of up to 20x1000x5x3 300 000 packets were sent for each of the four channel models PRR was used as benchmark in this work indicating link quality which is an important reliability metric for communication This ratio stands for the number of successfully received packets divided by number of total transmitted packets The higher PRR achieved the better the link performed Our results for channels through the body and inside the body are displayed separately in view of differences between through body and in body propagation characteristics 4 3 1 Implant to Body Surface Now we come to the results for communication performance evaluation first we will discuss the Implant to Body Surface channel model category Figure 26 shows that for any TX RX distance tested 2FSK high sensitivity modulation always achieves the best performance The 2FSK high rate modulation has poorer link quality compared to 2FSK high sensitivity with lower 44 PRR while 4FSK is the poorest of the three modulations When distance is greater than 100 mm BSM100 can hardly maintain connectivity with AIM100 using 4FSK modulation 4 2F SK high rate PS O EPO OOOO stetetetetetetetetetetetetetetetteteteteretetete s E WLLL HHHH eee ed a DOO OAT Se
63. letetatetetetetetetetetetetstetetetstetetetstetetetatetetetstetetetstetete s Bm MMMM KREE KR EEEE C tete R809 2F SK high sensitivity aD kaSe PEIRIERO ORIIRE OIO OIRO Mad r eee etree rrr crc rere OOO oO OO 6 O OO Oo Boo OO Oo o oO Ob oo co ELEEEEEEEEEREEEREEEEEEEEEREEEEOEEEREEEEESEE SHE Se eee her iD Deep Tissue Implant to Implant Modulation vs Link Quality a foe of O O O Soa km Ct peace Hi EEE aot a a a a E aaa a aha E E peaa a a eiai ao a ba ai ao tan a ao Sat Pa a a 5 LLL HHHH ete e ee ee ee E E A ee ee HO d a aan n n Papan S Seia n RHEA ORE e EER ni S ARN ae a AHi n T MALLE SENHE E E Ee eee amp ci i i doai mi pi ee oe toa ae ooo oe ioe oo ooo CELE EriiiiJ4FSk Okbps s00kb S R Fem E TNE EASi LT ni a Near Surface Implant to Implant Modulation vs Link Quality Co DH oo P LO Loy wry Lm Oo co aa a co co o mm 9 awy vondesay jayIeY mm D D Distance mm Distance mm Figure 26 Implant to Body Surface Modulation vs Link Quality On the left is Near Surface Implant on the right is Deep Tissue Implant 4 3 2 Implant to Implant In Figure 27 for Implant to Implant channel models similar results are observed with implant to body surface in Figure 26 We can easily judge that 2FSK high sensitivity modulation always outperforms the other two But in additional not only the PRR of 2F
64. mplants and two deep tissue implants placed as transmitters inside a virtual male human body Their general statistical path loss model is described in Table providing the NIST model parameters for different channel conditions For BANs nodes situated away from the human body should also include free space path loss and additional loss caused by apparel In the NIST work in body to in body and in body to out of body propagation effects were treated separately Their work was adopted by the IEEE 802 15 task group TG6 on BANs Therefore we have applied four NIST channel models in this thesis PL d PL dy 10nlog S 4 S N 0 0 d 50mm 5 In equation 5 PL represents path loss d is the distance between the transmitter and receiver S is shadow fading which subjects to Gaussian distribution here Their general statistical path loss model is described in Table providing the NIST model parameters for different channel conditions Figure 5 is the scenario from NIST to generate path loss model using Ansoft HFSS software 17 Table 3 NIST Path Loss Models IMPLANT TO BODY PATH LOSS dB SURFACE Deep Tissue 14 Near Surface na 81 IMPLANT TO IMPLANT PATH LOSS dB os dB Figure 5 NIST Path Loss Model Simulated Scenario Near Surface Implants defined by NIST are ICD and Pacemaker Left Pectoral Muscle Vagus Nerve Stimulation Right Neck amp Shoulder Motion Sensor Right Hand Right Leg Dee
65. n Channel 0 Channel 9 21701 for Emergency Regist Bytes Block 14 Bytes Block 14 Status Polling Data Gather Max Blocks Packet 31 Status Polling Data Gather Mex Blocks Pecket 31 Wake up Moe Spas ae Hoke aw Mod Seco Perom CCAMRES 2450 MHz 400 MHz State State Wake up Responses or Emergency Calls IMD ID Count Company Description System Messages System Messages Figure 21 User Interface of Receive Signal Strength Indicator for RSS based localization Left is BSM100 GUI transmitting carrier wave using 400MHz channel 0 Right is AIM100 GUI reading RSSI value from channel 0 38 Chapter 4 Performance Evaluation of Communication This chapter discussed the results and analysis of performance evaluation for communication of body area networks using the hardware platform which was introduced in Chapter 3 Section 4 1 verified and validated the accuracy of our hardware platform in both perspective of path loss and fading Section 4 2 described the scenario used in evaluating communication Detailed results are presented in Section 4 3 4 1 Hardware Platform Channel Model Validation Here we verify the ability of reproducing NIST channel models utilizing our platform before producing results The verification is very important because only if the accuracy of the platform is strictly confirmed the results are trustworthy The setup of the channel model consists of two parts in total setting the power of the impulses and choosing the
66. nce with NIST model After customizing close the window and save it as tap files 3 1 1 2 PROPSim Simulation Editor 26 4 SurtoOnOnePath Simulation Editor Continuous simulation Channel model Auto Create Output Settings Average input level dBm Information Center frequency 403 500 Sj ilar Mobile speed kmh CIR update rate 73 Output g ir dE Tap spacing n Figure 11 Simulation Editor User Graphic Interface Channel Model Editor is a tool help users define impulse responses Simulation Editor is a tool help users set up channel models define input and output and channel connection ways Simulation Editor User Graphic Interface is given in Figure 11 3 1 1 3 PROPSim Simulator Control The Simulator Control Tool is used for running simulation Figure 12 Its parameters are similar to Simulation Editor Because one channel is used for communication the channel number is 1 Model gain here is indicating the path loss caused channel model If it is static channel the model gain starts at 16 dB since we have Gaussian fading in NIST path loss model the model gain starts at 20 1 dB According to equation 3 and Table 3 if we apply the deep 27 tissue to the body surface channel model scenario then when the distance between transmitter and receiver is around 78 mm we have a 49 1 dB Add up such path loss to the Output end the total channel gain becomes 69 2 dB which means the real pow
67. ne is deeply implanted while the other is implanted near the body surface placing inside body tissue around the stomach Two black rings represent body wearable sensors one on wrist the other is above the heart upon the body surface Pairwise communication paths are in accordance with 42 Table 3 Implant Oo Body Surface Figure 25 Simulation Scenarios of Different Sensor Nodes Locations Include two body surface implant on arm and chest and two deep tissue implant located lower and upper the stomach 4 3 Performance Evaluation of BANs Communication To establish connection communication is initiated by the Zarlink AIM100 as indicated in Figure 24 The BSM100 works as receiver and counter in listening for emergency state watching constantly for any output packet from the Implant board To accurately evaluate link performance several parameters have to be fixed such as no retransmission 1 dB transmit power and same packet and block length Every emergency packet includes 31 blocks and every block contains 3 bytes for the User PID 36 43 Table 4 Modulation and Properties CHANNEL MODULATION RAW DATA SENSITIVITY RATE 2FSK high sensitivity 200 kbps 81 dBm 2FSK high rate 400 kbps 76 dBm 4FSK 800 kbps 70 dBm In sight of Zarlink system configuration three different modulations two kinds of binary FSK 2FSK at different data rates and 4 ary FSK 4FSK as shown in Table 4 are discussed Simu
68. network Monitoring your health with your mobile phone Imec Online Available http www imec ni nl nl_en press imec news wirelesshealthnecklaceinterface html Given Imaging Online Available http www givenimaging com en us Pages GivenWelcomePage aspx Li Yin lin Huang Zhong hua Modeling of Body Area Network in medical healthcare applications in IT in Medicine amp Education 2009 ITIME 09 IEEE International Symposium 14 16 Aug 2009 Muhammad Shuja Uddin Noohul Basheer Zain Ali and Nor Hisham Hamid Wave propagation and energy model for dynamic Wireless Body Area Networks in Electrical Control and Computer Engineering INECCE 2011 International Conference 21 22 June 2011 Sang Hun Han and Sang Kyu Park Performance analysis of wireless body area network in indoor off body communication Consumer Electronics IEEE Transactions 86 30 31 32 33 34 35 Vols 57 Issue 2 pp 335 338 May 2011 Puduru Viswanadha Reddy Viswanath Ganapathy Performance of multi user detector based receivers for UWB body area networks in health Networking Applications and Services 2008 HealthCom 2008 10th International Conference 7 9 July 2008 Raul Chavez Santiago1 2 Ali Khaleghi1 2 llangko Balasingham1 2 and Tor A Ramstad2 Architecture of an ultra wideband wireless body area network for medical applications in Applied Sciences in Biomedical and Communication Tech
69. nologies 2009 ISABEL 2009 2nd International Symposium 24 27 Nov 2009 F R a Regulations MICS Band Plan Part 95 Jan 2003 Kazunari Tai Hiroki Harada Ryuji Kohno Channel Modeling and Signaling of Medical Implanted Communication Systems and a Step to Medical ICT in 16thIST Mobile amp wireless communication Special session on medical ICT June 1 4 2007 Sean F Heaney William G Scanlon E Garcia Palacios Simon L Cotton Fading characterization for Context Aware Body Area Networks CABAN in interactive smart environments in Antennas and Propagation Conference LAPC 2010 Loughborough 8 9 Nov 2010 Kamran Sayrafian Pour Wen Bin Yang John and Kamya Yekeh Yazdandoost A Statistical Path Loss Model for Medical Implant Communication Channels in PIMRC IEEE 20th International Symposium Tokyo Japan 2009 87 36 37 38 39 40 ZL7010X Application Development Kit User s Guide Version 2 0 0 Zarlink March 19 2009 Xinghong Kuang Huihe Shao Maximum Likelihood Localization Algorithm Using Wireless Sensor Networks in Innovative Computing Information and Control 2006 ICICIC 06 First International Conference Aug 30 2006 Sept 1 2006 H Cram r Mathematical Methods of Statistics Princeton NJ Princeton Univ Press ISBN 0 691 08004 6 OCLC 185436716 Jie He Qin Wang Qianxiong Zhang Bingfeng Liu and Yanwei Yu A practical indoor TOA ranging error model for
70. o different WiFi RFID devices Reference 16 used Ray tracing and Elektrobit PROPSim C8 to simulate multipath effect for time of arrival TOA based indoor geolocation This reference discussed different indoor environment scenario such as line of sight LOS and non line of sight NLOS detectable direct path DDP and undetectable direct path UDP since the direct path being the most important for TOA based localization method A Trilateral Centroid localization algorithm is proposed and used in this contribution and many field test and simulation results are compared for both ranging error and localization error for four different scenarios The testbed results and field test results show excellent agreement With future study our work can be extended into a cyber physical system Cyber Physical Systems CPS is integrations of computation and physical processes Embedded computers and networks monitor and control the physical processes usually with feedback loops where physical processes affect computations and vice versa 17 This term refers to a new generation of systems with integrated computational and physical capabilities that can interact with humans through many new modalities 18 Unlike most traditional embedded systems a full fledged CPS is typically designed as a network of interacting elements with physical input and output instead of as standalone devices which the ability to interact with the physical world through
71. ocalization in future These conclusions are intuitive Therefore our hardware platform can have good performance evaluation for in body sensor communication It helps test and understand practical performance of medical devices implemented inside human body which software cannot Along with the repeatable multipath and real time properties it can be used extensively for various body area network scenario channel propagation models evaluation and even body area localization as well 47 Chapter 5 Performance Evaluation of Localization Technique This chapter discussed the performance evaluation for localization of body area networks in terms of localization accuracy using the hardware platform which was introduced in Chapter 3 Section 5 1 gives a general idea of RSS based localization Section 5 2 describes our localization scenarios used to evaluating our localization technique Section 5 3 presents the RSS based localization algorithm used in this thesis Section 5 4 discusses and analyzes the results from our hardware platform for BANSs localization 5 1 Overview of RSS Based Localization Receive Signal Strength RSS based localization has two major algorithm One is least square root another is maximum likelihood However least square root algorithm is known to all that have poor localization performance Thus here we focus on the introduction of Maximum Likelihood algorithm And the research method of bounds is discussed la
72. on methodology From experiment result we can see that REC algorithm has improved performance in typical indoor environment in a significantly manner comparing with LS CN TOAG and Nano localization algorithms 54 BS Tyo Fa Figure 29 RITEM based location area division BS x3 Figure 30 Maximum likelihood centroid algorithm with 4 estimated distances The above two figures Figure 29 and Figure 30 shows how the location area are divided according to RITEM algorithm and how the intersection occurs when applying multiple base station for distance estimating at the same time Most probable predicted location is determined by maximum likelihood based on location area division strategy 55 5 3 Body Area Localization Scenario As the platform has already been verified previously in simulation accuracy of communication performance evaluation here we do not need to verify again Thus first we will discuss the localization scenario used in this thesis As shown in Figure 31 MATLAB is able to generate body or organ meshes by linking every individual surface point from their txt The left of the figure is three dimensional body muscle the right of the figure is three dimensional small intestine The rings on the front of body surface indicate where the base stations locate There are four base stations on the back mirroring the front ones And red dot indicates where the implant resides I 00 450 A
73. onal features of the ZL70101 The link can be set up by using Start Session or Start listening to Emegency B 2 Application Implant Module AIM100 Main Form 77 ZL70101 Application Implant Module AIM100 Link Setup RSSI amp CAL Data Link Setup 400 MHz Normal Operation Company ID hex 0 1 IMD Transceiver ID hex 0 lo lo Jo jo 1 TX Modulation RX Modulation Test Channel Bytes Block Max Blocks Packet System Status and Control Session Control Send Emergency Enable HK Write Access Status Polling Data Gather Wake up Mode 2450 MHz 400 MHz System Messages 400 MHz Link Status 400 MHz Emergency Operation TX Modulation RX Modulation Channel Bytes Block Max Blocks Packet IMD Status Control AIM Compary ID IMD Transceiver ID Setting Measured TX Modulation RX Modulation Channel Bytes Block Max Blocks Packet Operational wams State C Track Figure 49 Application Implant Module AIM100 Main As shown in Figure 49 The AIM100 main form for the ADK is also divided into two main sections The upper section is comprised o configuration settings of the ZL70101 as w the device The lower section if comprised and control for the main operational featu Form it has several function tags as well f a tabular form that allows access to the different ell as providing for control of operational modes of of a static display that allows for basic system statu
74. onal property of PROPSim Because when evaluating channel performance and doing localizations the receiver and anchors have no need to send information back therefore we can use PROPSim channel directly There is no need to add circulator in between constructing feedback path e Second specification of the path loss model and how to configure channel models in PROPSim e Third how to perform RSS based localization and gather RSSI information We will address these considerations in details in the following section 3 2 2 Hardware Platform Implementation The steps of implement this hardware platform is as following 1 Startup PROPSim C8 in advance and make sure a one and half hour warm up 2 Cabling together various components DUTs Zarlink Boards PROPSim 33 3 Checking the power levels at PROPSim C8 inputs signal generator computer connecting with Zarlink Boards 4 Adjusting received power levels using attenuators to fit to path loss model 5 Load path loss model with shadow fading 6 Use a network analyzer to check the signal path and attenuation of the path verify the output signal is desirable 7 Switch the power on Zarlink Verifying that the Zarlink Boards was successfully connected with the computer 8 PROPSim C8 done by turning off the RF local oscillator terminated the link 9 Increasing the path attenuation using the PROPSim output attenuators 10 Performing measurements 3 2 3 Assembly Signal
75. oth However MICS gives a range of a couple of meters 32 Table 2 List of Scenarios and Their Descriptions m DESCRIPTION FREQUENCY BAND EE ar MODEL Implant to Implant 402 405 MHz E Implant to Body 402 405 MHz Surface Implant to External 402 405 MHz o M S Body Surface to 13 5 50 400 600 CM3 Body Surface LOS 900 MHz 2 4 3 1 10 6 GHZ Body Surface to 13 5 50 400 600 Body Surface NLOS 900 MHz 2 4 3 1 10 6 GHZ External LOS 2 4 3 1 10 6 GHZ External NLOS 2 4 3 1 10 6 GHZ 12 For wireless networks the received power can fluctuate widely due to fading effects Here we discuss general components of wireless path loss model for channel analysis and we describe in detail the specific channel model from National Institute of Standards and Technology NIST used in this thesis research 2 2 1 Path Loss Modeling Path loss or path attenuation is the reduction in power density attenuation of an electromagnetic wave as it propagates through space Path loss is a major component in the analysis and design of the link budget in any telecommunication system Path loss normally includes propagation losses caused by the natural expansion of the radio wave front in free space which basically takes the shape of an ever increasing sphere absorption losses sometimes called penetration losses when the signal passes through media not transparent to electromagnetic waves diffraction losses when part of the r
76. p Tissue Implants defined by NIST are Endoscopy Capsule Upper Stomach 95mm below body 18 surface and Lower Stomach 80mm below body surface This thesis brings into correspondence with NIST s path loss model in definition of sensors locations and types 2 3 Previous Works This section provides an overview of previous studies related to development of our real time hardware platform In our investigation into previous work we found very limited analysis of BANs using a hardware simulation emulation platform as we have conceived it here Therefore we have taken guidance in our present work from an earlier study carried out in WPI s CWINS Laboratory That study had been conducted under the direction of Prof Kaveh Pahlavan Real time performance evaluation of wireless local area networks WLANs is an extremely challenging topic The major drawback of real time performance analysis in actual network installations is a lack of repeatability due to uncontrollable interference and propagation complexities These are caused by unpredictable variations in the interference scenarios and statistical behavior of the wireless propagation channel This underscores the need for a Radio Frequency RF test platform that provides isolation from interfering sources while simulating a real time wireless channel thereby creating a realistic and controllable radio propagation test environment Such an RF isolated testbed is necessary to enable an empir
77. r Continuous RSSI measurements are performed on the receiver side It is a decimal number varying from 0 to 31 The larger the RSSI value the stronger the RSS strength Thus if we transmitted power from four base stations corresponding to the NIST channel model s result recorded RSSI ranging from implant at each distance respectively an RSS database can be setup for RSS based localization Link Setup CCA amp Cal Data Test Remote Implant HK Link Setup RSSI amp CAL Data Test TX Carrier Test RSSI Measurements Carrier TX Pwr i Freq MH2 Ch Wave Setting RSSI Filter7ADC Int Int Read 400MHz 402 1570 E co Rw mor MICS Channel 0 31 R 2450Mhz 2450 000 NA TXEN 2 x AA MICS Channel 1 31 R MICS Channel 2 30 R RX Carrier Test RSSI TNA MICS Channel 3 21 R Freq MHz Ch RX Only Setting MICS Channel 4 18 R 400MHz 4021570 RXEN aB Rw MICS Channel 5 23 R MICS Channel 6 24 R MICS Channel 7 23 R MICS Channel 8 20 R MICS Channel 9 14 R Stop RSSI Continuous RSSI System Status and Control System Status and Control Session Control 400 MHz Link Status Session Control 400 MHz Link Status IMD Status Company ID 0x01 Company ID 0x01 _ 400 MHz IMD Transceiver ID 0000001 IMD Transceiver ID 0000001 Si 3 C Ary Company TX Modulation 2FSKFB TX Modulation 2FSKFB Vsup 1 ee C Any Implant RX Modulation 2FSKFB RX Modulation 2FSKFB Vsup2 3 Start Listening C Auto Liste
78. r When it finished we can remove Zarlink ADK boards from the JTAG connector 3 1 2 4 5 ZL70101 ADK Programming Software ZL70101 ADK has provided the software includes three main categories Graphical User Interfaces GUI s Application Programming Interfaces API s and embedded software for the MSP430 boards ADP board base station mezzanine board and implant mezzanine board 80 Code Compose Studio v4 is used in our thesis as compiler and debugger To start with Zarlink s Source Code first starting CCE and if it prompts you to select a workspace browse to the workspace ZL70101 ADK has offered you Then be sure if it is the first time you start CCE for a new source tree you must update the TOP variable in the CCE workspace for the source tree To do so open Window Preferences General Workspace Linked Resources in CCE and change TOP to point to the top directory of the source tree Afterwards refresh all of the projects in CCE so it will update its links Now we can begin modifying the codes The User Graphic Interface of CCS4 is as below C C MicsCmd c Code Composer Studio Licensed File Edit View Navigate Project Target Tools Window Help S a Her g iP i Bi hle ea E3 5 Debug Fig C C Fy C C Projects 2 gt 6 Oicitwrerthale A ImAppMain c A MicsWrite c 9 MicsInit c ImAppLed c aa e a o OS So ieee 1 fRRAKAKAKAKTKAKAEAAKAATARKAKSKEK ERK K KEKE ERA A 7 e nis file contains functions for
79. s Sensor validation identifying possible weaknesses within the hardware and software design is very important Data consistency Constant Monitoring accuracy and completeness of a patient s information is crucial for understanding a medical condition Interference coexistence of sensor node devices with other network devices in the same environment Cost implementation feasibility is a must Constrained Deployment minimizing impairment of patient s daily activity Consistency of Performance sensor measurements must be accurate and regularly calibrated One of the major challenges in designing sensor devices for wireless communications inside the human body is the accessibility of the transmission medium for performance evaluation It is practically impossible to install a development module for a sensor inside the human body and when the sensor has been designed we need expensive procedures conducted under physician supervisions in order to evaluate the performance of the sensors In this thesis we address these issues by introducing an interference controllable repeatable real time hardware platform for performance evaluation of a typical in body sensor chipset used in most implant applications Zarlink ZL70101 ADK operating at 402 405 MHz This evaluation platform utilizes an existing multipath channel emulator Elektrobit PROPSim C8 to analyze the performance of the communication link between a sensor locate
80. s res of the ZL70101 Alike Base Station Module Main Link Setup Setup connection and display link properties e RSSI amp CAL Reading RSSI and calibration CAL e Data Send and receive data e Test Enable and disab le carrier wave set transmitter level 78 The Application Implant Module can function as transmitter using Send Emergency or as receiver using Direct Wakeup When the Base Station Module is trying to Start Session Implant will react automatically without the need to Direct Wakeup 3 1 2 4 3 Programming ZL70101 ADK Firmware M FET Pro430 FET MSP430 Flash Programmer Elprotronic Inc am File view Setup Serialization Tools About Help Open Code Fie gt Im ppi0l tst path C Program Files ZarlinkW2L 7010X Blow Security Fuse Enable Power Device tom Adapter Microcontroller Type Status 33v w Device l i Device Action i Voltage Group MSP430Flxx 4 ETT E Reload Code File POWER ON OFF MSP430F1611 ka EA Enable Blank Check Total 7 Check Sum Source OXOSTEDE E verity Security Fuse Selected Device Information RAM 10240 bytes FLASH 48 kB Memory ERASE FLASH Repor Beading Code Fie di Device Serialization E BLANKE CHECK Code size O46E4 18148 bytes C WRITE FLASH VERIFY FLASH Reads COPY Pork LSB Automatic Erase Write memory option JTAG 4 wires All Memor CN Figure 50 Programming ZL70101 ADK
81. s 30 VII Figure 14 Base Station Module BSM100 with Dual Band Printed Loop Antenna An ADP board is mounted in the other side upon AIM board as power provider and controller with power switch on the lower right COPNEL ccccccsssseececceesecceeeeescceseeeueeeessaeeeceesaeeeeeeas 31 Figure 15 Programmer Cable Adaprex tatiecicidiennishs csieuieie th cont seca A A A 32 Figure 16 MSPFET430 USB Debug Interlacessiwsteswiantlciciewecni ta aeiceienaanciie hehe wines 32 Figure 17 Assembled Hardware Platform Zarlink boards are at the bottom left PROPSim is in the WIIG Cl Css asstect ots es Sna A so anecctenziaaeceisavensaeantaaneaeG ease T GDA 34 Figure 18 Gaussian fading parameter CONFIQUrATION ccccecccccccssseeecceecsueesseeceessueeseeeeeessaaeaees 35 Figure 19 Doppler Spectrum of gaussian fading model in PROPSIM C8 ccccsssssseececeeeeeeeeeees 36 Figure 20 User Interface of Receiving Packets for Determine Packet Reception Left is BSM100 GUI receiving and counting received packets numbers Right is AIM100 GUI TRANSMITTING PICKET Sru a E See seaivaccasiea wasnede Gey Goanneetons aeeae 37 Figure 21 User Interface of Receive Signal Strength Indicator for RSS based localization Left is BSM100 GUI transmitting carrier wave using 400MHz channel 0 Right is AIM100 GUI reading RSS Valle fron Channel O einari 38 Figure 22 Relationship between RSS and RSSI RSSI are real received power in discrete dots RSSI line is th
82. s power ratio of the beams A B is In this thesis since only one normal distribution is needed both beam shifts are set to 0 Variances of both beams are set to zero to meet the normal distribution conditions The ratio of A B is set to O dB which is 1 in decimal unit meaning they are equal Accuracy verification is done to validate this Gaussian fading configuration Figure 19 Doppler Spectrum of gaussian fading model in PROPSim C8 3 3 Packets and RSSI Collection As we know Zarlink Application Implant Module works as transmitter starting sending signal Zarlink Base Station Module works as receiver respond to received signal In this section we will discuss how to setup wireless access and collect data 36 3 3 1 Packet Reception To receive packet first of all we need to let transmitter and receiver start communicate In link performance evaluation Zarlink Application Implant Module begin sending Emergency packets Zarlink Base Station Module is listening for Emergency Figure 20 is an example of emergency communication The Count in Emegency Calls session is indicating how many packets we have received The IMD ID is 000001 means packets are from Zarlink AIM100 with Company ID 01 Thus if we transmit a fixed number of packets from the transmitter at a time the more packets we received the better link performance is In this example 1000 packets are sent only 583 packets were successfully received which
83. s us a fitting plot and an equation derived from this fitting line The green spots are individual RSSI values blue solid line is fitting line X label is RSSI without unit y label is RSS in dB From this Figure we can observe a clear relationship and a good fitting result Therefore we can use the Y 1 9938 x X 106 28 24 as the converting function where X stands for RSSI Y stands for RSS In other words as long as we observe an RSSI value from Zarlink Implant board corresponding accurate RSS value can be derived Roo vs ROS Line Fitting Plot 1 99s564 106 20 Ros value dB Real Data Line Fitting 5 10 15 20 25 30 35 RSSI value no unit Figure 33 Relationship between Receive Signal Strength and Receive Signal Strength Indicator 5 5 Body Area RSS Based Localization Algorithm In this part an improved maximum likelihood algorithm is applied and introduced based on the localization scenario The principle of algorithm is discussed first its performance evaluation will be analyzed in Section 5 6 58 0 99949 0 999 U 94 0 95 Gg 0 75 0 5 0 25 0 1 0 05 Probability N 2000 0 0001 b 30 20 10 0 10 20 a0 Data Figure 34 Cumulative Probability Plot CDF of random Gaussian fading added in our path loss model X axis is the varying range of path loss in dB Y axis is corresponding probability In this figure the probability is 90 Because in equation 4 and 5 there
84. ssaeeseeeeesseeaaees 75 Figure 48 Base Station Module BSM100 Main FOr c ccsssecccccseeecceeeeeeeeeseeeeeeeseeeseeseseueessss 76 Figure 49 Application Implant Module AIM100 Main cccccccsssssseecceceeenesseecceeeeeeeneeeeeeseueees 78 Figure 50 Programming ZL70101 ADK Firmware Elprotronic FET Pro430 ccccccssseeceeessseeeees 79 Figure 51 Code Composer Studio version 4 User Graphic ssssssesssseeesssserrssseeresseseresserresseerres 81 List of Tabes Table isPOr Fre GUeney BanG masne T O oul danloly ATOA 11 Table 2 List of Scenarios and Their Descriptions cccccceesecceeescecceeececeuececseeceeeeseceteneeeeeeneeens 12 Tables NIST Path LOSS Models ceace A 18 Fable4 Wiodtil ation and Propertie Sueno a a base histens a A 44 Table 5 Sample RSSI localization values from hardware platform ccccccessccccsssecceceeseeeeeeaeeeees 68 XI Glossary 2FSK 4FSK BAN CCA CWINS DUT FDTD HFSS LOS MICS MIMO MLE NIST NLOS PRR RSS RSSI RX TOA 2 Frequency Shift Keying 4 Frequency Shift Keying Body Area Networks Clear Channel Assessment Center for Wireless Information Network Studies Device under Test Finite difference Time domain High Frequency Structure Simulator Line of Sight Medical Implant Communication Service Multiple Input Multiple Output Maximum Likelihood Estimation National Institute of Standards and Technolog
85. t in the structure of channel models for BANs In 33 the author claims that the multipath effects are negligible in body area networks by analyzing wideband channel characteristics in the body based on the measurements and the theory of EM wave propagation and properties of dialectic materials S21 parameters are measured between two antennas inserted in 48 different points of pig bodies Experimental results show that severe attenuation in high frequency band and smaller time delay take place in contrast to the case of free space propagation That work drew the conclusion from measurements made along the pig s body center line that as a lossy medium human tissue becomes a strong absorber of radio waves The channel exhibited such a small delay spread that the multipath effect was negligible 2 2 3 Shadow Fading Shadow fading is the long term average changes in the RSS caused by changes in the relative position of large objects such as buildings in urban areas between the transmitter and 15 the receiver The actual RSS will vary around this mean value This variation of the signal strength due to location is often referred to as shadow fading or slow fading The reason for calling this shadow fading is that very often the fluctuations around the mean value are caused by the signal being blocked from the receiver by buildings in outdoor areas walls inside buildings and other objects in the environment It is called slow f
86. table drug pumps bladder control devices and Implantable physiologic monitors The ZL70101 is very flexible and supports several low power wakeup options Extremely low power is achievable using the 2 45 GHz ISM Band Wakeup receiver option The high level of integration includes a Media Access Controller providing complete control of the device along with coding and decoding of RF messages A standard SPI interface provides for easy access by the application The features and benefits of Zarlink devices are e 402 405 MHz 10 MICS channels and 433 434 MHz 2 ISM channels e High data rate 800 400 200 kbps raw data rate e High performance MAC with automatic error handling and flow control typical lower than 1 5x10 e Very few external components 3 pcs antenna matching e Extremely low power consumption 5 mA continuous TX RX 1 mA low power mode Ultra low power wakeup circuit 250 nA Standards compatible MICS FCC IEC 29 3 1 2 1 Applications Development Platform ADP100 board The Application Development Platform ADP100 board is a bridge board with integrated USB2 0 support to allow for interfacing between a PC running the ADK GUI software and an implant or base station board It is mounted on both Implant and Base Station boards containing a power switch and battery Figure 13 Base Station Module BSM100 with Dual Band Helical Antenna And ADP board is mounted upon BSM as power provider and controller wit
87. ter 5 1 1 Maximum Likelihood In statistics maximum likelihood estimation MLE is a method of estimating the parameters of a statistical model When applied to a data set and given a statistical model maximum likelihood estimation provides estimates for the model s parameters The maximum likelihood function is defined as follows ECO Xa ssce sees pM E S Diy No aint Xnl Ta f x 6 48 where 6 will be the function s variable and allowed to vary freely by considering the fixed parameters of observed values Xj X2 Xn a x O Actual MT location A X Estimated MT location _ Data Fusion Center Figure 28 Illustration of a simple scenario for wireless localization In general for a fixed set of data and underlying statistical model the method of maximum likelihood selects values of the model parameters that produce a distribution that gives the observed data the greatest probability i e parameters that maximize the likelihood function 7 me argmaxges x Maximum likelihood estimation gives a unified approach to estimation which is well defined in the case of the normal distribution and many other problems However in some complicated problems difficulties do occur in such problems maximum likelihood estimators are unsuitable or do not exist 49 Paper 37 proposes the use of a wireless sensor network for estimating the location of a transmitter that propa
88. the internal attenuators inside PROPSim C8 and external variable attenuators supplement this by connecting between the implant board and PROPSim simulator Variable Attenuator J Zarlink Base Station Zarlink Base PROPSim C8 station Simulator Zarlink Base Transmit Packets Station Read RSSI Value Zarlink Implant Zarlink Base Simulate Channel Models Station Receive Packets Figure 7 Functional Block Diagram of Hardware Platform for Localization 22 Unlike the block diagram for communication Figure 7 is the functional block diagram for Localization For localization accuracy we used 4 base stations at the same time The AIM100 board remains serving as transmitter transmitting medical signals constantly Four base stations functions as receivers recording Receive Signal Strength Indicator RSSI values separately In PROPSim among the total 8 available channels four of them are used to simulate different channel models each connecting to an receiver AIM100 board The two main components of our hardware platform for Body Area Networks consisting the Elektrobit PROPSim C8 Wideband Multipath Simulator Figure 8 and Zarlink now a k a Microsemi ZL70101 Advanced Development Kit Figure 9 Figure 8 PROPSim C8 Wideband Multipath Simulator 23 ZL70101 Application Developm P N ZLE70101BADA S N 145 Figure 9 Zarlink Advanced Development Kit 3 1 1 Elektrobit PROPSim C
89. tory performance for in body localization Thus this hardware platform can be used more extensively for localization along different body organs and different sensor locations Due to the real time repeatable and controllable characteristics of our hardware platform future work can be done on developing new body area cyber physical system Our hardware platform is one with which one can easily interact and intuitional perceive results This capability can be further enhanced by adding visualization components to this hardware platform 71 Appendix A PROPSim Tutorial Here we list several notes before giving a detailed description of PROPSim software 1 Make sure the monitor and mouse is correctly connected Otherwise we will lose control of the PROPSim and have to force shut down which is harmful for the machine 2 Do not connect the simulator to an AC power source before verifying that the line voltage is correct 3 Allow the simulator warm up to room temperature before turning the power on Turning on a cold simulator may damage it because of possible water condensation 4 Ensure at least one and a half hour warm up before running simulation which demanding high accuracy 5 Ensure the simulator has unrestricted airflow for the fan and ventilation openings in the rear cover and bottom panels since the inner PC is dissipating heat 6 Always beware of the RF IN and IN RFLO signal If it is too high over O
90. vative of W 8 by 8 and I O is the Fisher information defined above Apart from being a bound on estimators of functions of the parameter this approach can be used to derive a bound on the variance of biased estimators with a given bias as follows Consider an estimator 6 with bias b E 6 0 and let W b 0 Therefore any 53 2 unbiased estimator whose expectation is 8 has variance greater than or equal to w 6 I 0 Thus any estimator whose bias is given by a function b satisfies 1 b 0 TO 22 var From the above we can see the unbiased version of the bound is just a special case of this result when b 8 0 It is trivial to have a small estimator that is constant has a variance of zero But according to the above equation the mean squared error of a biased estimator is bounded by E 6 0 gt e b 0 23 5 2 Empirical Indoor Localization Study Paper 39 presents a practical RSSI based TOA ranging error model RITEM for localization algorithm which can be used to estimate ranging error interval in real time In RITEM ranging error is classified into four classes by the RSSI value in TOA ranging process and ranging error of each class always within a certain interval RITEM is verified by field tests in two typical indoor environments After validation RITEM is applied into Ranging Error Classification REC based TOA localization algorithm to introduce its applicati
91. y Non Line of Sight Packet Reception Rate Receive Signal Strength Receive Signal Strength Indicator Receiver Time of Arrival XII TX Transmitter Ul User Interface WLAN Wireless Local Area Network WPI Worcester Polytechnic Institute XIII Chapter 1 Introduction This chapter is divided into four sections Section 1 1 gives some background knowledge and literature search of this hardware platform and Section 1 2 discussed the motivation of writing this thesis In Section 1 3 gives the major contributions of this thesis and Section 1 4 the thesis outline is provided 1 1 Background The ongoing development of Body Area Networks BANs in conjunction with advances in implantable medical devices is generating great interest in use of this interdisciplinary technology for improved health care including patient monitoring diagnostic procedures and emergency treatment 1 2 However the development of these applications has been hampered by the technical difficulties encountered in wireless transmission through and inside the human body Prior research shows that signal propagation through and within the body is influenced by many factors including differing dielectric properties of various organs body shape From paper 3 their result shows even the effect of body shape and gender on Wireless Body Area Network on body channel Consequently knowing how signals propagate through and inside the body is challenging So f

Download Pdf Manuals

Related Search

Related Contents

Doc 218-F PdS OOUU Atr02 - Sotto il Monte Giovanni XXIII  Nortel Networks 553-3901-200 User's Manual  Variostop  Téléchargez le guide d`installation!  V7 CAT6 UTP Network Cable 2 (RJ45m/m) red 2m  Ematic EGD170 8GB Black  Manuel du 5211  平成21年度自動車等安全性能評価実施要領の策定について Ⅰ．自動車  TA剛TA  日本語  

Copyright © All rights reserved. Failed to retrieve file