Home

Designing for TARA - Repository TU Delft

1. kkk kkk 4 PRE PROCESSING PREPARATION COCR 4 open the data file 4 fid fopen beatsig 69 4 create the window for range FFT hamming LENGTH z ones 1 FFTLEN LENGTH by FFTLEN FFTLEN sweeps of LENGTH samples each sweep in column winr kron z y 4 create window for Doppler FFT hamming FFTLEN z ones 1 LENGTH 2 7 FFTLEN by LENGTH 2 LENGTH 2 range gates with FFTLEN sweeps 4 in each to do FFTLEN point FFT for each gate wind kron z y lt nhea 784 2 header consists of 784 bytes 4 skip the header not yet useful information header fread fid nhea short AIK 4 POST PROCESSING PARAMETERS if rem NPROFS 2 1 NPROFS NPROFS 1 to ensure even number of profiles end calculation of useful vectors 4 range rangeres 3e8 2 FEQ 1e6 range resolution m range Zi 1 Z2 1 rangeres range m 4 Doppler velocity or frequency lambda 3e8 3 298e9 TARA w
2. 7 POST PROCESSING 2 4 from first step Doppler range spectrograms to processed Doppler range spectrograms Processed means removal of non atmospheric 4 echoes clutter and noise From the processed Doppler range 4 spectrograms the profiles of several parameters Z V are 4 estimated ooo YP FRR ak ak ak ak ak ak ak ak 2K RACK MAIN BEAM POLARIMETRIC Noise and clutter suppression in the spectrograms and first 4 calculation of the mean Doppler velocity Z V Moments sLdr2 rt V2 sZ dea noisefloor NPOL lvel Vint vel FFTLEN initial maximum unambiguous Doppler velocity 7 Doppler velocities interval reduction around mean Doppler velocity V noise clutter suppression and single of double missing data replaced in the spectrograms definitive calculation of the profiles Z V Rho Moments pola rt V3 sZ dea Fvh dea noisefloor NPOL lvel V Vint 4 Calculation of the final profiles OUTPUT Zh ns Z reflectivity mm6 m3 Vh ns V mean Doppler velocity m s Wh ns W 7 Doppler width m s Rhoh ns Rho 4 cross correlation coefficient VA FKK KK FK KK FKK FKK FK FK FK K FK FK K FK K FK FK FK FK FK FK FK FK FK FK FK FK FK FK FK FK FK FK K FK FK FK FK FK FK K FK FK FK
3. No anusue y jo 4035 31235 FAD Jo 09906 IIA 6 jeuon uny e ur sey 3y jo pue 03102 Y jo Sa gt uasaja4 24 1035 The GUI VI because of its size is split up in 2 pages GUI sBueu 9045 uis 1 pb Bra sjuene Jasn 10j 40 Update of the visualisation intensity charts Here all the Z scales are going to Here the data goes the the graphs abcontrol of the be set individuali Visuali 1sualisal Zhh PZScale MarkerVals T rZScale MarkerVals Zhh bZScale MarkerVals Vhh PZScale MarkerVals Vhh PZScale MarkerVals Vhh a PZScale MarkerVals Whh reo ZScale MarkerVals jp n PZScale MarkerVals Whh PZScale MarkerVals PZScale MarkerVals LDR ZScale MarkerVals Here all the properties for the x and y scales lof the graphs will be set le Get 7 EA PZScale MarkerVals stieor P IntChart strict Horizontal speed Y Scale Range Maximum VScale Maximum pO Y Scale Range Minimum YScale Minimum ZScale MarkerVals YScale Multipl
4. ETEA speej 0 ainjeladuua E Josues dw y d jaued Wa shS 51079 4 04409 wayskS di VLVG A ge e i EE 8 5 5 8 dojs ueijs juausinsea lt SIOJJO LOHEZIJENSIA lt 5 UOISI9A e lt 8 yeyoed uonezijensiA 5 jexoed Jepe JepeeH 08409 Japea E uogeziensiA 154 yndjno 5 N jexoed ejep SION O dmes IXd 4 suoljdo 25810 S Jjo uo Aejag 4 JUSWILUO D 4 4 Ot euin 4 JJ ujbue 44 21949 jueueinseeu 2 J pe H Ott basy ujnuize euuajuy ja 34 uoneA eje Ott eej 4 pesseoold sall 4 Jeping 1 2950 Ind 1105 90 46 uJeJBeip VsV L
5. IeAPIO jun 45 sbueu5 anie 2 SES unejeg 05 EI 31635 39591 sanje andur aj iy 46 5 HH SAD uonesueod 45 4 HH 2247 uonesuejog IeAPIO PYD 351105 ppe 37743 uonesuejod 47 3p 2144941120444 3 5 ujsuiuejod uy uonesuejog AH 27343 uonesuejog sar 127275 uonesuejog 48 4 4 HA 33 55 uonesuejog 43275 ujsuuuejodu even 0 PYD HA ppe j1edxe p gt uonesuejod sar HA PPA uonesuejog 5 pt 49 anjeA AA PAD 91 9p uonesuejog 3 4 uonesuejod
6. Wed paads pono JeuuoN pads ucc prane z Tox J woven porno E wn Lol HER E peer waz TASIGEUeA ESE pss qu Saati pune snc ue E parsa He y 987 uonenuoniy H pargesip iiy uorsinoxa fouanbau4 BULON Sf awg M TU 1065119 SSION T a e fase ME a euuayuy 3 TAPIO 1 j JURO HOMO pusu wane sun 5 5 uo 325 0 4 1013409 24 03 Oppinb anes 285 GR TUNI pa ues passa2oid anes E uespoog 2 anes ep anes euusjuy
7. 549 E E 4 rs MAH HH 5 gt 2 str T 3p euusquy euusjuy A 442 _ Ole 0 pu z psyo 14122932540 ae suondo juawainseayy pue 2 36 Appendix D Labview figures Radar control he GUT Will own to the user fe Main state machine of radar control 37 The additional states of the Main State Diagram REPORT E i ERROR Rerort message is sent to the GUI via the error muon porter that the data measurement is started Data measurement is started 38 Headerbuilder input cluster 39 221558 2 sun kepa EET
8. IeAPIO PYD 321105 50 sar 2245 uguinelodu 3 5 ujsuuuejod uy ZO uonesuejog 2 395450 ppe uonesuejod Buey 20 PAD uonesue od PS3uenpy 51 3 545 2149420444 33835 35 13747 e3ejs 4 jse j1edxe uonesuejod qu 33 3 15 131745 uonesuejog sar 52 Data processor uoissitusuel 5599 ejep puss UOIS13A lt 213 4 5 pAg eep 1 pug 5 od 104 dnyss suone ns e5 dnas xd SAA 533163p yynuuize suonesuejod jo JUAN 35 144 abuey 144 1919900 5 sun deans uonenuapy 214 uoisinsxe Guanbay jou asiou uOISJ3A apo 105592014 eje sjndano uai siaunou 2 eep passs oid jou 313436 passasoud 541553204
9. pasn aq ue y doo aup 210590 eye sapeay 243 sureqo s32163p euuazuy SJUILUSINSEILU Jo 6415522044 ag asn 10 513351631 3145 03 3495 25107 asiou aug 55220 3A 3835 uat pue Burssa oud e seu aug sup qoe a 39415412240 995ou 34 uo aq 03 jeubis 3e2q 10 spem pue ap ayy 54015 A 241 66 neas Phe pasn ag jeubis ayz yim 24615 3e2q different states for the different processing actions Noisebrocesing n this state the VI starts to process the noise via the Matlab code provided It will also check if the user wants a3 or 5 e measurement Measurement Type amp 9 Noise Measurement Noise E Check for 3 or 5 cycle measurement for the 54 55 VI will stay in this state until the user shuts down the measurement It will wait for a notifier with the beatsignal and process it and send the processed idata away to other VIs via a notifier Afterwards it will Measurement Type wait for a new beat signal Case around processor do nothing if false execute matlab if true J 56 VI will stay in this state until the user shuts down the measurement It will wait for a not
10. Contents Preface i List of acronyms iv Summary v 1 Introduction 1 2 Problem description 2 2 1 Problem definition 2 2 2 uos quee eee n ESO e 2 2 3 Framework of the project wem Ra hg CES I E E E 3 3 Program of requirements 4 3 1 Product USES ok EI Ee 4 3 2 Environment ecology 5 3 3 45 d d ado due s gs ice QU 5 Production uu tuas A ie open ra 8 5 4 Review of current systems 6 AT Atmospheric radars 2g Lee dee ve oe ee Bee dod 6 4 25 c eb Geo dor De dpt etu ath Dea hu Iob T 4 37 PX iugo cede fede E Ree eS eke o ue e 9 4 4 d 43m trc List URS deu de dus atu RR act dues 9 4 5 Matlab Re gue dr wiesen P dou deemed AO cir NU 10 5 Design process 11 5 1 Blockdiagram and signals 11 5 2 Dtatediagrams voe EOS ue E 5 de ert ed 14 5 3 Matlab and LabVIEW ERREUR ERE ke RR SUE RE MER 15 54 Choice motivations 2520570 uta be diae Sobral ey Ru boss 15 6 Implementation 18 6 1 Radar Re
11. 200 sin EL pi 180 habo find height lt hnear near size habo 2 height height 1 near Nran Nhe size height 2 Data in the radar near field are discarded Zh 10 logi0 Zh near Nran unit dBZ ZHH squeeze Zh 3 ZDR ZHH squeeze Zh 1 LDR squeeze Zh 2 ZHH VHH squeeze Vh near Nran 3 WHH squeeze Wh near Nran 3 Rhoh Rhoh near Nran t toc disp temps de calcul num2str t s figure plot ZDR height 75
12. The overall block diagram has gone through a number of iterations involving different choices motivated along with the state of progress of the system requirements final version can be found in Appendix It consists of the following seven blocks GUI The Graphical User Interface is made to allow the user to define the desired type of measurement via an intuitive and easy to use interface the front panel of the user interface is displayed in Appendix C The GUI contains e switch to start or stop the measurement Input boxes for all the variables the user needs to set for the measurement e Tabs to switch between normal and advanced measurement options Check boxes to choose which data the user wishes to save Input boxes to set a measurement schedule e text box to input user comments 15 graphs displaying the processed data parameters as a function of height and time 5 tabs grouping the 15 graphs by parameters for the main beam the 2 offset beams the polarimetric parameters and the wind parameters 11 e A text box displaying the system status errors e The GPS location and time All of the measurement settings are sent to the Header Builder and the signal from the on off switch and the measurement schedule are sent to the Radar Control block Header Builder The Header Builder takes all of the measurement settings that are set in the GUI and bundles them together so they can be easily transmitted variab
13. 33 Visualization block Processed data Header Stop end ofday Internal Done s Scale adapting Time scale Height scale Data scale Real visualization 1 Enable Real time visualization Control Quicklook Done 3 or Real time 5 cycle Enable Storage NetCDF writer Header Noise data Processed data Cycle type Buffer Quicklook generator Mix Buffer Clear buffer Enable 7 Raw data Quicklooks Buffer Clear buffer Enable T Buffer Clear buffer T Storage control NetCDF creator Stop end of day Saving options User input Saving options 4 bits every bit switches one file type on off Saving done 34 Done visualization Quicklook Create directory File control Create file Save data gt Done Make file type Make file typ 3 bit 000 make no file 100 make noise file 101 make processed data file 110 make raw data file 111 store quicklook Appendix B State diagrams Processing state diagram dii 3 cycle D Oves 4 processing J 41 A b Noise Processing 4 K Pp p gt Raw data A Dp e ico N
14. containing the Data Processor Visualization and Storage VIs will be set to false via the Start data processor storage and visualization 18 BEAT SIGNAL BEAT SIGNAL ore oo a b Figure 6 1 Beat Signal notifier notifier This is done so that whenever the user stops the measurement and the State Machine returns to its Idle state the three VIs in the case can be started again after new measurement options have been input When the State Machine enters the Noise Measurement State the VIs for the Data Processor Visual ization and Storage will start running by setting the case that contains them to true by making use of the Start data processor storage and visualization notifier The Noise Measurement VI is then run in this state which is used to interact with the PXI in order to put a beat signal containing 2 minutes of noise bundled with a time stamp and the measurement type see figure 6 1 a on the Beat Signal notifier This notifier will be obtained by the Data Processor where it is meant to be processed Afterwards the Noise Measurement VI will put another beat signal containing 1 minute of raw data on the notifier to be stored which should be done in parallel to the processing of the 2 minutes of noise collected before Should this VI encounter an error during execution this error will be output in State Machine and it will return to the Idle state If one of the notifier
15. FFTLEN Nran NPOL raw Doppler spectra spec_2 zeros FFTLEN Nran NPOL Doppler spectra sZ_dea and cross spectrum Fvh dea dealiased for all the ranges Nran and polarizations NPOL sZ_dea zeros NPOL FFTLEN Nran NPOL Fvh_dea zeros NPOL FFTLEN Nran 7 Moments not yet interpolated Z zeros Nran NPOL 71 V zeros Nran NPOL zeros Nran NPOL 7 Doppler spectra spol and cross spectrum Fvh dealiased 4 with clutter suppression and clipped they are not yet used 4spol zeros NPOL FFTLEN Nran NPOL Fvh zeros NPOL FFTLEN Nran end Initialization PRE PROCESSING from radar signal 3 to raw spectrograms Doppler spec per range read block of data 4allbeat fread fid LENGTH FFTLEN NPOL short 1 spectrogram allbeat beatsig 1 LENGTH ns 1 FFTLEN NPOL 2 1 ns FFTLEN NPOL 2 for np 1 NPOL NPOL number of polarizations and radar beams 4 extract the beat signal for one polarization state or beam polarization or beam changes from sweep to sweep beat 1 allbeat np NPOL FFTLEN NPOL beat 2 allbeat FFTLEN NPOL np NPOL FFTLEN NPOL 2 remove the radar DC component from the beat signal z ones LENGTH 1 bm mean beat 1 beat 1 beat 1 kr
16. FM CW Frequency Modulated Continuous Wave GUI Graphical User Interface IRCTR International Research Centre for Telecommunications and Radar ISA Industry Standard Architecture computer bus NI National Instruments PXI PCI eXtensions for Instrumentation STW Dutch Scientific Society RADAR Radio Detection And Ranging TARA Transportable Atmospheric RAdar VI Virtual Instrument UTC Universal Time Coordinated Summary The TARA scans through a column of atmosphere in order to observe and study cloud system behavior After 10 years of operation new knowledge of processing has been acquired which allows for an update that will improve measurement quality and processing speed The main question of this thesis is How to implement data processing visualization and storage for TARA system from the radar control to the visualization and storage needs to be built from scratch In order to most effectively achieve this the entire system has been split up into two parts that will be designed separately To be able to combine the parts the groups will work together to build an overarching structure in which the separate blocks can be easily implemented chapter 2 In chapter 3 the program of requirements is given in which all the requirements for the system are mentioned Different topics have been researched in order to be able to do this project this can be read in chapter 4 Over the years variety of radar types have been developed each
17. Matlab and should be easy to implement c The noise processing should be done in 1 minute in order to make parallel computing easier d The processing block should provide the correct scales for the visualisation e The visualisation block should temporarely save all the visualization data in a buffer and make a quicklook file at the end of the measurement or day f The storage block should automaticaly store files in the correct directory and create this directory if it does not exist already 0 storage block should put all the nessesary data in a file automatically The user should be able to choose which data should be stored ma SUE CMT functions and simple debugging 2 Production and commissioning a The software should be modular so that it is easily adjustable to other systems b The software should be well documented c After installation the software shoul immediately be ready for use d The system should be operational before the 7th of June 2011 e The install process should be done within 60 minutes 3 Liquidation a The un install process should be done within 60 minutes 3 4 Production 1 The user interface as well as the processing software will be built in a standard software package 2 Pieces of Matlab code should be inserted in the software 3 PXI system of National Instruments should be used for data collection The system should be
18. a ea 18 19 6 3 Header Builder eae el Sets eee E V d AVENA Eg 21 6 4 Processor amp e ede EE vL E ode A 4 21 6 5 Visualisation 225 xem bREIeIL AWE DES ed 23 10 23 7 System evaluation Product Use vaste Ld oes a cee de e 248 dide d A ee le ete Boh ee ood 7 2 Environment and ecology Wo OYSUOIEGOSIEIL 219 S ie doe ee die e d vus ot t oe TA Production asns REIR Xue datu teu dr ege unn deer dr et dono qeu d 7 5 Conclusions and recommendations 8 Discussion 9 Conclusions and Recommendations Conclusions 2 o deii Ae d SP eee od e ADC tee ella eS 9 V Seu 8120 dee ut 12224 92 Recommendations se neue a ae UR b Ar wl a eG Bibliography Block diagrams B State diagrams C User interface D Labview figures E Examples of the implemeted Matlab code iii 24 24 24 25 25 26 27 29 29 30 31 32 35 36 37 64 List of acronyms ADC Analog to Digital Convertor ATMOS Remote sensing of the atmosphere group part of the remote sensing of the environment department at Delft U CESAR Cabauw Experimental Site for Atmospheric Research DDS Direct Digital Synthesizer FFT Fast Fourier Transform
19. and when is it OK to use them Available http zone ni com devzone cda tut p id 5317 Felix Annan 2011 June Notifiers in LabVIEW Available http cnx org content m13762 latest LabVIEW manual 2011 June Types of Graphs and Charts Available http zone ni com reference en X X help 371361B 01 Ivconcepts types of graphs and charts 31 Appendix A Block diagrams Processing block Beat signal Header input re Noise BS Noise Noise or beat FEA processing Code version NPOL Mean noise Calc factor Ts Data type BS Data Zdr Ldr Height Noise data gt Processed data 3 Pola rt FEQ EL At NPOL Ts Start time Cycle type Specsmoother rt Moments sLdr2 rt Moments pola rt y Polawind rt Code version i Code version Code Versions Processed data 5 Z V W Zdr Ldr p Heigh Horizontal Start time Split Control header data 32 M x Clut offbeam rt Unfold rt Conran rt Vertical windspeed Direction Moments sLdr2 rt Moments pola rt Moments rt Moments var rt ObRescale rt wind3D rt i MANI JeMwod _ JeMod EUER 4
20. are set The y scales are fixed for all graphs while the z scales are set individually The y scale are set with information from the processing block The processing block sends a height vector in the processed data notifier From this vector the minimum and maximum are derived For the z scale 2D arrays are needed with the combination of color and the value for that color At the moment the maxima and minima for the z scales are fixed for each different parameter but later they can be made variable if necessary From the minima and maxima other values are calculated in order to make a full color scale These arrays of collors and values for each graph are then put in a larger array and this array is bundled with the values for the minimum and maximum of y This bundle is then sent to the GUI via the Azes notifier After this state the visualization block goes either to the 3 cycle visualization state or the 5 cycle visualization state The 3 cycle visualization state and the 5 cycle visualization state are very similar The data for the graphs comes from the data processor via the processed data notifier In the visualization block the data is given trough to the GUI via the Visualization for GUI notifier arrays are first transposed in order to match the vizualisation format After that all the information is put into a bundle which goes to the GUI via a notifier When a 3 cycle measurement is taken not all the graphs are used To be sure th
21. designed using an open approach allowing for easy implementation of new Chapter 4 Review of current systems In order to be able to design the system knowledge had to be gained about the topic first That is why the different topics have been researched This chapter describes the different atmospheric radars and specifically the TARA Furthermore the tools that will be used are shortly discussed First in section 4 1 an overview of atmospheric radars is given then in section 4 2 the TARA is described and a short description of calculations is given In section 4 3 the data acquisition is discussed followed by a description of the programming language LabVIEW in section 4 4 Finally in section 4 5 a short description of Matlab is given 4 1 Atmospheric radars Over the past few decades climate change has become a hot topic The earth and its climate is a complex system With a lot of variables of which for some of them the knowledge is lacking One of these variables are clouds and their behavior When the earth gets warmer it is likely that more water out of the oceans evaporates and more clouds will develop At the moment there is not much known about the influence of clouds Climate models are more or less just guessing on the effects but most of them are probably underestimating this effect 2 In order to gain knowledge on cloud systems it is necessarry to be able to measure them The most common way to measure clouds is with radar radi
22. elements outputs an error this means that the notifiers have been released because the user has pressed the exit program button The State Machine will go the Shutdown state afterwards In case there are no errors the next state will be the Data Measurement state Data Measurement state will start the Data Measurement VI This VI will keep running as long as the switch for the transmission in the GUI is switched on It will continuously collect beat signals and put them on the Beat Signal notifier every time it has collected 2 profiles bundled with a time stamp and measurement type see figure 6 1 b This beat signal can then be used in the Data Processor and Storage VIs The Data Measurement VI should only exit when the user has stopped the transmission So after it stopped running the Data Measurement state will check for errors and then return to the dle state ready for the user to start the next measurement Currently the only purpose of the Shutdown state is to exit the loop containing the Main State Machine when the user has pressed the exit program button When storage is implemented this state will be used to allow the Storage VI to finish storing data that is still in the buffers before shutting it down 6 2 GUI The GUI is the only part that the user can actively interact with In the GUI the user can input different settings The output of the system is presented in graphs and a comment box T he measurement is started with a sw
23. having its own advantages and disadvantages Four of the different properties are important in TARA That are monostatic vs bistatic FM CW radar vs pulsed radar Doppler radar and polarimetric radar Frequency modulation is used to be able to determine the range of the measured objects By comparing the received frequency with the current transmitted frequency the time that was needed for the signal to travel can be determined also the Doppler speed of the particles can be calculated For the implementation of this project we will work with LabVIEW and Matlab An overall block diagram has been made after which the more detailed block diagrams and state diagrams of the three blocks that group 2 has to design could be made The overall block diagram shows the signals that every block in the systems sends out or receives chapter 5 final block diagrams are shown in Appendix and the state diagrams in Appendix B The state diagrams are used to describe how the sub systems accomplish their tasks Incoming data is processed by Matlab code which is provided This is one of the main reasons for the choice of LabVIEW because Matlab code can be implemented in the program via Matlab nodes Variables that should not be changed during measurement and should be used by multiple different systems are put into global functional variables Variables that can be changed are sent via notifiers After everything was designed and some choices for implementati
24. is done the real measurements start and the data processor will process the data it receives so that conclusions can be drawn and visualizations can be made Because of the large amount of data coming from the radar control unit the processing block may end up struggling with processing all the data if the user has opted to record raw data as well which will consequentially slow down the entire system Therefore it needs to be able to skip data packets so that none of the raw data that is recorded will be lost the processing can then also be done afterwards The visualization will be done by the visualization block which will produce a real time visualization of fifteen different variables clustered in five groups two for the polarimetric measurement types and three for the combined polarimetric wind measurement types In order to do this the visualization block receives the processed data from the processing block which also provides information about the scaling of the graphs for the visualization The data will be displayed on 2D graphs in which different colors will provide information about the atmospheric particles as a function of time and height visualization block will also generate a quicklook file which will be stored so that the user can see what was measured at a given moment in the past Finally the processed data and the quicklooks need to be stored on a hard disk which will be done by the storage block This block should put
25. measurement via the GUI Therefore the signal from the Transmit ON OFF notifier also generates an error if the value is false in order synchronize the shut down operation with the other two errors that can cause the VI to shut down State Machine State Machine in the Data Processor is used to run the Matlab nodes which contain the code for the processing of the noise and measurement beat signals in the right order and choose the correct Matlab node for the type of measurement 3 or 5 cycles The State Machine starts in the Idle state where it waits for beat signal to be put on the Beat Signal notifier When the beat signal is retrieved from the notifier the Polarimetric Cycle is read from the Header Builder and compared to the standard polarimetric cycles for 3 and 5 cycle measurements If they match the next state will be the Noise Processing state If not it will remain in the Idle state When entering the Noise Processing state the beat signal cluster coming from the Beat Signal notifier is unbundled in order to separate within the cluster the beat signal from the time stamp corresponding to when the beat signal was taken and the measurement type which is of either the noise or transmission type Because noise measurement is required to get processed data the beat signal is checked if it first supplied the noise measurement type If not the state machine will return to the Idle state If the measurement type is correct the h
26. picture can be measured in real time Normally the program would filter this out but that function was put out on purpose since no clouds were present to be measured In this way it is shown that the system functions correctly As shown in chapter 7 except from the non implementation of the storage block most of the other requirements have been met Summarizing it has been demonstrated that real time measurement can be performed with this system 9 2 Recommendations system worked well during tests at the TU and at the site in Cabauw This doesn t mean there is no room for improvements Following are 5 recommandations for improving the system As mentioned the storage block has not been implemented but a block diagram and a state diagram have been designed These diagrams can be used to implement a storage block in order to make storage of the measured data possible It is recommended to do this as soon as possible since putting the data in a database is important for doing the research for which TARA is build One of the requirements was that the graphs are scaled according to the UTC time on the x axis This has not been achieved due to a lack of knowledge of LabVIEW at this point Research has been done but a solution has not been found yet Some further investigation could be driven at this point so that this feature can be implemented in the product At this moment inputs have been created on the GUI to set up a measurement sche
27. pp 22 27 2007 no 3 Ronald E Rinehart Radar for meteorologists fifth edition Nevada Missouri Rinehart Publications 2010 pp 1 14 Herman Russchenberg 2011 May TARA the S band Transportable Atmospheric Radar Available http home tudelft nl en research knowledge centres irctr research radar research facilities tara Transportable Atmospheric RAdar TARA User Manual TU Delft The Netherlands 2005 pp 1 7 S H Heijnen L P Ligthart TARA Transportable Atmospheric Radar 28th Europian Microwave Con ference 1998 pp 61 66 S H Heijnen L P Lighthart and H W J Russchenberg First Measurements with TARA An S Band Transportable Atmospheric Radar Phys Chem Earth B vol 25 no 10 12 pp 995 998 2000 Silvester H Heijenen et al dedicated computer system for FM CW radar applications Journal of telecommunications and information technology pp 21 25 2001 no 4 John Proakis Dimitris Manolakis Digital Signal Processing fourth international edition Upper Saddle River New Jersey Pearson Prentice Hall 2007 pp 416 425 MATLAB Function Reference The MathWorks Inc 2010 fft Y Dufournet Ice Crystal Properties Retrieval doctoral dissertation Dept Telecommunications T U Delft Netherlands 2010 pp 47 48 D Spinellis Choosing a Programming Language IEEE Software 62 63 2006 July August NI Developer Zone 2011 June Are LabVIEW global variables good or bad
28. sp Storage Dataflow M Stp Wait for data pa me 8 F gt _ A Check exists File exists v p A al m N directory r not full gt y Dir File b b 4 dea doesn t exist 7 Put data in netCDF Create Create fle y directory uiar a ue a Visualization Dataflow 3 me mS R i y i A b Ceci CU VA Idle Adapt scales check cycle gt Cyde 25 u Sa 428 cycle visualization Cad lt 35 Amplitude Amplitude Amplitude Appendix C User interface Wg 0 4 T 4 E NO anusuei e 2 4 m gt 5 5 5 f 5 0 0 Y 00 00 4 2 uo uongada uoneung sum 31235 8 4035 HEIS gt 5 6 4 4 j 1955 wed 8p ole syooppin cien ayy 25I0N 0 0 2 0 D suondo ue 55 2044 2 f as vor ndo Burnes pue E sajddo abue Tu 5 00 00 00 4 aaa ja i sup T daams uoisinoxe fouanbai4
29. the chart automatically draws a new vertical line when new information comes in When the chart is full the oldest line the one on the left slides out of the box and a new one is drawn on the right untill the real time buffer of the chart is full The lenght of this scrolling option can be set automatically but should be limited in order to not saturate the physical memory the charts receive different information The information is represented in an array which is sent by the visualization block in the Visualization for GUI notifier The axes are all scaled programmatically in the GUI block The information needed for this comes from the visualization block in the Axes notifier The scales are set with property nodes The y axes are set in 20 for loop because of the fact that they should all be the same For the y axes the property node gets 3 values a minimum a maximum and a multiplier The z axes are all set individually each node receives an array of colors and corresponding values System comments The system comments box displays all the information about the system This includes warnings errors and the version of the matlab code The information for the box comes from the system control where all the comments are collected and put in a queue In the GUI the strings should be collected and displayed This is done by concatenating the new string with the old one with in between an enter in order to display them one unde
30. the quicklooks so a buffer and quicklook generator are not present in the visualization block Production and commissioning The software is made modular it consists of a number of Vis so it is adjustable to other systems It is easy to put in new VIs or remove unnecessary ones also changes to the VIs can be made easily because of the modular structure After the software is installed it is immediately ready for use this installation process can be done in 60 minutes for the whole system including the parts of the other subgroup The system was operational at the 7th of June 2011 software is documented at some points but at other points there is a lack of documentation This is because of the fact that time was limited and the emphasis of the project was on making a product of good qaulity Liquidation The un install process can be done within 60 minutes again for the whole system 7 4 Production the requirements for production have been met software is build in LabVIEW which is a standard software packet for building virtual instrument in order to take measurements The pieces of matlab code are inserted in LabVIEW and can even be changed easily The PXI system is used for the data collection 25 7 5 Conclusions and recommendations Most of the requirements listed in chapter 3 are met And for the storage most of the requirements are met by the block and state diagram that have been made Also the storage bo
31. was made race issue is not present here because only one VI sets the data for the Header Builder but the performance of global functional variables is better than the performance of global variables A global functional variable called GUI refnum select is also used in the VI for the GUI the references to the properties of the graphs that display the measurement data are bundled and the references to the header are bundled and put on the global functional variable T he references can then be used in the two different while loops in the GUI This allows the GUI to clear the graphs in the while loop that waits for the start of the transmission and at the same time to constantly update the graphs in the while loop that waits for a visualization packet for the GUI The choice for notifiers in the design the variables that can be changed during system operation are sent and received via notifiers This gives the possibility to pause VIs pending on the data they need to continue which also means that the VIs within the system will be automatically synchronized For example when the Data Processor needs a beat signal to process it will wait for the PXI to send it and before the Visualization can start creating visualization packets for the GUI it will wait for the processed data to be received from the Data Processor Every VI will operate as soon as the data is available and wait for the next set of data after it s done o
32. DELFT UNIVERSITY OF TECHNOLOGY BACHELOR THESIS Designing for TARA Data processing visualization and storage Authors Enzo den Engelsen Ester Stienstra June 14 2011 Supervisors Yann Dufournet Tobias Otto Simone Placidi Christine Unal Preface This thesis is part of the Bachelor end project of electrical engineering at Delft University of Technology With a group of five people we are responsible for the programming of a new system for TARA which is owned by the ATMOS group The ATMOS group is part of the remote sensing of the environment depart ment of the electrical engineering mathematics and computer science faculty of delft university of technology Here research is done in order to gain better knowledge about the influence of atmospheric phenomena on climate change As subgroup of two also referred to as group 2 we designed the part that will process the incoming data visualize the processed data and store collected data on a hard disk This thesis describes how we have designed this part which design choices were made and how the system is finally implemented Group 1 a group of three people is responsible for the control of the radar Their work can be read in the thesis 1 The subgroups combined are responsible for the main structure of the system and the user interface We would like to thank all of the people of the ATMOS group For giving us the opportunity and helping us with this challenging project
33. In the design during the test there has been a noise data collection of 15 seconds and the processing of this was done in a few seconds There have been tests with the test VI where the results looked promising During those tests it took around 40 seconds to process the noise data which is well within the limit of 60 seconds However during implementation of the data processor in the system it became apparent that the communication with and calculations in Matlab slowed down The processing of 2 profiles in the test VI takes 2 5 seconds in the Test VI but takes 4 seconds in the implemented data processor This could possibly mean that the processing of the noise takes longer than the 60 seconds it should take but further tests will need to be done to be sure The processing block should provide the correct scales for the axes At this moment only the y scale is provided The z scale is fixed although the visualization block is made in such a way that flexible z scales would be possible The x scale is relative time instead of UTC time At this moment there is not enough LabVIEW knowledge to build graph with a scale depending on a timestamp storage block has not been build However checkboxes are present in the GUI in which the user can choose which kind of file he she wants to store With this information nothing is done at the moment Because the storage block has not been build in the visualization block there has been no attention for
34. K FK FK FK FK FK FK K K K FK FK K FK K FK 3K K K 4 END PROFILE LOOP CCBAC AK AA AAAI RRR k kkk kkk kk kkk END PROCESSING CBA VA y FKK KK FKK K K FK FKK FKK FK K FK FK K FK K FK FK ok ok ok FK ook ok ok ok ok FK K ok ok ok K FK FK FK FK FK FK FK K FK K ok ok OK OK OK Suspicious data still remaining after the processing Doppler width excessively small or large are discarded JL 2 K ak aK ak ak ak a 3 3 3K 3K 3K 3K K K aK aK ak ak ak aK 3K 3K 3K 3K 3K 3K K K aK K aK 2k ak 3K 3K 3K 3K 3K 3K K K aK K 2K 2K 2k 2K 3K 3K 3K 3K 2K K K K K K K K K 2K 3K 3K 3K 2K 2K 2K 2K K K WMB Wh mask using the Doppler width WMB WMB 1 velres NaN main beam WMB WMB 1 gt 3 NaN for ns 1 NPROFS 2 waboi find isnan WMB ns 1 main beam mask Zh waboi ns NaN Vh waboi ns NaN 74 Wh waboi ns NaN Rhoh waboi ns NaN end ok kk ke k RA k kkk k STORAGE height range sin EL pi 180 hnear
35. K aK K 2K aK 2k 2K 3K 3K 3K 3K 3K K K K K K K 2K 2 2K 3K 3K 2K 2K 2K 2K KK K for nr Nran 1 1 4 Doppler spectrum for each pol SZ zeros FFTLEN NPOL CALIBRATION AND SMOOTHING 7 sZ absolute calibration with calfactor for np 1 NPOL sZ np spec 1 nr np conj spec 1 nr np spec 2 nr np conj spec 2 nr np calfactor nr np 2 SZ np Specsmoother rt sZ np FFTLEN beta it 0 end 4 MAIN BEAM CROSS SPECTRUM CALIBRATION calfactor 4 PHASE COMPENSATION FOR THE NON SIMULTANEITY OF THE MEASUREMENTS 4 VV AND HH Compvvhh and phase calibration Phi cal Fvvhh spec 1 nr 1 conj spec 1 nr 3 spec 2 nr 1 conj spec 2 nr 3 Compvvhh exp i Phi cal Fvvhh Fvvhh sqrt calfactor nr 3 sqrt calfactor nr 1 2 Phase angle Fvvhh should be not smoothed for dealiasing Fvvhh amp Specsmoother rt abs Fvvhh FFTLEN beta it 0 Fvvhh pha Specsmoother rt Phase FFTLEN beta it 0 Fvvhh Fvvhh amp exp i Fvvhh pha 4 MAIN BEAM POLARIMETRIC DEALIASING The Doppler spectrum is placed at the right Doppler velocities for np 1 NPOL find Phase gt Phi np sig amp Phase Phi np sig itemp np 1 FFTLEN abo sZ_dea itemp nr 1 sZ abo 1 sZ_dea itemp nr 2 sZ abo 2 SZ dea itemp nr 3 sZ abo 3 phase correction when aliasing Fvh dea itemp nr Fvvhh abo exp i Phi np end 73 end end END RANGE LOOP
36. TLEN by LENGTH 2 LENGTH 2 range gates with FFTLEN sweeps 4 in each to do FFTLEN point FFT for each gate wind kron z y calnoise zeros NPROFS LENGTH 2 NPOL PROFILE LOOP RK for ns 1 NPROFS 4read in data block for one profile allbeat beatsig 1 LENGTH ns 1 FFTLEN NPOL 1 ns FFTLEN NPOL fallbeat fread fid LENGTH FFTLEN NPOL short folder version 4 Run through all polarization beam states for np 1 NPOL extract the beat signal for one polarization state beam polarization beam changes from sweep to sweep beat allbeat np NPOL FFTLEN NPOL remove the DC component from the beat signal 2 ones LENGTH 1 bm mean beat beat beat kron z bm 66 do range FFT with windowing beat beat winr chwin fft beat discard image half of matrix and take transpose since we will 4 now do FFT over other dimension Make both components zero mean and then apply the Hamming window over each column 2 ones FFTLEN 1 chwin 1 LENGTH 2 iqw iqw complex kron z mean real iqw kron z mean imag iqw iqw iqw wind now do FFT over all columns to get spectrum specch fft iqw specch specch conj specch FFTLEN finch fftshift specch 1 7 flip fft center O frequency determine the measur
37. This language should be easy to learn because of the time linitation and should be able to communicate with the PXI system NI has developed its own programming language to accompany their products which means that it is relatively easy to use that language together with the PXI system This language is LabVIEW which is a graphical programming language In this language programming is done by selecting blocks with the correct functionality and connecting those blocks with wires For most engineers this is more intuitive than a text based programming language One of the problems faced with LabVIEW is that there was almost no experience with it at the beginning of the project by both the project group and the department This implies that a lot of time was spent on learning how to deal with the language instead of implementing the desingns An advantage of LabVIEW is that Matlab code can be included in the project via a Matlab script node this was very useful because Matlab code for the calculations was provided 4 5 Matlab process the data coming from the PXI system a lot of calculations have to be done LabVIEW mostly deals with the dataflow from one block to another Small and simple calculations like adding or multiplying two numbers can easily be done in LabVIEW but bigger calculations become a problem very quickly This is why LabVIEW is offering the opportunity to implement Matlab code in a special designed block For TARA the signal
38. all the data it receives in a buffer during the measurement and when the measurement is stopped or the day is over the information needs to be saved to files on the hard disk The storage block should make sure that the files are saved in the right directory and that the files and folders are named correctly There are four types of files that need to be saved which are the noise data files the processed data files the raw data files and the quicklooks The user can indicate which type of measurement needs to be done and saved on the user interface For each file type except for the quicklooks different packages of data have to be created and stored into files The storage block needs to make sure that the different types of packages are put in the correct buffers After the measurement has been stopped or the day has ended the block should take care of saving the different files to the right location one at the time 2 3 Framework of the project As with every design there are limitations present with the biggest limit being the short period of time that is given until the project must be completed There are only seven weeks for both the design and implementation of the software Because of this there is a need for rapid prototyping because at the end of the 7 weeks there should at least be a proof of concept Furthermore the ATMOS group will provide Matlab code that is to be used for the processing block within the design which means that the s
39. ar constant K2 0 93 4 K 2 related to water permittivity beamw 2 1 pi 180 4 beamwidth deg GdB 38 8 4 antenna gain dB G 10 GdB 10 4 antenna gain linear PJ 512 log 2 4 Probert Jones correction factor FF 329866 TARA carrier frequency Hz CC 3 08 light speed m s L CC FF radar wavelength m Zc 1 18 dbZ conversion factor Dr 3e8 2 FEQ 1e6 range resolution m radar constant C with dbz conversion factor included Const PJ L 2 Zc pi 3 G 2 beamw 2 K2 Dr the radar resolution volume must be multiplied by exp C2 r 2 to account for the non overlapping of the antennas in the near range This correction 4 takes in account the spacing between the two antennas of Delta 5 52 m the parallax error 0 5 deg is assumed neglectable transmit and receive antenna are looking in the same direction and have the same 4 beamwidths Delta lt lt range minimum value 100 m Delta 5 52 C2 2 1og 2 Delta 2 beamw 2 4 Transmit power outside the antennas Pt 41 3 4 in watts Tnone 0 0001 Tsc Ts Tnone no data collected at the begin of a sweep 4Tsc 7 8 Ts 4 old version Calculation of the system noise bolt 1 38e 23 Boltzman s constant NfdB 1 noise figure dB Nf 10 NfdB 10 noise figure linear Ta 50 antenna temperature K Tr 290 room temperature K Tsys Ta Tr Nf 1 system temperature Bandw 1 Tsc s
40. at the unused graphs stay black an array filled with zeros is sent to the unused graphs After this the block goes back to the initialization state 6 6 Storage The design for the storage has been done a block and state diagram have been made see chapter 5 But implementation has not been realized because of time constraints and tasks re allocation between the two sub groups in order to compensate unexpected extra working load which would affect the design implementation Therefore in agreement with the user there has been decided to skip the storage part The implementation of the storage should be done based on the current block and state diagram in order to complete the program 23 Chapter 7 System evaluation In this chapter the result of the project is evaluated by looking at the system and the program of requirements the requirements made in chapter have been checked During the excursion to Cabauw at the last day of the project where the system has been implemented in the actual TARA This chapter follows the same outline as chapter 3 In section 7 1 the product use is discussed after that in 7 2 the environment and ecology of the system is evaluated Then in 7 3 the requirements for the system design are discussed and in 7 4 the production of the system is evaluated Finally in 7 5 conclusions are drawn and recommendations are made 7 1 Product use system processes the data and visualizes it in real tim
41. ation the calculations in formulas 4 1 and 4 2 can be done In order to compute the discrete Fourier transformation on a computer a variety of algorithms have been developed known as FFT algorithms Matlab has developed its own FFT algorithm 10 which is easy to use by using the command x measurement is always started with a noise measurement From this noise measurement two important variables are calculated the mean of the noise and a calibration factor The mean of the noise is later used to filter out noise from the measurement Roughly speaking all the signals that are lower than the mean of the noise are removed The calibration factor is used to convert the processed data from power to reflectivity values This factor is calculated from the ratio between the average noise measurement and the computed theoretical noise reflectivity 11 4 3 PXI System Technology is going fast so a new data acquisition system will be a big improvement of the capabilities of TARA The system should be able to acquire the data and perform some preprocessing tasks The choice has been made to use the PXI system from NI This choice was mainly based on the fact that the department already had a good relationship with NI and that NI offers good quality of service Also this PXI system is modular which means that new modules can be added later on 4 4 Labview When developing a software program first of all a programming language has to be chosen
42. aved for instance what the weather is like Also the user can input a schedule for measurements but at this moment it is not functional Finally there are three buttons one to start or stop the measurement one to reset all the settings to their default values and one to exit the program When the measurement starts the input fields are all disabled This is done via property node in for loop A property node is a block in labVIEW in which you can input a large amount of properties for the fields on a GUI One of these properties is the disable property As soon as a 717 is sent to the node the corresponding field is disabled A property node can either belong strictly to one object or different objects can be linked to it This is done in a for loop by putting different references in it one after each other Also when the measurement starts the header should be built For this purpose the inputs are bundled and then sent from the GUI to the header builder Also some calculations are done in the calc matlab block to check whether the settings for the measurement are valid If they are not the measurement will not start When the measurement is stopped by the user all the other blocks are informed and all the inputs are enabled again using the same method as when disabling them When the exit program button is pressed the measurement should be stopped and all the other VIs in the system should stop in order to be able to close the program To
43. avelength m nyguist lambda 4 Ts NPOL max unambiguous Doppler velocity m s velres 2 nyguist FFTLEN 7 Doppler velocity resolution m s vel velres nyquist velres nyquist Doppler velocities m s 4 Extension of the Doppler velocity interval after dealiasing lvel velres NPOL nyquist velres NPOL nyquist m s nyfeq 1 2 Ts NPOL max unambiguous Doppler frequency Hz feqres 2 nyfeq FFTLEN 7 Doppler frequency resolution Hz feq feqres nyfeq feqres nyfeq Doppler frequencies Hz 4 parameters for spectral smoother beta 0 1 4 moderate smoothing it 50 max number of iterations 4 NOISE FILE radar noise level calibration factor 4 scaling of the calibration factor Atlin 10 At 10 calfactor Atlin calfactor 4 determine the clip level for all polarizations 70 DBaboveNoise SNR threshold acceptable SNR in db for clipping factnoise 10 DBaboveNoise 10 4 linear 4 in the Doppler domain noisefloor meannoise calfactor FFTLEN factnoise DEALIASING MAIN BEAM use phase diff between signal vv and hh 4 Doppler phase compensation for the cross spectrum because of non 4 simultaneity of the measurements vv and hh Compvvhh exp i 2 pi feq 2 Ts Tm NPOL Ts 4 sampling time s tt 2 Ts 4 offset time between VV and HH measurements s Phimax 2 pi tt Tm 4 absolute phase increment due to Doppler aliasing imax fix NP0L 2 mean values of the cross spectr
44. calfactor 2 mean calfactor 1 calfactor 2 calfactor 1 off2 4 calibration range independent for Zdr ZHH ZVV off3 mean calfactor 3 mean calfactor 1 calfactor 3 calfactor 1 off3 TIHHBBHHHHHHHHHHHHHEHE end of noiseprocessor ht toc 4disp temps de calcul figure num2str t s plot 10 l1og10 meannoise 3 Polarimetric processor V4 function ZDR LDR ZHH VHH WHH Rhoh height cvp Pola rt V4 beatsig FFTLEN NPOL Ts FEG EL At meannoise calfactor 4 Polarimetric processor V4 4 C Unal March April 2011 Polarimetric processor 4 This script processes one datafile This data file is part of a time 4 series of data files measured with the same radar specifications 4 From the radar signal spectrograms Doppler spectra per range and 4 profiles of radar observables are obtained profile shows the variation 4 of one radar observable versus range or height at a specific time The 4 profiles are output of this script 7 Modifications C Unal and Y Dufournet 2011 real time interface 7 C Unal 2011 two spectrograms are averaged at the beginning to decrease time constraint INPUT 4beatsig 2D array raw data with transmit on VA LENGTH number of sweeps FFTLEN Doppler FFT length NPOL Number of polarisation beam states 4Ts Sweep time s frequency excursion MHz EL elevation angle deg frequen
45. ck has been taken into account in every design and implementation step of the other blocks When the block is implemented it is thus easy to connect it with the rest of the product Some requirements have not been met due to a lack of LabVIEW knowledge At the start of the project there was little knowledge about this programming language so everything had to be figured out This caused for some not implemented functions However the software is made flexible so it is easy to put in new elements After evaluation with the staff and the students it was considered that more time and manpower would have been required to fully fulfill the project tasks The product that has been designed can be used although no storage is done Someone of the ATMOS group will build the storage block to make the system complete this can be done according to the block and state diagrams described in chapter 5 The VI can then be put into the rest of the project and when connected correctly the data can be saved Also in order to make it easier to make future changes the documentation should be expanded 26 Chapter 8 Discussion In this chapter three points are discussed on which the project or the product could be improved Things that went wrong are pointed out and possible solutions are given Time As mentioned earlier the Bachelor end project has a tight time frame and deadlines have to be met Because of this decision are made that would perhaps be diff
46. cy excursion MHz attenuation dB 0 dB 100 10 10 W 20 10 etc meannoise related mean noise level measured ouput of noise processor 4calfactor calibration factor ouput of noise processor FFTLEN length Doppler FFT standard at 512 LENGTH length range FFT standard at 1024 NPOL Number of polarisation states Ts Sweep time s 68 AOUTPUT 4 ZDR LDR ZHH VHH and Rhoh are 2D array Nran near 1 NPROFS 2 ZDR differential reflectivity dB LDR linear depolarization ratio dB ZHH reflectivity value of the main beam for HH pol dBZ VHH Mean Doppler velocity of the main beam for HH pol m s 1 Doppler width of the main beam for HH pol m s 1 Rhoh complex cross correlation coefficient height height vector m above ground level code version called functions 4 Specsmoother rt signal smoothing 7 Moments sLdr2 rt first profile calculation main beam polarimetric 7 Moments pola rt definitive profile calculation main beam polarimetric K publications 4 Spectral polarimetric dealiasing 4 Unal C M H and D Moisseev 2004 Combined Doppler a
47. data This allows the user to easily check which Matlab code was used the processing of the data that is being viewed 12 Visualization Visualization block has to create packets of visualized data which the GUI can interpret and display in graphs It also needs to create the quicklooks that need to be saved The detailed block diagram of the Visualization can be found in Appendix A The block receives the header variables and the processed data as inputs The processed data is sent to the scale adapting block and the 73 or 5 cycle block The scale adapting block checks the minimum and maximum height that is contained within the processed data packet and determines how to scale the y axis of the graphs according to those values and sends them to the real time visualization blocks The 73 or 5 cycle block checks of which measurement cycle the processed data is and sends it to the correct real time visualization Both the real time visualization blocks send their signal to the next 3 or 5 cycle block that checks again of which measurement cycle the data is and then outputs it processed radar spectral polarimetric parameters coming out of Data Processor block combined with the scale adaptation for the graphs are then sent out as a visualization packet to the GUI Each of the processed parameters is range and time dependent represented as 2D matrices They are therefore visualized on a 2D graph time and range in th
48. data processor Matlab nodes were set up in such a way that it simulated how it needed to work in the Data Processor VI get a sense of how long the processing of the noise and data would take a timing test was setup in the same test VI Tests with this VI showed that there was a significant overhead when communicating between LabVIEW and Matlab When a the processing of a measurement beat signal takes 1 5 seconds in just Matlab it takes a total of 2 5 seconds in Labview which is quite significant The 2 5 seconds it takes for the processing was however still within the time it takes the PXI to receive and send a beat signal So with those numbers the data processor should be able to process all the data received from the radar without dropping a packet When the Matlab nodes were implemented in the Data Processor and the data was sent to the nodes via notifier simulating the actual working of the system it turned out that there was additional slowdown It takes an average of 4 seconds to process the data which is more than the 3 seconds it takes for the PXI to receive and send a beat signal One packet is therefore dropped every two packets that are sent to the Data Processor While implementing the Matlab code in the Data Processor there was a choice between adding the Matlab node directly into the LabVIEW environment or to create a subVI containing the Matlab node The latter choice makes the code easier to interpret and change later on by be
49. do this a signal is send so that the other blocks can finish their work to prevent loss of information After all the blocks have stopped working the program is shut down In advanced mode the user can make his own measurement cycle To do this there are 7 buttons 6 for the different polarimetric measurement types HH HV VH VV 01 O2 and one button to remove the last added type in the sequence When a button is pressed the corresponding type is added to a string In order to do this 7 cases were made one case for each button The string is a local variable which is necessary to keep it while going trough different cases Each time a button is pressed the corresponding case is executed In this case the correct string is added to the final string When the measurement is started the string is put in the header builder and thereby passed on to the radar control Graphs There are 15 graphs all of the same type These graphs show the processed data in 2D graph the x and y axes always represent relative time and height respectively The z axis is a color representation displaying the value of the variable shown in the graph The 15 graphs are grouped in 5 groups of 3 displayed on different tabs The 5 groups are main beam offsetbeam 1 offsetbeam 2 polarimetric and wind This way the GUI stays orderly without throwing away measurement information For the graphs intensity charts are chosen An intensity chart is a chart that shows values by color
50. done in the noise processing state After the processed noise data is obtained the Data Processor will need to either start processing a 3 cycle measurement a 5 cycle measurement or return to doing nothing depending on whether the user has selected to do 3 or 5 cycle measurement or just store raw data Two states are added in order to do this Both a 3 cycle signal processing state and a 5 cycle signal processing state to do the respective processing Data Processor needs to keep processing data whenever a new beat signal is received so it needs to remain in its current state be it the idle 3 cycle signal processing or 5 cycle signal processing state until the measurement is stopped Visualization State Diagram The Visualization part is started alongside the Data Processor and needs to wait until it receives processed data it can display So it remains in an idle state until it receives the required data After it receives the data it needs to scale all the graphs using information from the data packet so the processed data will be displayed correctly When this is done the Visualization part needs to check whether a 3 or 5 cycle measurement needs to be on the GUI There is a scale graphs and check cycle state created for that because it needs to be done before anything is visualized When the graphs are set up to visualize the processed data it will move to the 3 cycle visualization and create quicklook state or the 5 cycle visualizati
51. dule This schedule is not implemented nor is there design to implement them within the system full investigation of the possibilities in LabVIEW for creating such a schedule is required After which implementation can be done The software should run stable on a PC This is not yet the case since for an unknown reason the PC has to be restarted every now and then approximately every 15 min because it becomes very slow It could be that the memory becomes full and old data is not removed A possible solution would be to limit the amount of data stored in the graphs but further investigation on the way the overall system is programmed should be done in order to accurately detect and resolve this issue Finally the LabVIEW code Matlab code and the communication between them should be investigated in order to see if improvements can be made considering the timing At this moment the processing block is the slowest of the system so improvements should start from there It is not known if the noise processing is done quickly enough when integrated into the complete design so this should also be investigated and afterwards perhaps improved 30 Bibliography 10 11 12 13 Jeroen van Gemert Martijn Janssen and Satoshi Malotaux Designing for TARA The radar control unit Bsc Thesis Electrical Engineering TU Delft The Netherlands 2011 Muller M Climate change clouds remain the misty factor Delft Outlook
52. e The reflectivity value of the offset beam 1 in dBZ e Zog2 The reflectivity value of the offset beam 2 in dBZ Vopi The mean Doppler velocity of the offset beam 1 in m s e The mean Doppler velocity of the offset beam 2 in m s e Wog1 The Doppler width of the main beam for HH polarization in m s e Wog The Doppler width of the main beam for HH polarization in m s e V The vertical mean Doppler velocity in m s e V The horizontal wind speed in m s e D horizontal wind direction in degrees As with the previous two Matlab nodes this Matlab node also provides the code version which will be put on the code version notifier 22 6 5 Visualisation The visualization block takes care of visualizing the measured data First the scales of the graphs are set correctly Then the data is sent to the graphs At the end of a measurement or a day quicklook should be made but this has not been implemented yet because of the lack of time State Machine At startup the visualization block starts in the initialization state In this state the polarimetric cycle that is used for the measurement is extracted from the header builder for later use Furthermore this is the only state in which the visualization sub VI can be shut down by the system in order to prevent the loss of quicklooks After the initialization state the block goes to the Adapt scales state In the Adapt scales state the y scales and the z scales
53. e in other words the user has the feeling that everything is happening immediately after measuring The user can select which measurement needs to be taken and all the settings stay visible when the measurement is started so the user can see if everything is set up correctly The graphs are scaled automatically and the units of the radar parameters are shown next to the graphs Due to a lack of time because of a change in task division and the complexity of the system to be implemented the storage part has not been implemented these requirements are therefore not met However a block diagram and a state diagram have already been made for the storage block and is therefor ready to implement In these diagrams it has been taken into account that different types of files have to be stored on the same hard disk Also knowledge of the standard file formats is present at the department A header has been made in order to hand all the necessary information to the different blocks so this element can be added to the files when data is saved radar parameters should be plotted as function of UTC time and height height is used to scale the graphs but the x axis are not showing the UTC time The x axis are now showing relative time compared to the starting time The demand of showing UTC time is most important when making a quicklook because when a measurement is taken it is known which time it is at that moment but when data is taken out
54. e GUI Storage Storage block needs to store the data so it can easily be found back and extracted The detailed block diagram of the Storage can be found in Appendix A The noise data processed data beat signal and the quicklooks can all be sent to this block along with the header information depending on what the user desires For the sake of clarity and in order to easily trace back the measurement specification the header data are always combined together with the different data stored There should be 4 types of data packages that can be stored e noise package which consists of the header processed noise data from a measurement of 2 minutes and 1 minute of raw data e transmission package which consists of the header and the processed data e raw data package which consists of the header and raw data e quicklook package be able to different packages at the same time several buffers are considered to be used in parallel after the merging of the required data has been done When the buffers are full the data should be written in NetCDF format Finally the data should be stored into several package dedicated folders present on the hard disk These folder should also be placed within daily classified directories This folder classification tree will be handled and created by the Storage block itself The Storage block also has its own control block This block will be notified if the day is at an end so that
55. e and only one of the two would have their data added to the global variable Functional global variables also prevent the slowdown caused by global variables when large amounts of data are retrieved every time the global variable is called Notifiers work quite differently Similar to global variables and functional global variables they can be used to access and pass data between simultaneously running VIs However instead of simply calling the function the notifier is obtained before a while loop and released after the while loop has stopped Within the while loop the notifier will be constantly available and updated with the latest variables that have been put on there by another VI that is running at the same time The main advantage of using notifiers is that a loop within a VI can be paused and made to wait for the data on the notifier to update This makes it easy to run a loop within a VI only when new data is available drawback of using notifiers is that in order to not lose any data the rate at which the notifications are sent and the rate at which they are received needs to be synchronized The notifiers can also not be buffered and the data is therefore lost if it is overwritten before the VI has been able to take it out 14 The choice for global functional variables in the design variables the Header Builder need to be accessed by a lot of the VIs so the choice to put the variables on a functional global variable
56. eader data required for the processing along with the noise beat signal that needs to be processed will be used as input for the Matlab node that contains the Matlab code to process the noise 21 Two variables will be computed that will be bundled together as processed noise data The code version of the Matlab code that was used in the calculations will also be provided as an output of the Matlab node in order to keep track of the type of processing performed When the noise is processed both the code version and the processed noise data will be put on notifiers for use in other VIs Processed Noise Data will also be sent to shift registers to be used for processing of the raw data in the next state When the next state is activated a first selection is done by comparing the input polarimetric cycle to the polarimetric cycle for 3 cycle measurements If these are equal the next state will be the 3 cycle state If not the next state will be the 5 cycle state In the 3 cycle state there will be another check to see if the measurement type of the provided beat signal is of the correct type i e Data measurement type If not it will not use the matlab code to process the data but instead skip it and remain in the state and check the measurement type of the next beat signal If the measurement type is of the Data Measurement type it will take the variables extracted from the header the beat signal and the processed noise data that was calcu
57. ed noise floor it is range dependent for TARA because of a hardware filter calnoise ns np sum finch end 4 end polarization beam loop end end profile loop end PRE PROCESSING 4 ABSOLUTE CALIBRATION 4 Absolute calibration is obtained by calculating the ratio between the 4 theoretical noise reflectivity and the measured noise calfactor 4 Calfactor depends on range and polarization beam Calfactor should not 4 contain the statistical fluctuations of the measured noise need for averaging and smoothing Average noise measurement meannoise squeeze mean calnoise 4 Smooth noise measurement beta 0 1 4smoothing parameter it 50 smoothing paramter for np 1 NPOL meannoise np Specsmoother rt meannoise np LENGTH 2 beta it 0 end tranZcalc Zcalc 7 Measured noise reflectivity Zcalc calfactor zeros LENGTH 2 NPOL for np 1 NPOL calfactor np tranZcalc meannoise np end correction for transmitter imbalance 67 calfactor calfactor 2 1 VV 4 for HV only not VH 10 0 35 10 calfactor 1 10 0 35 10 calfactor 2 4 calibration range independent for Ldr v ZHV ZVV off2 mean
58. erent if there would be more time Also some choises had to be made in order to keep the qaulity of the delivered product These choises where the cause of a smaller product LabVIEW At the start of this project the language LabVIEW has been chosen to program in This was mainly done because of the fact that with this language it was possible to have a prototype within a short time which is also very flexible for later adjustments In the beginning of the project knowledge of this language was limited both in the project group and in the department This meant that a lot of time was spend on getting to know LabVIEW The languages well known by the project group are Matlab and java both not considered suitable to use for the main frame of the radar system Although Matlab has been used for most calculations it is not suitable for building a main structure Another language that could have been used is C This language has not been chosen because of the fact that it is harder to learn and harder to keep a good overview of what has been programmed With C it would be possible to use Matlab code There are even two options to do that Matlab can be called in C code or Matlab code can be translated into C which can easilly be done in the Matlab compiler If Matlab is translated in C and then put in the project the program will probably run faster because of the fact that only one platform is running In LabVIEW two platforms are running that should c
59. erwards the GUI and the subroutines for the Data Processor Visualization and Storage are disabled by setting their case to false when Radar Control first runs The GUI should not be displayed until the radar is set up correctly and no errors are found and the three subroutines should not run without the header data that is provided by the GUI The Main State Machine The Main State Machine is the part of the Radar Control that controls when certain VIs should run and when notifiers are sent to the different subVIs of the system The first state it enters after it starts up is the Radar Initialization state In this state the radar system checks for errors and sets the initial values for the PXI When this is done and no errors are found a signal is sent via notifier to the case containing the GUI The GUI will then run and pop up on the screen allowing the user to input the desired measurement options The State Machine will then enter into an Idle state The State Machine will wait in this idle state until it receives notification from the GUI on the GUT control of Radar Control notifier This notification indicates by sending along the next state that either the measurement options have been set and the switch has been turned on to start the measurement or that the exit program button has been pressed Regarding this notification the next state will either be the Noise Measurement state or the Shutdown state In the Idle state the case
60. f the notifiers are obtained in the Radar Control VI and then wired to the inputs of the other VIs that use them This means they are only obtained once and can all be released in the same VI Now if a wait for notification element in one of the subVIs is waiting when the notifier it belongs to is released it will generate an error within that element This error signal can then be used in order to shut down a subVI 16 This decision is a response to the problem that arises when a subVI that creates and releases a notifier within itself it can end up getting stuck on the wait for notification element This happens when this element is waiting for a notifier that is supposed to be sent by another VI but the VI sending the notifier has already been shut down The loop in which the wait for notification element is present will then continue running causing the VI to never shut down which will result in system instability or data corruption A LabVIEW control file was created to store the names of the different notifiers that are called in the entire program This prevents spelling errors when obtaining a notifier and makes it easy to look up if a VI is receiving all the required notifiers Timing Choices ensure the correct working of the Matlab code in the LabVIEW environment a test VI was created to understand how the Matlab node needs to have the data delivered in order for it to work correctly and how it outputs its data The noise and
61. from a database this is harder to know At this moment no quicklooks are made when this is implemented it also might be possible to make quicklooks with time 7 2 Environment and ecology The software is running on a normal PC under Windows The PC that is used is a high end consumer PC with Microsoft Window 7 as the operating system The system is user friendly and has been created in close interaction with the people working in the ATMOS RSE group who are the future users The software should run stable but this is not entirely the case yet Every couple of runs the computer has to be restarted 24 the likely problem is that parts of the system mainly graphs and processor uses too much memory or does not release all of it 7 3 System design Usage features system is designed using an open and flexible approach so it allows for easy implementation of new functions and simple debugging The user is able to update the code easily and the matlab code written is easily implementable in the program noise processing should be done in 1 minute to make parallel computing easier This hasn t been tested in the final design because the code to collect 2 minutes of noise data which is necessary had not been implemented in the design during the final test This code wasn t there because it takes a lot of time during the test but also because it is uncertain if it is possible to transfer such amounts of data in LabVIEW
62. her add the data to the file and directory if they exist or create them and then add the data This is done in the check directory create directory check file create file and add data to file state After the data has been added the Storage part will return to wait for new data 14 5 3 Matlab and LabVIEW Important factors to consider when choosing a programming language are programmer productivity main tainability efficiency portability tool support and hardware and software interfacing 12 The different programming languages that were considered are C Matlab and LabVIEW These languages all fulfill the requirement for software interfacing in that they can communicate with Matlab The choice for LabVIEW was made because the programming language allows for rapid prototyping where everyone can design their own independent parts and easily combine them allowing full design flexi bility It is already capable of interfacing with the PXI so the only thing that needed to be learnt was how to use the elements which interface with the PXI in LabVIEW Another benefit of LabVIEW is that it is a graphical programming language which makes it easier to understand when beginning to work with the language without prior experience LabVIEW also makes it possible to run different sub systems in parallel which is required to make real time visualization possible These sub systems are defined as Virtual Instru ments VI in LabVIEW These VIs are si
63. ier Vertical speed ZScale MarkerVals Orientation ZScale MarkerVals Axess notifier in System comments A1 The state the state machine will enter when the start stop measurement switch has been switched off Enabled Disabled c o g 2 m o 5 5 are enabled here inp Transmit On Off ble GUI control elements 42 The state for the normal measurement settings for the Polarization Cycle Normal Default 7445 Doppler FFT ELE Strings P Doppler wi Range FFT Strings P v163 5 5 io cw Sweep time AD Sweep Spann 016 Frequency excursion DBt Attenuation 43 Measurement settings are not valid measurement is not started user is informed within the calc matlab vi Transmit ON 2 Pise Start Stop measurement rmm n mss 44 The events for when the user presses specific buttons pp 45 52 a 13 A uonezijeriur epey o sbueu anje wesboid 243393 0 103005 1epey jo 013403 119 23235 umopinus aui gt sepes uiu ee o m CEA 4015 sped ay 3x3
64. ifier with the beatsignal and process it and send the processed data away to other via a notifier Afterwards it will Measurement Type wait for a new beat signal height cvpw Polawind rt V3 beatsig FFTLE calfactor PhiN calfactor calfactor 1 57 58 Visualisation Adapt scales Adapt the scales of the graphs to get the correct visualization land decide whether a 3 or a 5 cycle should be visualized Maxima for the color scales 40 3333333 Here all the color scales calculated Height is extracted from the Minima for the color scales 3333333 processed data packets to be able to scale the y axis of the graphs Bg Color scale for orientation processed data Scycle notifier in processed data Gcycle notifier in Visualization for GUI 5 The color scales and the height for the scaling of the y axis are bundled Axess notifier in and put on a notifier for the GUI Exit program notifier in K Transmission on off 59 4 E itialize the block E get all the header info Polarimetric Cycle gt Get a 60 3 cycle Visualize the 3 cycle processed data parameter
65. iled block diagram of the Data Processor can be found in Appendix A The signals that enter the Data Processor are the header variables and the beat signal The header variables are used to process the raw data using the correct parameters The beat signal contains either noise data 3 or 5 cycle measurement data and is sent to either noise processing or 3 or 5 cycle measurement processing depending on which type of data it contains noise processing will provide 2 sets of variables of which the first is the mean noise and the second is a calibration constant These 2 variables will be used in the upcoming 3 or 5 cycle polarimetric measurement processing and also sent to the Storage block as Noise Data The 73 or 5 cycle block as the name implies checks whether the beat signal is of a 3 cycle polarimetric measurement or a 5 cycle polarimetric measurement The beat signal is then sent to either the Pola rt block if the measurement is of the 3 cycle variant or to the Polawind rt block if the measurement is of the 5 cycle variant These blocks will process the beat signal and provide up to 15 processed radar spectral polarimetric parameters that will be output as processed data This processed data that will be sent to the output of the block so it can be sent to Storage to save the data and to Visualization to display it in graphs The sub blocks of the Data Processor will also send a string with the code version along with the
66. ing able to just find the subVI containing the Matlab node in the project folder and changing the function it needs to call However after some tests it seems that when processing 10 profiles the processing takes 100 300ms longer when the Matlab node is placed inside subVI and then called Therefore in order to speed up the processing as much as possible the choice for just using Matlab nodes was made Choice for the use of intensity charts In LabVIEW there are several types of graphs each for a different type of data For the graphs in this design a display is wanted that displays 3D data on a 2D plot using colors Because of the fact that the data should be displayed as function of time and height a normal 2D graph cannot be used In LabVIEW there are 2 types of graphs that can be used for plotting 3D data with the use of colors These are an intensity graph and an intensity chart 15 The main difference between these two is that in a graph new data results in a completely new figure every time were as in a chart every time new data comes in a new vertical line is drawn When a measurement is taken with TARA a new array should always result in a new line on the graph not in a new picture This is why the intensity chart is chosen Also this chart has a history which saves data that is not visible anymore making scrolling back possible This is an advantage but the memory size should not be set to high because then the system will be slowed dow
67. itch At this point all the underlying systems are started in order to complete the measurement with the settings of the user picture of the GUI is given in appendix C The software behind the GUI is running in three separate while loops in order to keep the different processes running in parallel There is a loop for the inputs one for the graphs and one for the system comment box 19 User inputs The user has a variety of inputs that he she can give to the system Most important are the setting for the measurement For these settings the user can choose between normal settings and advanced settings These two categories are separated by a tab On the tab with normal settings settings can be made only in a way that the data processor can process the data On the tab with advanced settings the user can also make settings that are not supported by the data processor In that case only raw data will be saved In both cases the user can input the polarimetric measurement cycle the frequency excursion sweep time range FFT length Doppler length and attenuation At the advanced settings the user can choose if the delay line is on or off and input a fully customized polarimetric measurement cycle Besides these settings the user should input the GPS coordinates the elevation of the antennas and the azimuth of the radar For storage the user is able to choose which types of files he wants to save and the user can add a comment that will be s
68. l combinations of sending and receiving can be made if wanted If the reflected electric field is measured in a different direction than it has been sent the shape of an object can be determined 4 2 TARA is custom built by the ATMOS department and became operational in April 2000 4 TARA is a FM CW Doppler polarimetric radar which scans trough a column of atmosphere The radar has two antennas of 3 meters in diameter each antenna is equiped with three beams main beam and two offset beams that each are 15 degrees off axis see figure 4 1 The main beam is fully polarimetric in other words it can transmit receive both horizontally and vertically offset beams are used to measure wind and are single polarized 5 Combining the different polarizations there 6 types of measurement possible 4 using the main beams and two using offset beam 1 and 2 The 4 main beam measurement types are HH HV VH and VV using the mean beam in different polarizations where the fist letter expresses the receiving polarization and the second expresses the transmitting polarization measurement consists of a cycle of differtend types after each other lt Figure 4 1 Offsetbeams of TARA History The Project for designing and building the TARA system started in the beginning of 1995 at the Interna tional Research Center for Telecommunications transmission and Radar IRCTR The radar became oper ational in April 2000 This five
69. lated in the Noise Processing state and use them as input variables for the Matlab node containing the Matlab code for 3 cycles polarimetric data processing 7 variables are computed which will be bundled together as 3 cycle processed data cluster The cluster consists of the following variables e The differential reflectivity in dB e Lpr The linear depolarization ratio in dB e The reflectivity value of the main beam for HH polarization in dBZ e The mean Doppler velocity of the main beam for HH polarization in m s e Wau Doppler width of the main beam for HH polarization in m s e p The complex cross correlation coefficient e Height The height vector in meters above ground level This cluster will be put on the Processed Data 3 cycle notifier to be used by the Visualization and Storage VIs The code version of the Matlab code that has been used will be put on the code version notifier also for keeping track of the processing type performed on the data The 5 cycles state acts similar to the 3 cycle states the differences being that the Matlab node with the Matlab code for 5 cycle polarimetric data processing require an extra input and gives additional outputs The outputs are 15 variables that are bundles as 5 cycle processed data cluster which will be put on the Processed Data 5 cycle notifier This cluster has the following additional variables compared to the 3 cycle processed data cluster
70. lements within LabVIEW that can be used to access and pass data between several VIs that run simultaneously They are however subject to race and performance issues The race issue occurs when two different VIs try to write to the same global variable only one of the messages will be added 15 Meaning that one of the messages will be lost The performance issues occur when a large amount of data is present on the global variable and it is called by a lot of VIs Every time a VI calls the global variable a new copy of the data will be put on the wire and sent to the VI So when the global variable is put in a loop every iteration it will retrieve a new copy of the data which will slow down the program 13 Functional global variables are similar to global variables with respect to their ability to access and pass data between simultaneously running VIs The main difference between the two is that functional global variables need to be called with either an initialization or signal When the functional global variable is called with the initialization signal it takes the variables that the user wants to pass between the VIs and stores it in the functional global variable When it is called with the get signal it takes the data that has been stored before with the initialization signal and outputs it so it can be used in the VI that calls it This prevents race issues with global variables where two VIs would write to it at the same tim
71. les present in the Header Builder block are fixed and set only once before the measurement Therefore they can easily be transferred to the other blocks The Radar Control needs the measurement settings to tell the PXI and consequentially the radar what type of measurement must be done The Data Processor needs them in order to process the beat signal correctly the visualization needs to know what exactly to display the processed 3 or 5 cycle measurement data and the Storage block needs to store the measurement settings together with the data that is to be saved Radar Control The Radar Control block is designed to control and receive signals from the radar itself and start each of the other blocks when they are needed Also through the connection with the radar connector panel it is able to send signals to the PXI which then sends the signals to the radar By doing this the Radar Control block can instruct the PXI to create the type of beat signal that is required for the noise measurement or the 3 5 or customized cycle measurement Data Processor task of the Data Processor is to process the beat signal in order to extract atmospheric echoes that are received by the radar so they can be stored and displayed The processing is only performed in case of specific 3 cycle and 5 cycle measurement configurations If the beat signal is of a custom measurement configuration it will not be processed and only raw data will be stored The deta
72. milar to user made functions in other programming languages in that they can be called to execute specific parts of the code One of the main benefits is that LabVIEW allows for easy implementation of Matlab codes in the pro gram These can just be run via so called Matlab nodes where data input from LabVIEW are used as variables for the function you call within Matlab The outputs from Matlab are then converted to LabVIEW so it can be used in the LabVIEW environment In order to process the raw data coming from the radar Matlab code is provided that needs to be implemented in the program This unique design including Matlab code in LabVIEW environment provides great flexibilities that are required for continuous improvement and keeping the current system as a state of the art system The Matlab functions that need to be called are Noiseprocessor_rt Pola_rt and Polawind_rt These functions will call other functions for their calculations but only these 3 need to be called in LabVIEW Examples of the Matlab code provided can be found in Appendix E The Noiseprocessor rt function takes a raw beat signal containing 2 minutes of noise measurement data arranged as a 2D array and specific header variables needed for the calculations the function needs to perform It then performs the processing on the beat signal and outputs 2 variables both 2D arrays containing the mean noise and a calibration factor for use in the 3 and 5 cycle p
73. n 17 Chapter 6 Implementation In this chapter it is explained how the different subroutines have been implemented in LabVIEW Section 6 1 details the implementation of the Radar Control as an overarching system followed in section 6 2 with how the GUI was implemented Section 6 3 briefly explains the how the Header Builder works After which section 6 4 deals with the implementation of the Radar Processor and this chapter will end with the implementation of the Visualization in section 6 5 Finally in section 6 6 some future work related to the Storage implementation is briefly discussed The LabVIEW code for these Vis can be found in Appendix D 6 1 Radar control make sure that all the different subroutines run at the appropriate time and get the right signals for their actions a design of an overarching VI was created that would run and link these subroutines together During development it became apparent that the actions of this overarching VI could be interpreted as those of the Radar Control block that was created during the design of the Main Block Diagram This overarching VI was then rebranded as the Radar Control VI The VIfor the Radar Control starts with obtaining and naming all the notifiers that are used in the systemand then wiring them to the input connectors of the subVIs that also need them After the notifiers are obtained the System Control VI is started which will keep running as long as the program is running Aft
74. nal to return With the time that that takes the distance of the object can be computed Another type of radar is the continuous wave radar this radar continuously transmits and at the same time receives a signal As expected this is only possible with a bistatic radar Also a little bit more effort is needed in order to be able to tell the distance of the target Most of the time this is done by frequency modulation If this is the case the radar is called FM CW radar By changing the frequency of the transmitted signal the time between transmission and reception can be derived from the signal How this is done will be discussed at the end of the next section very useful radar is the Doppler radar This radar uses the effect discovered by Doppler that a moving source of sound causes a frequency shift l his shift is directly proportional to the speed of the source For radio waves the same effect is present And the effect works exactly the same if the source is not moving but the object is So the velocity of the object can be calculated when the transmitted and received frequency is compared This property can be very useful when the behavior of particles inside a cloud system studied final type is polarimetric radar In this radar the transmitted electric field can oscillate horizon tally and vertically the user can choose which one to use at any point Also the receiving antenna can receive horizontally and vertically al
75. nd polarimetric radar measurements correction for spectrum aliasing and non 4 simultaneous polarimetric measurements J Atmos Oceanic Technol 21 4 443 456 4 Spectral polarimetric clutter suppression 4 Unal C 2009 Spectral polarimetric radar clutter suppression to 4 enhance atmospheric echoes J Atmos Oceanic Technol 26 1781 1797 Y 2K 2K 2K aK aK aK ak ak a 2 3K 3K 3K 3K 3K 3K 3K K aK aK K K K K K K tic dk ak ak ak ak ak ak aa K aK aK ak I dk 3K 3K 3K K aK dk 3K K I I 21 21 21 1 3K K K AC K K Rk 2 2 2 2K 2K 2K 2K K K 4 Measurement specifications 4 FFTLEN 512 number of samples or sweeps for Doppler processing LENGTH 1024 number of samples for range processing within one sweep NPOL 3 number of polarizations Ts 0 001 4 sweep time s 4number of profiles determination NPROFS 2 4 TO be determined LENGTH sizey size beatsig NPROFS fix sizey FFTLEN NPOL determined from the radar control block how much data is sent by the radar control 21 1 lowest range bin 0 m Z2 LENGTH 2 highest range bin SNR 5 signal to noise ratio for the Doppler spectrum clipping Nran Z2 Z1 1 number of range bins pola_V4 oe oko
76. new file for a new day of measurements can be created It will also receive a signal which indicates what packages the user wants to save as well as a signal that tells the Storage block that the measurement has been stopped and it needs to save the data System Control System Control block is there to receive and process errors When one of the sub blocks of the system runs into problems it will output an error to the System Control block This block will then output the error to the user through the GUI and in the worst case it will send a signal to the Radar Connector Panel to shut down the radar 13 5 2 Statediagrams State diagrams are used to describe how the sub systems would accomplish their tasks This gives a good indication of what specific events are required for the sub system to transition to its next task The state diagrams of the 3 different subsystems detailed in this section can be found in Appendix B Data Processor State Diagram When the Data Processor is started it needs to do nothing as long as it has not received a beat signal Therefore an idle state is created in which it will wait for a beat signal and only after it is received it will advance to the next state next thing the Data Processor needs to do after a measurement cycle is started is to take a beat signal containing noise data and process it be obtain noise data that can be used to filter out the noise in the measurement itself This will be
77. nt data processing visualization and storage for TARA In this chapter the conclusion will be drawn in section 9 1 After that in section 9 2 recommendations are done 9 1 Conclusions In order to design the full system block and state diagrams were made First the main structure was designed in the main block diagram and after that the blocks were worked out in detail and state diagrams were made This task required almost three weeks but proved to be extremely useful during implementation in LabVIEW Radar Control was implemented Data Processor and Visualization were implemented and worked correctly during the tests with the PXI system at the TU Delft Storage block however has not been implemented in the system in LabVIEW This was decided in agreement with the supervisors because after a while it became clear that it would not be possible to implement all the designed blocks within the given timespan Because of the fact that storage is at the end of the chain other block can be very well implemented without it On the 7th of June 2011 there has been an excursion to TARA where the new system has been installed and Figure 9 1 Measurements taken on the 7th of June at Cabauw tested At that day the system worked as was expected see figure 9 1 In the graphs noise radar artefacts 29 horizontal lines in the top picture and cars driving by at the nearby road oblique lines in the mean Doppler velocities in the middle
78. nto two parts that will be designed separately To be able to combine the parts the groups will work together to build an overarching structure in which the separate blocks can be easily implemented At the end of the project the system should be able to execute and control all processing activities necessary for taking measurements and studying the results without using any part of the old system 2 2 Objectives A main block diagram where the separate blocks can be easily implemented needs to be designed by the two groups together and will serve as the back bone of the project In this diagram all the necessary signals and communication between the sub blocks will be clearly defined so that there will be no miscommunication between the two groups When this has all been done the sub group will start to work on its sub blocks where the data processing visualization and storage aspects will be developed into individual blocks The processing block needs to process the raw data it receives from the radar control unit which is built by the other group 1 The block will also receive information on the measurement that is currently running which it will receive from the control block and the header A noise measurement needs to be done first which then needs to be processed into noise data that can be used to filter out the noise that will be present in the measurement data This information will also be provided to the user After the noise measurement
79. o detection and ranging techniques A radar sends out a radio signal this signal is reflected by any object in the sky including cloud particles raindrops insects birds and aircrafts From the signal that is reflected the power is detected by the radar Through signal processing there can be determined what the range of the object is what size it is and how large its speed is from the Doppler velocity Radars have been developed for a range of applications but in this thesis only atmospheric radars will be discussed Over the years a variety of radar types have been developed each having its own advantages and dis advantages Four of the different properties are important in TARA and therefore discussed below These properties are monostatic vs bistatic FM CW radar vs pulsed radar Doppler radar and polarimetric radar difference between a monostatic and a bistatic radar is the number of antennas The most common monostatic radar uses one antenna for both transmitting and receiving a bistatic radar on the other hand uses two antennas one only for transmitting and one only for receiving The antennas for a bistatic radar can be standing side by side but they also can be placed at different locations Of course in the last case it is necessary to aim both of the antennas at the same volume of air 3 Most of the radars used nowadays are pulse radars This means that the radar transmits a pulsed signal and then waits for the sig
80. of cloud systems and how to observe them using radar and satellite observations They do this in cooperation with other institutes in the CESAR project CESARs goal is to find out what the effects are that different types of weather have on the climate The ATMOS group owns two different radars to conduct their research TARA has been developed 10 years ago after 10 years of operation new knowledge of processing has been acquired which allows for an update that will improve measurement and processing speed The current system will be replaced with a data acquisition system which primarily uses signals in the digital domain rather than the analogue which is currently the case Also the two computers used at the moment will be replaced with one single computer which makes synchronization a lot easier main question of this thesis is How to implement data processing visualization and storage for TARA The project started framing up the programe of requirements after which a small study on the cur rent state of the art technology was done Followed by the designing of the system which took most of the time After the design was finished an implementation was made and tested on the actual radarsystem The visualization and storage should happen in the way that is usual so everyone can easily understand it The dataprocessing code will be provided by the ATMOS group and is writen in Matlab This thesis is constructed in the following way The
81. oftware for the radar needs to be created in a programming language that can understand and execute Matlab code Chapter 3 Program of requirements Modern computers are capable of doing complex calculations even in parallel at a much higher rate than the current dedicated hardware in the radar This allows for an update to improve measurement speeds as well as processing speeds To implement this update a computer that is connected to a data aquisition card is used together with a DDS This computer needs to be programmed in such a way that it can replace the function of the existing dedicated hardware and has a design to allow for new functions to be implemented easily In this program of requirements all the requirements for the data processing visualisation and storage part of the system are mentioned This program has been realized after close consultation with the future users of the system The system will be evaluated following all the requirements mentioned afterwards First in section 3 1 the requirements for the use of the product will be given after that in section 3 2 requirements regarding environment and ecology are given In section 3 3 requirements for the design will be given and finaly in section 3 4 requirements considering production will be named In a program of requirements martketing and company strategies are also mentioned but because this product is not a commercial one these requirements are not given here 3 1 Prod
82. olarimetric measurement processors and a string that indicates which version of the Matlab code was used when reviewing the processed data it is desired to know which version of the code was used for the calculations The rt function takes a beat signal containing 2 profiles of cycle polarimetric measurement data arranged as a 2D array so it can be processed In order to process the signal it makes use of the 2 noise variables output by the Noiseprocessor rt function and specific header variables It outputs 6 radar spectral polarimetric parameters the height that corresponds to the data and the version of the Matlab code The Polawind rt function works similar to the Pola rt function but is used for 5 cycle polarimetric measurement data It uses an additional variable from the header and outputs an additional 9 radar spectral polarimetric parameters making a total 15 parameters plus the height that corresponds to the data and the version of the Matlab code 5 4 Choice motivations Because the Data Processor Visualization and Storage all need to run in parallel a way to deliver signals to constantly running subVIs needed to be found The main problem with parallel tasking is a timing issue which arise from the non ability of the subVIs to handle new sets of data while they are still busy running with previous ones Solutions for this problem are to use global variables functional global variables or notifiers Global variables are e
83. on and create quicklook state depending on which measurement cycle was selected by the user In these states the Visualization will send data packages containing the data to be visualized to the GUI and because all the visualization data is already in the Visualization block these states will also create the quicklooks that need to be stored As with the Data Processor it needs to keep visualizing as long as processed data is being provided so it will stay in the same state until the measurement stops after which it will return to the idle state Storage State Diagram As soon as the Data Processor and Visualization part are started the Storage part will start waiting for data it needs to store When it receives data it needs to put it into a buffer and keep adding to that buffer until it is full The Storage part will stay in the wait for data state until it receives the data it will then move to the put data in buffer state where it will then send the data to a buffer and return to the wait for data state if buffer is not full When the buffer is full the data it has gathered needs to be stored and the buffer needs to be emptied for the next sets of data The storage of the files needs to be in the netCDF format which will be created in the create netCDF clear buffer state The buffer will also be emptied in this state Storage part will then need to check if the directory and file it wishes to save to have already been created and then eit
84. on were made the implementation could begin chapter 6 First an overarching system was made but during the development it became apparent that the actions of this system could be interpreted as those of the Radar Control block a block that was originally designed by group 1 After the choices were made on how the Radar Control operates the design process of the GUI started and the data processing and visualization block were implemented Implementation of the storage block has however not been realized because of time constraints and task re allocation between the two sub groups Finally the system has been implemented in the actual TARA First real time measurements have been succesfully performed evaluation about what has been achieved compared to the requirements discussed in chapter 3 can be read in chapter 7 As already stated before the storage has not been implemented so requirements considering storage have not been met Besides that most requirements are met although improvements can still be made In chapter 8 a discussion is given about the points on which the project could be improved In chapter 9 the conclusion is drawn that most of the requirements are met and recom mendations are given on how to improve the system further Chapter 1 Introduction The TARA scans trough a column of atmosphere in order to observe and study cloud system behavior TARA is owned by the ATMOS group which is tasked to do research in the field
85. on z bm bm mean beat 2 beat 2 beat 2 kron z bm 4 do range FFT with windowing beati beat_1 winr chwin_1 fft beat1 beat2 beat_2 winr chwin 2 fft beat2 discard image half of matrix and take transpose since we will 4 now do FFT over other dimension Make both components zero mean and then apply the Hamming window over each column 2 ones FFTLEN 1 igw chwin_1 1 LENGTH 2 iqw complex kron z mean real iqw kron z mean imag iqw iqw iqw wind now do FFT over all column to get spectrum specch fft iqw specch specch sqrt FFTLEN finch fftshift specch 1 7 flip fft center O frequency spec 1 np finch raw Doppler range spectrogram chwin_2 1 LENGTH 2 iqw complex kron z mean real iqw kron z mean imag iqw iqw iqw wind 72 specch fft iqw specch specch sqrt FFTLEN finch fftshift specch 1 spec_2 np finch second raw Doppler range spectrogram end end loop end PRE PROCESSING POST PROCESSING 1 from Doppler range spectrograms to first step processed Doppler range spectrograms First step processed means smoothing calibration correct Doppler velocities dealiasing k k k ak aK K K K aK aK ak ak aK aK 3 3K 3K 3K 3K 3K 3K K aK aK ak 2k ak 3K 3K 3K 3K 3K 3K K
86. onstantly exchange information which can slow down the process With the main limitation being the time to finish the project LabVIEWs advantage of being relatively easy to learn was bigger than the disadvantage of slowing down the system Also if LabVIEW is well documented it is easier for someone else to understand than C code Making it easy for someone to change parts of the system n Task division During the project some tasks have been shifted from one group to another The main task of designing a main block diagram and state diagram have been executed by both of the groups combined But building the overarching system and the GUI have mainly be done by this group because the task of the other group turned out to be more difficult than thought at first Because of this there was not enough time to complete the task of building a storage block The designing has been done but there has been no implementation 27 It is likely that the storage block could have been implemented successfully if more time was allocated for this task Because design of the blocks was already achieved and LabVIEW knowledge was already gained However in order to keep the quality of the delivered work it has been chosen to deliver two properly finished blocks rather than three unfinished The two block that now were to be done are indeed of good quality 28 Chapter 9 Conclusions and Recommendations question of this thesis was How to impleme
87. problem with the current state of the TARA radar is described in chapter 2 program of requirements follows in chapter 3 after which an overview of state of the art systems is given in chapter 4 The process of designing the system and relevant subblocks will then be explained in chapter 5 After that the final implementation is described in chapter 6 and the evaluation of the system in chapter 7 Afterwards there will be a discussion in chapter 8 followed by the conclusion in chapter 9 Chapter 2 Problem description Knowledge on the field of data acquisition and processing is constantly improving so it s no surprise that the ten year old systems in TARA are starting to show their age The radar will therefore be upgraded with new hardware which will then need to be programmed in order to work as intended This chapter describes what needs to be implemented so that the software will accurately process the data the radar receives visualize the processed data in a useful manner and store the processed data the visualizations and the raw data In the first section the problem is described after that in the seccond section the objectives of the project are named And finally in the third section the framework of the project is given 2 1 Problem definition The system from the radar control to the visualization and storage needs to be built from scratch In order to most effectively achieve this the entire system has been split up i
88. processing is already done in Matlab and this code will also be used in the new design Matlab is a high level programming language specialized in making calculations with matrices Matlab does not have a big influence on the design process because the code is already written The only thing it has influenced is the choice for LabVIEW because of the fact that it can be implemented in LabVIEW 10 Chapter 5 Design process This chapter describes the design process of the project First the main block diagram of the system is presented in Section 5 1 in order to put the sub system designs handled by group 2 into context In section 5 2 the state diagrams of the Data Processor Visualization and Storage blocks are exposed in details those diagrams being the main design task of group 2 In Section 5 3 the choice for LabVIEW is explained and a small explanation of the provided Matlab codes is given Finally each of the design choices made during the implementation in LabVIEW are motivated and commented in Section 5 4 5 1 Blockdiagram and signals In order to obtain an overview of all the sub systems a top down design approach was chosen for the system started by discussing an overall block diagram This was necessary because the work was divided between two groups and the overall block diagram with carefully made signal definitions ensured that the sub blocks although designed and implemented independently from each other would work together
89. r the other In order to do this the value of the string is read out by a property node and then put into a concatenate block 6 3 Header Builder The Header Builder VI is a basic Functional Global Variable When it is called in the GUI with the initialize state it puts the bundled variables it receives from the GUI on a shift register Other VIs can then call it with the get state and it will then take the bundled variables from the shift register and output them to the VI that calls the Header Builder 6 4 Data Processor Data Processor has to receive the measurement options set in the GUI to process the beat signal provided by theradar The processing itself is done in Matlab by making use of the Matlab nodes in the LabVIEW environment This allows a great flexibility in updating processing codes which are all written in Matlab In the Data Processor VI the notifiers are wired to the notifiers outside of the while loop throught the references This is also where the measurement settings are extracted from the Header Builder and then unbundled within the while loop in order to use them for processing of the beat signals The notifier elements generate errors when the notifiers are released while they are waiting The Matlab Script Nodes also generate an error when there is a problem with the input variables In both cases where an error is produced the VI should shut down When no errors occur the VI is then shut down when the user stops the
90. s re transposed so they can col di 61 Visualize the 5 cycle 62 Matlab prosessing test test VI used to test the Matlab nodes containing the Processing codes with some test data i 5 EHE P P iC 3 g H ims 8 8 E TE EE Ic i A i 4 8 IntChart strict 5 History i P _YScale Ofst amp Mult eig Ts FEQEL At factor ZDR LDR ZHH VHH WHH Rhoh hi 100010000018 PURI EB EE 9 5 fabe Noise Processing Noise Processing as Noisep 78211984 kaltactor en Extract Transmission Beat Signal MATLAB script FFTLEN beatsig cve extract rawdata V3 LENGTH FFTLEN NPOL Ts itype 1024 B 001 H Estract Noise Beat Signal ve 512 HFFTLEN beatsig cve extract rawda m gt 1024 63 Appendix E Examples of the implemeted Matlab code Noiseprocesor real time V4 function meannoise calfactor cvn Noiseprocessor rt V4 beatsig FFTLEN NPOL Ts FEQ Tnone 4 Noiseprocessor real time V4 4 C Unal March April 2011 Noiseproces
91. sor 4 Reads raw data of TARA noise measurements with the transmitter switched 4 off 7 Uses the standard processing scheme windowing and FFT s to obtain 4 a the calibration factor b the mean noise 4 The calibration factor is polarimetrically corrected as follows 1 The transmitter imbalance is post corrected H gt V 0 35 dB 4 2 The receiver imbalance is obtained from the measured data gt V 4 0 19 dB 7 Modifications Y Dufournet en C Unal May 2011 real time interface INPUT 4beatsig FFTLEN NPOL Ts FEA OUTPUT ymeannoise calfactor hevn 2D array raw data no units LENGTH number of sweeps Doppler FFT length Number of polarisation beam states Sweep time s frequency excursion MHz time s within one sweep s where no data are collected fixed for the moment measured mean noise level not calibrated calibration factor version of the code used warning off all ftic kkk k kk 2k 64 4number of profiles determination LENGTH sizey size beatsig NPROFS fix sizey FFTLEN NPOL determined from the radar control block 4 much data is sent by the radar control noise V4 4 Calculation of the Rad
92. t one sweep takes also known as sweep time fy is the beat frequency and Bsweep is the bandwidth of the sweep Also the TARA is able to perform Doppler measurements By this the velocity of particles can be determined shift of the Doppler frequency is assumed to be lot smaller than the effect of the fm modulation so that formula 4 1 still holds Since the frequency shift is too small to use the change of the phase of the received signal in time at a fixed range R is used 66 Ro vrt 42 with vr being the radial speed of the particle t being the time and A the radar wavelength Because and t are known v can be calculated Of course the atmosphere contains a lot of particles which makes the calculations more complicated than described above but the basic formulas are still used As stated before in order to know the frequency a Fourier transform has to be done Since the signal is digital signal this is a discrete Fourier transform With this transform a transformation from a digital sequence into a digital sequence is done The transformation is almost the same with the biggest difference that the integral is replaced by a summation in order to be able to input a discrete signal The formula of the discrete Fourier transform is given 9 N 1 Ane PP k 0 1 2 N 1 4 8 n 0 The outcome of this formula is a sequence representing the frequency domain of the signal With this inform
93. the system operational until a proper upgrade could be realized April 2011 2004 BAP project Maintenance TARA upgrade January 1995 April 2000 and ERA aign Start of TARA project Completion TARA Replacing of Data processor pa g Cabauw The Netherlands 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 January 1994 January 2012 Jun 2007 Aug 2007 COPS measurement campaign Black Forest Germany Apr 2000 Jan 2012 On duty at Cabauw for KNMI Figure 4 2 Timeline of TARA Data processing Frequency modulation is used to be able to determine the range of the measured objects Each time a sweep is transmitted which consists of a sine with a frequency that is either a ramp up or a ramp down By comparing the received frequency with the current transmitted frequency the time that was needed for the signal to go away and come back can be determined be able to compare the two signals the signals are mixed with as result the beat signal From that beat signal the frequency is determined by using the Fourier transformation the frequency of the beat signal is exactly the difference of the frequency of the two signals Since it is known that the speed of radio waves equals the speed of light calculations can be made to determine the range The basic formula used for that is CT fo a 2Bsweep 4 1 where R represents the range in meters c is the speed of light is the time tha
94. uct use 1 The system has to process the acquired radar data real time 2 system has to visualize the processed data real time e Graphs should automatically be scaled correctly gt The user should be able to select the radar measurements he she wants to monitor 5 The radar parameters should be plotted as function of UTC time and height 6 The unit of the radar parameters has to be mentioned 7 The important measurement specifications have to be mentioned for the user to check whether his her measurement is the right one 8 raw data has to be stored on a hard disk 9 processed data has to be stored on hard disk 10 The quicklooks have to be stored on a hard disk in order to allow the user to see quickly what was measured at a given moment in the past 11 The system has to add a header consisting of measurement specifications to all the saved data 12 Files have to be saved in the atmospheric data standard format Netcdf 13 The quicklooks have to be saved in a standard format 3 2 Environment and ecology 1 The software has to run on normal PC under the Windows operating system 2 software should run stable without causing any instability to the PC running it 3 The system should be user friendly 3 3 System design 1 Usage features a The user should be able to update the code easily b The code written by the user containing the processing algorithms is in
95. um phase for meteorological targets in 4 case of Doppler aliasing after Doppler phase compensation Phi dea imax Phimax Phimax imax Phimax 4 conversion between minus sign due to measurement specifications 4 of TARA has to be checked for the first measurements which sign has the measured phase for 3VDmax Vdmax Phi angle exp i Phi dea NPOL 3 sig 60 deg rho co gt 0 72 sig mean diff sort Phi 2 4 max phase std permitted All the targets with co lt 0 72 are suppressed with this dealiasing Phase diff between VV and HH due to radar Phi_cal pi ISOC CC CO k kkk kkk k kkk kkk 2k OO k kkk kkk k kkk kkk 2k OUTPUTS Moments interpolated Zh zeros Nran NPROFS 2 NPOL reflectivity mm6 m3 Vh zeros Nran NPROFS 2 NPOL mean Doppler velocity m s Wh zeros Nran NPROFS 2 NPOL 4 Doppler width m s Rhoh zeros Nran NPROFS 2 4 cross correlation coefficient vv hh PROFILE LOOP k kkk kk kkk 4 for is 1 2 NPROFS ns is 1 2 1 profile number Yi Initialization spec 1 zeros
96. year project was financed by the STW 6 and the final system was to be a transportable atmospheric profiler that could be used for the study of the static and dynamic behavior of atmospheric phenomena at different geological locations 7 Real time processing was originally done on a dedicated DSP based computer system 8 This processing included filtering noise correction clutter suppression and statistical quantification of the Doppler spectrum Displaying of the measurement and the storage of the data are both done in real time In 2008 the performance of the Plessey ICs responsible for the data processing had deteriorated to such an extent that this unit needed to be replaced Also there was a need for some maintenance to keep the system running Because of the fact that FFT algorithms were now commonly available C and because of the increased processing power of the common PC the need for custom made data processor was no longer there So the choice was made to implement the data processing on a PC Because the computer used for the data processing did not have a slot for a ISA card another computer was needed for the user input and radar control That is the reason why the TARA is currently controlled by two PCs In order to provide the connectivity from the original ADC to the PC a new I O device was needed For this functionality a PCI6534 was ordered from National instruments This change in configuration was more of a patch up to keep
97. ystem bandwidth Pcalc bolt Tsys Bandw Calculated noise power 4 Calculation of the measured noise reflectivity Zcalc x 1 LENGTH 2 range x Dr Dr range starts at O m 4 range bin 1 artificially put to 1 m to avoid 0 value for Zcalc range 1 1 range2 range range 65 Zcalc Pcalc Pt Const range2 4 First 100 m of range backscattered signal not reliable zabo find range gt 100 br zabo 1 4 near range correction for the physical separation of the antennas Zcalc br LENGTH 2 Zcalc br LENGTH 2 exp C2 range2 br LENGTH 2 2K K K aK ak ak ak ak ak 3 3k 3K 3K 3K 3K K K aK aK aK ak aK aK 3 3K 3K 3K 3K 3K K K aK aK aK ak 2k 3K 3K 3K 3K 3K 3K K K aK aK aK 2K 2 2K 3K 3K 3K 3K 2K K K K K K K K K 2K 2K 3K 3K 2K 2K 2K 2K K K JL NOISE PRE PROCESSING ak ak a 3 3 3K 3K 3K 3K K I I 3K 3K K K 3K 3K 3K K K I I I K K K K K K 4 The noise pre processing should be identical to the atmospheric data pre processing in order to absolutely calibrate 4 create the window for range FFT Hamming hamming LENGTH z ones 1 FFTLEN LENGTH by FFTLEN FFTLEN sweeps of LENGTH samples each sweep in column winr kron z y create window for Doppler FFT Hamming y hamming FFTLEN z ones 1 LENGTH 2 7 FF

Designing for TARA - Repository TU Delft

Contents

Download Pdf Manuals

Related Search

Related Contents