AWR1243BOOST: Wrong results in RadarStudio PostProcessing(4.75)

Naveen kanta

Part Number: AWR1243BOOST
Other Parts Discussed in Thread: AWR1642, AWR1243, DCA1000EVM, , TDA2

Respected,

As you advised me to focus on azimuth angle calculation first i tried calculating by doppler compensating with above logic and below logic which is given in doxygen document (simply math) but not able to estimate the angle properly

Que:1

Doppler Compensation

X'(m,k)=X(m,k) exp (-j* m*delta) , m= 1......,N_Tx -1 , k=0,....., N_Rx -1

Where , delta = phi_doppler /3 ----- doppler induced phase shift between consecutive sets of Rx antennas

phi_doppler ----doppler induced phase shift between subsequent chirps of the same Tx antenna

phi_doppler =2*pi*idopp / N_FFT_dop , idopp is detected doppler bin index , - N_FFT_dop /2 <idopp<N_FFT_dop /2-1

with reference of this i am trying to compensate the doppler and written a matlab code for this which is

DetectionValues = [ P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 ];

where P1,P2,P3,P4 are corresponds to Tx1 Rx1,Tx1 Rx2,Tx1 Rx3,Tx1 Rx4 & P5 P6 P7 P8 P9 are corresponds to Tx2 Rx1,Tx2 Rx2,Tx2 Rx3,Tx2 Rx4 & P9 P10 P11 P12 are corresponds to Tx3 Rx1,Tx3 Rx2,Tx3 Rx3,Tx3 Rx4 ......................................Here Tx1 and Tx3 are azimth antennas ..................

dopplerInd=CurrentDopplerInd;

phi=(2*pi*dopplerInd/ DopplerFFTsize);

deltaPhi=(phi/3); %%%%% working with 3 Tx antennas

DetectionValues(5:8)=DetectionValues(5:8) *exp(-1i*1*deltaPhi);

DetectionValues(9:12)=DetectionValues(9:12) *exp(-1i*2*deltaPhi);

but no use of this math which is mentioned in doxgyen document ................................

And I wanted to bring another point to your notice ................ today i tried simulating another static target ( X= 4 , Y =30 ,Z =10 ) by test source and captured the RAW IQ data

The RAW IQ data is processed using internal RadarStudio PostProcessing(4.75) matlab GUI verison and plotted 1642 EVM TDM MIMO XY Results where the Results should ( X= 4 , Y =30 ,Z =10 ) but the results is

X= -1.13

Y= 31.8

Z= 0

Even in TI Inbuilt Matlab GUI the results of TDM MIMO is wrong .....................

So please reply as soon as possible with best solution that you always advise !

Thanks in advance

Que :2

can you please tell me for below logic

Should the current_doppler_index in range of -dopplerfftsize/2:1:dopplerfftsize/2-1 or 1:dopplerfftsize

Please verify and tell me

dopplerCompensationIdx =current_doppler_index;

where current_doppler_index is lies between -dopplerfftsize/2:1:dopplerfftsize/2-1 ( for now the fft size is 64)

If (dopplerCompensationIdx >=numDopplerBins/2)

{

dopplerCompensationIdx -=(int32_t) numberDopplerBins;

}

%% for azimuth 1/2 and for elevation 1/3 is refered

dopplerCompensationIdx =dopplerCompensationIdx /2 ;

if (dopplerCompensationIdx < 0)

{

dopplerCompensationIdx +=(int32_t) numDopplerBins;

}

expDoppComp=azimuthModCoefs[dopplerCompensationIdx];

if (mod((dopplerIdx/2 ),1)~=0)

{

expDoppComp= azimuthModCoefsHalfBin;

}

X(5:8)=X(5:8)*expDoppComp;

X(9:12)=X(9:12)*expDoppComp;

With the above chirp configuration

Case1:

i enabled test source and generated one static target ....................

Where for static target we need not do doppler compensation right ?

Even for this also not able to get correct azimuth and elevation angle .....................anglefft(P1,P2,P3,P4,P5,P6,P7,P8)

Can you please advice me why ?

Case 2:

When the target is moving perpendicular to the Tx plane then the doppler compentation factor will be 0 ...... it means in this case also we need not perform doppler compensation

Here i enabled test source and generated one target moving away from the radar where the velocities are Vx=0 ; Vy= 1; Vz=0 ...........

For this case also i am not able to get proper azimuth and elevation angles ...........

********* You might suspect that i may do wrong in raw data collection and separation and to simplify bug i captured 64 chrips per 256 samples and one frame ......................verified properly and i found nothing wrong in my code , still the azimuth and elevation angles are wrong ! .............................

For angle estimation below procedure is followed

For separate the azimuth and elevation calculation - save the 1D FFT results from each receiver (you should do this anyways)

For each range bin with a detected point:
- construct a 1D array consisting of the 8 virtual antennas in azimuth and zero fill the rest (64 or 128 size array)
- FFT across this array to get azimuth
- this can be repeated in the elevation direction

Thanks in advance !

Note:

If needed I can provide you with adc.bin and adc_data_LogFile

over 5 years ago

0 Raghunandan Kamath over 5 years ago

TI__Mastermind 22155 points

Naveen,

Are you using the AWR1243 or the AWR1642 ? Both these devices have change in the data format. In AWR1243, the data is in the interleaved format and in the AWR1642 it is in the non-interleaved format.

Suggest you to check the section 2.1.2.1 of DataPath App note to understand the interleaved format of the AWR1243. If you change your processing according to this format you will find the solution.

Thanks,

Raghu

0 Naveen kanta over 5 years ago in reply to Raghunandan Kamath

Expert 1020 points

Here i am trying to build my own radar post processing algorithm in matlab .For algorithm input i was capturing the real time ADC data of AWR1243boost EVM with DCA1000 EVM .

Using AWR1243boost +DCA1000EVM with mmwave studio , after opening the mmwave studio 02_00_00_02 application i connected the device and updated the firmware BSS,MSS .

When setting up TDM MIMO in mmWave Studio. I defining a chirp with only a single TX antenna associated with it for that chirp. Then defining subsequent chirps with different single TX enabled.

For example, let's say that I define three chirps starting at chirp #0 and go to chirp #2. Each chirp only enables one TX. In mmWave Studio, this would look like the following:

--TDM MIMO (all three TX enabled)
ar1.ChirpConfig(0, 0, 0, 0, 0, 0, 0, 1, 0, 0)
ar1.ChirpConfig(1, 1, 0, 0, 0, 0, 0, 0, 1, 0)
ar1.ChirpConfig(2, 2, 0, 0, 0, 0, 0, 0,0, 1)

Now when I go to configure my frame to use these three chirps, this would look like the following:

ar1.FrameConfig(0, 2, 20, 128, 64,40, 0, 1)

From the above settings , I know that in adc.bin file the chirps are Tx1,T2,Tx3,Tx1,T2,Tx3,Tx1,T2,Tx3,Tx1,T2,Tx3,............................... Tx1,T2,Tx3. for all four antennas .

where Tx1 and Tx3 are azimuth antennas

Tx2 is elevation antenna

______________________________________________________________________________________________________________________________________________________________

Now the adc_data.bin file contains 128*64*256*3 = 6291456 samples in each row ( 4 rows for 4 antennas ) ...... %%% NFrame =128, Nchirps=64, Nadcsamples=256,NTxantennas=3;

Here my algorithm starts

reading data from adc_data .bin %%% using the TI code for data read ...... 16 bit ADC data , numlanes =4 , IQ data and 4 Rx antennas....................

%%% http://www.ti.com/lit/an/swra581b/swra581b.pdf in Section 9.1 DCA1000 with xWR12xx and xWR14xx MATLAB Example code

Separation of chirps configuration in accordance with Tx

Tx1,T2,Tx3,Tx1,T2,Tx3,Tx1,T2,Tx3,Tx1,T2,Tx3,............................... Tx1,T2,Tx3. for all four antennas.

for i=1: Nframe

for j=1:Txantennas

1DFFT

2DFFT

end

CFAR detection

Detection matrix

for n=1:detection objects list

doppler compensation for TX2 i.e., X(5:8)

doppler compensation for TX3 i.e., X(9:12)

3DFFT

end

angle plot

end

even processing the data as mentioned above i am not getting correct azimuth angle

When i am doing angle fft on 8 antenna peaks i am getting error .i.e., X=[P1,P2,P3,P4,P9,P10,P11,P12] combined angle fft result showing error

if i do angle fft separately for X=[P1,P2,P3,P4] and X=[P9,P10,P11,P12] i was getting result correct ........................

If needed i can also send you the adc.bin file and adc_log_txt file to understand the problem and help me through ........................

0 Naveen kanta over 5 years ago in reply to Raghunandan Kamath

Expert 1020 points

Still waiting for your reply ...

Please respond to this because to cross verify the results with RadarStudio processing GUI is also not working properly .

The results of TI Inbuilt Matlab GUI the results of TDM MIMO is wrong

Version : mmWave Studio 2.0.0.2

0 Raghunandan Kamath over 5 years ago in reply to Naveen kanta

TI__Mastermind 22155 points

Naveen,

Attached is the matlab script to perform the processing on the ADC data collected from DCA1000EVM +AWR1243BOOST.

https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/1023/Straight_5F00_Forward_5F00_Processing_5F00_with_5F00_chirp_5F00_configs_5F00_12xx_5F00_DCA1000.m

BR,

Raghu

0 Naveen kanta over 5 years ago in reply to Raghunandan Kamath

Expert 1020 points

Raghunandan Garu,

It is very nice of you for responding to my problem .

I was did little more analysis of the bug as mentioned below

when i use internal test source i was not getting proper Relative symbol phases at 8 antennas

If you observe properly the phases of Receivers is shown below.

Rx1 ----- 0

Rx2 ----- wx

Rx3 ----- 2wx

Rx4 ----- 3wx

Rx5 ----- 0

Rx6 ----- wx

Rx7 ----- 2wx

Rx8 ----- 3wx

The actual expected phase of receiver antennas are

Rx1 ----- 0

Rx2 ----- wx

Rx3 ----- 2wx

Rx4 ----- 3wx

Rx5 ----- 4wx

Rx6 ----- 5wx

Rx7 ----- 6wx

Rx8 ----- 7wx

So i understood that test source in mmWave Studio is not supporting TDM MIMO config and it will work for 1 Tx antenna only .

It is also verified with TI expert where he mention same that test source in mmWave Studio is not supporting TDM MIMO config and it will work for 1 Tx antenna only in some other thread ....

So now i am capturing the practical data by placing targets ( corner reflector ) in front of radar and it is working fine ....

One more main problem that i am facing is not able to validate my results ( Z position ).

I will explain you very clearly .Please help me as you did before .

If i don't have ideal / perfect data (test source ) to test my algorithm it will be hard to answer many questions . In mmWave Studio , Test source part has mentioned as targets can be emulated with different positions but now to get elevation data it was not supporting.

I am able to estimate X, Y and Z position of a target .

X and Y position are validated by using Radar Studio post processing GUI .But Z position i didn't find any option/setting to check even placing a target with certain height above the ground and i am not able to say it confidently.

When i downloaded the mmWave Studio 2.1.0.0 version i got matlab example codes for AWR1243 cascaded version (cascade MIMO signal processing ) for TDA2 dataPlatform but i dont have any idea to edit the code for single 12xx with DCA1000

Can you please help me and this means a lot to me .

I am using the below algorithm

I am doing separate azimuth and elevation calculations - save the 1D FFT results from each receiver

For each range bin with a detected point:
- construct a 1D array consisting of the 8 virtual antennas in azimuth and zero fill the rest (64 or 128 size array)
- FFT across this array to get azimuth
- this can be repeated in the elevation direction

The X position is very good and i am very confident , because i verified with Radar Studio post processing GUI results where two results are almost equal.

Only problem is Z position validation ....... I cant tell because in the received signal i find many scatters/ clutter signal at same range bin ....

So to validate i dont have any ideal / test source data ...

So I request you to help me in this .

I want and wish you to answer my problem.

Thanks in advance .

0 Naveen kanta over 5 years ago in reply to Raghunandan Kamath

Expert 1020 points

Still seeking for your help !

0 Naveen kanta over 5 years ago in reply to Naveen kanta

Expert 1020 points

Sir,

Any suggestions ?

0 Jitendra Gupta over 5 years ago in reply to Naveen kanta

TI__Guru 69960 points

Hi Naveen,

Latest mmWave Studio v2.1 provides few matlab script to do the post processing which may help you here for a reference.

http://software-dl.ti.com/ra-processors/esd/MMWAVE-STUDIO/latest/index_FDS.html

Look into this directory- C:\ti\mmwave_studio_02_01_00_00\mmWaveStudio\MatlabExamples

Regards,

Jitendra

0 Naveen kanta over 5 years ago in reply to Jitendra Gupta

Expert 1020 points

That is what i mentioned in my previous post ,

I thought if you people help it will be better but referring to something really won't solve the problem.

Thank you very much .

Sensors

Sensors forum

AWR1243BOOST: Wrong results in RadarStudio PostProcessing(4.75)