• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 3303 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

IWR1642BOOST: CFAR configuration for noise reduction

Dimitrios Petridis
Prodigy 50 points
Part Number: IWR1642BOOST

Hi everyone,

I am currently working on setting up the .cfg file for the IWR1642BOOST. I already changed the chirp profile according to the data that the estimator gave me and used it with the demo .exe file via Matlab. The gui seems to be working fine as I get reasonable images and it also seems to recognise my movement. But the images still seem to be noisy. Apparently I will have to change the configurations on the CFAR. My current configuration looks like this:

dfeDataOutputMode 1
channelCfg 15 3 0
adcCfg 2 1
adcbufCfg 0 1 1 1
profileCfg 0 77 7 6 51.67 0 0  67.51 1 259 5000 0 0 48
chirpCfg 0 0 0 0 0 0 0 1
chirpCfg 1 1 0 0 0 0 0 2
frameCfg 0 1 27 0 60 1 0
lowPower 0 0
guiMonitor 1 0 0 0
cfarCfg 4 1 18 16 8 4 0 63 0
doaCfg 3 1047 3 600 10 50
dbscanCfg 1 4 12 13 20 3 256
trackingCfg 1 10 10 2 3 25
sensorStart

where the cfarCfg is still set up as the default one. According to the user's guide the cfarCfg should look like this:

cfarCfg 0 2 8 4 4 0 5120
cfarCfg 1 0 8 4 4 0 5120

If I understood it right one configuration is for detection in range direction and one for detection in Doppler direction. Somehow if I use the above the program is running but I don't get anything displayed. Is there maybe a difference between the cfarCfg for the actual gui and the one for use with Matlab?
In the configuration from the guide I can also set the number of samples in which the data is averaged, which I suppose would help to reduce the noise. I can't really find that option for my use, as I refer to the excel sheet inside the demo folder to change the .cfg, which gives me the structure of cfarCfg.

Maybe someone could explain the meaning of each value so it becomes more clear to me.

Thank you in advance for your help!

Best regards

Dimitrios

  • 0
    TI__Genius 11405 points

    Hi Dimitrios,

    I tested your configuration. The error was not related to cfarcfg setting. It's the profileCfg setting with sample rate setting too low (try setting to 6MHz). In mmw/profiles folder, there's the default profile_2d.cfg which has a working configuration with cfarCfg setting as 

    cfarCfg 0 2 8 4 4 0 5120
    cfarCfg 1 0 8 4 4 0 5120

    Another thing I would like to point out is cfar noise averaging will not alter the 1D/2D output noise level in the display. It's only affecting number of objects to be detected. When cfarCfg is updated, the display of range output noise level is not expected to change.

    As for the excel sheet you mentioned, I don't find it in the released mmwave sdk folder. Can you point to the source or upload so I can access?

    Thanks and Regards,

    Michelle

  • 0 in reply to Michelle Liu
    Prodigy 50 points

    Hi Michelle,

    Thank you for your quick answer!

    Michelle Liu said:

    cfarCfg 0 2 8 4 4 0 5120
    cfarCfg 1 0 8 4 4 0 5120

    As already mentioned I tried this configuration, but somehow when I run the demo program with Matlab with the configuration it doesnt display anything. I dont know if the cfarCfg for matlab is different to that from the profile_2d.cfg, as it also looks different from that. Do you run your example with Matlab? 

    Michelle Liu said:

    It's the profileCfg setting with sample rate setting too low (try setting to 6MHz).

    Thank you for the advise! But I am not sure which parameter you mean exactly. Do you mean the number of samples per chirp?

    Michelle Liu said:

    As for the excel sheet you mentioned, I don't find it in the released mmwave sdk folder. Can you point to the source or upload so I can access?

    I attached the excel sheet. It is located in my gui folder. This is basically how I tried to understand the .cfg file and tried to adapt it.

    Thanks again!

    Best regards,

    Dimitrios

    chirpParams_TMdemo.xlsx

  • 0 in reply to Dimitrios Petridis
    TI__Genius 11405 points

    Dimitrios,

    What I tested is mmw demo. profileCfg is set to

     0 77 7 6 51.67 0 0  67.51 1 259 6000 0 0 48

    6000 means 6MHz, if using 5000 (5MHz) sample rate, I got error.

    I just realized that you are running traffic monitoring demo. I haven't done that part of test yet.

    I'll test it first and get back to you. 

    Thanks and Regards,

    Michelle

  • 0 in reply to Michelle Liu
    Prodigy 50 points
    Hi Michelle,

    Thanks again for your effort and help!

    I will have a look if the 6MHz will effect my results. Let know if you found out anything new!

    Best regards,

    Dimitrios
  • 0 in reply to Dimitrios Petridis
    TI__Genius 11405 points

    Dimitrios,

    Sorry for the delay. We got feedback from the designers. 

    The configurations included in the TM demo are tune to perform the best in the traffic monitoring scenario (including the chirp config, as well as the algorithm config). Any change to the configuration will impact the performance one way or another, or even crash the system. We do not encourage any changes to the configurations we have included in the demo. 

    Let me know if you have further question

    Regards,

    Michelle

  • 0 in reply to Michelle Liu
    Prodigy 50 points
    Hi Michelle,

    No problem, thanks again for the effort!

    Well the thing is that I tried the configuration that was delivered with the demo, but unfortunately it was not showing any reasonable results, so I decided to change the configuration. With my Profile the result at least looks better than before.
    I already talked to a contact from TI and he basically said that I might have to change the cfar configuration to get rid of the noise. I will give it some other tries and hope to get better results. Until then I will leave this topic unresolved, so maybe someone else can also provide some information!

    Thanks you!

    Best regards

    Dimitrios
  • 0 in reply to Dimitrios Petridis
    TI__Mastermind 23725 points

    Hello Dimitrios,

    This demo is tuned for Far range and objects like vehicles.

    And the accompanied configs might not yield the best results for closer distance and relatively smaller objects.

    Please give the following config a try : TM_TI_design_Near Range_clustering_tracking_0914.cfg

    You might want to tune the CFAR threshold value suit the scenario.

    As on now this TI design is provided as is to exemplify the implementation of high level algorithm such as grouping and clustering paired with detected object.

    We will be taking this as a feedback for customization needs on this design for future updates.

    Thank you,

    Vaibhav

  • 0 in reply to Vaibhav Mahimkar
    Prodigy 50 points
    Hello Vaibhav,

    thank you for your answer. The configuration file you posted seems to be working fine for close range.

    As you already mentioned, I tried to tune the threshold, but when I change the value It doesn't seem to effect my output. Is there anything else I have to change in the cfar configuration?

    Thanks a lot!

    Dimitrios