AWR2243: question to bistatic issue in tdm mimo

seung ha Ryu

Part Number: AWR2243

Tool/software:

Hello,

I have a question regarding DOA estimation in a multipath environment where the Direction of Arrival (DOA) is not equal to the Direction of Departure (DOD), also known as the bistatic case.

I am using the post-processing MATLAB code provided by TI.

To my understanding, it uses an 86-element virtual array to estimate the DOA in the horizontal direction using an FFT-based method, while ignoring the overlapped elements.

I understand that virtual array synthesis assumes DOA equals DOD. Therefore, in a multipath environment, the estimated DOA of ghost targets may be incorrect when DOD ≠ DOA.

So, does this MATLAB code simply assume that DOA equals DOD and not consider multipath scenario?

Best regards,
Seung Ha

5 months ago

0 Samhitha Bobba 5 months ago

TI__Mastermind 20770 points

Hi Seung Ha,

seung ha Ryu said:
So, does this MATLAB code simply assume that DOA equals DOD and not consider multipath scenario?

Yes, we do not differentiate between a signal received directly from the target and the signal received after multipath reflections. You can refer to a previous discussion on the forums (+) AWR1843: Multi-path/Refraction/Diffraction while Radar @77GHz and install on the side of the bus for BSIS functionality - Sensors forum - Sensors - TI E2E support forums.

Regards,

Samhitha

0 seung ha Ryu 5 months ago in reply to Samhitha Bobba

Prodigy 10 points

Hi Samhitha,

Thank you for your reply.

I understand that the virtual array assumes that the direction of departure (DOD) is equal to the direction of arrival (DOA). So, in the case of multipath reflections where DOD ≠ DOA, I assume the virtual array post-processing would result in DOA distortion in the estimated target location.

Is my understanding correct?

Best regards,
Seung Ha

0 Samhitha Bobba 5 months ago in reply to seung ha Ryu

TI__Mastermind 20770 points

Hi Seung Ha,

Yes, your understanding is correct.

Regards,

Samhitha

0 seung ha Ryu 5 months ago in reply to Samhitha Bobba

Prodigy 10 points

Hi Samhitha,

Thank you again for your kind and consistent responses — I truly appreciate your support.

If I understand correctly from your explanation, the TI post-processing MATLAB code seems to tolerate the position error of ghost targets caused by multipath (i.e., when DOD ≠ DOA) as a trade-off for improved angular resolution using the virtual array approach.

Assuming that's the case, I'm curious:
Did TI determine that the DOA error of ghost targets is negligible due to the small angular degree, or is it simply not considered critical due to a lack of interest in ghost positions?

Looking forward to your thoughts.

Best regards,
SeungHa

+1 Samhitha Bobba 5 months ago in reply to seung ha Ryu

TI__Mastermind 20770 points

Hi Seung Ha,

After multipath reflections, if the signal strength is too weak, we will not consider it as a valid detection. As mentioned in previous reply, we do not differentiate between a signal received directly from the target and the signal received after multipath reflections. We do the same processing in both the cases.

seung ha Ryu said:
If I understand correctly from your explanation, the TI post-processing MATLAB code seems to tolerate the position error of ghost targets caused by multipath (i.e., when DOD ≠ DOA) as a trade-off for improved angular resolution using the virtual array approach.

Irrespective of how the signal is received, you will be able to resolve two targets if you have fine angular resolution. If the angular resolution is not very fine, both the targets won't be resolved. Both these topics are different. If the target detected after multipath reflections and the actual target are not closely spaced, you will see two detections (one is actual and other is the ghost target).

In the current solutions, we do not focus on multipath propagation and there is no ghost target removal technique.

Regards,

Samhitha