• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 30 subscribers
  • Views 555 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

IWR6843: Speed disambiguation in 3Tx/4Rx MIMO TDM

Timofey Savelyev
Intellectual 560 points
Part Number: IWR6843
Other Parts Discussed in Thread: TIDEP-0090

Dear experts,

I see two solutions from TI to resolve speed ambiguity in MIMO TDM:

1. Extended Vmax CLI command that enables a correction procedure based on 3 hypotheses and corresponding angular spectra. The one with the biggest peak gives both correct speed and azimuth. This allows to unroll speed up to 3*Vmax and delivers a correct point cloud within one measurement cycle. 

2. G-track automatically unrolls the aliased speed by range rate analysis over 3-5 measurement cycles.  However, in case of MIMO the raw point cloud with aliased speed contains also incorrect azimuth due to unmatched Doppler compensation. Does G-track resolve both interrelated issues then? Is there any global parameter that tells G-track to take care of speed ambiguity?

Solution 1 can fail in case of multiple targets in the same range-Doppler bin or in case of large Doppler spread from a single target at a short range. Can G-track do the disambiguation all alone in case of MIMO at the cost of a few measurement cycles?   Thanks in advance for shedding light on this.

Best,

Tim

  • 0
    TI__Mastermind 30950 points

    Hi Tim,

    I'm not quite sure what you mean by the following:

    Timofey Savelyev said:
    However, in case of MIMO the raw point cloud with aliased speed contains also incorrect azimuth due to unmatched Doppler compensation

    If you had a target that was moving faster than the maximum velocity, why do you think the azimuth measurements would be incorrect too? I agree there will be error in the doppler measurements due to comparing multiple antennas, but I don't know what azimuth measurements you think are incorrect. Can you elaborate?

    Best,

    Nate

  • 0 in reply to Nathan Block
    Intellectual 560 points

    Hi Nate,

    in case of two or three Tx antennas transmitting sequentially we get a constant time shift between the corresponding Tx/Rx channels, which translates to a unique phase shift at each Doppler frequency. Normally we compute those phase shifts for every Doppler bin beforehand and compensate them in range-Doppler matrices coming from Tx2/4Rx or/and Tx3/4Rx. Then for complex detections in the same range-Doppler bins we perform an angle-FFT over Tx1/Tx3/4Rx virtual array to measure azimuth. Without Doppler compensation angle-FFT gives an angular spectrum with either large or small error depending on a scenario.

    In case of ambiguous speed the signal shows up in a wrong Doppler bin and becomes compensated incorrectly. For example, -80 km/h speed will show up at +40 km/h for Vmax = +/-60 km/h, and then its signal will be corrected by a wrong phase shift precomputed for +40 km/h. Apparently that results in erroneous azimuth after angle-FFT.

    I am reading all this in a TI Design Guide: TIDEP-0090 Traffic Monitoring Object Detection and Tracking Reference Design Using Single-Chip mmWave Radar Sensor, rev. Sep. 2019. It explains both solutions on pp. 26-30 for 2Tx/4Rx MIMO TDM. Furhermore, Fig. 42 shows an measured scenario with ambiguous speed but it's not clear if the correct track was obtained by the combined solutions or by G-track only. Here is the link:

    https://www.ti.com/lit/ug/tidud31b/tidud31b.pdf?ts=1660214449309&ref_url=https%253A%252F%252Fwww.google.com%252F

    Thanks,

    Tim

  • +1 in reply to Timofey Savelyev
    TI__Mastermind 30950 points

    Hi Tim,

    Thank you for elaborating. I have consulted experts internally, who have reported that the gtrack algorithm does NOT do this compensation. It assumes it's already been taken care of in earlier processing.

    Best,

    Nate

  • 0 in reply to Nathan Block
    Intellectual 560 points

    Hi Nate,

    OK, so let me conclude the following:

    1. G-track resolves ambiguous speed automatically if there is any but it considers azimuth and other parameters measured next to speed to be correct. This might explain, by the way, why I was getting ghost tracks at speeds higher than Vmax in my measurements of a single unambiguous car.

    2. G-track doesn't have any setting that disables speed disambiguation. I looked through all TI's documentation I could find and found only two cases when "velocity unrolling" was mentioned in the tracker description (Tracking radar targets with multiple reflection points, rev. 1.8).

    3. G-track as is can resolve ambiguous speed in SIMO where Doppler compensation is not needed.

    4. In MIMO TDM it's best to enable Extended Vmax in order to feed G-track with corrected point clouds.

    Agree?

    I must say it's great that TI provides such advanced algorithms to facilitate development of radar applications. Very insightful!

    Thanks,

    Tim

  • +1 in reply to Timofey Savelyev
    TI__Mastermind 30950 points

    Hi Tim,

    Agree with 1-3. With 4, it really matters how fast the objects are moving. For cars yes, for people, maybe not. I'm glad you appreciate our radar offering. I am also consistently impressed with the amount of great work that comes out of the radar department.

    Best,

    Nate

  • 0 in reply to Nathan Block
    Intellectual 560 points

    Thanks a lot for helping, Nate!

    Best,

    Tim