IWR1843: Optimization of configuration/parameters

Hi,

We are working on this. Please give some time to get back.

Thanks

Yogesh

HI, Yoshiharu:

Here is some answers regarding to your question.  

(1) The sensor seems to detect signals that appear as ground reflections or just random noise.

These "ground reflections" appear in a range very close to the sensor, up to 50cm in front direction.

We would like to know how to avoid/reduce this. (in terms of parameters or post-processing)

[ZY]: the ground reflection is part of the radar reflection, we did not have a solution to filter this out.  When the sensor is moving, the ground reflection will keep the same distance, but your other target will move relatively.   This difference may be useful to separate the floor reflection from other target. 

(2) We have observed a phenomenon that the sensor occasionally misses the target obstacle during a single or a few consecutive frames.

Probably, it depends on threshold parameters and S/N or intensity of the target signal. We would like to know how to improve this.

[ZY] If the obstacle has a weak RCS, or placed in a wide angle relative to the sensor, then it can result in a miss detection.   To improve SNR, you can set a higher RX gain, a full TX power, and a longer chirp (longer chirp ramp time) and a larger number of chirps.  This can improve the signal to analog noise ratio, but if the performance is limited by the nearby clutters (like ground reflection, to nearby other strong reflectors), you probably need to tune or modify the detection algorithm.   

(3) We are currently using a modified version of the default parameters of TI demo (profile_3d.cfg).

We would like to make sure our configuration is optimized for a near-range application.

How can we optimize the parameters to our application ?

[ZY] for near range application, you need to make sure that you have the best range resolution and smaller number of numADCSamples in a chirp (such as 64), and you can then increase the number of chirps in the frame (such as 128) to increase Doppler resolution.  

(4) The current velocity resolution is 0.43m/s. How can we improve this ?

[ZY] Increase the number of chirp in the frame to increase Doppler resolution.

(5) How can we reduce ambiguities on speed ?

- We observe speed values of +/- 1.2 m/s when a machine which is at stop, and facing a fixed scenery

- We need around 0.1 m/s resolution

[ZY] the max non-ambiguous speed is controlled by chirp duration, decrease the chirp ramp time to improve the max Speed.  

(6) How can we improve the z-axis resolution ? 

The z values we receive have a wide spread range which does not match the height of the objects in front of the sensor.

We observe ghost describing a circle around the doppler axis, causing lot of incoherence on the angle axis.

We'd expect a z-resolution around 5/10 cm.

[ZY] Need more details to understand what do you mean. Please add some plots and more detail descriptions. And what do you mean by z-resolution around 5/10 cm?   Do you mean required accuracy on z dimension is 5/10cm?

The ISK board has only two antenna height in z direction, which will results in relatively lower accuracy for elevation angle estimation compare to azimuth angle estimation.   You can also increase SNR threshold in CFAR detection to only detect the target with strong SNR to improve the accuracy.  If you see a several detected points for one target, and results in accuracy reduction.  You can try to enable peak grouping option inside cfarCfg to improve the accuracy.

There are some training videos and slides provided for some of your topic. https://training.ti.com/mmwave-training-series

Performance tuning is application based.  The out of box demo is just a reference for you to start.  You can tune some of the parameters, but you may also need to change the detection algorithm to get better results for your application. 

Best,

Zigang