Ray Tracing Module - CRC fails for non-zero speed values

[neilb22]

Hello Everyone.

I'm going through the Sionna tutorial on Ray Tracing. I've followed the tutorial example up until the line:

a, tau = paths.cir()

I've used sionna.channel.cir_to_ofdm_channel and sionna.channel.cir_to_time_channel. For encoding and decoding, I've used the Transport block encoder and decoders.

The point where I am confused is wrt the lines here:

l_min, l_max = sionna.channel.time_lag_discrete_time_channel(RESOURCE_GRID.bandwidth)
l_tot = l_max- l_min + 1

# Apply Doppler shifts
paths.apply_doppler(sampling_frequency=15e3, 
                        num_time_steps=RESOURCE_GRID.num_time_samples+l_tot-1, # Number of OFDM symbols
                        tx_velocities=[0, 0, 0], # We can set additional tx speeds
                        rx_velocities=[3.0, 3.0, 0]) # Or rx speeds       

I get a CRC fail if I specify non-zero speed values for RX. The CRC passes if I specify zero speeds for rx_velocities. Is there a reason why this happens?

For comparison, I tried with Sionna's TDL model instead of the "paths.cir()" for RT, and in that case, I would get a CRC pass even with non-zero min and max speeds.

[neilb22]

Hi everyone.

It's been a few weeks since I made this post. I have not been able to resolve the issue. May I request for assistance/suggestions regarding this?

Kind Regards,
Neil

[faycalaa]

Hi,

Could you share a minimal code snippet that reproduces the issue ?

[neilb22]

Hi Fayçal.
PFA the notebook file.
rt_crc_issue.zip (Edited to remove file.)

The issue I am having is with the paths.apply_doppler() function, specifically with rx_velocities.

Kind Regards,
Neil

[faycalaa]

Hi Neil,

In the code you shared, when invoking the apply_doppler function, the sampling frequency should match the bandwidth and not the subcarrier spacing.

Also, please note that there is currently a bug in apply_doppler() (see #222).

[neilb22]

I see. Thank you very much for pointing out my mistake. Much appreciated. The program works now as expected. Thank you for also mentioning the issue 222. I will keep that in mind.

Kind Regards,
Neil