Channel Sounding: counter-intuitive repetition behavior + non-configurable event interval (nRF54L15)

[ELavmap]

Hello,

we are developing a Channel Sounding application starting from the Nordic samples "Channel Sounding Initiator with Ranging Requestor" and "Channel Sounding Reflector with Ranging Responder" (SDK v3.3.0). We are using the nRF54L15 development board.

ISSUE 1 - CS_CHANNEL_MAP_REPETITION counter-intuitive behavior
Comparing two configurations, single channel map repetition (72 steps, repetition=1) gives better accuracy results than repetition=2 with the same number of steps, contrary to the expectation that more repetitions (= more IQ samples) should improve the estimate. This behavior was confirmed also in an obstacle-free LOS environment, across repeated tests.

Config A (repetition=1) - result:

Config B (repetition=2) - result:

Questions:

1. Is this behavior plausible, or does it indicate a configuration error on our side?
2. Which other parameters significantly affect accuracy and should be varied to verify a potential benefit from repetition>1?
3. Any suggestions on how to set the CS parameters to obtain real benefits from multiple channel map repetitions?

ISSUE 2 - Event interval fixed at 2, not configurable
Regardless of the CS configuration set, the logs always show event interval = 2 after CS procedures enabled.

Why is this not controllable and does not go down to 1? This affects the procedure duration and therefore indirectly the power consumption of the application, so we would like to be able to reduce it.

We remain available for any further details, thank you in advance.

Emilio

[Simonr]

Hi

This is not the expected behavior when adding more channel map repetitions AFAIK, so I have asked internally for any ideas as to what could cause this. Could you upload your project's .conf file and the .config file output in your build folder so we can take a look at what's set when you add the channel map repetition?

Best regards,

Simon

[ELavmap]

Yes.
This is my parameter setting:

#define CS_MIN_MAIN_MODE_STEPS 72
#define CS_MAX_MAIN_MODE_STEPS 72
#define CS_MAIN_MODE_REPETITION 0
#define CS_CHANNEL_MAP_REPETITION 2
#define CS_SUBEVENT_LEN_US 35000
#define CS_PROCEDURE_INTERVAL_RT 4
#define NUM_MODE_0_STEPS 3
#define CONN_INTERVAL_MIN 80
#define CONN_INTERVAL_MAX 80

I send also my .conf and .config files.

conf_files.zip

Thank you for the support.

Emilio

[Simonr]

Hi Emilio

Okay, so what could happen is that if the channel sounding procedure doesn't fit into one subevent, there will be a magnitude and phase shift between results of different subevents.

Modifying the CS configuration to make it fit in one event should make that go away.

Best regards,

Simon

[ELavmap]

With this configuration all the steps are executed in a single subevent. I checked.

The lenght of subevent is negotiate by the controller and is less than Subevent_lenght_us.

To verify that the measurement was performed entirely in one subevent, I checked the current meter and sent appropriate logs to the terminal to understand how many steps and subevents were executed.

How can I reliably verify that it actually performed all the measurements?

So, why did you tell it could be that it doesn't fit? Are any particular change that I have to make?

Thank you,

Emilio

[Simonr]

Hi

I'm saying that behavior like this would be expected if the channel sounding doesn't fit in a sub event. Do you have the logs of your measurements so we can take a look there?

Best regards,

Simon