Channel Sounding Distance Inconsistency

[Triscuit]

Hi,

I have been testing out the channel sounding feature on the NRF54L15. Each device has a bias (difference between the readings and the actual distance) and I was trying to see if I can fit a function onto each device to correct the readings and calibrate them. However, I realized that the estimations are really inconsistent. I put an initiator and a reflector in the same spot and turned them on and off multiple times and almost every single time the readings were different. Sometimes the readings were pretty accurate sometimes they were way off. This makes it hard for me to perform calibration. There is no interference either as I conduct the tests in an EMI chamber. Is there anything I can do to improve the consistency or is the technology just too new right now? I know it takes time sometimes for it to settle down when restarted but this is after a few minutes already.

Sincerely,

Triscuit

[Edvin]

Hello,

Sometimes the readings were pretty accurate sometimes they were way off

What error do you see when the estimates are way off? 1m? 5m? 10m?

And what method were used in the samples that are way off? Can you show some logs?

And lastly, what sample did you use? Did you use Zephyrs:

zephyr\samples\bluetooth\channel_sounding

Or did you use the one from the nrf folder:

nrf\samples\bluetooth\channel_sounding_ras_initiator
nrf\samples\bluetooth\channel_sounding_ras_reflector

If you didn't use the last one, can you try that one, and look for the samples marked with "ifft"?

Best regards,

Edvin

[Triscuit]

Hi Edvin,

I realized it was probably because I was in an EMI chamber which maybe led the device to pick up longer paths, which led to an increase in distance readings. I moved to another quieter area and the readings were a lot more consistent. However, I compared Mode 2 and Mode 3 and I found Mode 3 to be very inaccurate at times, could it be just more sensitive? I placed two nrf54l15 boards 2m apart, and the distances averaged out to be around 3m, sometimes with readings going up to 6-7m. For Mode 2 I was able to consistently get results within 0.5m. Would you recommend using Mode 2 if the goal is to make an indoor positioning system? Also I am also trying to calibrate the distance readings. I recorded the difference between the readings and the actual distance at each distance and I am trying to fit a function on there. I found that the difference between them does not go up in a complete linear or quadratic function. There are some sudden spikes that make it hard to fit. Do you think this is the best way to calibrate it or should I take the different readings, rtt, ifft, and phase slope and find a relation between those using? Thank you so much.

Sincerely, 

Triscuit

[Edvin]

Triscuit said:

Would you recommend using Mode 2 if the goal is to make an indoor positioning system?

I recommend that you use the method used in NCS\nrf\samples\bluetooth\channel_sounding_ras_initiator, and the one that is printed in the log as "ifft".

Regarding mode 1 and 2:

/** Channel sounding main mode */
enum bt_conn_le_cs_main_mode {
	/** Mode-1 (RTT) */
	BT_CONN_LE_CS_MAIN_MODE_1 = BT_HCI_OP_LE_CS_MAIN_MODE_1,
	/** Mode-2 (PBR) */
	BT_CONN_LE_CS_MAIN_MODE_2 = BT_HCI_OP_LE_CS_MAIN_MODE_2,
	/** Mode-3 (RTT and PBR) */
	BT_CONN_LE_CS_MAIN_MODE_3 = BT_HCI_OP_LE_CS_MAIN_MODE_3,
};

Then RTT (Round Trip Time) is the least accurate. So that means that both 1 and 3 will use this, and it is least accurate. But Mode 2 is also not as accurate as the IFFT (Inverse Fast Fourier Transformation) method used in the nrf\samples\bluetooth\channel_sounding_ras_initiator sample.

Best regards,

Edvin

[Triscuit]

Hi, I am a little confused, because all modes in the NCS\nrf\samples\bluetooth\channel_sounding_ras_initiator display ifft distance, even Mode 2. Regardless of which mode I am using I see three distances, ifft, phase slope, and rtt. Since you said rtt is the least accurate and both mode 1 and 3 use it, does this mean that mode 2 would be the most accurate since it only uses pbr?

[Triscuit]

Hi, I also have another question. Before I was working on sequential channel sounding procedure with 3 reflectors and 1 initiator and it was working fine. However, when I increased the amount of reflectors to 6 I was not getting any distance readings. The first thought that came to mind was interference. I then reverted my program back to 3 reflectors and just left the other 3 reflectors on but not connected, I was still running into the same issue, where I was only receiving data from 1 - 2 of the devices, sometimes none at all. When I turned the other 3 reflectors off, it went back to normal and I started receiving data again. Could this really be interference or is there another cause I am not aware of? Thank you for your response in advance.

Sincerely,

Triscuit