NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

Amanda Hsieh — Tue, 02 Apr 2024 14:44:13 GMT

Hi,

Did you enable logging of error return values?

Could you check if the devices try to range?

You can check the value returned here: https://github.com/nrfconnect/sdk-nrf/blob/main/subsys/dm/dm.c#L184
( I think it is logged here: https://github.com/nrfconnect/sdk-nrf/blob/main/subsys/dm/dm.c#L440 )
also, the quality of the report could be interesting, if the ranging does not result in an error.

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

AlexVrubel — Tue, 26 Mar 2024 19:08:17 GMT

No, I don't. Can you prove, that it works with 10+ devices in role without scanning? For example if I use PCA10040 boards and your implementation from the box? Did you tested it in that way?

Thanks

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

AlexVrubel — Tue, 26 Mar 2024 18:17:22 GMT

Hi,

First of all thanks for reply.

1. I use NCS version 2.5.0, with some modifications with logs

2. Devices try to range and it ends with error - I even comment this debug message because it spams a lot

So, I took an original project nrf_dm_WithoutScanning from this thread and revert my changes with logs in SDK, but save my prj.conf file to work with custom board. Full diff of config files:

And also change mode from MCPD to RTT:

#define DEFAULT_RANGING_MODE    DM_RANGING_MODE_RTT /* DM_RANGING_MODE_MCPD */

Then I build and flash one device with SCAN_ENABLE option and 15 without. And get only errors

Full logs:

devzone.nordicsemi.com/.../3247.txt

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

Amanda Hsieh — Fri, 22 Mar 2024 14:06:54 GMT

Hi,

Were you able to resolve this issue?

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

Amanda Hsieh — Mon, 18 Mar 2024 21:11:09 GMT

Hi,

Sorry for the delay. I was out of the office.

From the logs, it looks like the timeslot allocation works now.

What NCS version are you using?

Could you check if the devices try to range?

You can check the value returned here: https://github.com/nrfconnect/sdk-nrf/blob/main/subsys/dm/dm.c#L184

( I think it is logged here: https://github.com/nrfconnect/sdk-nrf/blob/main/subsys/dm/dm.c#L440 )

also, the quality of the report could be interesting, if the ranging does not result in an error.

Do you have any modifications to the sample? Could you share the project?

-Amanda H.

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

AlexVrubel — Fri, 08 Mar 2024 17:51:13 GMT

Hello, so for I tried:

- Increase advertise interval from BT_GAP_ADV_FAST_INT_MIN_2 to BT_GAP_ADV_SLOW_INT_MIN and BT_GAP_ADV_FAST_INT_MAX_2 to BT_GAP_ADV_SLOW_INT_MAX

- CONFIG_DM_TIMESLOT_QUEUE_COUNT_SAME_PEER: 1 (10 by default)

- CONFIG_DM_RANGING_OFFSET_US reduce to 120000 (120 ms, was 1.2s)

- CONFIG_DM_TIMESLOT_RESCHEDULE tried y and n

- CONFIG_DM_TIMESLOT_QUEUE_LENGTH tried 1, 3, 5, 8, 40 (by default), 100

- Tried decrease CONFIG_BT_BACKGROUND_SCAN_WINDOW from 1024 up to 4

Nothing helps, and get only worse. Instead of 5-8 rangings from the start, I get only 1-2-3 at the beginning and only 1 every 2-3 seconds after ~5 minutes.

I use 1 devise with SCAN_ENABLE and 14 without scanning

Also I added logs to data_cb function in main.c:

And in logs I can see up to 8 packets per second:

devzone.nordicsemi.com/.../1643.log.txt

My config file:

devzone.nordicsemi.com/.../1732.prj.conf

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

Amanda Hsieh — Fri, 01 Mar 2024 14:40:00 GMT

Hi,

The issue is probably that you try to schedule a lot more rangings than what you have air-time for.

You could try to reduce the number of CONFIG_DM_TIMESLOT_QUEUE_COUNT_SAME_PEER. And if you do that, you can probably reduce CONFIG_DM_RANGING_OFFSET_US to get some latency improvement.

Reducing the advertising/scan interval should also help avoid having that many requests in the queue.

Is there a reason you use CONFIG_DM_TIMESLOT_RESCHEDULE=n?

Regards,
Amanda H.

RE: NrfConnect SDK nrf_dm example with dedicated roles with several Reflectors and one Initiator degrades with time

Michal — Tue, 27 Feb 2024 20:55:59 GMT

Hello,

I will look into it.

Best regards,

Michal