Synchronizing timout when using Periodic Advertising with Responses (PAwR)

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

ATla5 — Fri, 29 Mar 2024 07:38:00 GMT

Hello,

Thank you for your explaination.

I've checked out the fix (zephyr sample repository) of the PaWR sample and it works.
Thank you!

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

dejans — Tue, 26 Mar 2024 12:28:29 GMT

Hi,

With regard to previously observed warning, the sample was trying to set the subevent data to an invalid subevent index, and this causes the warning to be reported by the controller.

The root cause of the issue seems to be scheduling conflict between subevents and passive scanner in the periodic_adv_rsp sample. In a previously suggested fix, scan_window (30ms + scheduling overhead) is added into periodic advertising interval.

(subevent_interval + scan_window) * NUM_SUBEVENTS = (0x30 + 0x20) * NUM_SUBEVENTS = (60ms + 40ms) * NUM_SUBEVENTS

Another (better) solution can be found in this PR - fix PAWR sample.

Best regards,
Dejan

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

dejans — Fri, 22 Mar 2024 14:30:02 GMT

Hi,

[quote user="ATla5"]The advertising interval was too short (0xFF * 1.25ms = 318.75ms) since the given slot spacing (0x50 * 0.125ms = 10ms) and the number of subevents (5) and the subevent interval (0x30 * 1.25ms = 60ms) led to a situation where the given subevents could not be processed within the given advertising interval, right?[/quote]

I will get back to you regarding this probably during next week.

[quote user="ATla5"]which included ASCII characters such as ":" (colon) and "-" (dash). I have not found any restrictions given the Bluetooth device name. Do you know why these characters cause an issue?[/quote]

Regarding the Bluetooth device name, some changes (shown below) are needed in the periodic_adv_rsp sample.

static void device_found(const bt_addr_le_t *addr, int8_t rssi, uint8_t type,
			 struct net_buf_simple *ad)
{
  ...
	if (strcmp(name, "PAwR: sync-sample")) {
		return;
	}
  ...
}

Best regards,
Dejan

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

ATla5 — Fri, 22 Mar 2024 10:46:52 GMT

Hello,

I have tested your suggestion and these are the results and it looks like the expected outcome.
Correct me if I am wrong - the adjustment for both the min. and max. interval depends on the

.response_slot_spacing = 0x50,

parameter. The advertising interval was too short (0xFF * 1.25ms = 318.75ms) since the given slot spacing (0x50 * 0.125ms = 10ms) and the number of subevents (5) and the subevent interval (0x30 * 1.25ms = 60ms) led to a situation where the given subevents could not be processed within the given advertising interval, right?
Therefore, the new advertising interval is (0x50 * 5 * 1.25ms = 500ms) is large enough given the parameters

static const struct bt_le_per_adv_param per_adv_params = {
	.interval_min = 0x50 * NUM_SUBEVENTS, /* 300 * 1.25ms = 500ms */
	.interval_max = 0x50 * NUM_SUBEVENTS, /* 300 * 1.25ms = 500ms */
	.options = 0,
	.num_subevents = NUM_SUBEVENTS, /* 5 */
	.subevent_interval = 0x30, /* 48 * 1.25ms = 60ms */
	.response_slot_delay = 0x5, /* 5 * 1.25ms = 6.25ms */
	.response_slot_spacing = 0x50, /* 80 * 0.125ms = 10ms */
	.num_response_slots = NUM_RSP_SLOTS, /* 5 */
};

Bluetooth® Core Specification Version 5.4 - Technical Overview

This is the output of the RTT Viewer log files of both the periodic_adv_rsp and periodic_sync_rsp application.

devzone.nordicsemi.com/.../rtt_5F00_log_5F00_pawr_5F00_adv_5F00_rsp_5F00_with_5F00_fix.log

devzone.nordicsemi.com/.../rtt_5F00_log_5F00_pawr_5F00_sync_5F00_rsp_5F00_with_5F00_fix.log

One more thing - The initial issue why no synchronisation was even possible was caused by the modified Kconfig option

CONFIG_BT_DEVICE_NAME="PAwR: sync-sample"

which included ASCII characters such as ":" (colon) and "-" (dash). I have not found any restrictions given the Bluetooth device name. Do you know why these characters cause an issue?

Best regards,

ATla5

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

dejans — Fri, 22 Mar 2024 09:42:40 GMT

Hi,

Thank you for providing the logs.

Could you please try to manually modify 2 lines in main.c (per_adv_params) of the periodic_adv_rsp sample as shown below and run the sample again?

//.interval_min = 0xFF,
.interval_min = 0x50 * NUM_SUBEVENTS,
//.interval_max = 0xFF,
.interval_max = 0x50 * NUM_SUBEVENTS,

Best regards,
Dejan

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

ATla5 — Fri, 22 Mar 2024 08:26:18 GMT

Hello,

I have copied the original samples again to a separate folder, build and flashed the applications to the respective devices and captured the packet transfer at the periodic_sync_rsp device with a sniffer (nRF52840 Dongle) and logged the terminal output of both applications.

The synchronization works now but the HCI related warnings on the periodic_adv_rsp device remains. Let me know if you have any suggestions what causes these warnings.

Are the "Received empty indication: subevent 0" indications at the periodic_sync_rsp terminal expected?

Best regards

ATla5

devzone.nordicsemi.com/.../7633.pawr_5F00_sync_5F00_rsp_5F00_sniffer_5F00_log.json

devzone.nordicsemi.com/.../term_5F00_log_5F00_pawr_5F00_sync_5F00_rsp.log

devzone.nordicsemi.com/.../term_5F00_log_5F00_pawr_5F00_adv_5F00_rsp.log

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

dejans — Thu, 21 Mar 2024 17:14:25 GMT

Hi,

[quote user="ATla5"]Does this issue occur when you test these samples too?[/quote]

I have observed the same behavior as shown in your initial post on adv side, and somewhat different on sync side probably because I used NCS v2.6.0.

[quote user="ATla5"]Is anything required besides the two samples (periodic_adv_rsp, periodic_sync_rsp) each running on a separate nRF52840 DK?[/quote]

We will need to look into this further, but It would be useful if we could have a sniffer log. Could you provide it?

Best regards,
Dejan

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

ATla5 — Thu, 21 Mar 2024 08:47:58 GMT

Hello Dejan,

Even with another NCS/Zephyr version such as v2.6.0 as you have suggested the problem remains.
Does this issue occur when you test these samples too?

Is anything required besides the two samples (periodic_adv_rsp, periodic_sync_rsp) each running on a separate nRF52840 DK?

Best regards

ATla5

RE: Synchronizing timout when using Periodic Advertising with Responses (PAwR)

dejans — Mon, 18 Mar 2024 18:06:21 GMT

Hi,

Could you please try the same samples using NCS v2.6.0?

Best regards,
Dejan