How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Wed, 27 Oct 2021 10:31:57 GMT

Hi Elfving,

I tried to tinker with the code as you wrote and eventually I can implement the multi synchronization + multiple cte ! Thanks a lot! I will continue to work on my project in this demo. And I may come back to you with further questions, since my project may require multi-sync with about over ten beacons. I am afraid that I will counter more bugs during the process.

Anyway, this is a big jump! Thanks ！

Best regards,

Zitong

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Elfving — Tue, 26 Oct 2021 09:01:54 GMT

Hey Zltong!

I believe I am seeing the same error as you.

So it seems that you also saw that adding a bit of a time delay between creating the two syncs would make a difference? If I replaced your for loop with a k_msleep(1000); With that I can sometimes lose the error.

But a better solution would be to use the semaphore:

printk("starting sync creation usually resulting in err EBUSY...\n");
err = k_sem_take(&sem_per_sync, TIMEOUT_SYNC_CREATE);
if (err) {
        printk("failed semaphore with err: (err %d)\n", err);
        return;
}
err1 = bt_le_per_adv_sync_create(&sync_create_param1, &sync1); 
k_sem_give(&sem_per_sync);
printk("sync creation done!\n");

Does this work for you?

Best regards,

Elfving

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Sat, 23 Oct 2021 11:24:47 GMT

From my understanding, scan_recv function only provides the per_addr information of the beacons. Though this information is random, it is fixed since the locator first detect the beacons. Besides, in create_sync function, I see that the only variable to consider is just per_addr. So I supposed that we can just record two Bluetooth devices' addresses and then create synchronization with them one by one.

I actually see a variable something like xxx_pool[BT_PER_ADV_SYNC_MAX] in a C header file. I think you officially support synchronous connections with multiple beacons but just failed.

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Thu, 21 Oct 2021 13:05:21 GMT

[quote userid="103347" url="~/f/nordic-q-a/78850/how-to-reduce-the-time-to-reduce-the-time-to-establish-periodic-sync-in-direction-finding/334818#334818"]I tried [/quote]

static void scan_recv(const struct bt_le_scan_recv_info *info,
		      struct net_buf_simple *buf)
{
	char le_addr[BT_ADDR_LE_STR_LEN];
	char name[PEER_NAME_LEN_MAX];

	(void)memset(name, 0, sizeof(name));

	bt_data_parse(buf, data_cb, name);

	bt_addr_le_to_str(info->addr, le_addr, sizeof(le_addr));

	printk("[DEVICE]: %s, AD evt type %u, Tx Pwr: %i, RSSI %i %s C:%u S:%u "
	       "D:%u SR:%u E:%u Prim: %s, Secn: %s, Interval: 0x%04x (%u ms), "
	       "SID: %u\n",
	       le_addr, info->adv_type, info->tx_power, info->rssi, name,
	       (info->adv_props & BT_GAP_ADV_PROP_CONNECTABLE) != 0,
	       (info->adv_props & BT_GAP_ADV_PROP_SCANNABLE) != 0,
	       (info->adv_props & BT_GAP_ADV_PROP_DIRECTED) != 0,
	       (info->adv_props & BT_GAP_ADV_PROP_SCAN_RESPONSE) != 0,
	       (info->adv_props & BT_GAP_ADV_PROP_EXT_ADV) != 0,
	       phy2str(info->primary_phy), phy2str(info->secondary_phy),
	       info->interval, adv_interval_to_ms(info->interval), info->sid);

	if (!per_adv_found && info->interval && device_number == 0) {
		bt_addr_le_copy(&per_addr, info->addr);
		printk("1 [DEVICE]: %s\n", le_addr);
                per_sid = info->sid;
		device_number++;
		return;
	}
	//if (!per_adv_found && info->interval) {
	if (!per_adv_found && info->interval && device_number == 1) {
		per_adv_found = true;
		per_sid1 = info->sid;
		strcpy(name_all, name);
		bt_addr_le_copy(&per_addr1, info->addr);
		printk("2 [DEVICE]: %s\n", le_addr);
		device_number = 0;
		k_sem_give(&sem_per_adv);
	}
}

This is what I do with the scan_rec function, where I added one more global variable

static bt_addr_le_t per_addr1 together with per_addr to record the address of two broadcasting devices with cte extensions.

After that, in create_sync function

static void create_sync(void)
{
	struct bt_le_per_adv_sync_param sync_create_param;
	struct bt_le_per_adv_sync_param sync_create_param1;
	int err;
	int err1;
	printk("Creating Periodic Advertising Sync...");
	bt_addr_le_copy(&sync_create_param1.addr, &per_addr1);
	bt_addr_le_copy(&sync_create_param.addr, &per_addr);

	sync_create_param.options = 0;
	sync_create_param.sid = per_sid;
	sync_create_param.skip = 0;
	sync_create_param.timeout = 0xa;

	sync_create_param1.options = 0;
	sync_create_param1.sid = per_sid1;
	sync_create_param1.skip = 0;
	sync_create_param1.timeout = 0xa;

	err1 = bt_le_per_adv_sync_create(&sync_create_param1, &sync1);
        for(int j=1;j<10000;++j)
        j=j;
	err = bt_le_per_adv_sync_create(&sync_create_param, &sync);
	if (err) {
		printk("err failed (err %d)\n", err);
		return;
	}
	if (err1) {
		printk("err1 failed (err1 %d)\n", err1);
                return;
	}
	printk("success.\n");
}

I tried to create 2 synchronizations with such two devices using the per_addr and per_addr1. My question is: it will always get error messages in the second times to do

bt_le_per_adv_sync_create function.

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Elfving — Tue, 19 Oct 2021 10:40:46 GMT

Hey again Zltong!

Once again, I am sorry it has taken this long. Our direction finding expertise has been very busy lately.

I will take more of a look at it myself. Any more updates on how far you have gotten in the debugging on your side?

[quote user="Marcous"]Hence I tried to get and record two per_addr(which is per_addr and per_addr1 )[/quote]

Would you mind showing me how you've modified the scan_recv function? I'll see if I can recreate your issue.

Best regards,

Elfving

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Sun, 26 Sep 2021 03:12:04 GMT

Hi, Elfving.

Providing more debug details:

In the scan.c -> bt_le_per_adv_sync_create -> get_pending_per_adv_sync function, I added some printk cmd to just show the error message.

Noting that I add one more bt_le_per_adv_sync_create to create_sync function in main. And the log information is like this:

I have set the parameter CONFIG_BT_PER_ADV_SYNC_MAX to 3. And the code goes well in the first bt_le_per_adv_sync_create, and does not enter the if judgement. However, in the second bt_le_per_adv_sync_create, it goes wrong with per_adv_sync_pool[0].flag = 536873404 and SYNC_SYNCING = 2

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Elfving — Thu, 23 Sep 2021 09:04:07 GMT

Yes, hopefully we will!

I've had a talk with the expert about your case, and what you have done should in theory be working. He is planning on taking a closer look.

Thank you for your patience. I'll get back to you when I have more information.

Best regards,

Elfving

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Wed, 22 Sep 2021 01:00:30 GMT

Hi, Elfving.

It is already sometime after 20th. Hopefully we will deal with this problem soon!

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Tue, 07 Sep 2021 11:30:26 GMT

Hi, Elfving,

Glad to see you again, since the feature of multiple sync is every important in my current project, I am looking forward to your updates!

Hope the problem would be solved!

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Elfving — Tue, 07 Sep 2021 11:13:17 GMT

Hey again Marcous!

This is getting a bit technical. Our expert on direction finding is out of office until the 20th, so I will have to get back to you sometime after that.

Best regards,

Elfving

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Mon, 06 Sep 2021 07:36:46 GMT

Hi, Elfving.

Updates:

From my debugging process, the code always encounters problems when traversing sync_pool, causing the problem of not generating a new sync. Do you have a solution?

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Fri, 03 Sep 2021 09:22:19 GMT

Hi Elfving,

I have been trying to tinker the code as you said. And followings are what I have done.

-I tried to add extended advertising points to scan_recv but got no idea how to deal with it. Hence I tried to get and record two per_addr(which is per_addr and per_addr1 ) in scan_recv in order successfully and then jump out the k_sem(&sem_per_adv).

- Next I tried to create two synchronizations in create_sync(), the code is as followed:

static void create_sync(void)
{
	struct bt_le_per_adv_sync_param sync_create_param;
	struct bt_le_per_adv_sync_param sync_create_param1;
	int err;
	int err1;
	printk("Creating Periodic Advertising Sync...");
	bt_addr_le_copy(&sync_create_param1.addr, &per_addr1);
	bt_addr_le_copy(&sync_create_param.addr, &per_addr);

	sync_create_param.options = 0;
	sync_create_param.sid = per_sid;
	sync_create_param.skip = 0;
	sync_create_param.timeout = 0xa;

	sync_create_param1.options = 0;
	sync_create_param1.sid = per_sid1;
	sync_create_param1.skip = 0;
	sync_create_param1.timeout = 0xa;

	err1 = bt_le_per_adv_sync_create(&sync_create_param1, &sync1);
	err = bt_le_per_adv_sync_create(&sync_create_param, &sync);
	if (err) 
	{
		printk("err failed (err %d)\n", err);
		return;
	}
	if (err1) 
	{
		printk("err1 failed (err1 %d)\n", err1);
		return;
	}
	printk("success.\n");
}

This code goes wrong when it run at the line of

err = bt_le_per_adv_sync_create(&sync_create_param, &sync);

and the wrong information is err = -16(EBUSY in the errno.h).

Actually I added the CONFIG_BT_PER_ADV_SYNC_MAX = 3 to prj.conf file and this parameter in the code is changed to 3.

Just as a reminder, If I chang the order of

err1 = bt_le_per_adv_sync_create(&sync_create_param1, &sync1);
err = bt_le_per_adv_sync_create(&sync_create_param, &sync);

err = bt_le_per_adv_sync_create(&sync_create_param, &sync);
err1 = bt_le_per_adv_sync_create(&sync_create_param1, &sync1);

The error information becomes err1 = -16, which means it always meets trouble in the latter one.

Can you help me to create multile sync? Thx a lot !!!

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Elfving — Mon, 30 Aug 2021 08:53:52 GMT

In the prj.conf or Kconfig file of the project.

Open the project folder of the demo, then find prj.conf, then add the line CONFIG_BT_PER_ADV_SYNC_MAX = 3.

Best regards,

Elfving

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Marcous — Mon, 30 Aug 2021 06:19:33 GMT

Thanks for your reply, but How do I change the parameter CONFIG_BT_PER_ADV_SYNC_MAX to two or more in nrf connect sdk demos?

RE: How to reduce the time to reduce the time to establish Periodic sync in direction finding ?

Elfving — Tue, 24 Aug 2021 11:08:36 GMT

Hey Marcous!

I would advise you to use multiple advertising streams instead of having to resync. The zephyr controller allows you to have several so that you don't need to switch between beacons. The config is called CONFIG_BT_PER_ADV_SYNC_MAX, it is set to 1 by default.

You might have to tinker a bit more with it in order to get it to work though. For instance:

- When scanning callback in scan_recv you have to check for extended advertising points, then call the API to create a new sync if that is the case, and enable CTE sampling for that sync.

- Create_sync needs to be updated too, it uses globals for sync establishment.

- cte_recv_cb must match IQ report to appropriate sync instance.

Best regards,

Elfving