LTE-M Reconnect after LoS

RE: LTE-M Reconnect after LoS

ibiglari — Sat, 23 Jul 2022 02:48:54 GMT

Hi,

Just wondering if your team's discover needs to be implemented in lwm2m library rather than the application that uses it?

RE: LTE-M Reconnect after LoS

ibiglari — Wed, 06 Jul 2022 22:23:00 GMT

Hi,

Just had a chance to read your response carefully as we've been dealing with floods here in Sydney. I'm not sure I understand what this exactly means, and whether I can modify Asset Tracker v2 in any way to get around the problem. If the bug is with reporting connection status and the connection status itself, it should, in theory, have no effect on communications whatsoever, right?

RE: LTE-M Reconnect after LoS

Øyvind — Mon, 04 Jul 2022 08:45:05 GMT

I've finally got some feedback from our team. They have identified a bug in the modem fw due to which the CEREG status may show invalid status during PSM and cell lost crossover. It should still be possible to send user data assuming that the cell is available at that time. %XMODEMSLEEP can be used for monitoring sleeps, including PSM sleep.

Kind regards,
Øyvind

RE: LTE-M Reconnect after LoS

Achim Kraus — Fri, 01 Jul 2022 14:46:52 GMT

Sure, if this is the case, it "must" be implemented in that lwm2m library.

The idea pointing to my examples was more, to have a short cut to find out, if this ENETDOWN is really the issue also in your case and the "reopen socket" is then the fix.

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 01 Jul 2022 13:16:36 GMT

Will try to over the weekend... Just got a big project off my back... Had a quick look at the repo and well done!

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 01 Jul 2022 13:15:20 GMT

Very interesting notion. However, if this is the case (and I haven't had a chance to test it yet - switched jobs last week and it's been a very busy first week), I think it might be best if the handler for ENETDOWN is implemented somewhere in the LwM2M library, and not in the application that uses it, as the library encapsulates network communications (sorry if this sounds silly as I have yet to completely understand LwM2M implementation in nRF SDK).

RE: LTE-M Reconnect after LoS

Achim Kraus — Tue, 28 Jun 2022 12:57:40 GMT

Maybe, if you're interested, if that approach would help also in your case, try out Zephyr - coaps client / tinydtls CID. if that works before your garage, and stops working in the garage, you will see, if it reconnects afterwards.

Good luck.

RE: LTE-M Reconnect after LoS

Achim Kraus — Tue, 28 Jun 2022 12:47:48 GMT

That's a lot of information, so I'm not sure, if my experience matches this case. Hope, I don't mix up more, than I help.

I was faced a similar behavior just using a UDP socket for CoAP with my own DTLS CID implementation (eclipse/tinydtls), see How to recover a nRF9160 from sporadic ENETDOWN .

if the mobile network gets lost, and data tried to sent, you get a ENETDOWN error. If so, you need to close the socket (seems to apply to UDP as well) and wait until the network is back (callback with LTE_LC_EVT_NW_REG_STATUS). Then open the socket again.

That gets a little more complicated, if PSM is enabled and active. Then you may first wait for the LTE_LC_EVT_NW_REG_STATUS callback after starting to send data. If that times out, i switch into flight mode, and try it about 5 minutes later and double that time on every retry. After a couple (e.g. 2h) I reboot my device in order prevent it from stick in some bugs.

That works for me now quiet well, a 2 weeks test with some areas without connectivity have all been mastered and the thingy always connected back.

RE: LTE-M Reconnect after LoS

Øyvind — Mon, 27 Jun 2022 10:00:36 GMT

Thanks for providing this information. I have forwarded to our nRF9160 team to have a look. Will get back to you when I have more information.

Kind regards,
Øyvind

RE: LTE-M Reconnect after LoS

ibiglari — Sun, 26 Jun 2022 03:41:09 GMT

Finally managed to grab the log, wrapping my laptop and device in aluminium for 5 minutes. Trace ran for 35 minutes from device boot. Last 30 minuts of the trace, device was not communicating with my server.

devzone.nordicsemi.com/.../trace_2D00_2022_2D00_06_2D00_26T02_2D00_59_2D00_41.604Z.bin.7z

RE: LTE-M Reconnect after LoS

ibiglari — Sat, 25 Jun 2022 00:57:18 GMT

Tried to install nrf Desktop on a Raspberry Pi (was thinking to power the Pi using a power bank) but the installer did not run despite me having installed 64 bit of Raspbian. Next thing to try is to start the trace in my apartment, go into the parking for 5 minutes (defenitely no cellular coverage there), and come back up. Unfortunately I have a behemoth of laptop that doesn't fit into my microwave oven...

If that doesn't work, I'll try driving through that tunnel while recording the trace.

RE: LTE-M Reconnect after LoS

Øyvind — Fri, 24 Jun 2022 13:57:26 GMT

Are you able to bring it in your car through the tunnel as you mentioned? Or you could try inside a larger microwave (preferably not turning it on), as this acts as a faraday cage when closed.

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 24 Jun 2022 13:33:47 GMT

That will take some doing as I have to find a laptop small enough that I can wrap in foil along with the device. My old surface would do the trick but its battery capacity is doubtful. I will update this post with the resolution.

RE: LTE-M Reconnect after LoS

Øyvind — Fri, 24 Jun 2022 13:28:39 GMT

We need to see the issue from reset in order to find the cause.

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 24 Jun 2022 12:25:09 GMT

I connected the cable after I unwrapped the foil.

RE: LTE-M Reconnect after LoS

Øyvind — Fri, 24 Jun 2022 12:23:51 GMT

How did you collect modem trace earlier?

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 24 Jun 2022 11:06:28 GMT

There seems to be a complication. The device does not loose network connection if I wrap it with the USB cable is connected. Seems like the cable becomes part of the antenna.

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 24 Jun 2022 10:52:44 GMT

Hi,

I'm using the USB port, and Trace Collector v2.

I'll try again and send the logs across.

RE: LTE-M Reconnect after LoS

Øyvind — Fri, 24 Jun 2022 10:45:37 GMT

Ok, thanks for clarifying.

I think we need to see new traces, as these traces seem to be missing vital information. Did you use Trace Collector v2?

Do you connect your device via external debug probe? If so, you need to wrap the device in aluminum foil, start modem trace and reset the device. When the device looses connection, remove the alu foil and leave the device running modem trace for at least 30 minutes after removing the foil.

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 24 Jun 2022 09:17:26 GMT

Probably 10~15 minutes...

The device (at least the UI module) behaves as you'd expect it to in ACTIVE mode. Every two minutes or so, LEDs flash green, then teal, then off. If I press SW3, LEDs flash green, and then teal. If I register an observer from my LwM2M server for object 6 (location) when the connection is initially established, the modem turns purple as in it is trying to get a location fix in between green/teal cycles.

RE: LTE-M Reconnect after LoS

Øyvind — Fri, 24 Jun 2022 09:00:20 GMT

The modem trace doesn't make much sense. It looks like the device is resetting. How long did you leave the modem trace running?

RE: LTE-M Reconnect after LoS

ibiglari — Fri, 24 Jun 2022 08:25:15 GMT

Hi,

Few changes to support SUPL binary:

prj.conf:
```
CONFIG_LOG=n
CONFIG_AT_HOST_LIBRARY=n
```

overlay-lwm2m.conf

CONFIG_LWM2M_TLS_SESSION_CACHING=n
CONFIG_LWM2M_RD_CLIENT_SUPPORT_BOOTSTRAP=n
CONFIG_LWM2M_CLIENT_UTILS_LOG_LEVEL_DBG=n
CONFIG_LWM2M_CLIENT_UTILS_DEVICE_OBJ_SUPPORT=y
CONFIG_LWM2M_CLIENT_UTILS_LOCATION_OBJ_SUPPORT=y
CONFIG_NRF_CLOUD_AGPS=y
CONFIG_SUPL_CLIENT_LIB=y

src/cloud/CMakeLists.txt

target_sources_ifdef(CONFIG_SUPL_CLIENT_LIB app
                     PRIVATE ${CMAKE_CURRENT_SOURCE_DIR}/lwm2m_integration/assistance_supl.c)

src/cloud/Kconfig.lwm2m_integration

config LWM2M_CLIENT_UTILS_SERVER
        string "LwM2M server address"
        default "coaps://mylwm2mhostnamehere:5684"

/src/main.c

.
.
.
#if defined(CONFIG_SUPL_CLIENT_LIB)
#include "cloud/lwm2m_integration/assistance.h"
#endif
.
.
.
void main()
{
    .
    .
    .
    self.thread_id = k_current_get();
    
#if defined(CONFIG_SUPL_CLIENT_LIB)
    assistance_init();
#endif
    .
    .
    .

src/cloud/cloud_codec/lwm2m/lwm2m_codec.c

int cloud_codec_encode_agps_request(struct cloud_codec_data *output,
                                    struct cloud_data_agps_request *agps_request)
{
#if defined(CONFIG_SUPL_CLIENT_LIB)
       /* SUPL doesn't usually provide satellite real time integrity information. If GNSS asks
        * only for satellite integrity, the request should be ignored.
        */
       if (agps_request->request.sv_mask_ephe == 0 &&
           agps_request->request.sv_mask_alm == 0 &&
           agps_request->request.data_flags == NRF_MODEM_GNSS_AGPS_INTEGRITY_REQUEST) {
               printk("Ignoring assistance request for only satellite integrity\r\n");
       } else {
               printk("Assistance data needed, ephe 0x%08x, alm 0x%08x, flags 0x%02x\r\n",
                       agps_request->request.sv_mask_ephe,
                       agps_request->request.sv_mask_alm,
                       agps_request->request.data_flags);
               assistance_request(&agps_request->request);
       }
       return -ENODATA;
#else
// ORIGINAL METHOD
.
.
.
#endif

I also copied assistance.h and assistance_supl.c from GNSS sample for nrf91670

Would you like me to zip my project folder and send it to you somehow?

RE: LTE-M Reconnect after LoS

Øyvind — Fri, 24 Jun 2022 08:03:25 GMT

Have you done any other changes to the application other than adding lwm2m?

RE: LTE-M Reconnect after LoS

Øyvind — Thu, 23 Jun 2022 10:14:13 GMT

Thanks for providing the modem traces. I will have a look at them and get back to you.

RE: LTE-M Reconnect after LoS

ibiglari — Wed, 22 Jun 2022 22:05:04 GMT

Trace #2, this morning. This was after driving with Thingy 91 to work, and realising it had stopped communicating when I went through a tunnel.

devzone.nordicsemi.com/.../2772.trace_2D00_2022_2D00_06_2D00_22T21_2D00_58_2D00_16.233Z.bin