gNB does not work as shown in Documentation

[oguztudo]

Hello everybody,

I am running the gNB and i dont know if this is an issue or if this works as intended but my gNB sends signals irregularly unlike the screenshot posted here:

https://docs.srsran.com/projects/project/en/latest/_images/maia_sdr.png

Mine looks like this:

Detecting the gNB works on my Samsung Galaxy S23 Ultra in the NR only mode as described in the tutorial. However i can not connect successfully. The first two lines of the log confuse me a little because one says rlc_mode=UM and the other one says AM (I suppose the rlc mode gets changed to AM).
There are some lines which say "criticality": "reject" but I don't know if that's a Problem causing factor.
I think the connection could be established solving the the Real-time failure in FAPI issue which can be seen further down in the log.

Here is my config and the log to gNB with detail level debug for all:

cots_gnb.txt

gnb.log

[suttonpd]

Your gnb log shows real-time failures for every DL TTI - ~70 real-time failures in <2ms.
What are the specs of the machine you're trying to use?

[oguztudo]

Hello @brendan-mcauliffe,
thanks for the info, that did do the trick to make it work BUT i still encounter the issue of not being able to connect.
I realized that the log became more detailed on the info setting so i will post the log file here:

gnb.log

[brendan-mcauliffe]

This log only seems to be running for a few seconds. I recommend you try connecting the phone and sending the logs from that test.

Are you using an external source for clocking the gNB?

[oguztudo]

I do not have a GPSDO but I do have a gps antenna on the GPS ANT Port and a 10MHz Signal from a Signal generator in the REF IN Port of my USRP x310.
Despite not having a GPSDO I still tried all possible variations on the parameters (internal, external, gpsdo) clock and sync but none of those provided any change in the outcome.

This time i tried to run it for a longer period of time and i am trying to give as much detail as possible:
with clock and sync set to internal.
I ran the gNB for 10 mins now. After some tries i could find the gNB, which was almost at min from start of the gNB. Then i tried to connect to the gNB. The whole process of trying to register until the connection fail last about 2 and a half minutes. Then i made another try to register to the gNB with my phone and again nothing. Then i made another search, found the gNB again. Trying to connect happened at minute 10 since starting the gNB, lasted another 2min 15s and ended in a connection fail yet again. This is the log, the amount of information is the same.

gnb.log

I made another run with clock and sync on external but the outcome is the same...

[ismagom]

Can you try commenting out the parameters time_alignment_calibration: -300 and otw_format: sc12.

Also try reducing the tx_gain: to 20.

Which is IMSI you have configured in the SIM card?

I think most likely there is a clocking issue. Using the external REF IN is not enough without a PPS (I think), so I'm not sure you can use external parameter.

You say you don't have a GPSDO, but I see in the console the message: Found an internal GPSO: LC_XO so in theory setting gpsdo to both sync and clock should pick it up.

Do you have some external instrument to measure the LO offset?

Regards

[oguztudo]

I deleted those parameters already in my earlier tries when Brendan suggested me to do it.
Here is the latest config i used:

cots_gnb.txt

Changing the tx gain did not provide any solution either (i also varied the distance of the UE to the gNB).

I used he following IMSI, starting with a 00101 and then used something random: 001010000082271

Regarding the GPSDO: Yes the USRP has an internal GPSDO but the precision of 210^⁻8 seemed low to me when compared to the Leo Bodnar GPSDO, which is being advertised with running at a precision of 110^-12. Also i can only set the clock to GPSDO, when i try to set the sync it says "Error: failed to read sensors" which appears to be a hardware problem (?). So for clock only internal and external are possible.

My spectrum analyzer gives me a maximum for the local oscillator at the set dl_freq. I also could get my hands on a fieldfox today which can show me 5G NR cells. There i could see the SSB offset being mostly at -60kHz and sometimes at 0kHz. The fieldfox could only see the cell properly after setting the frequency offset to +60kHz. In this case i could see the whole spectrum shift by 60kHz (LO included) and the fieldfox could find the cell instantly. I could even detect a change in pci, but eventually set it back to 1. Then i tested and tried to see if i could find the gNB with my phone but could not find anything...

[oguztudo]

Update:
I went to a non Roaming scenario, setting the gNB PLMN to 00101 and also turned Roaming and VoLTE off on my phone. I could manage to establish a connection!
Nonetheless i can not access the internet via my phone, despite having the settings from the open5gs site active... I guess thats the next step.

[brendan-mcauliffe]

If you cannot access the internet, it is probably a misconfiguration with the APN. Double-check how the APN is configured on your phone and in the core. Make sure they match exactly.

[oguztudo]

My phone APN is exactly the same as the one given in the docu. The core APN has the following settings. When i set a UE IPv4 Address, i can ping that corresponding address successfully so maybe i am missing something here...

heres my APN setting in the core:

[brendan-mcauliffe]

Check the consistency between the phone and APN again, once you are sure they're the same try restarting the open5GS services and try to connect again.

[oguztudo]

I set the UE and SMF IPAdresses and managed to make it work! Thank you very much for the help!!!