srsRAN_4G icon indicating copy to clipboard operation
srsRAN_4G copied to clipboard

5g SA amarisoft using USRP N310

Open ayamoheddine opened this issue 3 years ago • 7 comments

Issue Description

I’m trying to duplicate the scenario presented in the latest version of srsRAN using 5g SA configuration. I followed the guidelines mentioned in the latest release in the app notes. However, the USRP is not able to connect to the network and stuck in the UE attaching step.

Link to the latest version of srsRAN followed:

https://docs.srsran.com/en/latest/app_notes/source/5g_sa_amari/source/index.html#g-sa-srsue

Setup Details

The setup used is: N310 USRP as user equipment, connected through 10Gbit interface to ubuntu 18.04 PC. Amarisoft Classic Callbox as 5G core network.

The USRP and amarisoft are connected through a set of antennas.

Expected Behavior

Successful attachment of the UE to 5G network generated by the amarisoft callbox.

Actual Behaviour

Stuck in the UE attach phase with lots of RF underflows.

Steps to reproduce the problem

Changed the ue.conf file as mentioned in the app notes and as attached in ue.txt file and the execution is given within the screenshot. ue.txt srsue

Additional Information

The ue.log file can be found here.

usrp_dl_earfcn=2850_ue.txt

Your help is highly appreciated.

ayamoheddine avatar May 20 '22 10:05 ayamoheddine

We've not tested the N310 with the UE. But your sample frequency offset and srate params look odd to me.

andrepuschmann avatar May 20 '22 10:05 andrepuschmann

We've not tested the N310 with the UE. But your sample frequency offset and srate params look odd to me.

The srate and sample frequency offset are those of N310 as can be seen from this table n310-sample-rates

ayamoheddine avatar May 20 '22 10:05 ayamoheddine

But you're trying to receive a 10 MHz wide signal. So sampling at 7.68Msps isn't enough. Try e.g. 15.36e6

andrepuschmann avatar May 23 '22 07:05 andrepuschmann

Nothing changed. Here is the ue.log file. ue-15.35e6.txt

ayamoheddine avatar May 23 '22 07:05 ayamoheddine

Could you try dropping the rx_gain and tx_gain to a (much) lower number that is not 0? The CRC on the frames you receive is KO, and when I had that happen on my end (admittedly with 5G SA instead of NSA) dropping the gains helped a lot. I still see a lot of KO's, but I also see regular OK's with correlations large enough to be relevant.

TallGuy74 avatar May 24 '22 07:05 TallGuy74

Thanks for your reply, but I don't think this is the main problem. I'm trying something different now to see if it's gonna work.

ayamoheddine avatar May 24 '22 08:05 ayamoheddine