srsRAN gNB with srsUE

[nadiayoza]

Hello,

I am trying to implement an end-to-end 5G connection using srsRAN Project for the gNB and srsRAN 4G for the UE, as indicated in the tutorial recently released [1]. The UE gets RRC connection to the gNB and it gets an IP address. However, when I try to ping between gNB and UE, the round trip time of the ping takes too long:

sudo ./srsue ue.conf
Active RF plugins: libsrsran_rf_uhd.so libsrsran_rf_soapy.so libsrsran_rf_blade.so libsrsran_rf_zmq.so
Inactive RF plugins:
Reading configuration file ue.conf...

Built in Release mode using commit 655a098b7 on branch master.

Opening 1 channels in RF device=uhd with args=clock=external
Supported RF device list: UHD soapy bladeRF zmq file
[INFO] [UHD] linux; GNU C++ version 9.4.0; Boost_107100; UHD_3.15.0.0-74-ge35f66e8
[INFO] [LOGGING] Fastpath logging disabled at runtime.
Opening USRP channels=1, args: type=b200,master_clock_rate=23.04e6
[INFO] [UHD RF] RF UHD Generic instance constructed
[INFO] [B200] Detected Device: B210
[INFO] [B200] Operating over USB 3.
[INFO] [B200] Initialize CODEC control...
[INFO] [B200] Initialize Radio control...
[INFO] [B200] Performing register loopback test...
[INFO] [B200] Register loopback test passed
[INFO] [B200] Performing register loopback test...
[INFO] [B200] Register loopback test passed
[INFO] [B200] Asking for clock rate 23.040000 MHz...
[INFO] [B200] Actually got clock rate 23.040000 MHz.
Setting manual TX/RX offset to 300 samples
Waiting PHY to initialize ... done!
Attaching UE...
Random Access Transmission: prach_occasion=0, preamble_index=0, ra-rnti=0x39, tti=5934
Random Access Complete. c-rnti=0x4601, ta=0
RRC Connected
PDU Session Establishment successful. IP: 10.46.0.36
RRC NR reconfiguration successful.

The amf.yaml is as follows and, based on it, I am using 127.0.0.2 as the AMF address and 127.0.0.100 or 127.0.1.1 as the AMF bind_addr in the gnb.yaml file:

I tried both ways: conducted tests using physical radios (B210s) connected through a 30 dB attenuator or, instead, using zmq, but I obtain the same results. I am using Ubuntu 20.04 LTS and Open5GS core. I didn't have any problems using the same parameters with the previous srsRAN version. The subscribers parameters indicated in Open5gs are according to the ue.conf parameters.

ue.txt
gnb.txt
upf.txt
amf.txt

[1] https://docs.srsran.com/projects/project/en/latest/tutorials/source/srsUE/source/index.html#

If you have any suggestions, please, let me know.

Thanks!

[robertfalkenberg]

Hi,
please share the gNB logs, UE logs, and - if possible - PCAP files from the gNB (both, MAC and NGAP).

Best regards

[nadiayoza]

Hi Robert,

Thanks for your response. Please, find below the logs and pcap files:
https://drive.google.com/drive/folders/1aB0OIdCQFcb9r50o4JoEaTxq02DCMr5h?usp=share_link

The ue.log show messages like this:
"2023-05-02T19:30:32.195404 [PHY_LIB] [I] [ 0] PDCCH DMRS Rx: Measure L=2; ncce=4; RSRP=-47.8 dBfs; EPRE=-34.1 dBfs; Corr=0.043
2023-05-02T19:30:32.195406 [PHY_LIB] [I] [ 0] Discarded PDCCH candidate L=2;ncce=4; Correlation is too low (0.0<0.5); EPRE=-34.15; RSRP=-47.83;"

The computers are set to performance mode using sudo ./scripts/srsran_performance
Also, if I just replace the gnb from srsRAN Project with the enb from the new srsRAN 4G 23.04 (the same I'm using for the tests above), the ping round trip time is fine for 5G.

Best regards,

Nadia

[nadiayoza]

Hello,

I installed the new srsRAN Project 23.5 released yesterday. Thank you all for the great work! Now the ping round trip time is comparable to the one I got with srsRAN 4G with a maximum throughput of ~23 Mbps. I just wanted to confirm, the adaptive MCS capability is not included yet, right?

Thanks!

[frankist]

Hey @nadiayoza, where did you get the dl_snr value above? Is this the value reported by the UE?

[nadiayoza]

Hi Francisco! Yes, the dl_snr value in the ue_metrics.csv file is reported by the UE. So I was wondering why we see a change in dl_snr and dl_brate, but not in the dl_mcs, which remains in 28. In LTE using srsRAN, the MCS would change as well.
Thanks!

[frankist]

Can we retry the same experiment with the latest srsgnb release, and sharing both srsgnb and srsue logs? Having both logs will help me understand if it is not a problem in the PUCCH decoding.

[nadiayoza]

Sure, I'll try that and let you know. Thanks!

[nadiayoza]

I retried the experiment with the latest versions of srsRAN Project (gNB) and srsRAN 4G (UE). Like in the previous tests, I'm increasing the attenuation 10 dB every 5 seconds. I see a similar behavior, where the MCS doesn't get updated in ue_metrics.csv. Please, find below the gNB and UE logs and the ue_metrics.csv file:
https://drive.google.com/drive/folders/1A853g_bgYMrj5pJCNW9SE1zR1FmfSU4I?usp=sharing
Is it that the MCS is adaptive, but it's just not reported in the ue_metrics.csv? The throughput is decreasing and the BLER is increasing with lower SNR:

Thanks!