[srsran-users] Retain COTS UE connection at Low RSRP

Silvio silwireless23 at gmail.com
Wed Mar 26 23:07:17 UTC 2025


Hi!

I was working with srsRAN 23.11 and noticed something interesting.

I can never keep my COTS UE connected whenever the RSRP drops lower than
-105dBm. My test is mainly on Band 5 with 50 PRBs.

I get this by running an USRP B210 at gain 60 with a 20dB attenuator added
to the TX chain. The UE throws Radio Link Failures with Cause:Other
Failure, disconnects, then comes back up, etc.

My own troubleshooting so far:

   1. I measured the phase noise on a CW at 880MHz. The results are in the
   neighborhood of -95 to -100 dBc/Hz up to 1 MHz. This should be okay for
   LTE.
   2. I measured the frequency offset and compensated for it (roughly
   +500Hz offset was needed). A 500 Hz offset wasn't good for LTE. I suspect
   this brought my RSRP sensitivity down from -95dBm to -105dBm.
   3. I added an external 10MHz reference signal to the USRP generated by a
   50ppb oscillator.
   4. I set the following SIB parameters:
      1. In SIB1 and SIB3, I have q_rx_lev_min = -70.
      2. I set rs_power in SIB2 to the actual RSRP measured by my Spectrum
      Analyzer on a conductive test (~-50 dBm). I was worried the UE would
      interpret a higher pathloss and report wrong CQIs.
      3. I set time_alignment_timer to "sf500" to force timing alignment to
      happen as often as possible. I think this makes the system timing more
      tight at low RSRP.
      4. I set the ue timers to be way more permissive:
          ue_timers_and_constants =
          {
              t300 = 2000; // in ms
              t301 = 100;  // in ms
              t310 = 1000; // in ms
              n310 = 10;
              t311 = 10000; // in ms
              n311 = 1;
          };

Nothing seems to allow my RSRP to reach ~-115 to -120 dBm. I am in a deep
rabbit hole. Please help me figure this out.

Thank you!

ᐧ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20250326/aaf136d7/attachment.htm>


More information about the srsran-users mailing list