[srslte-users] eNB Output Power @SDR / PAPR disparity ?

[Cristian Balint]

Dear All,

  I noticed that the srsenb signal have very low gain on various SDRs. It
looks to me as not being related to SDR hardware. This was also noticed [1]
and measured [2][3] recently on the list.

  I tried "cell_gain <cellid> <db>" in console and found that values of
10-15 might normalize the signal from SDR to some acceptable levels
reaching -20dBm, but any value >3 (3dB is maximum) lead to degradation with
UE.

---------------------<<<-----------[TESTS]---------------------<<<-----------

cell_gain <cellid> 3 (OK, no quality degradation on UE)  -> x1.41
cell_gain <cellid> 5 (NOK, quality degradation on UE)  -> x1.77
cell_gain <cellid> 10 (NOK, , UE unable to sync)  -> x3.16
cell_gain <cellid> 15 (NOK, UE unable to sync)  -> x5.62

1. Gain magnification of  <cell_gain> looks to be done by scaling up the
complex I/Q vectors on the final stream:
https://github.com/srsLTE/srsLTE/blob/45486b6e2c622ab1d47906521662f087816d7d03/srsenb/src/phy/cc_worker.cc#L276

2. The +10dB would be reasonable, scales with only 3.1x, but it degrades
the signal badly for UE.

3. Tried to clip up to 1.0f region by extracting and selecting maximas of
the buffer chunks (@ cc_worker.cc#L276) even by suppressing the
non-maximas, signal looks much better on the scope but very lossy with real
UE.

--------------------->>>-----------[TESTS]--------------------->>>-----------

*  LTE signals have high PAPR, but also might be that within srsenb the
signal tied together from different chunks (out of subroutines) might have
higher "norm" disparity for some parts.

* Also noticed that pdsch power allocation scheme is disabled:

https://github.com/srsLTE/srsLTE/blob/45486b6e2c622ab1d47906521662f087816d7d03/lib/src/phy/enb/enb_dl.c#L245

* Trying to get the srsenb signal broadcasted >30dBm looks quite
challenging in its actual form.


Do people here confirm or have some similar experience ?



Thank You !

Cheers,
~Cristian

[1]
http://www.softwareradiosystems.com/pipermail/srslte-users/2021-April/004179.html
[2]
https://forums.ni.com/t5/RF-Measurement-Devices/Low-output-power-on-USRP-2901/m-p/4142182
[3] https://forums.ni.com/ni/attachments/ni/290/3563/1/spectrum_analyzer.png
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20210418/b07d03ca/attachment.htm>