[srslte-users] AGC Mode

[Charbel Lahoud]

Hi Xavier,
Thanks a lot for your response. The AGC_DEFAULT_TARGET is 0.25 and not 2.5 (sorry for that).
The signal amplitude is less than 1 and the AGC is not always settling to the same gain.

I have observed that the AGC is settling for a short time and then resulting bad performance again. So it is not settling on a matching Rx gain value even while system is stationary. Please have a look on the logs attached.
Here are my AGC config:
#define SRSLTE_AGC_CALLBACK(NAME) void(NAME)(void* h, float gain_db)
#define SRSLTE_AGC_DEFAULT_TARGET (0.25f)
#define SRSLTE_AGC_DEFAULT_BW (0.2f)
#define SRSLTE_AGC_HOLD_COUNT (200)
#define SRSLTE_AGC_MIN_MEASUREMENTS (100)
#define SRSLTE_AGC_MIN_GAIN_OFFSET (1.0f)

the AGC Gain value can also be observed from the log file.

Best regards
Charbel

From: Xavier Arteaga <xavier at srs.io>
Sent: Wednesday, August 18, 2021 9:22 AM
To: Charbel Lahoud <charbel.lahoud at etit.tu-chemnitz.de>
Cc: srslte-users at lists.softwareradiosystems.com
Subject: Re: [srslte-users] AGC Mode

Hi Charbel,
With AGC_DEFAULT_TARGET greater than 1, I have the following questions:
- Is the signal coming from the N321 having a greater amplitude than 1?
- Is it always settling in the maximum gain?

Regarding AGC_HOLD_COUNT, it looks like a lot of latency. I wonder if the Rx gain settling takes that long. This makes the AGC take a while to settle. Does the UE recover the performance after moving?

I guess the best is to add some traces in `agc.c` to understand how the AGC is interacting with the RF and see if it settles into a bounded value.

Regards,
Xavier

On Wed, 18 Aug 2021 at 09:04, Charbel Lahoud <charbel.lahoud at etit.tu-chemnitz.de<mailto:charbel.lahoud at etit.tu-chemnitz.de>> wrote:
Hi Xavier,

Thanks a lot for your email.
I have work it out by modifying the AGC parameters as you mentioned (AGC_HOLD_COUNT = 200 or 300; AGC_DEFAULT_TARGET = 2.5). While the system is stationary, it works well but while UE is moving the performance drops again resulting in high BER.
Any suggestions?
Thank you.

Best regards
Charbel

From: Xavier Arteaga <xavier at srs.io<mailto:xavier at srs.io>>
Sent: Tuesday, August 17, 2021 8:19 AM
To: Charbel Lahoud <charbel.lahoud at etit.tu-chemnitz.de<mailto:charbel.lahoud at etit.tu-chemnitz.de>>
Cc: srslte-users at lists.softwareradiosystems.com<mailto:srslte-users at lists.softwareradiosystems.com>
Subject: Re: [srslte-users] AGC Mode

Hi Charbel,
Many thanks for your email. You could try the following:
- It could be the gain setting is slow, could you try increasing SRSRAN_AGC_HOLD_COUNT to >100;
- Lower SRSRAN_AGC_DEFAULT_TARGET

Regards,
Xavier

On Mon, 16 Aug 2021 at 16:07, Charbel Lahoud <charbel.lahoud at etit.tu-chemnitz.de<mailto:charbel.lahoud at etit.tu-chemnitz.de>> wrote:
Hi,

Recently, I have been testing using srsRAN AGC mode. I have recognized that on srseNB and srsUE, the AGC mode isn’t working properly.
While AGC is enabled, the RSRP is changing a lot and I have a high BER. On the other hand, while specifying the RX gain on both eNB and UE, I am able to reach max performance.

I am using USRP N321 with UHD 3.15 LTS on both sides eNB and UE. As well, I have modified the max gain inside AGC.c  to 60.0 and increased the number of measurements, but still getting the same results from high BER and low performance.

Any suggestions will be helpful. Thanks a lot.

Best regards,
Charbel
_______________________________________________
srslte-users mailing list
srslte-users at lists.softwareradiosystems.com<mailto:srslte-users at lists.softwareradiosystems.com>
https://lists.softwareradiosystems.com/mailman/listinfo/srslte-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20210818/23e646f1/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ue_metrics.zip
Type: application/x-zip-compressed
Size: 2760913 bytes
Desc: ue_metrics.zip
URL: <https://lists.srsran.com/pipermail/srsran-users/attachments/20210818/23e646f1/attachment.bin>