AIR6468 High DL BLER

Hi,

Anyone with an issue regarding High DL BLER ~30-40% at AIR6468 64x64 LTE TDD?
Deployment has been done in hot spot area. The issue appeared after SW upgrade however the SW has been checked so far and LA works properly. No issue visible in moshell (especially when it comes to CPRI) - ‘sdi’ command, no alarms. From BB logs it is visible that UEs have low DL SINR. We have enabled both TM7ModeSwitching and TM8ModeSwitching (a little improvement however the issue still exists and can only be recovered by cell lock/unlock or unit restart).

Any proposals what can be checked further?

Thank you!

How are the RSSI values? If you have an external interference it could be the cause of the high values in the BLER. I have also seen cases where the antenna may have a hardware failure, and replacing the antenna fixes the problem.

We didn’t check DL RSSI only reported CQi however we took a look at UL RSSI counters assuming channel reciprocity and no issues there. Basing on the latest analysis we only correlated the issue with that the bad UEs support more bands (UE Capability info). I guess this is interesting.

Regarding HW we excluded it coz the issue appears at different AIR sites in different period of time that is tricky.

To find out the real reason of high bler , we used tti trace from vendor in slot scale for 3.5ghz.
From tti trace, which slots cause high bler is visible. Make sure also if you have 3.5 ghz tdd around 5G which can interfere 4G TTI if slot format overlaps in some slots. If you totally sure it is not a baseband or aau issue I suggest you to move on tti trace analysis.
.

I agree with @LordGm, you should move on tti trace analysis, because DL low SINR should come from the interferences of adjacent cells due to time advancement setup mismatch, or clock synchronization issue between cells. You mentioned TM7 and TM8 transmission mode switching didn’t help, but cell lock/unlock and restart unit helps.

@LordGm @Wendy

We’ll check it. Thanks for tips. :grinning:

Hi,

Just to inform you there was RADIOSW fault