Hi All.
Question: in 5G how is SSB with 4 csi-rs beams interact with PMI based beam selection?
Seems it can be either GoB with SSB with csi-rs beams. Or GoB with PMI beam selection.
Now the next question is: how do we recognize in UE logs which method is used by gNodeB: GoB with SSB and CSI-RS beams or GoB with PMI?
Does It depends on report quantity IE?
Are you sure that we can get this information from UE logs?
Because UE may be reporting PMI and CSI-RS measurement simultaneously.
So it’s up to gNB what criteria to use.
gNB would just instruct which beam UE should decode.
Isn’t it?
If report quantity is “CRI-RSRP” then it is based on csi-rs beams.
If report quantity is “CRI-RI-CQI-PMI” then it is GoB with PMIs.
There are many other options for report quantity.
Do you want to say with this trigger quantity UE will not send CQI PMI RI?
I don’t think so.
This is why we have several csi-rs resources. Each one is with a different purpose.
For that csi-rs resource it will send just cri and rsrp but for other csi-resource it will send cqi pmi ri.
Most of the logs are with report quantity is “cri-RI-CQI-PMI”.
Damn confusing for me. 
I need to read more about it.
The whole secret is inside reportquantity.
This is what UE reports.
Also it’s needed to understand weights.
And precoding for MIMO (PMI) and precoding for beamforming.
Extra info: I think this Eigen Based beamforming based on SRS is the best as it provides the weights without restrictions so gNodeB could use the most accurate values providing the highest gain for beams. This would not be possible with PMI where there are less options for PMI value. Only few limited like in the codebook. The “book” is limited to few codes with PMI not to all options codes like with SRS. Anyway SRS based is possible only on TDD.
In LTE mMIMO what seen weights decided on base of Srs only.
For NR it’s change to csi reporting where fixed weights used.
Now in later NR release again weights decide on basis of SRS report.
By weight I mean long term and short term weights called 2 steps weight algorithm.
Srs is very acurate, pmi is not so accurate.
Yes, true.
Once we have MU-MIMO with FR1 algorithm ready then it will change to srs based.
Reason being srs tell position of user as well ie AoA.
So gnb can come to know which users to keep in pairing which to be excluded.
As per spatial seperation of users.
It’s purely gNB decision as per inbulit algorithm on basis of AoA driven from SRS.
Also once needs to ascertain how reliable SRS is in poor SINR so gNB doesn’t make any wrong decision.
It works even for srs sinr in UL at -12 dB.
Amazing.