Does Huawei eNB use MU MIMO for the lightly loaded cell?
Is there any threshold like PRB usage or RRC users for MU MIMO?
Nope there is no threshold.
Once enabled it will work even for light load.
You can see this by their counters.
N.ChMeas.MIMO… DL or UL… MuPairing… Layers… etc…
Tthere can be MU-MIMO even with 2 paired layers.
By the counters you see that pairing can be even if in that TTI is only one layer used.
I am not able to understand one layer pairing.
For say 8 layer pairing I can serve 8 UEs which are geographically separated using the same RB with RI = 1.
But with one layer MU MIMO pairing I am not able to correlate it.
Nope for 8 layer pairing there are 4-8 UEs paired.
1 layer pairing means that although the group of paired UE is with 4-8 UEs only one layer is sent due to lack of data.
Only1 UE has data enough to be sent on 1 layer.
But scheduler work sin terms of 2 layers, 4 layers, 6 layers, 8 layers, etc.
You will seldom see 1 layer, 3 layers, 5 layers, 7 layers, etc.
Are there any specific parameters to define scheduler on how many layers?
Like 2, 4 and so on or is it rf based?
Like epf in LTE with regards to agg achieved in LTE.
Yes there is for Huawei.
It is sold in pairs of 2 pdsch layers per cell.
But poor rf can require more layers.
So you have to buy 2 layers or 4 layers or 6 layers or 8 layers, etc.
Then more layers will be used.
Thanks. So data is fixed of one layer or as per rf more can be required.
To see 8 layers required all 8 UEs be close to gNodeB and spread out quite well.
One UE in a beam.
Very particular position of users is required.
Very unlikely to see in reality except demo tests.
Okay, got It.
So in summary although gNB has paired 4 or 8 UEs, may be in some particular TTTs, it has scheduled only one UE, remaining 3 or 7 UEs are not scheduled in that particular TTI.
Am I correct?
Yes, it is like this.
There are parameters in RAN5.1 how often UEs are evaluated to get in our out of the pairing grouping.
Also threshold for correlation is a parameter.
When you say UE pairing requires UEs on a different CSI RS beam, is there an angular distance between the adjacent CSI RS beams used by gNB pairing calculation for othogonality?
No there is no angular distance.
Correlation is done based on beamforming weigths of each beam.
UE report CSI-RS beam set, gnb check if beam id of UEs are different, they can be paired.
Can you please elaborate more on this?
For eg: gNB transmit 32 csi-rs beams totally.
UE1 report beam set is beam 1+2, UE2 is beam 3+4 → these UE can be paired, used same PRB.
If UE2 report beam 2+3 → they can not be paired.
In which message it reports beam set id?