It is mentioned that for MU-MIMO maximum number of layers for one UE is 2 (rank2).
But Huawei says they can deliver 4 layers.
How is this possible? What is correct?
This is 4 users MU-MIMO, but dont know how many layers for each users.
In some MU-MIMO of Huawei & E//, I see they test 16 layers with 8 UE.
Right, this means maximum 2 layers per UE in mu-mimo scenario.
Why Huawei says it can support 4 layers per user in MU-MIMO scenario?
16 layers so 16 DMRS ports, right?
Which DMRS type 1 or 2 do you think for 16 layers?
Max 12 DMRS ports possible.
I remember someone said they can use othogonal ports.
12 are orthogonal. And 4 are reutilized to get to the value of 16.
As per MU MIMO testing experience, UE uses only max rank2 when participating in pairing. The moment when MUBF layers reduced and no pairing on those TTI, same UE will use rank-4. So practically, we have seen only rank-2 under MU-MIMO. HW docs are not appropriate most of the time.
Thanks @ajaymsfather, it makes sense.
How do you monitor MU-MIMO UE only rank2?
Can you draw how DMRS ports for 8 UEs using MU-MIMO 16 layers?
It is easy. You look at codebook type:
- If type I means it is SU-MIMO.
- If type II means it is MU-MIMO.
And rank is reported by UE all the time.
In DCI is reported antenna ports and number of layers.
Some DMRS are reused in different spatial location.
I mean we monitor at OMC tracing or UE log?
Type 2 for MU-MIMO.
In Huawei there are 16 pdsch layers possible in mu-mimo scenario.
That means some dmrs ports will be reused in different spatial locations.
Yes seems so. In my case as of now max 8 layers, i.e. 4 ssb beam so 4 UE can be in pairing.
But @ankgitm, why do you always says that pdsch beams are ssb beams?
They are different, right?
PDSCH beams are part of GoB (Grid of Beams) that can be dynamic where as SSB beams are static.
As per my understanding without refined beam we need layers to transmit codeword to users that come from ssb beam only.
1 ssb = 2 layers
Is it like we have some different layers for pdsch.
In MU-MIMO, Huawei suggest RSRP of best beam > second best beam at least 10 dB?
This is only for csi-rs beams.
For regular PDSCH beams it is not required this.
So by reporting PMI UE basically indicated what would be the best beam that gnodeB has to schedule data on.
Each antenna port 3000s has a unique csi-rs reference signal that is un-precoded.
PDSCH beam also static per radio module; position in the amplitude and phase of each antenna element for refined beam and corse beam is fixed per radio module in Nokia.
PDSCH dynamic means is it changing based on UE location?