In Huawei LTE, which parameters can decided it is SU-MIMO or MU-MIMO?
There are separate switch for MU-MIMO.
Does it give real gain as compared to SU-MIMO in LTE in terms of DL Throughput?
Only Capacity gain, no Throughput gain.
So in high users it can be considered not on less PRB utilisation and Volume.
MU-MIMO increases Cell Throughput so it also increases User Throughput because more PRB will be available and more DCI will be available.
Never seen Throughput gain in live it’s nightmare with MU-MIMO.
In case of high PRB utilization, MU-MIMO may also increase User Throughput.
But for Single user peak throughput, I think MU-MIMO feature require more overhead control channel, so peak su-thrp may be decreased.
To enable MU-MIMO there are 3 switches in nrducellalgoswitch, that indicates whether to enable the functions of the MU-MIMO feature in an NR DU cell.
UL_MU_MIMO_SW: uplink MU-MIMO function.
PDCCH_MU_SW: spatial multiplexing function on the PDCCH.
DL_MU_MIMO_SW: downlink MU-MIMO function.
For high traffic sites MU-MIMO brings significant benefit in terms of throughput and capacity.
Oh, I’ve forgotten: Huawei always show User Throughput gain.
Then you’re right @RFSpecialist !
Gain in average user throughput.
It is recommended for cells where average PRB utilization is over 90%.
Even Huawei shown for bbh 9 Mbps where other vendor just in 1 Mbps range
But mu-mimo performance is strongly dependant on users location in the cell.
And this isn’t in MNOs control.
If mu-mimo improve throughput for one user?
General consent is yes.
But there is a case when it does not improve user throughput: mu-mimo can schedule max 2 layers per UE so not possible rank 3 or rank 4 for MU-MIMO UE.
For NR we can have 4 ues in pairing each with 4x4 DL MIMO mode considering 8 SSB used on site.
Theoretically I read but practically not sure.
For MU-MIMO max 2 layers per UE.
For LTE yes, with tm8 mode.
For NR it can be 4, I read it.
I’ve read 2 (even in 5G Bullets book), and all vendor trial MU-MIMO also just 2 layers per UE.
To reach 16 layers their test used 8 UE x 2layers.
I think if can support 4layer/UE, using 4UE x 4 layer may have better total throughput than 8 Ue x 2 layers because of less interference between UE.