How does UE know how many layers to use in DL using DCI in 5GNR?

Hi All,
In NR, how does UE get to know about how many layers to use in DL using DCI?

More details here:

Thanks @RFSpecialist. In UL it shoul be Precoding info which decides this, right?

This is for DL.
Precoding matrix is acting in DL.

I meant to decide UL layers… precoding Information element sent in UL DCI is used, right?

Precoding in DL is not supported in NR it seems

I do not find this information in 3gpp.

And here there is information about precoding matrix in NR:

Thanks @RFSpecialist.

In TS 38.211, it talks about precoding in PUSCH, but nothing on PDSCH.
That is where I am trying to get my understanding from.
I do read in some docs that downlink MIMO transmission in NR is mainly based on non-codebook-based precoding.
So it might just be that it doesn’t support codebook based precoding in DL.

Try 38.214 chapter
Physical procedures for data.

In some books you will find that preferred sets of beamfoming weights are both used for beamfoming and precoding weights. It doesn’t say that precoding is not used in NR.

Just that not standardized, that’s why couldn’t see much info in specs.

BTW, in LTE… we could move from spatial to transmit diversity in TM3 based on codebookSubsetRestriction IE.

How would that be done in NR?

No TM modes defined for NR …so no concept of switching here.
Transmission scheme 1 there.

Yes, correct.