Precoding — The matrix the transmitter applies — and why NR hides it
Space — layers, ports, reporting
Where it sits
What it is
What it is. The matrix the transmitter applies to map layers onto physical antennas — choosing the amplitude and phase from each element so the streams arrive separable at the device.
In NR's downlink it is invisible, and PDSCH §6.4 makes the argument: TS 38.211 clause 7.3.1.4 is an identity, TS 36.211 clause 6.3.4 (LTE) is a full clause with a codebook, and the difference is that NR precodes the DM-RS with the data so the device never needs to know the matrix.
Where the codebooks survive. On the reporting side only — TS 38.214 clause 5.2.2.2 — as the vocabulary a device uses to suggest a precoder. The base station is free to ignore the suggestion entirely.
And on the uplink, where precoding is not transparent: the network tells the device which precoder to use, because the network is the one that must separate the streams (ref-channels §9).
The asymmetry has one cause. The end that receives must be able to separate the streams; the end that transmits must know how to make them separable. Downlink: base station transmits, device receives, and the device manages with DM-RS. Uplink: device transmits, base station receives — and the base station, having many antennas and a scheduler, would rather dictate than infer.
Read on
This concept was first written up in ref-mimo, which reads the whole group as one argument.
Before this concept, the hierarchy says to learn the following — the full chain, in order: