Link adaptation — Choosing a rate the link will actually carry
Procedures over time
Where it sits
What it is
What it is. Choosing $Q_m$, $R$ and rank so the transmission is as fast as the channel will bear. PDSCH §10.3 shows the gap: 3GPP defines the CQI tables and the MCS tables, and never defines the mapping between them.
Inner loop — take the reported CQI, convert it to an MCS using the table correspondence of Table 10, schedule.
Outer loop — maintain an offset applied to that conversion, decreased on every NACK and increased on every ACK, so the measured block error rate converges on the 10 % the CQI definition assumed (PDSCH §10.1).
Why the outer loop must exist. The report is stale, it describes a reference resource rather than the actual allocation, and it is the device's opinion of its own receiver. The outer loop corrects all three without needing to know which caused the error.
This is standard engineering practice rather than a 3GPP requirement, and the usual citation is given in PDSCH §15.2. Leaving the control law unspecified is what lets implementations differ.
Read on
This concept was first written up in ref-system, which reads the whole group as one argument.
Before this concept, the hierarchy says to learn the following — the full chain, in order: