LDPC — Low-density parity-check coding — NR's data code
Bits into a codeword
Where it sits
What it is
LDPC — low-density parity-check coding — is the code NR uses for data. Not one code but two base graphs, chosen by payload size and target rate.
BG2 if the payload is at most 292 bits, or at most 3824 bits at rate ≤ 0.67, or at any size with rate ≤ 0.25. BG1 otherwise (TS 38.212 clause 7.2.2).
Two codes because one cannot be efficient at both ends. A code is punctured upward from its mother rate to reach a high rate and repeated downward to reach a low one, and both cost performance. BG1 is 46×68 with mother rate ≈ 0.32, for throughput; BG2 is 42×52 at ≈ 0.19, for coverage, and its smaller matrix is cheaper to decode — which matters because the coverage case is also the far-away, low-power one.
Read on
The full treatment is in nr-pdsch — this page is the definition and the map reference; that note is where the subject is actually developed.
Before this concept, the hierarchy says to learn the following — the full chain, in order: