Knowledge‐aided informed dynamic scheduling for LDPC decoding of short blocks

Abstract

Low-density parity-check (LDPC) codes have excellent performance for a wide range of applications at reasonable complexity. LDPC codes with short blocks avoid the high latency of codes with large block lengths, making them potential candidates for ultra reliable low-latency applications of future wireless standards. In this work, a novel informed dynamic scheduling (IDS) strategy for decoding LDPC codes, denoted reliability-based residual belief propagation (Rel-RBP), is developed by exploiting the reliability of the message and the residuals of the possible updates to choose the messages to be used by the decoding algorithm. A different measure for each iteration of the IDS schemes is also presented, which underlies the high cost of those algorithms in terms of computational complexity and motivates the development of the proposed strategy. Simulations show that Rel-RBP speeds up the decoding at reduced complexity and results in error rate performance gains over prior work.