Generalized Two-Magnitude Check Node Updating with Self Correction for 5G LDPC Codes Decoding

Abstract:
The min-sum (MS) and approximate-min∗ (a-min∗) algorithms are alternatives of the belief propagation (BP) algorithm for decoding low-density parity-check (LDPC) codes. To lower the BP decoding complexity, both algorithms compute two magnitudes at each check node (CN) and pass them to the neighboring variable nodes (VNs). In this work we propose a new algorithm, ga-min∗, that generalizes the MS and a-min∗ in terms of number of incoming messages to a CN. We analyze and demonstrate a condition to improve the performance when applying self-correction to the ga-min∗. Simulations on 5G LDPC codes show that the proposed decoding algorithm yields comparable performance to the a-min∗ with a significant reduction in complexity, and it is robust against LLR mismatch.c