Ensemble Belief Propagation Decoding for Short Linear Block Codes

Abstract:
The upcoming communication standard 6G is expected to specify high demands on the efficiency of current communication systems, especially on the latency and error correction performance of the decoding unit. A well-investigated algorithm that allows a low-latency implementation is Belief Propagation (BP) decoding on the Tanner graph, which is the state-of-the-art decoding algorithm for Low-Density Parity-Check (LDPC) codes. In this paper, we present a methodology to apply this BP algorithm to non-LDPC codes while achieving a good error correction performance. In addition, we propose a method to generate a diverse ensemble of parity-check matrices for short linear block codes to enable the use of Multiple-Bases Belief Propagation (MBBP), an extension to BP, further improving the error correction performance of LDPC and non-LDPC codes. At the example of LTE Turbo codes, our results show that this approach is capable of achieving the error-correcting performance of the Turbo code’s state-of-the-art max-Log MAP algorithm, while being a universal approach for decoding any kind of short linear block code.