Near-optimum Results on Improved Belief Propagation-Based Decoding of LDPC Codes

摘要

In this paper, we provide several near-optimum results on scaled and offset belief propagation (BP)-based decoding of low-density parity-check (LDPC) codes. A practical method for searching the best suboptimal factors of these two improved BP-based algorithms is presented. The simulation results for the (3,6) regular LDPC code of length 8000 on the additive white Gaussian noise (AWGN) channel show that both algorithms with carefully chosen factors perform well, and approach the BP algorithm within 0.05dB.