基于置信传播和波束搜索的LDPC联合译码算法

Combination of Belief Propagation and Beam Search Algorithms for the Decoding of LDPC Codes

当置信传播算法BP(Belief Propagation)应用于LDPC短码时,由于传递消息的不独立性,使得译码性能与最大似然译码有较大差距。考虑到LDPC码分布的稀疏性,以及在BP失败译码中仅有少量错误位的统计事实,提出了一种波束搜索算法,把它级联于BP算法来对失败译码进行后处理。在付出一定计算复杂性的代价下,以期提高整体的译码性能。在波束搜索算法的实现中,译码伴随式的信息以及码字位的可靠性信息被融合进对搜索个体的代价函数中。它的目标是在给定的波束宽度下,寻找到失败译码对应的错误样式,从而纠错的目的。仿真结果表明对LDPC短码,此联合算法在译码性能与计算复杂度之间取得较好的折衷。

For short low-density parity-check codes, due to the message gap between the belief propagation （BP） and the maximum likelihood d dependency, there exists a performance ecoding （MLD） algorithms. Aiming at bridging the gap at the cost of some complexity,we put forward a beam search （BS） postprocessing scheme, based on the fact that LDPC codes commonly have sparsely distributed codewords and only a small portion of the codeword bits are erroneous in a BP decoding failure. In the implementation of BS to search for the error pattern, when evaluating the intermediate individuals given a limited beam width, we take advantage of the information about the decoding syndrome and the reliability metric of each bit fnlly. Simulation results show that the proposed method achieves a good tradeoff between performance and complexity.