一种改进型LLR BP算法的LDPC译码研究

A New Modified LLR BP Algorithm Research for LDPC Decoders

采用经典LLR BP算法对低密度奇偶校验(Low Density Parity Check,LDPC)码译码时,译码复杂度非常高,主要表现在译码时迭代次数过多和每次循环时校验节点的计算复杂度过高。基于以上问题提出一种新的改进型LLR BP译码算法,自适应地实时监控译码是否稳定来降低在低信噪比条件下的译码迭代次数,同时采用函数逼近的方法降低每次循环时校验节点的计算复杂度。仿真结果表明,该算法在译码性能损失不大的情况下可大幅降低LDPC码的译码复杂度。

Classical LLR BP algorithm to decode LDPC involves high complexitybecause of iterations and calculation complexity of horizontal directions in every iteration process. This paper proposes a new modified LLR BP algorithm to monitor whether decoding output is stable or not to decrease iteration times in low signal noise rate. At the same time,Taylor function approximation is adopted to reduce calculation complexity at check codes. Simulations reveal that this proposed method can reduce decode complexity widely for LDPC in the condition of a small decoding performance loss.