UMP-BP算法中置信度被高估的改进

Improvements on overestimation of reliabilities for UMP-BP algorithm

分析了用于低密度奇偶校验码迭代概率译码的最大噪声功率归一化置信度传播算法(UM P-BP算法)中置信度被高估的主要原因,提出了改进算法(MUM P-BP算法).算法中减小置信度更新值的最佳乘性系数与低密度奇偶校验码变量节点的度数成反比关系.在稍许增加译码计算量的代价下,MUM P-BP算法的误码性能超过了UM P-BP算法,接近或略优于BP算法.

In this paper, a main reason for overestimation of reliabilities in uniformly most power based belief propagation （UMP-BP） decoding algorithm is analyzed. Overestimation of reliabilities in iterative decoding algorithm for low-density parity check （LDPC） codes may results in performance degradations. An modified UMP-BP （MUMP-BP） algorithm is proposed. The optimal multiplicative factor to decrease reliabilities is in inverse proportion to variable node degrees. Simulation results show BER performance improvements for MUMP-BP over conventional UMP-BP algorithm and sometimes over belief propagation （BP） algorithm at the cost of slightly increasing computational complexities.