摘要： 短低密度校验(LDPC)码的Tanner图中通常存在环路，变量节点之间的信息不再相互独立，导致LLR BP算法译码性能的下降。针对上述问题，提出一种改进型LLR BP译码算法，推导出有环时变量节点的真实信息，利用最小均方误差准则计算出有记忆的变量节点信息的权值，通过调整变量节点信息的迭代过程降低变量节点之间信息的相关性。仿真结果表明，改进型LLR BP算法具有比LLR BP算法、归一化BP算法及偏移量BP算法更好的LDPC译码性能。

关键词: 低密度校验码, 置信传播, LLR BP译码算法, Normalized BP译码算法, Offset BP译码算法

Abstract: The Tanner graph of short Low Density Parity Check(LDPC) codes usually has loops, so the information within variable nodes is not mutually independent and the decoding performance of LLR BP decoding algorithm will decrease. A modified LLR BP decoding algorithm is proposed. The true information of variable nodes with loops is derived. The weights of variable nodes information with memory are calculated according to minimum mean square error rule. It can reduce the relativity of variable nodes by adjusting the iterative process of variable nodes information. Simulation results show that the LDPC decoding performance of modified LLR BP decoding algorithm is more effective than that of LLR BP decoding algorithm, normalized BP decoding algorithm and offset BP decoding algorithm.

Key words: Low Density Parity Check(LDPC) codes, Belief Propagation(BP), LLR BP decoding algorithm, Normalized BP decoding algorithm, Offset BP decoding algorithm

中图分类号: 

• TN919