一种基于改进线性规划的LDPC码混合译码算法

A hybrid decoding algorithm based on improved linear programming for LDPC codes

与基于消息迭代的置信传播译码相比,线性规划(linear programming,LP)译码分析有限长LDPC码性能更为有效。然而,传统LP译码算法运算量非常大,不利于系统实现。本文结合LDPC码校验矩阵的特点,去掉传统LP译码中不必要的约束,得到一种低复杂度LP内点译码算法。为了降低译码延时,将LP内点译码算法与置信传播译码算法结合,提出LDPC码混合译码算法。仿真结果表明,混合译码算法的误码性能优于传统LP译码和BP译码算法,而译码延时低于传统LP译码。

linear programming（LP） decoding is more flexible and effective than message passing iterative decoding in analyzing finite-length LDPC code.However,LP decoding is too complex to implement.Considering the structure of the parity check matrix of the LDPC codes,this paper proposes a new interior point decoding algorithm with low complexity for LP by removing all of the unnecessary restrictions in traditional LP decoding.In order to reduce the decoding latency,a hybrid decoding which combines the interior point decoding and the message passing iterative decoding is presented.Simulation results show that new method is superior to traditional LP decoding and BP decoding in BER,while the decoding latency are lower than that of the traditional LP decoding.