针对低密度奇偶校验(LDPC)码的BP译码算法在每一次迭代过程中,都要对全部比特和校验信息进行更新,存在计算量大、译码效率低的问题.提出了一种改进的BP译码算法.由于不同的比特节点和校验节点,其可靠程度不同,对BP译码的贡献也不同,为此给出了一种新的可靠性判断准则:采用每个比特的非法校验数和每次迭代过程中比特的伪后验概率的差来判断比特的可靠性,认为非法校验数小、伪后验概率差大的节点具有较高的可靠度.对可靠性较高的比特,下一次迭代过程中不参与更新,只更新那些有可能发生错误的比特.仿真结果表明,改进的BP译码算法在损失极少译码性能的情况下,大大地减少了迭代过程中的计算量,提高了译码效率.
An improved belief propagation (BP) decoding scheme was put forward to solve the high complexity and low decoding efficiency problem of BP in low density parity check (LDPC) codes. Because different bit nodes and check nodes have different reliability and contribute differently to BP decoding; the proposed decoding scheme employs a differentiate rule based on lawless check number and pseudo posterior probability to evaluate every bit's reliability before each step of iteration; The nodes have little lawless check number and large pseudo posterior probability are supposed to have higher reliability. The bits with high reliability need not to be updated by the decoder. Simulation shows that the proposed BP decoding algorithm can reduce the computational load and increase the decoding efficiency significantly.