使用误码率仿真的方法,比较了基于Viterbi算法(VA)的最大似然序列估计(MLSE)电色散补偿技术和低密度奇偶校验(LDPC)纠错译码技术的VA+BF译码、VA+BP译码简单级联和VA+BP译码置信级联3种级联方法。仿真结果表明,本文提出的置信级联能够有效地判断出VA给出的序列估计值是否可信,并在此基础上进行LDPC译码需要的对数似然比(LLR)信息的初始化。与简单级联方式相比,置信级联的方法能够获得更好的LDPC软译码性能,而且随着光纤传输距离的增加和码间串扰(ISI)效应的恶化,置信级联比简单级联具有更高的性能增益。
This paper compares three concatenation methods which are Viterbi algorithm(VA)-based maximum-likelihood sequence estimation(MLSE),and low-density parity-check(LDPC) decoding:VA+BF decoding,VA+BP decoding with simple concatenation,and VA+BP decoding with belief-based concatenation.Simulation results show that the belief-based concatenation proposed by this paper can effectively determine whether a bit from the VA is trustful and how to use it in the LLR(Log-likelihood ratio) initialization for decoding.Comp...