中文摘要：

多元LDPC码采用扩展最小和（EMS）算法进行译码时,若消息向量长度取值过小,则性能相对其采用多元和积算法（QSPA）有很大损失.针对该问题,提出了一种动态扩展最小和（D-EMS）译码算法.首先,基于Monte Carlo方法研究了消息向量中有效似然值在各GF（q）符号间的分布,得出随着译码迭代次数的增加,有效似然值逐渐集中于少部分符号.因此,D-EMS译码算法先将消息向量长度设为nm1,一定迭代次数后再将其截短为nm2,这样译码复杂度可得到有效降低.同时,为了降低译码器实数比较运算复杂度,D-EMS算法校验节点基本步骤采用检泡（BC）算法.复杂度分析和仿真结果表明,在合理的参数设置下,D-EMS算法在有效降低EMS算法译码复杂度的同时,其性能在AWGN和Rayleigh衰落信道下均逼近相应EMS算法,因此可有效应用于基于多元LDPC码的实际通信系统.

英文摘要：

Decoded with the extended min-sum（EMS） algorithm,q-ary LDPC codes suffer great performance degradation compared with that decoded with the q-ary sum-product algorithm（QSPA） in case the length nm of message vectors is too small.To solve this problem,a dynamic EMS（D-EMS） algorithm is proposed in this paper.First,we examined the distribution of likelihood values among GF（q） symbols in the message vectors,and concluded that the likelihood values would concentrate to a small portion of symbols as the iteration number increases.Therefore,the D-EMS algorithm first set the length of message vectors to nm1,then truncate it to nm2 after certain decoding iterations,thus the decoding computational complexity can be efficiently reduced.Meanwhile,in order to reduce the complexity of real comparisons in the decoder,the proposed algorithm employs the bubble check（BC） algorithm during the check node elementary steps.Complexity analysis and simulation results indicate that,under appropriate parameter configurations,while efficiently reducing the decoding complexity of the EMS algorithm,the D-EMS algorithm performs nearly as well as the corresponding EMS algorithm over both AWGN and Rayleigh fading channels,and thus can be efficiently applied to practical communication systems based on q-ary LDPC codes.