基于有环因子图的密度进化理论分析  

Analysis of Density Evolution on Cycled Factor Graph

作　　者：童胜勤[1] 邓勇强[1] 

机构地区：[1]华中科技大学电子与信息工程系,武汉430074

出　　处：《计算机科学》2007年第11期41-43,共3页Computer Science

基　　金：国家自然科学基金重大项目"未来移动通信系统基础理论与技术研究"(No.60496315);国家高技术研究发展计划(863计划)(No.2003AA12331005)

摘　　要：密度进化理论是分析低密度校验码的迭代译码性能的有效工具。本文在对密度进化理论进行研究的基础上,探讨了基于有环因子图的密度进化方法。首先讨论了有环因子图中环存在的情况,得到了环存在的概率表达式。然后研究了迭代译码算法中误码率的进化情况,在加入环存在对译码的影响因素后得到了迭代译码中误码率的进化表达式。在对该式讨论中,获得了有环情况下密度进化对信道条件的要求,即译码门限。本文的研究表明,在考虑因子图中存在环的情况下进行密度进化分析时,其获得的译码门限要低于不考虑环存在的情况。Density evolution （DE） is one of the most powerful tools for analyzing the performance of low-density paritycheck （LDPC） codes. With cycle-free factor graph as one of its fundamental assumptions, density evolution has been widely and successfully applied to different channels. An improved density evolution based on cycled factor graph is proposed in this paper. Research on the cycled factor graph and then the probability formula of cycles in the factor graph is presented. Then the error bit probability of iterative decoding algorithm based on cycled factor graph is studied, and the formula of error bit probability is developed. The thresh hold is got based on the analyzed this formula. It is shown that thresh hold of cycled factor graph is lower than the value of cycle-free case.

关 键 词：低密度校验码 密度进化 环 因子图 迭代译码 

分 类 号：TN911.22[电子电信—通信与信息系统]