频率选择性慢衰落信道下的消息传递迭代均衡算法  

Turbo Equalization for Frequency-Selective Slow Fading Channels Using Combined Message Passing Algorithm

作　　者：曹荷芳 张传宗 王忠勇[1] 王行业 CAO He-fang1, ZHANG Chuan-zong2 ,WANG Zhong-yong1, WANG Xing-ye3(1. Zhengzhou University, Zhengzhou, Henan 450001, China;2. Nanyang Institute of Technology, Nanyang, Henan 473004, China;3. North China University of Water Resources and Electric Power, Zhengzhou, Henan 450045, Chin)

机构地区：[1]郑州大学,河南郑州450001 [2]南阳理工学院,河南南阳473004 [3]华北水利水电大学,河南郑州450045

出　　处：《信号处理》2018年第3期312-318,共7页Journal of Signal Processing

基　　金：国家自然科学基金资助项目(61571402)

摘　　要：针对频率选择性慢衰落信道,现有的迭代接收机性能与最优估计性能有较大差距,且复杂度高。为了提高频率选择性慢衰落信道均衡器的性能,提出了BP-EP-MF消息传递迭代均衡算法。该算法包含置信传播(Belief Propagation)、期望传播(Expectation Propagation)和平均场(Mean Field),其中MF算法处理非线性因子节点,EP算法降低计算复杂度。消息传递算法首先利用系统中的未知变量及关系建立信道的因子图模型,然后根据因子图上各部分的特点选择合理的、有效的消息更新规则和合适的消息更新机制,最后依据最大后验估计准则得到数据估计值。仿真结果表明,与非线性卡尔曼(EKF)迭代均衡器相比,消息传递算法(BP-EP-MF)均衡器的性能大幅提升、收敛速度加快、复杂度略微降低。The performance of the existing iterative receiver with high computational complexity had to be a big difference from optimal receiver for frequency-selective slow fading channels. In order to improve the performance of the equalizer, the BP-EP-MF messaging passing iterative equalization algorithm was proposed in this paper. The proposed message passing algorithm contained belief propagation （BP） , expectation propagation （EP） and mean field （MF）. MF was used to handle the nonlinear model of the factor, meanwhile EP made a great contribution for the low complexity implementation of the proposed message passing algorithm iterative receiver. Firstly, a channel factor graph model that based on global posterior probability density function was established. In the channel factor graph model, all unknown variables in the system and the relationship between factor nodes and variable nodes were made clear. Then, according to the characteristics of the factors, reasonable and effective message passing rules and the appropriate message update mechanism were chosen to improve the performance of iterative receiver and reduce the computational complexity. Finally, the data symbols can be estimated based on maximum a posteriori estimation. Simulation results turn out that our message-passing algorithm with little less computational complexity and faster convergence speed leads to better performance than fixed-lag soft input nonlinear Kalman filtering EKF） for frequency-selective slow fading channels.

关 键 词：信道估计 符号间干扰 消息传递算法 频率选择性慢衰落 迭代均衡器 

分 类 号：TN911.23[电子电信—通信与信息系统]