变速率部分更新盲均衡算法  

A Variable Step-size Partial Update Blind Equalization Algorithm

作　　者：杨大龙[1,2] 任亚博[1,2] 张健[2] 陈志强[1] 陈大海[2] 

机构地区：[1]清华大学工程物理系,北京100084 [2]中国工程物理研究院电子工程研究所,四川绵阳621900

出　　处：《电讯技术》2014年第7期869-875,共7页Telecommunication Engineering

摘　　要：对于长抽头系数自适应算法,基于最大化自适应滤波器系数误差向量原则的变速率部分更新算法,能够在大幅度降低算法实现复杂度的同时,解决部分更新算法收敛速度慢的问题。但是,该变速率算法仅适用于LMS结构,对于具有非线性代价函数的部分更新自适应盲均衡算法并不适用。基于同样的最优化思想,通过替换步长计算表达式中的部分统计量,提出了能够适合于部分更新多模盲均衡算法(MMA)的确定性变步长控制算法,并通过递归的方式计算步长值,简化了实现过程。对固定信道和时变信道的数值仿真结果表明,新算法相比传统基于收敛误差的经验性变步长算法具有更快的收敛速度和更好的跟踪性能,有效解决了部分更新自适应盲均衡算法的确定性变速率控制问题,提升了算法的收敛速度和跟踪性能。For adaptive filters with large amount of filter taps,the variable step-size（VSS） partial update algorithm can significantly reduce the computational complexity and solve the problem of slow convergence speed of partial update algorithm at the same time by maximizing the filter tap error vector’s mean square deviation. However,this VSS method can only be applied to LMS structure,not to the partial update blind equalizers with non-linear cost functions. Based on the same optimization ideas,a new deterministic VSS algorithm is proposed by replacing some statistic expression,which is suitable for multi-modulus algorithm （MMA）. And the proposed algorithm is simplified by recursive calculation. Numerical simulation results under fixed and time varying channels show that the new algorithm can realize faster convergence speed and better tracking performance than the traditional empirical VSS based on the mean square error. The pro-posed VSS algorithm effectively solves the deterministic VSS control problem for partial update adaptive blind equalization algorithm,and improves the convergence speed and tracking performance.

关 键 词：盲均衡 多模算法 变速率算法 部分更新 收敛速度 跟踪性能 

分 类 号：TN911.5[电子电信—通信与信息系统]