一种新颖的抗阶数过估计的增强子空间盲辨识算法  

An enhanced subspace method for blind identification robust to order overestimation

作　　者：郭士旭[1] 蒋建中[2] 刘世刚[2] 崔波[2] 

机构地区：[1]总参工程兵科研三所,河南洛阳471000 [2]解放军信息工程大学信息工程学院,河南郑州450002

出　　处：《电路与系统学报》2013年第2期261-268,共8页Journal of Circuits and Systems

基　　金：国家863计划资助项目"超宽带无线通信系统研发与应用示范"(2009AA011205)

摘　　要：研究了子空间盲辨识算法在信道阶数过估计时失效的问题,提出一种新的鲁棒性强的增强算法。首先针对过估计模型,提出了"加权矩阵定理",理论证明通过构造一个加权矩阵来修正子空间算法,可消除阶数过估计的影响;然后,给出一种用指数序列构造近似的加权矩阵的增强算法。与子空间盲辨识算法相比,增强算法复杂度相当,对阶数过估计鲁棒性强。仿真实验验证了该算法的有效性。This paper addresses the problem of Subspace method that it may fail when the channel order is overestimated.An enhanced Subspace method is presented,which is robust to channel order overestimation.Firstly,overmodel is studied,and a new theorem is derived and proved,which indicates that a weighting matrix can remove the influence of the overestimated order to Subspace method.Then,the weighting matrix can be approximated by an exponential sequence.Compared to Subspace method,the proposed algorithm has the same computational cost and is more robust to channel order overestimation.Simulations demonstrate the effectiveness of the algorithm.

关 键 词：子空间算法 信道阶数过估计 信道盲辨识 单输入多输出 二阶统计量 

分 类 号：TN911[电子电信—通信与信息系统]