一种改进的基于秩检测的信道有效阶数估计算法  

Improved Channel Order Estimation Algorithm Based on Rank Test

作　　者：郭士旭 郭军利 蒋建中 刘世刚[1] 

机构地区：[1]信息工程大学信息系统工程学院,河南郑州450002

出　　处：《信息工程大学学报》2012年第5期552-557,共6页Journal of Information Engineering University

基　　金：国家863计划资助项目(2009AA011205)

摘　　要：信道阶数估计是SOS类盲辨识算法的前端步骤。针对Liavas阶数估计算法对信噪比要求高的问题,通过对接收序列自相关矩阵进行"降噪"处理和改进秩检测目标函数,提出一种改进算法,有效降低了对信噪比的要求。分析了在不同信道条件下,各种阶数估计算法的性能,阐明随着信噪比降低,阶数估计误差不可避免,该结论是进一步研究对阶数估计误差鲁棒的盲辨识算法的重要依据。Channel order estimation is a basic step for blind identification algorithms based on second order statistics. Since the Liavas algorithm demands high SNR, an improved algorithm is proposed which can effectively reduce the SNR requirement by removing the noise from the autocorrelation ma- trix of the received signal sequences and improving the objective function of the rank test. The per- formance analysis of various order estimation algorithms under different channel conditions leads to the conclusion that as the SNR decreases, the order estimation error is inevitable. This conclusion is an important basis for further study on robust algorithms of blind order estimation errors identifica- tion.

关 键 词：阶数估计 单输入多输出 二阶统计量 信道盲辨识 秩检测 

分 类 号：TN911[电子电信—通信与信息系统]