结合扩展卡尔曼的单通道盲分离PSP算法  

PSP Single-channel Blind Source Separation with Extended Kalman Filtering

作　　者：严雅岚 高勇[1] YAN Yalan;GAO Yong(College of Electronics and Information Engineering,Sichuan University,Chengdu Sichuan 610065,China)

机构地区：[1]四川大学电子信息学院,四川成都610065

出　　处：《通信技术》2024年第4期347-353,共7页Communications Technology

摘　　要：针对非平稳信道的GMSK混合信号的单通道盲分离问题,提出了一种可适用于瑞利信道的单通道盲分离算法。采用逐幸存路径处理算法对信道与符号进行联合估计,引入扩展卡尔曼滤波以优化算法对变化信道的跟踪能力。该算法结合了扩展卡尔曼滤波对非线性系统的近似线性处理,以及逐幸存路径算法在无先验信息情况下对符号和信道参数良好的估计性能。仿真结果表明,该算法在平稳信道条件下能够实现较好的符号序列估计,对比传统PSP算法有近1.5 dB的性能提升,在非平稳信道条件下对比传统PSP算法,对混叠信号分离性能大幅提升约2 dB。To address the problem of single-channel blind source separation of GMSK mixed signals over non-stationary channels,a single-channel blind source separation algorithm that can be applied to Rayleigh channels is proposed.The per-survivor processing algorithm is used to joint estimate channels and symbols,and extended Kalman filtering is introduced to optimize the ability to track channel conversion.The algorithm combines the approximate linear processing of extended Kalman filtering for nonlinear systems and the excellent estimation performance of the per-survivor processing algorithm for symbols and channel parameters in the absence of a priori information.The simulation results indicate that the algorithm can achieve better symbol sequence estimation under stationary channel conditions,with a performance improvement of nearly 1.5dB compared with the conventional PSP algorithm,while the separation performance of the aliased signals is greatly improved by about 2dB compared with the conventional PSP algorithm under non-stationary channel conditions.

关 键 词：单通道盲分离 逐幸存路径处理 扩展卡尔曼滤波 GMSK 瑞利信道 

分 类 号：TN911.7[电子电信—通信与信息系统]