CAJ | 학술논문

多信道估计时，如果利用信道的稀疏性和多信道的相关性，可以提高信道估计性能。本文利用阵列信道的结构性稀疏特性，提出了一种多路分组稀疏LMS算法（Group Sparse LMS，GS-LMS）。该算法将多路信道作为一个整体同时进行自适应信道估计，通过引入l2，1范数，将结构性稀疏先验引入到稀疏LMS算法的代价函数中，导出新的滤波器权系数更新公式。仿真结果表明了在不同信道条件下，本文算法的稳态误差性能明显优于若干现有的稀疏LMS算法。
다신도고계시，여과이용신도적희소성화다신도적상관성，가이제고신도고계성능。본문이용진렬신도적결구성희소특성，제출료일충다로분조희소LMS산법（Group Sparse LMS，GS-LMS）。해산법장다로신도작위일개정체동시진행자괄응신도고계，통과인입l2，1범수，장결구성희소선험인입도희소LMS산법적대개함수중，도출신적려파기권계수경신공식。방진결과표명료재불동신도조건하，본문산법적은태오차성능명현우우약간현유적희소LMS산법。
When we do multichannel estimation,if the sparsity and relevance of multichannel are used,the performance of channel estimation can be improved.This paper proposes a Group Sparse LMS algorithm which is based on structural sparse priori of the array channel.The algorithm takes these channels as a whole for adaptive channel estimation,a gradient de-scent recursion of the filter coefficient vector is deduced through introducing l2,1 norm,which introduce structural sparse pri-ori to the criterion function of sparse LMS algorithm.The simulation results show that,under different path conditions, steady state error performance of the proposed algorithm is obviously better than several existing sparse LMS algorithms.