Application of Uncorrelated Leaning from Low-Rank Dictionary in Blind Source Separation

Abstract: This paper proposes a kind of method about signal BOA estimation from the aspect of sparse decomposition. The whole interested space is divided into several potential angles of arrival to establish a over-complete directory to convert the estimation problem of signal DOA to sparse representation problem. A MMV array is formed by data received from multiple snapshots, then using optimization method of joint sparse constraint to solve the problem. First, make singular value decomposition on received data array to connect the each snapshot data, then using the sparse representation problem of ݈଴ bounded to solve the problem. To improve the anti-noise performance of algorithm, the paper applies similar Sigmoid function of two parameters to approximate ݈଴ norm. This method applies to the DOA estimation of narrow-band andbroad-band signal.  ଴ܮܵܬെ ܸܵ ܦshall be used for solving MMV problem, which achieves joint sparse constraint of all frequency of reception matrix of broadband signal, to make array elements spacing break through the limitation of half wavelength and improve resolution of DOA signal.
Keywords: Blind source; Source separation; Joint sparse; Smooth norm
Author: Liu Sheng
Journal Code: jptkomputergg160057

Artikel Terkait :