Abstract:

For the direction of arrival (DOA) estimation, traditional sparse reconstruction methods for wideband signals usually need many iteration times. For this problem, a new method for two-dimensional wideband signals based on block sparse reconstruction is proposed. First, a prolate spheroidal wave function (PSWF) is used to fit the wideband signals, then the block sparse reconstruction technology is employed for DOA estimation. The proposed method uses orthogonalization to choose the matching atoms, ensuring that the residual components correspond to the minimum absolute value. Meanwhile, the vectors obtained by iteration are back-disposed according to the corresponding atomic matching rules, so the extra atoms are abandoned in the course of iteration, and the residual components of current iteration are reduced. Thus the original sparse signals are reconstructed. The proposed method reduces iteration times comparing with the traditional reconstruction methods, and the estimation precision is better than the classical two-sided correlation transformation (TCT) algorithm when the snapshot is small or the signal-to-noise ratio (SNR) is low.