Abstract: For the problem of two dimensional underdetermined direction of arrival (2-D DOA) estimation, three parallel sparse arrays are designed and a novel DOA method based on continuous sparse recovery is proposed. The dimension of received data is reduced and the cross covariance matrix is vectorized to improve freedom degree. According to the objection function of the minimization issue based on total variation norm constraint, the space frequency is achieved by solving corresponding dual problem. The center frequency, azimuth and pitch angles are paired by capon method. Theoretical analysis shows the proposed method has a higher array aperture extension and can neglect basis mismatch effects. Extensive simulation results demonstrate the effectiveness of the proposed method.

Key words: array signal processing, broadband DOA estimation, continuous sparse recovery, total variation norm