High resolution direction-of-arrival estimation based on compressive sensing with noval compression matrix

In this paper, the narrowband DOA estimation problem is studied in compressive sensing (CS) perspective. A novel compression perception model is proposed making use of the spatial sparsity. Two novel approaches for constructing the compression matrix are also presented. The one is to design a new random compression matrix; the other is to apply approximate QR decomposition to form a main diagonal compression matrix. Moreover, singular value decomposition (SVD) is explored on the data matrix in order to lighten computational burden. We also propose two different kinds of methods for direction-of-arrival (DOA) estimation based on new compression matrices: I.CS recovery: QR-SVD-MFOCUSS; II.CS beamforming: Random-SVD CS beamforming (RSVD-CSB) and QR-SVD CS beamforming (QRSVD-CSB).Simulation results demonstrate that the proposed methods possess high resolution, robust to additive noise, reduction computational burden and so on.