Compressive beamforming for direction-of-arrival estimation of cyclostationary propeller noise

Abstract

A compressive beamformer is proposed for direction-of-arrival estimation of cyclostationary ship-radiated propeller noise sources. The sources are recovered by iterative hard thresholding algorithm which is adapted for seeking the sparsity spatially and in cyclic frequency domain. Exploiting this double sparsity, the proposed cyclic compressive beamformer enables high-resolution localization of sources with different cyclic frequencies. Spectral averaging is suggested to reduce the adverse effect of the spatial Dirichlet kernel when there are multiple sources. Numerical examples demonstrate that the proposed scheme outperforms the well-known methods especially for low signal-to-noise ratios, numerous closely-spaced sources and few sensors. Performance of the proposed estimators are verified by explicit analytical derivations for several array and signal parameters.