An improved angular super-resolution deconvolved beamforming via virtual aperture transformation

Deconvolved conventional beamforming (DCBF) utilizes the deconvolution structure provided inherently in the beam power based on conventional beamforming (CBF) and effectively improves the angular resolution and estimation accuracy performance. However, the existing methods suffer from the wide beam of CBF, which leads to a significant decrease in resolution capability for closely spaced sources. To address these problems, a deconvolved beamforming based on the virtual array is proposed. The virtual array structure contained in the local covariance matrix is firstly employed to realize the aperture extension, which the degree of freedom is about twice that of the physical array. Then, the derivation proves that the mainlobe beam based on the virtual array is narrower than that of CBF. Meanwhile, the linear shift-invariance property for the virtual array structure is further demonstrated, which can be extended to deconvolution processing. Finally, the dimension reduction idea helps save complexity in implementations without losing the effective freedom degree of the virtual array. Theoretical analysis and numerical simulation verify the effectiveness of our proposed method for closely spaced sources due to virtual array processing, even in cases where the low input signal-to-noise (SNR) and small snapshot number, the proposed method can still provide better performance on estimation and resolution. In addition, performance is also evaluated with real data.