DOA Estimation of Arc Arrays Using Deconvolution with Shift-Variant PSF

Arc arrays are widely used in practical sonar system as they have some advantages which linear arrays do not have, such as approximately constant beamwidth. Deconvolved conventional beamforming (DCBF) provides remarkably low sidelobes (SLs) and narrow mainlobes (MLs) beamwidth, and guaranteed robustness in direction of arrival (DOA) estimation. Nevertheless, the Richardson-Lucy (R-L) algorithm of deconvolution used in original DCBF method is only applicable for those arrays with shift-invariant point spread function (PSF), e.g., linear arrays. Therefore, the DCBF method is unable to achieve expected performance when applied to an arc array with shift-variant PSF. This paper proposes an improved deconvolved conventional beamforming (IDBCF) method which utilizes a modified R-L algorithm to deconvolve the conventional beamforming (CBF) beam power. Furthermore, an automatic acceleration method is used to accelerate the deconvolution processing. Simulations apply the CBF, DCBF and IDCBF methods to an arc array. The results suggest that the modified R-L algorithm can effectively deconvolve the beam power of the arc array; therefore, the IDCBF method yields lower SL levels and narrower ML beamwidth than CBF even DCBF methods did in DOA estimation.