Sparsity-based direction-of-arrival and polarization estimation for mirrored linear vector sensor arrays.

• An improved MLVSA model considering the polarization characteristics of incident EM waves is constructed. • The RAL of the manifold curve based on the differential geometry is derived for performance analysis of the MLVSA. • A group-lasso-based algorithm is proposed for joint DOA and polarization. • Proposed algorithm outperforms existing signal-separation-based algorithms. In this paper, we first derive an improved reflection model of arbitrarily polarized electromagnetic (EM) waves for mirrored systems. Then, a mirrored linear vector sensor array (MLVSA) model is constructed. The rate of change of the arc length (RAL) of the manifold curve is subsequently derived based on the differential geometry to analyze the parameter estimation accuracy of MLVSA. Furthermore, the Cramer–Rao bound (CRB) is derived under the framework of the manifold curve. Moreover, the direction-of-arrival (DOA) and polarization parameters are jointly estimated by solving a group-lasso problem. Compared with the signal-separation-based algorithms solved by undetermined linear equations, the proposed algorithm is superior, regarding the parameter estimation accuracy and multitarget resolution. Numerical simulations are conducted to further compare the performance of MLVSA and that of the nonmirrored array. [ABSTRACT FROM AUTHOR]