Improved resolving capabilities of linear array using 2 q th order non‐circular statistics

Abstract

Resolving more sources than sensors is always of interest to researchers. For this purpose, generation of the virtual array from the non-uniform linear array has recently gathered a lot of attention. The covariance/cumulant lags of the array output define virtual sensors and thereby virtual array. The aperture of the designed virtual array is much more than the aperture of physical array. This large aperture provides highest degrees of freedom to solve the underdetermined system. In the case of non-circular signals, pseudo-covariances/cumulants are significant, and this additional information can further be used to increase the virtual array aperture. Herein, a framework is proposed to extend the virtual array aperture by additionally using pseudo-/non-circular cumulants along with the circular cumulants of the array output for non-circular signals. The suggested framework not only increases the resolvability but also improves the DoA estimation accuracy. With fourth-order statistics, the virtual array aperture becomes almost double, and the increment is much more with further higher order statistics. Numerical simulations demonstrate the efficacy of the claims.