Enhancing New ASED Basis Function Method With Dual‐Level Compressive Sensing Scheme for Fast Analysis of Finite Periodic Arrays

ABSTRACT

This letter presents a novel dual‐level compressive sensing (CS) scheme to accelerate the new accurate subentire‐domain (ASED) basis function method for electromagnetic scattering problems of finite periodic arrays. On level one, a multiple measurement vectors (MMV)‐CS model is employed to reconstruct the whole basis functions in a subarray with very few measurements, which greatly reduces both the computation time and the number of basis functions after singular value decomposition. On level two, the overdetermined sparse representation based on the MMV‐CS or single measurement vector (SMV)‐CS model is constructed to solve the induced currents for monostatic or bistatic scattering problems, respectively. With either the MMV‐CS or SMV‐CS model, only very few rows of the impedance matrix need to be calculated, and the time‐consuming construction of the reduced matrix equation is replaced by the CS equation, which can be more efficiently solved. Numerical simulations indicate that the proposed framework attains a substantial acceleration over the new ASED basis function method while preserving comparable accuracy for both discrete and electrically connected arrays.