Abstract:

The performance of two-dimension subspace estimation algorithms degrades substantially because of the presence of mutual coupling by the perturbance of steering matrix and rotational invariance property. The classical two-dimension direction of arrival (2D-DOA) estimation and decoupling algorithms need multidimensional search, optimization and iteration, so it need large calucation quantity. Based on the special structure of the uniform rectangle array and the character istics of mutual coupling matrix, a decoupling two-dimension estimating signal parameter via rotational invariance techniques (2D-ESPRIT) algorithm in the presence of mutual coupling is proposed. The algorithm chooses three sub-arrays which maintain rotational invariance property in the presence of mutual coupling. An extended covariance matrixis is constructed. Decoupling and estimation of 2D-DOA can be accomplished via the eigen-decomposition of the matrix. The feasibility of rotational invariance techniques for application in 2D-DOA estimation in the presence of mutual coupling is proved. The algorithm requires neither 2D-searching spectrum peak nor the value of mutual coupling matrix, and the calculation quantity is reduced. The simulation results show that the algorithm can accurately estimate 2D-DOA and restrain mutual coupling. The estimation performance of the algorithm is equivalent to the standard ESPRIT algorithm without mutual coupling.

Key words: array signal processing, direction of arrival, mutual coupling, array calibration, parameter estimation