ADAPTIVE EIGENDECOMPOSITION ALGORITHMS FOR EIGENSTRUCTURE-BASED ARRAY SIGNAL PROCESSING
Lena Chang · Journal of the Chinese Institute of Engineers · 1997
This paper first presents a forward-only algorithm for adaptive eigendecomposition of a continuously updated sample correlation matrix for arbitrary geometry arrays. The proposed forward-only adaptive eigendecomposition (FOAE) algorithm is based on an alternate iteration technique which can alleviate the computation complexity of the standard eigendecomposition. For an MxM matrix, the FOAE algorithm requires O(Mq) complex operations per iteration to adjust the eigencomponents of the signal subspace, where q is the signal subspace dimension. Then, for a symmetrically distributed antenna array, we exploit the Hermitian persymmetric property of the correlation matrix to develop a real-valued forward-backward adaptive eigendecomposition (RFBAE) algorithm. The proposed RFBAE algorithm requires only O(Mq) real operations in each iteration. Moreover, the FOAE and RFBAE algorithms are applied to estimate or track the angles of arrival and design the eigenspace-based beamformer. Simulation results confirm the fast convergence rate of the proposed algorithms and the RFBAE algorithm performs better than the FOAE algorithm.