Optimizing Eigenvector-Based Frequency Estimation in the Presence of Identical Frequencies in Multiple Dimensions

Recently an eigenvector-based algorithm has been developed for multidimensional frequency estimation. Unlike most existing algebraic approaches that estimate frequencies from eigenvalues, the eigenvector-based algorithm can achieve automatic frequency pairing without joint diagonalization of multiple matrices, but it is not applicable if there exist identical frequencies in certain dimensions. In this paper, we propose to use weighting factors to extend the eigenvector-based algorithm to handle identical frequencies in one or more dimensions. The weighting factors are optimized by minimizing the error variance. Simulation results demonstrate the effectiveness of the proposed approach