Frequency Estimation by Principal Component AR Spectral Estimation Method without Eigen Decomposition

摘要

For accurate frequency estimation, Principal Component Autoregressive (PC-AR) spectral estimation methods have received considerable attention in the recent literature. Explicit computation of the Eigen-decomposition of the autocorrelation matrix is required to obtain the PC-AR solution. An alternative approach called the Eigenvalue filtering method (EFM) where the eigenspace need not be computed, is proposed in this paper. The proposed method utilizes the geometry of the distribution of the eigenvalues in a matrix function so that it closely approximates the pseudoinverse of the autocorrelation matrix. It is shown via computer simulation that compared with the Forward/Backward method, the proposed method enhances the threshold in SNR by about 6-8 dB. Further improvement is obtained by a simple subset selection method and a second eigenvalue filtering iteration.