An Improved Algorithm of Orthogonal Vector Spectral Estimation Method

Orthogonal Vector Spectral Estimation method - OVSS method, with high statistical stability, high resolution, is much better than other existing spectral estimation methods (such as the MUSIC method). But while the method is used for dealing with superposition of multiple sinusoidal signals, the pseudo-peaks will often appear. In this paper, the principle of the OVSS is investigated in detailed. A new matrix is formed by using the elements of the primary eigenvectors and the right singular vector corresponding to the zero singular value of the matrix is orthogonal with signal space. Actually, the signal information can be extracted by extending the processing of autocorrelation matrix. However, when there are several signals, pseudo-peaks will appear inevitably because of spurious zeros. From this view, a new method for eliminating the pseudo-peaks should be found. Based on the matrix above, a modified matrix is formed through reusing the specific elements on specific positions, and then following the steps of OVSS method, a modified OVSS method can be gained. A large number of simulation experiments show that, the pseudo-peaks can be eliminated by using the modified OVSS method, with a little loss of resolution in low SNR cases, and without significant loss of statistical stability in high SNR cases.