A TOPS Improved Algorithm Based on Eigenvalue Discrimination

Abstract

Under low SNR, if the signal subspace estimation of TOPS (Test of orthogonality of projected subspaces, TOPS) algorithm reference frequency point is inaccurate, the error will spread to the whole frequency point, thus causing the problem of algorithm performance degradation. In this paper, a TOPS improved algorithm based on eigenvalue discrimination is proposed. This method selects the optimal reference frequency point by judging the distinguishing degree of characteristic value between signal subspace and noise subspace at each frequency point, and then uses the transfer matrix to project the signal subspace at the optimal reference point to each frequency point, reducing the error of the whole frequency band caused by the inaccurate selection of reference frequency point. This algorithm has higher resolution accuracy under low SNR, and the threshold of 100% resolution probability SNR is about 4dB lower than the original algorithm. Simulation results verify the effectiveness of the proposed algorithm.