A robust source enumeration algorithm for low SNR coherent signals based on eigenvectors

A robust source enumeration algorithm is proposed for coherent signals at low signal-to-noise ratio conditions. The approach determines the source number by exploiting the modified covariance matrix and its eigenvectors other than the eigenvalues used in traditional methods. By using a series of transformation of the snapshot vectors and eigenvectors, new decision variables which have better immunity to noise than eigenvalues are obtained. The theory analyses and the simulation results show that the proposed algorithm can achieve improved performance at lower SNR.