Abstract:

Detecting the source number is an important step to eigensubspace spatial spectrum estimation algorithms.Information theory methods are not applicable to cope with the situation where both uncorrelated and coherent sources exist.A novel method based on matrix reconstruction is proposed to solve this problem. A Toeplitz equivalent covariance matrix is reconstructed by using crosscorrelation information of the receiving data from arrays.To some extent,the constructed matrix can reduce the spreading of the noise eigenvalues.Compared with the conventional method using the first vector of the covariance matrix to construct a decoherence Toeplitz matrix,the proposed algorithm has better decoherence performance and avoids the loss of array physical aperture.Ultimately,the proposed algorithm utilizes weighted eigen subspace projection method to estimate the source number.Computer simulation confirms that the algorithm can accurately estimate the number of sources mixed with uncorrelated and coherent sources and has better performance compared with spatial smoothing Akaike information criterion and spatial smoothing minimum description length algorithms.