An improved signal subspace decomposition method in low SNR case

The SNR (Signal-to-Noise Ratio) of the targets in HF radar is usually not high enough and the array data length is short. Therefore, the Toeplitz construction method is applied to achieve better azimuth resolution and higher azimuth estimation accuracy than the classic covariance matrix construction method of the spatial spectrum estimation technique. However, its signal subspace and noise subspace could not be separated clearly by the conventional subspace decomposition method. In this paper, an improved signal subspace decomposition method based on the cross-correlation coefficients analysis of the eigenvectors is presented. This method uses an adaptive three sigma rule to select the eigenvectors corresponding to the smaller eigenvalues as the signal subspace. Thus, the signal subspace is divided more clearly. Then, combined with the Multiple Signal Classification (MUSIC) algorithm, the estimate performance for the coherent targets is improved.