Improved DOA Estimation Algorithms Using Modified Covariance Matrix

Despite the high resolution that the min-norm algorithm has proven in estimating the directions of arrival of non-coherent signals, this subspace based method is fail to work when signals are highly correlated or coherent such as in the multipath scenario. In this paper, a new improved min-norm method is developed to estimate both coherent and non-coherent signals basing on covariance matrix reconstruction. The reconstructed matrix can highly de-correlate the spatial coherency between signals due to its ability in eliminating rank loss. The rooting version of the min-norm approach basing on the new reconstructed covariance matrix is also developed and tested. The effectiveness of the proposed modification is verified through computer simulations, and it is shown that the both improved methods perform well in the situation of low SNR and small angular separation.