An Improved Propagator Method for DOA Estimation of Coherent Sources Without the Source Number

Subspace-based methods are very effective to estimate the direction of arrival (DOA) of sources with a high accuracy. In this paper, a method is proposed to estimate DOA of coherent signals without a priori knowledge of the source number, which is also suitable for independent or partly correlated sources, making the family of subspace-based methods more practical. The low-complexity autocovariance matrix is considered, and the spatial smoothing technique will be adopted to restore the column rank. In the case that the source number is unknown, the noise effect is eliminated with a specific noise estimation for coherent signals. A method combining the source number estimation and source localization is firstly presented. The corresponding analysis is given, and the weakness is shown. Then, another method is introduced to overcome the weakness, in which an improved propagator can be constructed to estimate the DOAs without the source number. The effectiveness is evaluated through simulations.