EFFECT OF SPATIAL SMOOTHING ON THE PERFORMANCE OF SUBSPACE METHODS IN THE PRESENCE OF ARRAY MODEL ERRORS

In this paper, the effect of spatial smoothing (Forward smoothing and Forward-Backward smoothing) on the performance of unweighted and weighted subspace methods in the presence of Array Model errors for Direction-Of-Arrival (DOA) estimation is studied. Theoretical expressions for the Mean Squared Error (MSE) in DOA are obtained, based on a common framework of analysis. For weighted subspace methods, optimal weighting matrices are presented which minimize the MSE in DOA. Two typical errors, Random errors and Gain and Phase errors are considered as examples for illustration. Simulations are carried out to substantiate the theory developed. For the cases considered, smoothing improves the performance of ESPRIT and Minimum-Norm method while it is not so for MUSIC. Forward-Backward smoothing improves the performance of all methods compared to that of Forward smoothing.