A temporal smoothing approach to direction of arrival estimation of coherent signals in fading channels

In this paper we describe a temporal smoothing approach for estimating the direction of arrival of coherent signals in a fading channel. Over a period of several frames, the multipath signal phases in a fading channel give rise to randomly varying phases. By temporally averaging over these phase terms in a noise subspace estimator such as MUSIC, the received signal array covariance matrix will become full rank, even in the presence of coherent signals. Theoretical results demonstrate that, given K directions of arrival, temporal smoothing requires K+1 sensor elements compared with spatial smoothing which requires 2K elements, and forward/backward spatial smoothing which requires 3K/2 sensor elements.