DOA estimation in multipath: an approach using fourth-order cumulants

We consider the problem of direction of arrival (DOA) estimation in the presence of multipath propagation. The sensor elements are assumed to be linear and uniformly spaced. To perform DOA estimation, we combine two existing algorithms which are often used for other purposes. The first algorithm exploits fourth-order cumulants to perform blind estimation of the steering vectors and the second algorithm estimates the directions of arrival using the estimated vectors. We refer to this two step approach as the SV-DOA estimation algorithm. We also present an algorithm independent performance analysis for the DOA estimation problem based on fourth-order cumulants. We summarize the algorithms and present Monte-Carlo simulations demonstrating the effectiveness of the SV-DOA algorithm as well as verifying the performance analysis using an optimal (but computationally expensive) DOA estimation algorithm.