Asymptotic analysis of manifold ambiguity resolution in DOA estimation with nonuniform antenna arrays

The asymptotic properties of direction-of-arrival (DOA) `manifold ambiguity' resolution for m uncorrelated sources is analytically investigated. Manifold ambiguity arises when a DOA estimation technique generates some number κ of additional spurious directions. The main idea of our approach is to find the best single match between the set of all estimated covariance lags on the one hand, and the set of power estimates for the (m+κ) ambiguous DOA's on the other hand. For a finite sample volume N, selection of the m greatest power estimates generated by a linear program `fitting' algorithm has been previously proposed to identify the m true DOA's among the set of (m+κ) ambiguous estimates. In this paper, we present an analytic study into the asymptotic behaviour of this technique, and support it with simulation results.