Performance bounds for subspace estimation in array signal processing

Estimation of unknown parameters using arrays of passive sensors is a well-known problem in signal processing. This problem is studied via subspace estimation using geometric representations on Grassman manifolds. The variability on Grassman manifolds is modeled by a transitive action of a special unitary group and by using a Bayesian formulation on the space of unitary matrices. An a posteriori is used to derive an MMSE estimator and a lower-bound on the expected squared error. Empirical analysis using stochastic gradient processes is considered for numerical computation of the estimator and the lower bound