"Almost blind" steering vector estimation using second-order moments

The problem of separation and reconstruction of superimposed signals using an array of sensors attracted considerable interest. The estimation of the steering vectors of an uncalibrated array is considered. We identify a cost function whose minimizer is a statistically consistent estimate of the unknown parameters. Next, we present an iterative algorithm for finding a local minimum of that cost function. The proposed algorithm is guaranteed to converge, the performance of the algorithm is compared with the Cramer-Rao bound (CRB).