Direction finding with fewer receivers via time-varying preprocessing

The present a new technique for localizing multiple narrowband sources by a passive sensor array based on dimensionality-reducing time-varying preproc essing of the sensor outputs. The technique allows a significant reduction in the number of receivers required for the implementation with only two re ceivers sufficing in the extreme case. The estimation method we use is base d on approximating the corresponding maximum likelihood estimator by a comp utationally simpler generalized least squares (GLS) estimator that is prove d to be both consistent and asymptotically efficient. Simulation results co nfirming the theoretical results are included.