Array signal processing for recursive tracking of multiple moving sources based on LPA beamforming

The windowed linear local polynomial approximation (LPA) of the time-varying direction-of-arrival (DOA) is developed for nonparametric high-resolution estimation of multiple moving sources. The method gives the estimates of instantaneous values of the directions as well as their first derivatives. The asymptotic variance and bias of these estimates are derived and used for the optimal window size selection. Marginal beamformers are proposed for estimation and sources visualization. These marginal beamformers are able to localize and track every source individually nulling signals from all other moving sources. Recursive implementation of estimation algorithms are developed for two different tasks: estimation of DOAs with varying number of sources and multiple source tracking in time.