Planar array for accelerated sources tracking using local polynomial approximation beamformer

The main goal of this paper is to develop a planar array model for beamforming and estimating movement parameters of accelerated targets using a nonparametric approach, based on a windowed 2D local polynomial approximation (LPA) beamformer. Exploiting the acceleration of the source is used to distinguish between many nonstationary sources located in the same direction, moving with the same angular velocity. Using the 2D LPA beamformer decreases the mean square error (MSE) of the angle estimation with a small window size. It can also be used to scan the main beam towards any point in space. A model for the output signal of the receiving sensor array is derived. Asymptotic expressions for the bias and variance of the estimate, using the short-time asymptotic behavior of the DOA, are used for the optimal window size selection. This is valid for a wide class of accelerated time-varying single sources, as well as multiple sources.