Source localization performance and the array beampattern

Passive localization of sources in space is done using data collected by an array of sensors. The performance of all source location estimation (SLE) algorithms depends on the source-array geometry. In this paper we study this dependence and we show that array localization ability is uniquely determined by a single scalar function - the array beampattern. By analyzing absolute lower bounds on the mean-square-error (mse) of the source location estimates in different scenarios we exploit the relation between features of the array beampattern and the best achievable array localization performance. The beampattern is characterized by its sidelobes level and smoothness, its beamwidth and its mainlobe curvature. We study the effect of these features on source localization properties such as sensitivity to the presence of other sources, resolution ability, sensitivity to ambiguous estimates, etc.