Passive target tracking by unscented filters

Tracking of multiple targets by using only the bearing information is an important problem in passive sensing scenarios such as passive sonar or infrared seeker applications. In this paper, a new algorithm is presented to adaptively estimate the DOAs of multiple moving targets when linear equispace array (LES) is used for making the measurements. These algorithms are based on the differential MUSIC method. The results obtained by these algorithms are then applied as input to a nonlinear state linear measurement unscented filter. The unscented filter based tracker is compared with the EKF based tracker. Simulation results show that the former is superior to the latter.