Design and implementation of a novel real time target tracking scheme for passive SONARs using MVDR beam forming and Kalman filtering

Target detection and tracking are the essential functions of modern SONAR systems. Once the direction of arrival is estimated the target is tracked continuously in bearing and the target spectral information is used for classifying the target. A continuous and reliable tracking method is essential for getting the spectral information and classification of the target. Conventional tracking methods use the split beam correlation technique by forming two half beams and correlating between them to detect the path delay between the two beams and then estimate the accurate bearing. This scheme does not offer good performance at low signal to noise ratios and fails when a scenario has two targets crossing each other. In this paper we propose a scheme which addresses these problems by employing (Minimum Variance Distortionless Response) MVDR based tracking and a Kalman smoother to refine the track. The kalman filter gets input from the MVDR based tracker and may also be used to predict a target trajectory when there is a cluttered target scenario when the MVDR based tracker does not provide reliable inputs.