Tracking of acoustic source in shallow ocean using field measurements

We investigate the problem of tracking a moving source in shallow ocean in a Bayesian framework, using acoustic field measurements which are more informative than the commonly employed bearings-only or time-delay measurements. The acoustic field measurement model is described and compared with the bearings-only measurement model. A general approach to Bayesian filtering based on the Gaussian approximation model is then presented. Within this framework, we consider the unscented Kalman filter (UKF), fifth-degree cubature Kalman filter (CKF5), and quasi-Monte Carlo Kalman filter (QMC-KF) algorithms that employ different numerical integration procedures. Simulation results indicate that acoustic field measurements yield a significantly lower root mean square error (RMSE) than bearings-only measurements, and that the RMSEs for QMC-KF and CKF5 are much lower than those for UKF and extended Kalman filter (EKF).