Adaptive Kalman Filter Based Single Beacon Underwater Tracking With Unknown Effective Sound Velocity

The basic principle of single beacon underwater tracking is to estimate the unknown vehicle position based on acoustic signal measurements and onboard sensors. The extended Kalman filter (EKF) is the most widely adopted single beacon underwater tracking method, which assumes known noise probability distributions. The setting of the measurement noise parameters has significant impact on the estimator performance. In practice, these are usually determined based on experience or data post-processing. However, due to the uncertainty of underwater acoustic propagation, the probabilistic characteristics of acoustic measurements noise are not only unknown but also varying both with time and location. Therefore, EKF which runs with presupposed measurement noise parameters cannot describe the practical noise specifications, which in consequence will influence the estimation accuracy, or even lead to practical divergence. To overcome the divergence issue of EKF caused by violation of known measurement noise assumption, this paper implements the adaptive Kalman filter (AKF) for single beacon underwater tracking. Numerical results confirmed that the proposed algorithm can estimate the unknown measurement noise parameters, and could have significantly improved tracking accuracy over EKF.