Performance Analysis of Gaussian Optimal Filtering for Underwater Passive Target Tracking

The problem of passive target tracking in the underwater environment is usually handled with nonlinear filtering algorithms, in which nonlinear measurement model is combined with linear system dynamics. The primary goal in passive target tracking is to extract accurate information about real-time state of the target from noisy nonlinear observations obtained from sensors. In this study, performance analysis of Gaussian optimal filtering is proposed for accurate state prediction of an underwater dynamic object. This paper delicately analyzes the state estimation performances of nonlinear version of Kalman filter like Gauss Hermite Kalman Filter (GHKF) and discrete-time Kalman smoother, called Gauss Hermite Rauch-Tung-Striebel (GHRTS) smoother. This analysis is done through variation in standard deviation of white Gaussian measurement noise which is a key feature in the target tracking framework. This performance-based study is conducted in the context of Bearings Only Tracking (BOT) phenomena by using two and three acoustics sensors installed on observer base station. All the experiments are performed for finding the Root Mean Square Error (RMSE) among true and predicted state of the object. Independent Monte Carlo simulations based numerical results demonstrate that GHRTS smoother provides better performance from GHKF for given circumstances.