A Novel Bayesian Kalman Filter Under Non-Gaussian Noise for Underwater Integrated Navigation

Affected by the complex underwater environment, underwater integrated navigation usually has to face the problem of measurement loss and non-Gaussian noise. To solve these problems, two kinds of novel statistical similarity measure-based Bayesian Kalman filter are proposed. In order to handle with the measurement loss, the proposed SSMBKF-I uses maximum a posterior estimation (MAP) to determine whether a measurement loss has occurred or not. The proposed SSMBKF-II introduces the probability of measurement loss into minimum variance estimation (MVE). Meanwhile, the statistical similarity measure method is employed for handling with the issue of non-Gaussian noise. Through utilizing fixed-point iteration approach to optimize the cost function to its maximum value, the suboptimal state estimation can be obtained. The simulation examples are utilized for evaluating the performance of the SSMBKF-I, SSMBKF-II relative to the Kalman filter, Bayesian Kalman filter and statistical similarity measure based Kalman filter.