Research on Invariant Extended Kalman Filter Based 5G/SINS Integrated Navigation Simulation

Abstract

The construction and improvement of 5G can empower navigation and positioning. 5G/SINS integrated navigation can broaden application scenarios of the traditional GNSS/SINS integrated navigation. Extended Kalman Filter (EKF) has poor convergence when the initial error is large, resulting in poor positioning accuracy. Compared with EKF, the Invariant EKF (InEKF) constructs the system state on the matrix Lie group, and its dynamic equation can describe the motion characteristics of objects more naturally and essentially. At the same time, InEKF obtains state-independent Jacobians at any linearization point. Therefore, we propose a 5G/SINS integrated navigation system based on InEKF. Furthermore, the accuracy and convergence are compared with result of EKF. The simulation experiment is carried out from two dimensions of different accuracy levels of SINS and different initial errors. Experimental results show that the performance of InEKF is significantly better than that of EKF when the SINS accuracy level is high; When the noise is large, although the model no longer satisfies the group affine property, the performance of InEKF is still better than EKF while the advantage is not as obvious as that with small noise. In addition, the larger the initial error is, the better performance of InEKF has than EKF.