A Study of a Kalman Filter for Improving the Accuracy of Stationary GPS Positioning

Abstract

A Kalman filter has been applied to a stand-alone Global Positioning System to improve its positioning accuracy. A conventional Kalman filter optimized for stationary states can not follow dynamic motion, and one optimized for dynamic motion has a high error in the stationary state. A method is proposed to improve the accuracy in the stationary state. The state-transition matrix and the variance matrix of the measurement and process errors are changed according to the state. The speed of motion obtained from the Kalman filter is used to judge the receiver state. Simulation showed that the proposed method yields a positional error of 0.58 m in standard deviation as compared with 1.18 m with a conventional method. Using actual data, the proposed method was shown to reduce the variation in position 50-75% compared to a conventional method.