Higher-Order GPS Noise Models, with Applications to Position Inference and Vehicle Navigation

Abstract

This thesis presents an experimental and theoretical investigation of GPS errors. Data from stationary GPS units were gathered from several locations and an analysis of the noise present was undertaken. Several noise models were proposed and their relative performance compared using the Akaike Information Criterion. A vehicle navigation system was implemented using a dynamical vehicle model which combined GPS, accelerometer and gyroscope measurements via a nonlinear Kalman filter to enable superior position and velocity estimation, especially during shortterm losses of GPS signal.