Nonlinear Filtering for Autonomous Navigation of Spacecraft in Highly Elliptical Orbit

This study examined the accuracy to which autonomous navigation might be realized in a Molniya orbit. Using appropriate physical force models and simulated pseudorange signals from the Global Positioning System (GPS), a navigation algorithm based on the Extended Kalman Filter was demonstrated to achieve a three-dimensional root-mean-square accuracy of 58.9m over a 500km x 40 000km Molniya orbit. Algorithms based on the Unscented Kalman Filter and the Cubature Kalman Filter were not found to improve this result, due to a high frequency of measurements during periods of highly nonlinear dynamics. During this study, detailed models were developed for GPS pseudorange errors, including ephemeris errors, transmitter clock errors, and ionospheric delay. Receiver clock bias error was shown to be a significant source of navigation solution error; for reasons of geometry, the navigation algorithm is not able to differentiate between this error and a radial position error.