GPS Precise Absolute Positioning via Kalman Filtering

Brian W. Tolman

Abstract:	Applied Research Laboratories, The University of Texas at Austin (ARL:UT), under sponsorship of the National Geospatial-Intelligence Agency (NGA), has developed precise positioning software that uses dual frequency GPS carrier phase and pseudorange data to obtain 2-3 centimeter level positioning accuracy for a single static receiver. The paper describes the design of the algorithm, its implementation in platform-independent software, and the results of a large test effort, including positioning accuracy and convergence behavior. Dual frequency pseudorange and carrier phase data are combined in preprocessing using a new smoothing algorithm that cleanly separates the pseudorange noise, carrier phase biases and first order ionospheric delay from the geometric range plus non-dispersive errors. This algorithm produces a best estimate of the pseudorange noise and removes it without the use of any filtering. It also produces optimal bias estimates for the carrier phases without otherwise altering them, specifically without changing their noise characteristics. The positioning estimator is a full implementation of the extended Kalman filter algorithm. The filter processes one ionosphere-free phase measurement per satellite; the filter state includes receiver position, clock bias, a residual zenith wet tropospheric delay, and a phase bias for each satellite. Adaptation of the filter to realtime operation and single frequency data processing will be straightforward. Several standard corrections must be applied to the data before filtering to ensure the accuracy of the result. These include phase windup, antenna phase center offsets, three relativity corrections, and the site displacement effects of solid Earth tides, ocean loading and atmospheric loading. The software implementation is built on The GPS Toolkit (GPSTk), an extensive open source platform-independent C++ library of GNSS-related algorithms. The positioning program is a single stand-alone executable that runs on Linux, Windows, Solaris, Mac and UNIX platform. The results of extensive testing of the ARL:UT software reiterates the well known result that the accuracy of the position solution is dominated by the accuracy of the satellite ephemeris. Filter convergence is very fast, typically less than 2-3 hours. Averaging over a sufficient timespan (about 1 day) can reduce the residual error to below 2 cm in each position horizontal component. These results have been verified with over 360 days of data at 12 different sites worldwide.
Published in:	Proceedings of the 21st International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2008) September 16 - 19, 2008 Savannah International Convention Center Savannah, GA
Pages:	1864 - 1874
Cite this article:	Tolman, Brian W., "GPS Precise Absolute Positioning via Kalman Filtering," Proceedings of the 21st International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2008), Savannah, GA, September 2008, pp. 1864-1874.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In