Kalman Filtering USNO's GPS Observations for Improved Time Transfer Predictions

Capt Steven T. Hutsell

Abstract:	The GPS Master Control Station (MCS) performs the UTC time transfer mission by uploading and broadcasting predictions of the GPS-UTC offset in subframe 4 of the GPS navigation message. These predictions are based on only two successive daily data points obtained from USNO. USNO produces these daily smoothed data points by performing a least-squares fit on roughly 38 hours worth of data from roughly 160 successive 13-minute tracks of GPS satellites. Though sufficient for helping to maintain a time transfer error specification of 28 ns (1 Sigma), the MCS's prediction algorithm does not make the best use of the available data from USNO, and produces data that can degrade quickly over extended prediction spans. This paper investigates how, by applying Kalman Filtering to the same available tracking data, the MCS could improve its estimate of GPS-UTC, and in particular, the GPS-UTC A1 term. By refining the A1 (frequency) estimate for GPS-UTC predictions, error in GPS time transfer could drop significantly. Additionally, the risk of future spikes in GPS's time transfer error could similarly be minimized, by employing robust Kalman Filtering for GPS-UTC predictions.
Published in:	Proceedings of the 27th Annual Precise Time and Time Interval Systems and Applications Meeting November 29 - 1, 1995 The Doubletree Hotel at Horton Plaza San Diego, California
Pages:	269 - 278
Cite this article:	Hutsell, Capt Steven T., "Kalman Filtering USNO's GPS Observations for Improved Time Transfer Predictions," Proceedings of the 27th Annual Precise Time and Time Interval Systems and Applications Meeting, San Diego, California, November 1995, pp. 269-278.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In