GPS-Based Time Error Estimates Provided by Smoothing, Wiener, and Kalman Filters: A Comparative Study

Y.S. Shmaliy, A.V. Marienko, M. Torres-Cisneros, O. Ibarra-Manzano

Abstract:	GPS timing plays a critical role in modern practice of time errors estimate and synchronization. Big noise of the GPS-based measured data and inherent non-stationarity of a time error cause major difficulties here. In spite of theoretical separation of the application fields for the filters (stationary and non-stationary signals), GPS-based time error processes require more explicit practical answer. Indeed, what process may be practically treated as a stationary one and, to opposite, how to recognize a non-stationary case? In this report we answer these questions by numerically and show that for the same transient time the following filter should be used to get the best accuracy for the known initial fractional frequency offset yo (time error rate) of oscillator, namely an average smoother for \yo I < r, , the Wiener filter for r, 5 (yo 1 5 r, , and the Kalman filter for r2 < 1 Yo 1, where r1 and r2 are coordinates dependent on the required accuracy. We prove this conclusion by the example of a time error estimate of the rubidium standard based on the reference tinting signals of the Motorola GPS UT+ Oncore Timing receiver.
Published in:	Proceedings of the 32th Annual Precise Time and Time Interval Systems and Applications Meeting November 28 - 30, 2000 Hyatt Regency Reston Town Center Reston, Virginia
Pages:	157 - 170
Cite this article:	Shmaliy, Y.S., Marienko, A.V., Torres-Cisneros, M., Ibarra-Manzano, O., "GPS-Based Time Error Estimates Provided by Smoothing, Wiener, and Kalman Filters: A Comparative Study," Proceedings of the 32th Annual Precise Time and Time Interval Systems and Applications Meeting, Reston, Virginia, November 2000, pp. 157-170.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In