Integrity and the Myth of Optimal Filtering

James Chaffee, Karl Kovach, Greg Robel

Abstract:	This paper carefully reviews the theory of Kalman filtering, highlighting the requisite hypotheses which are frequently ignored. It then shows that the so-called Extended Kalman filter (EKF) is not a Kalman filter at all, and is indeed suboptimal and unstable. Finally, the important fact that the residuals of the EKF are not white and not Gaussian is shown, and discussed in conjunction with their misuse in designing &i-square tests for GPS receiver autonomous integrity monitoring @AIM). The necessity of using more robust statistical procedures is addressed.
Published in:	Proceedings of the 1997 National Technical Meeting of The Institute of Navigation January 14 - 16, 1997 Loews Santa Monica Hotel Santa Monica, CA
Pages:	453 - 461
Cite this article:	Chaffee, James, Kovach, Karl, Robel, Greg, "Integrity and the Myth of Optimal Filtering," Proceedings of the 1997 National Technical Meeting of The Institute of Navigation, Santa Monica, CA, January 1997, pp. 453-461.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In