Improvement of Modular Terrain Navigation Systems by Measurement Decorrelation

J. Metzger, G.F. Trommer

Abstract:	Modular navigation systems are investigated which are based on the combination of a central navigation Kalman filter with a modular terrain referenced navigation (TRN) system. The need of such divided navigation systems lies in the desired reuse of already existing central navigation filters, when new features like TRN systems have to be implemented. The TRN system described in this paper incorporates a non-linear stochastic particle filter, which is able to calculate horizontal position estimates based on the measurement of the terrain height currently flown over and a digital reference map covering this area. Due to the fact that this non-linear filter, with its iterative algorithm, uses not only current but also previous measurements to determine a position fix, the output of the TRN system is corrupted with a time-correlated noise. This output is used in an adjacent central navigation Kalman filter and therefore a decorrelation process is necessary to hold the assumptions of the Kalman filter. A Kalman filter requires input data with uncorrelated noise to work in an optimal state. Thus the main focus of this paper lies in a decorrelation procedure and a system concept of the combined filter structure. The decorrelation is divided into three steps. The output signal of the TRN system contains the wanted information about the current horizontal position and a noise process with unknown parameters. To be able to estimate these parameters it is necessary to separate the noise process from the information content. In the first step weighted differences of the output signal achieve such a separation. In the second step the parameters of the time correlated noise process, which is modeled as a vector auto regressive process of first order, are estimated adaptively. This is necessary because of the fact that the noise parameters are changing with time due to changing flight situations and changing terrain roughness. In the third step the correlation effect has to be removed from the TRN output by building measurement differences. By adapting the measurement matrix of the adjacent central navigation Kalman filter it is possible to process the decorrelated output signal provided by the TRN system. Simulation results of different TRN systems, like point mass filters, bootstrap filters and kernel filters demonstrate the superiority of this setup compared to a simple cascaded TRN system or static decorrelation procedures using shaping filters. Not only the quality of these filters is improved but also the robustness and reliability is increased drastically.
Published in:	Proceedings of the 59th Annual Meeting of The Institute of Navigation and CIGTF 22nd Guidance Test Symposium (2003) June 23 - 25, 2003 Hyatt Regency Hotel Albuquerque, NM
Pages:	353 - 362
Cite this article:	Metzger, J., Trommer, G.F., "Improvement of Modular Terrain Navigation Systems by Measurement Decorrelation," Proceedings of the 59th Annual Meeting of The Institute of Navigation and CIGTF 22nd Guidance Test Symposium (2003), Albuquerque, NM, June 2003, pp. 353-362.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In