Neural Network Loran-C Calibration Using GPS

Alexey Cherepakhin, Yaokun Zhong, and Daniel Greenwood

Abstract:	Current approaches to reliable and accurate navigation are systems that include one or more inertial navigation systems, and also one or more navigation reference sensors such as a global positioning system (GPS). a long range navigation system (Loran-C), a Doppler radar, a depth sounder, and others. Usually nnvigntion sensor outputs are input to a nonlinear estimation algorithm that jointly optimizes the navigation processing and any associated signal- tracking functions. One of the main requirements of such sensor integration subsystems is enhanced navigation reliability. Systems which combine pseudo ranges from Loran-C and GPS would be more reliable than an unaided GPS receiver, or an unaided Loran-C receiver, and they can compensate for outages caused by satellite failures and/or bad satellite geometry. Similruly, such system can remove most of the coverage outages caused by Loran station shutdowns, high atmospheric noise levels or precipitation static. A hybrid GPS/Loran system could also reliably provide self contained fault detection and isolation ,otherwise, known as receiver autonomous integrity monitoring (RAIM). Loran-C is currently being expanded with new transmitters in the middle of the North American continent and plans exist for expanding in Northern Europe and Venezuela. The main problem in using Loran-C position sensors (receivers) results from the distortions caused by time dependent and time- independent factors. This problem can be mitigated by tables provided by Defense Mapping Agency (DMA), but even with the corrected measurements there still are errors in absolute position accuracy ranging from 200-500 ft. This report demonstites the feasibility of ;L neural network enhanced system which is able to automatically correct the Loran-C signal distortions caused by traveling over areas with different land conductivity by GPS data. Our system will also allow accurate position estimation when GPS accuracy is insufficient, or GPS positioning is. not continuously available. In addition Lhis system can enable permanent monitoring of the integrity of GPS positioning. In our approach to integrating Loran-C and GPS we constructed a neural network system capable of learning automatically the functional dependency of Loran-C distortions on geographical position and time dependent factors. We chose the neural network architecture to best correspond to an established mathematial mode1 of Loran-C signal propagation. Numerical experiments were performed based on real simultaneous GPS and Loran-C measurements taken in Dixon Entmnce, Canada, that were provided to us by the Canadian Hydrogmphic Service and the University of Calgary. These experiments show that after training on specially preprocessed daa our neural network output gives a smooth and precise calibration of the Loran-C signal.
Published in:	Proceedings of the 7th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1994) September 20 - 23, 1994 Salt Palace Convention Center Salt Lake City, UT
Pages:	407 - 416
Cite this article:	Cherepakhin, Alexey, Zhong, Yaokun, Greenwood, Daniel, "Neural Network Loran-C Calibration Using GPS," Proceedings of the 7th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1994), Salt Lake City, UT, September 1994, pp. 407-416.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In