An investigation of decentralized Kalman filtering as applied to integrated navigation

An investigation into decentralized Kalman filtering as applied to integrated navigation is presented. Two decentralized Kalman filtering schemes are given, and the federated decentralized Kalman filter is chosen for analysis. Computer burden calculations are used to show that the hypothetical implementation of the filter for a INS/GPS/DVS integrated navigation system using an Intel 80960MC microprocessor can be operated in real-time. Results from computer simulations reveal suboptimality of the decentralized Kalman filter to be comparable to that of the standard Kalman filter. Measures of the degree of fault tolerance are obtained using the simulated example. The results reveal an increase in potential fault tolerance of the decentralized Kalman filter with respect to the standard Kalman filter, but a decrease in accuracy after a GPS position ramp failure is reconfigured. Finally, two fault detection schemes are presented. Results from their simulation reveal no enhancement in fault detection through decentralization. It is concluded that there are no advantages to using a decentralized Kalman filter over the standard Kalman filter for fault detection and fault tolerance enhancement. (Abstract shortened by UMI.)...