Navigation sensor, filter, and failure mode simulation results using the distributed Kalman filter simulator (DKFSIM)

Abstract

Results obtained using the distributed Kalman filter simulator (DKFSIM) are given. It is a FORTRAN software package, and was run on a PC. The sensors modeled were a medium accuracy strapdown inertial navigation system (INS), a barometric pressure altimeter (BARO), the GPS, a SAR, and a terrain-aided navigation (TAN) system. The SAR system includes an electro-optical type imaging model and a precision velocity update model. The TAN system combines radar altimeter measurements with digital terrain elevation data. The mission profile for each simulation included a low-level terrain-following segment and a high dynamic combat maneuver segment typical for tactical fighter aircraft. The filter implementations used during this simulation sequence included (1) a single centralized Kalman filter incorporating all of the sensor measurements, (2) a federated Kalman filter with each sensor assigned to an independent local filter and the INS included as the system reference sensor, and (3) a cascaded Kalman filter where the GPS Kalman filter fed estimation information directly to a centralized Kalman filter. The failure modes modeled were a GPS satellite clock failure and an INS accelerometer failure. Observation of filter performance under varied conditions revealed advantages and other characteristics for each filter architecture modeled.