UD factorization applied to airborne Kalman-filter-based fusion

To ensure numerical accuracy and stability for real-time Kalman filter implementation, Bierman´s upper diagonal (UD) factorization is used. The use of multiple sensors to form a more accurate state vector has included combining infrared search and track (IRST), electronic support measures (ESM), and radar sensor data, with applications to track initialization/deletion, association correlation, and track-update fusion functions. Each area of fusion is discussed and the interfaces between sensors and fusion are given. Different fusion architectures are shown and their impact on state vector estimation accuracy is shown to vary. All three methods use the extended Kalman filter (EKF) as the base. Correlation, association, and track initialization are examined relative to the different fusion architectures. The correlated process noise which exists for the multisensor application is examined. Root-mean-square position and velocity plots versus time for aircraft are given which incorporate a six-state EKF