Multi-Sensor Data Fusion via Federated Dual-Band Filter

A federated dual-band filter is developed for use in multi-sensor systems for target tracking. Filter architecture consists of local processors and global processor to describe the distributed fusion problem due to correlation across track estimates for the same target. Each local processor incorporates modified probabilistic neural network with multiple model filter (MMF) to develop switching capability as adaptive manner to respond the target dynamics. The global processor combines local processor outputs via weighted least square estimator which can be implemented in a parallel structure to facilitate estimation fusion calculation. To determine dual-band process noise levels with MMF, the high level is selected by use of a conservative matrix upper bound to handle the track-to-track correlations and another is chosen as proper low level for tracking when target is not maneuvering. The resulting filter has better tracking performance improvement than traditional information matrix filter. Simulation results are included to demonstrate the effectiveness of proposed algorithm.