Multi-sensor data fusion via federated adaptive filter

The aim of this paper is to present a federated adaptive filter for use of multi-sensor data fusion systems for target tracking. The computational architecture of the filter consists of several local processors and a global processor. Each local processor incorporates a scheme of Bayesian decision theory into the multiple-model filter to develop a switching capability to react against the same target dynamics, inclusion of both maneuver and non-maneuver phases. The global processor combines the local estimates via a weighted least squares estimator for generating a global estimate. Simulation results show that proposed filter has better tracking performance than the information matrix filter.