Decentralized information filter in federated form

In this paper, a decentralized information filter in federated structure is developed in multi-sensor environments. Being equivalent to the Kalman filter algebraically, the information filter of Mutambara is extended to N-sensor distributed dynamic systems. In multi-sensor environments, the information-based filter is easier to decentralize, initialize, and fuse than a KF-based filter. The structural features and information sharing principle of the federated information filter is discussed. It is proved that the information state and its information matrix of the information filter, which are weighted in terms of information sharing factor, are equal to those of the centralized information filter under the regular conditions.