A Decentralized Parallel Kalman Filter for Multi-sensor System with State Constrains

Based on the decentralized computing platform, a decentralized Kalman filter with state constraints is presented in this paper. The decentralized sensing architecture takes the form of a network of wtransputer-based sensor nodes, each with its own processing facility. So it does not require any central processor, central communication facility or common clock. Based on that, the starting point is an algorithm that allows fully decentralization of the multisensory Kalman filter equations with state equality constraints among a number of sensing nodes. The algorithm is developed from a centralized method named projection method to minimized the communication among nodes and can take place without any prior synchronization between nodes. By designing a scheme, SFDD task can be achieved. Theory derivation is provided to the decentralized algorithm. Finally, simulation results of the navigation system illustrate the effectiveness of the proposed method.