The Study Of The Data Fusion Technology In Multi-sensor Network

Comprehensive utilization of data form multiple sensors for tracking move target is a long-discussed problem of multi-sensor data fusion. State estimation and fusion is the main problem of target tracking in wireless sensor network. Distribution is the essential characteristics of wireless sensor network, so with the development of wireless sensor network, distributed state estimation problems are brought to the attention of researchers.Kalman filter glaorithm is a classic method for state estimation in linear systems, the average consensus strategy is an effective method of network-wide distributed computing tasks. In recent years, researchers combined Kalman filter and consensus protocol and proposed Kalman Consensus Filter. KCF improved the consensus of different nodes and estimation performance, and it has aroused widespread concern for its simple distributed architecture, scalability and robustness of algorithm.Extended Kalman filter and Unsented Kalman filter is widely used in the state estimation in nonlinear systems. Based the second KCF algorithm in the third chapters, this paper propose Extended Kalman Consensus Filter and Unsented kalman Consensus Filter, by means of combining Extended Kalman filter and Unsented Kalman filter with the consensus protocol. The two nonlinear filtering algorithm based on consensus protocol have good estimation performances which are verified by simulation.