Global Anomaly Detection-based Distributed Estimation for Constrained Systems over Sensor Networks

The problem of distributed estimation for constrained systems over sensor networks is addressed, in which global anomaly detection strategy is developed in distributed fusion process. The prior constraint model and innovation vector are introduced to construct a feature vector in abnormal detection. Then a distributed power method is used to perform anomaly detection based on the global information of each sensor node. Distributed fusion scheme is proposed based on the detection report and the constraint information in which the fused estimate is proved to satisfy the priori constraint. Finally, two example of random anomaly measurement and time-varying anomaly measurement are provided to demonstrate the effectiveness of the proposed distributed estimator.