Trust-Based Distributed Set-Membership Filtering for Target Tracking Under Network Attacks

For target tracking problems in wireless sensor networks subject to malicious network attacks, this paper proposes a distributed set-membership filtering algorithm based on trust dynamic combination strategy. The algorithm has a prediction-correction recursive updating structure similar to Kalman filtering, by introducing the clustering fusion step of received data from other nodes between the prediction step and the measurement correction update step, the clustering fusion step uses K-means to cluster and classify the data of trusted and untrusted nodes, the target state is updated by the fusion of trusted received data set, to improve the resistance to various wicked network attacks. Simulation results show that compared with the traditional distributed set-membership filtering method, the proposed method has better target tracking performance in the face of wicked network attacks such as random attacks, false data injection, replay attacks, and hybrid attacks.