Distributed Kalman filtering for robust state estimation over wireless sensor networks under malicious cyber attacks

We consider distributed Kalman filtering for dynamic state estimation over wireless sensor networks. It is promising but challenging when network is under cyber attacks. The compromised nodes are likely to influence system security by broadcasting malicious false measurements or estimates to their neighbors, and result in performance deterioration. To increase network resilience to cyber attacks, in this paper, trust-based dynamic combination strategy is developed. The proposed distributed Kalman filtering scheme is resilient to random, false data injection and replay attacks. Furthermore, it is efficient in terms of communication load, only instantaneous estimates are exchanged between the neighboring nodes and compromised nodes localization is a byproduct.