Consensus-based filtering under false data injection attacks

摘要

In this paper, we consider the attack detection issues for consensus-based distributed filtering. Assume that a malicious attacker exists who can tamper the data transmitted on the communication channel. First, we design a residue-based detector for each sensor to decide whether the received data is malicious or not. Then, we design an optimal estimator for the networked system with the proposed detector. Further, we prove that the proposed detector based on a stochastic rule does not destroy the Gaussianity of the innovation, and provide a sufficient condition to guarantee the convergence of the estimation error covariance. Finally, we provide some examples to verify the effectiveness of the proposed detector under integrity attacks. (C) 2018 European Control Association. Published by Elsevier Ltd. All rights reserved.