Distributed Secure Estimation Over Wireless Sensor Networks Against Random Multichannel Jamming Attacks

Abstract

This paper addresses the distributed secure estimation problem over wireless sensor networks subject to random multichannel jamming attacks. Each sensor’s measurement is divided into $n_{y}$ (the dimension of measurement signal) components and transmitted via $n_{y}$ relevant wireless channels. The attacker is an active adversary in the sense that sensors’ measurements through wireless transmission channels are randomly dropped if the corresponding channels are successfully jammed. By employing a piecewise homogeneous Markov chain, a sophisticated two-level switching multichannel jamming attack model is developed. From the perspective of the attacker, this attack model is promising and makes the wireless channels highly vulnerable, because the attacker can randomly and arbitrarily decide when and where to launch the attacks. We then focus our attention on the secure estimation of a target signal with the caveat that some of the measurements can be incomplete induced by the attacks. A system theoretic framework is then developed to cast the network-based security problem into an $H_{\infty }$ estimation theory problem of a piecewise homogeneous Markov jump system. Criteria for analyzing $H_{\infty }$ estimation performance and designing resilient estimators against noises and attacks are also presented. The effectiveness of the proposed results is illustrated through a military F404 aircraft engine system.