A Dynamic Watermarking Scheme to Attack-Detection-Based Resilient Set-Membership Estimation for 2-D Systems Over Sensor Networks

Cyberattacks significantly compromise the security of sensor networks, which largely hinder the accuracy and privacy of the information interaction process. However, it is tough to identify the attacks with unknown statistical models by a traditional detection scheme. This article is concerned with the resilient set-membership estimation for 2-D systems over sensor networks subject to cyberattacks and unknown-but-bounded noises. First, using the dynamic watermarking (DW) technique, a novel attack detection method is proposed to guarantee the security of information transmission among sensors. Compared to existing results on attack detection, the developed mechanism is capable of adapting to the ellipsoid-dependent set-membership estimation framework that incorporates the set ideas. Second, a group of scalable resilient set-membership estimators are derived for large-scale sensor networks. Each sensor has access to an ellipsoidal estimation set containing the true state of the system, in which the proposed design procedure is independent of global information about the network topology. Third, a set of parameter optimization algorithms is introduced in a recursive manner, which allows the sensor network to obtain a satisfactory estimation performance utilizing the designed estimation gain and detection mechanism. Lastly, a simulation example is shown to evaluate the availability and superiority of the proposed technique.