Inversion Attack and Countermeasure for Event-Based Sensor Schedule in Remote State Estimation

Abstract

This article investigates a novel attack strategy, called inversion attack, against event scheduling that is used in remote state estimation. The scenario is that sensors transmit innovations to remote estimators, and event-based schedulers are equipped to save energy. Unlike the existing powerful cyber-attacks, the inversion attack is proposed for attackers with very limited capabilities. We demonstrate that a malicious attacker can degrade system performance simply by directly injecting an inverter into the event scheduler without any system knowledge. The stealthiness of the inversion attack is quantified by two types of failure detectors. To mitigate the attack effect, we further construct an alternative estimator as a countermeasure, based on which a defense strategy is designed for systems at risk of inversion attacks. Simulation examples are provided to validate the obtained results.