A switching-based Moving Target Defense against sensor attacks in control systems

Moving Target Defense (MTD) prevents adversaries from being able to predict the effect of their attacks by adding uncertainty in the state of a system during runtime. In this paper, we present an MTD algorithm that randomly changes the availability of the sensor data, so that it is difficult for adversaries to tailor stealthy attacks while, at the same time, minimizing the impact of false-data injection attacks. Using tools from the design of state estimators, namely, observers, and switched systems, we formulate an optimization problem to find the probability of the switching signals that increase the visibility of stealthy attacks while decreasing the deviation caused by false data injection attacks. We show that the proposed MTD algorithm can be designed to guarantee the stability of the closed-loop system with desired performance. In addition, we formulate an optimization problem for the design of the parameters so as to minimize the impact of the attacks. The results are illustrated in two case studies, one about a generic linear time-invariant system and another about a vehicular platooning problem.(c) 2022 Elsevier Ltd. All rights reserved.

Automated summary

This paper presents a Moving Target Defense (MTD) algorithm that enhances security by randomly changing the availability of sensor data. It aims to make it difficult for adversaries to predict the effect of their attacks and minimize the impact of false data injection attacks. Through optimization problems, the algorithm finds the probability of switching signals to increase the visibility of stealthy attacks and reduce the deviation caused by false data injection attacks. The algorithm guarantees system stability with desired performance and demonstrates its effectiveness through case studies.