A Moving Target Defense for Securing Cyber-Physical Systems

This article considers the design and analysis of multiple moving target defenses for recognizing isolating attacks on cyber-physical systems. We consider attackers who perform integrity some set sensors actuators in a control system. In such cases, it has been shown that model aware adversary can carefully attack vectors to bypass bad data detection identification filters while causing damage To counter an attacker, we propose defense which introduces stochastic, time-varying parameters The underlying random dynamics system limit attacker's knowledge inhibit his or her ability construct stealthy sequences. Moreover, nature thwarts adaptive adversaries. explore three main designs. First, hybrid where within existing plant are switched among modes. demonstrate how approach enable both malicious nodes. Next, investigate addition extended with coupled original but do not affect performance. Here, will authenticating subsystem turn be revealed by measuring plant. Finally, use sensor nonlinearities enhance effectiveness defense. nonlinear act conceal normal operational behavior from attacker tampered state, further hindering glean information about dynamics. all cases mechanisms proposed. analyze detectability each investigating expected lower bounds statistic. Our contributions also tested via simulation.