Observer‐based attack‐resilient control for linear systems against FDI attacks on communication links from controller to actuators

Abstract

For the adversarial attacks on the communication links from the controller to the actuators, most of the existing attack-resilient control results focus on denial-of-service attacks. Unlike the existing results, this paper studies the observer-based attack-resilient control problem for linear systems with false data injection attacks and process disturbances. Due to limited resources, the malicious attacker is assumed to only manipulate a certain number of communication links from the controller to the actuators. A novel control architecture is proposed, which consists of an attack-resilient state observer, a controller gain scheme, and a supervisory switching strategy. The observer is developed based on the maximin strategy, and state estimation will be used to construct the controller. The switching strategy is designed to pick an appropriate controller gain and prevent the attack signals from entering the plant automatically. It is shown that the closed-loop system is stable with an attack-resilient performance.Finally, to verify the effectiveness of the proposed controller, simulation results on a linearized reduced-order aircraft system and an IEEE six-bus power system are provided. KEYWORDSattack-resilient control, cyber-physical systems, observer, supervisory switching strategy INTRODUCTIONCyber-physical systems (CPSs) integrate physical processes, computational resources, and communication capabilities 1 and merge the cyber-world with the physical world. CPSs have become prevalent in multifarious real-wold systems, including communication systems, power systems, and aerospace systems. Since CPSs are increasingly connected to the intranet and internet for remote monitoring and control, they become more vulnerable to malicious cyberattacks. For this reason, security issues of CPSs subject to cyberattacks have attracted considerable attention in the last decades. [2][3][4] In a networked control system, cyberattacks to the communication links can be broadly divided into two categories: denial-of-service (DoS) attacks and deception attacks. DoS attacks refer to obstructing/blocking the communication between networked agents 5,6 and causing packet losses. Deception attacks (including false data injection [FDI] attacks, 7,8 sparse attacks, 9 replay attacks, 10 zero-dynamic attacks, 11 etc) are intended to manipulate transmitted data packages and result in false feedback information. 8 For the immense field of CPSs' application, various problems have been investigated, eg, sensor scheduling, 12,13 secure state estimation, 14-16 and secure control. 17 4382It is also worth noting that there are many research challenges toward achieving a secure control objective, as argued in the work of Cardenas et al. 17 During recent years, some researchers have tried to solve the challenging secure control problem for CPSs.Related works on secure control: The secure estimation and control problem was addressed in the work of Fawzi et al 18 for the discrete-time systems with FDI sensor attacks. The main idea is to first design ...

Adaptive control for nonlinear cyber‐physical systems under false data injection attacks through sensor networks

“…When the sensor and actuator attacks are frequency constrained, the reliable control problem for CPSs was studied in the work of Huang and Dong . Considering the attacks on communication links from controllers to actuators, the observer‐based attack‐resilient control architecture for linear systems was proposed in the work of Xie and Yang . In addition, the state estimation problem against sensor attacks was considered in the works of Yang et al However, these methods are mostly implemented for linear systems.…”

“…the observer-based attack-resilient control architecture for linear systems was proposed in the work of Xie and Yang. 15 In addition, the state estimation problem against sensor attacks was considered in the works of Yang et al 16,17 However, these methods are mostly implemented for linear systems.…”

Static output feedback secure control for cyber‐physical systems based on multisensor scheme against replay attacks

“…3,4 Therefore, how to curb cyber attacks has become an urgent issue. [5][6][7][8][9] As we all know, cyber attacks can be mainly classified into three categories: denial of service (DoS) attacks, 10,11 deception attacks, 12,13 and replay attacks. 14 For instance, a resilient dynamic event-triggered controller was designed to cope with the DoS attacks.…”

On modeling and secure control of cyber‐physical systems with attacks/faults changing system dynamics: An average dwell‐time approach

“…By exploiting the game theory, the resilience of networked control systems and wireless networked control systems subject to DoS attacks was analyzed in the works of Yuan et al Moreover, in the work of Zhao et al, the design, implementation, and impact evaluation of undetectable attacks were discussed via a geometric approach. For linear system under the false data injection attacks, an observer‐based control strategy with an attack‐resilient performance was presented in the work of Xie and Yang . A dynamic output feedback controller was given for discrete‐time stochastic nonlinear systems subject to deception attacks in the work of Ding et al The robustness of the networked cooperative systems against the malicious injections was improved via competitive interaction of the cloud network with the original systems in the work of Gusrialdi et al A data‐driven sliding‐mode compensator was developed in the work of Huang et al such that the effect of the actuator attacks on CPSs was eliminated.…”

“…For linear system under the false data injection attacks, an observer-based control strategy with an attack-resilient performance was presented in the work of Xie and Yang. 26 A dynamic output feedback controller was given for discrete-time stochastic nonlinear systems subject to deception attacks in the work of Ding et al 27 The robustness of the networked cooperative systems against the malicious injections was improved via competitive interaction of the cloud network with the original systems in the work of Gusrialdi et al 28 A data-driven sliding-mode compensator was developed in the work of Huang et al 29 such that the effect of the actuator attacks on CPSs was eliminated. The adaptive control policy was developed to mitigate the sensor and actuator attacks in the works of Jin et al 30 and An and Yang,31 and in order to resist the frequency-constrained sensor and actuator attacks, the proposed control equipped with an adaptive attack compensator could guarantee the stability of CPSs in the work of Huang and Dong.…”