(545d) Encrypted Model Predictive Control Design for Security to Cyber-Attacks
AIChE Annual Meeting
2022
2022 Annual Meeting
Computing and Systems Technology Division
Networked, Decentralized, and Distributed Control
Wednesday, November 16, 2022 - 4:27pm to 4:46pm
The security of process control systems has become crucially important since control systems are vulnerable to cyber-attacks, which are a series of computer actions employed by the attacker to compromise the security of control systems (e.g., integrity, stability and safety). Cyber-security and cyber-defense have garnered increasing research interests with the rise of virtualization and big data. Among cyber-attacks, intelligent, targeted attacks are severe threats for control systems because of their designs with the aim of modifying the control actions applied to a chemical process (for example, the Stuxnet worm aims to modify the data sent to a Programmable Logic Controller [1]). Additionally, targeted attacks are
usually stealthy and difficult to detect using classical detection methods since they are designed based on some known information of control systems (e.g., the process state measurements). Therefore, designing optimal, yet secure, control schemes for nonlinear processes in the presence of intelligent cyber-attacks remains an important, fundamental research issue. Furthermore, in the presence of cyber-attacks, it is important to achieve secure communication in the sensor-controller and controller-actuator links via encryption of the communication signals.
This work focuses on the development of a secure and private communication design using semi-homomorphic encryption to ensure cyber-security of model predictive control systems. Specifically, Paillier encryption [2] is used whose security guarantees rely on standard cryptographic principles. Implementing encryption to encrypt-decrypt the communication signals involves quantization of the signals and calculations using large integers, which may result in significant delays in order to ensure error-free signal encryption-decryption, and thus, the MPC system is designed to ensure a certain degree of robustness with respect to potential encryption process errors as well as delays. Finally, the encryption-decryption scheme is tuned to ensure that the calculations can be done with the available computational resources for a specific operating region in the state-space. In addition to MPC, other control schemes like classical controls and linear/nonlinear explicit control techniques will be discussed in this context. A chemical process example will be used
to demonstrate how the new MPC design and the encryption process can be implemented and evaluate their performance and robustness.
[1]. A. A. Cárdenas, S. Amin, Z. S. Lin, Y. L. Huang, C. Y. Huang, and S. Sastry. Attacks
against process control systems: risk assessment, detection, and response. In Proceedings of the
6th ACM symposium on information, computer and communications security, pages 355–366.
ACM, 2011.
[2]. P. Paillier. Public-key cryptosystems based on composite degree residuocity classes. In Proceed-
ings of the 17th International Conference on Theory and Application of Cryptographic Tech-
niques, pages 223–238, 1999.
usually stealthy and difficult to detect using classical detection methods since they are designed based on some known information of control systems (e.g., the process state measurements). Therefore, designing optimal, yet secure, control schemes for nonlinear processes in the presence of intelligent cyber-attacks remains an important, fundamental research issue. Furthermore, in the presence of cyber-attacks, it is important to achieve secure communication in the sensor-controller and controller-actuator links via encryption of the communication signals.
This work focuses on the development of a secure and private communication design using semi-homomorphic encryption to ensure cyber-security of model predictive control systems. Specifically, Paillier encryption [2] is used whose security guarantees rely on standard cryptographic principles. Implementing encryption to encrypt-decrypt the communication signals involves quantization of the signals and calculations using large integers, which may result in significant delays in order to ensure error-free signal encryption-decryption, and thus, the MPC system is designed to ensure a certain degree of robustness with respect to potential encryption process errors as well as delays. Finally, the encryption-decryption scheme is tuned to ensure that the calculations can be done with the available computational resources for a specific operating region in the state-space. In addition to MPC, other control schemes like classical controls and linear/nonlinear explicit control techniques will be discussed in this context. A chemical process example will be used
to demonstrate how the new MPC design and the encryption process can be implemented and evaluate their performance and robustness.
[1]. A. A. Cárdenas, S. Amin, Z. S. Lin, Y. L. Huang, C. Y. Huang, and S. Sastry. Attacks
against process control systems: risk assessment, detection, and response. In Proceedings of the
6th ACM symposium on information, computer and communications security, pages 355–366.
ACM, 2011.
[2]. P. Paillier. Public-key cryptosystems based on composite degree residuocity classes. In Proceed-
ings of the 17th International Conference on Theory and Application of Cryptographic Tech-
niques, pages 223–238, 1999.