Design of privacy-preserving control strategy for data opacity of industrial internet (面向工业互联网中数据不透明性的 隐私保护控制策略设计)

• Main Authors: Shen, Yi, Zhou, ChunJie, Hu, XiaoYa, Cheng, Cheng, He, Xin, Tai, RuoChen
• Other Authors: School of Electrical and Electronic Engineering
• Format: Article
• Language: English
• Published: 2022
• Subjects: Engineering::Electrical and electronic engineering; Industrial Internet; Networked Discrete-Event Systems
• Online Access: https://hdl.handle.net/10356/163007
• Institution: Nanyang Technological University

With the continuous development of information and communication technologies (ICT) as well as their applications in the industrial environment, industrial internet comes into being. In comparison to the relatively closed and secure production environment of traditional industries, the industrial internet faces significant security challenges in terms of data privacy. This paper investigates the privacy-preserving control problem for data opacity of the industrial internet to provide a resilient defending strategy against cyber threats, where the system to be controlled is modeled as a networked discrete-event system (DES) and malicious eavesdroppers exist to infer the system secret based on its observations. We here aim at designing a networked supervisor to achieve the following two objectives: (1) eavesdroppers cannot infer the system secret based on its observations, which is formulated as the requirement of opacity enforcement, (2) under the control of the designed networked supervisor, the system could still achieve the predetermined control requirement. To solve the previously mentioned privacy-preserving supervisory control problem, this paper first proposes a privacy-preserving control architecture for networked DES, in which each component is modeled as a finite-state automaton to simulate its dynamics. We design an algorithm based on the constructed models to synthesize the networked supervisor by transforming the original privacy-preserving control problem into a classical Ramadge-Wonham supervisory control problem. Furthermore, this paper proves (1) the designed synthesis algorithm is sound and complete, and (2) the existence of the supremal networked supervisor for this privacy-preserving supervisory control problem.