Secure and Resilient Control of Constrained Cyber-Physical Systems Subject to Network Attacks

点击翻译按钮获取中文摘要

Cyber-physical systems (CPS) embody a tight integration between network-based communications, software, sensors and physical process and they are expected to play a significant role in the design and development of future engineering systems. The Stuxnet malware designed to disrupt the operations of an Iranian's nuclear enrichment plant or the Industroyer malware affecting the Ukrainian power grid are two examples of cyber-attacks affecting CPS. State of the art shows that many solutions have been presented for the problem of robust and resilient state estimation of CPS under attacks, but few control remediation strategies have been proposed to ensure the safety of plants subject to state and input constraints. Classical fault-tolerant and networked control solutions are capable of guaranteeing plant safety against specific physical faults and network issues. However, they are not capable of facing malicious adversary. On the other hand, cyber-security tools alone are not sufficient to deal with CPS because they are not aware of physical implications of cyber-attacks. For such reasoning, there is a need for new CPS security and resiliency methods overcoming existing approaches.The long-term objective of this program is the investigation of promising ideas and solutions for a secure and resilient control of constrained CPS in the presence of cyber-attacks on the communication channels. From a methodological perspective, novel model predictive control (MPC) schemes will be developed to deal with constrained plants and attack scenarios. Different key research challenges will be faced in the short-term: modeling and characterization of cyber-attacks, development of set-theoretic estimation tools, and design of unifying control architecture capable of detecting and mitigating attacks coordinately. The developed control schemes will be responsible for the identification of cyber-attacks and ensuring plant resiliency. As long as the problem of identification is concerned, first the set of information and resources available to the attackers will be characterized to identify attack vectors and disruptive resources. Then, active set-theoretic state estimator modules, based on watermarking ideas, will be developed to prevent undetectable attacks. Physical countermeasures will be obtained by resorting to networked robust MPC tools. The main innovative aspects of this research proposal are: capability to jointly detect and mitigate cyber-attacks, plant constraints fulfillment regardless of any admissible attack occurrence and real-time affordability of the resulting control algorithms.The outcome of the proposed research will have a significant impact on the development of novel control architectures capable of improving the security of future engineering systems which will unavoidably have a broad impact on Canadian economy and will enhance the way we will live.