Predictive Modelling Based Approaches for Secure and Energy Efficient Cyber Physical Systems

Abstract

In recent years, Cyber-Physical Systems (CPSs) have drawn significant attention from newlineacademia and industry due to their potential for developing efficient and reliable applications across different domains. CPSs incorporate various computation, communication and control technologies, thus leading to complex cyber-physical interactions. These interactions, along with the increasing threat of cyber attacks, present significant difficulty to ensure the security and reliability of CPSs. In addition, many CPS applications newlinenecessitate energy-efficient operation, as they are often used in many batterypowered newlinedevices with stringent energy consumption requirements. However, ensuring newlinethe design objectives of security and energy efficiency in CPSs simultaneously results newlinein a trade-o_, as increasing security negatively impacts energy efficiency. More recently, novel approaches based on predictive modelling-based techniques have emerged as a promising alternative to address the various challenges in developing CPSs. Predictive modelling-based techniques offer a comprehensive methodological approach to newlineaddress the fundamental scientific and engineering challenges in the design, synthesis and analysis of CPSs. This dissertation focuses on developing secure, energy-efficient and reliable cyber-physical systems based on predictive modelling-based approaches. In particular, we address the following concerns regarding the design and development of cyber-physical systems. Cybersecurity Measures and Sustainable Energy Management: We perform various cyber attacks, such as DoS and Deception attacks, in CPSs and develop mechanisms to secure CPSs against such attacks. Towards the goal of newlinesecure and energy-efficient CPSs, we perform quantitative trade-o_ analyses between newlinecybersecurity and energy efficiency of CPSs, considering the available crypxix newlinetographic techniques and constraints imposed by limited energy budgets. Using newlinethe mathematical models of cyber attacks, we develop a three-stage Markov Decision newlineProcess (MDP) model for CPS...