Intelligent and secure framework for critical infrastructure (CPS): Current trends, challenges, and future scope

Abstract

Cyber–Physical Systems (CPS) are developed by the integration of computational algorithms and physical components and they exist as a result of technological advancement in embedded systems, distributed systems, and sophisticated networking technologies. Typically, these systems are intended to monitor and manipulate real-environment processes and objects. There had been advancements in CPS that affect various aspects of human lives and enable a wider range of applications and services. However, the interconnectivity between the cyber and physical world in CPS introduces new threats and security challenges and also increases the attack surface. The CPS is more prone to cyber-attacks than physical attacks because of the high accessibility of cyber components than physical ones. If we consider the last decade, it is no longer a myth to see industries relying on CPS being a victim of security attacks. In this paper, considering the applicability and dangers of CPS, we thus presented the key security aspects of CPS by reviewing published literature since 2018. We have examined the CPS requirements, challenges, security methods, security standards, threats, vulnerabilities, and past attacks. Moreover, we have considered the role of machine learning and deep learning in the enhancement of the security of CPS and reviewed the performance of existing works, and analyzed the challenges faced, and possible improvement techniques to enhance performance. We identified some key issues and challenges (such as CPS constraints, the performance of learning models, and security of learning models) in the adaption of learning-based methods for CPS security and accordingly proposed a security framework for CPS. Finally, several suggestions and recommendations are proposed considering the lessons learned from the comprehensive review.