Machine learning for securing Cyber–Physical Systems under cyber attacks: A survey
Cheng Fei, Jun Shen
Abstract
With the proliferation of cyber attacks targeting cyber–physical systems (CPS), ensuring the security of CPS under such attacks has become increasingly challenging. Machine learning technology has been widely applied in fields such as classification and data analysis in recent years. Furthermore, as CPS architecture continues to become more intricate and the amount of data generated exponentially expands, implementing machine learning (ML) approaches have become essential to obtain exclusive advantages. Therefore, ML technology is suitable for detecting attacks and using data-driven control capabilities to secure CPS under attacks. This paper aims to provide a comprehensive overview of common cyber attacks and analyze their impacts on CPS. Additionally, we will explore various ML techniques and their applications in securing CPS under attack. Finally, future research directions for CPS security are also presented.