Security for Cyber-Physical Systems Using Machine Learning-Based Anomaly Detection: A Survey

Abstract

A Cyber-Physical System (CPS) is a hybrid system that uses both digital and physical parts. IoT, smart power grids and remote laboratory environments, online medical care, intelligent manufacturing, vehicles that are autonomous, the Internet of Things, control systems for industries, and many more have all contributed to CPS's explosive expansion over the last decade. Malicious attacks have increased dramatically due to the broad usage of Cyber-Physical Systems in modern life.

The increased access to the public internet has greatly increased the vulnerability of critical infrastructure, making incidents targeting oil pipelines and electrical power grids more prevalent and concerning. An extensive literature overview on recent developments in anomaly detection methods for Cyber-Physical System security threat identification is presented in this article. Within industrial control networks (ICS), resolving issues related to life safety is given top priority. Reading through a few articles allows us to spot trends and gaps in the literature. Resource limitations, there is a lack of established methods for communication, and the business is highly diverse, which makes it difficult to reach an agreement, and conflicting information security priorities between IT and OT networks are some of the significant outstanding issues highlighted in the article. Identifying possible answers and/or avenues for future study is done to address this.