Anomaly-Based Intrusion Detection System for Cyber-Physical System Security

TL;DR

This paper presents a machine learning-based anomaly detection system for cyber-physical systems, specifically applied to water tank control, to identify cyber attacks and prevent dangerous states like overflow.

Contribution

It introduces a novel machine learning approach for detecting cyber attacks in industrial control systems, enhancing security in cyber-physical environments.

Findings

Effective classification of normal and anomalous behaviors

Successful detection of cyber attacks in water tank system

Prevents dangerous states such as tank overflow

Abstract

Over the past decade, industrial control systems have experienced a massive integration with information technologies. Industrial networks have undergone numerous technical transformations to protect operational and production processes, leading today to a new industrial revolution. Information Technology tools are not able to guarantee confidentiality, integrity and availability in the industrial domain, therefore it is of paramount importance to understand the interaction of the physical components with the networks. For this reason, usually, the industrial control systems are an example of Cyber-Physical Systems (CPS). This paper aims to provide a tool for the detection of cyber attacks in cyber-physical systems. This method is based on Machine Learning to increase the security of the system. Through the analysis of the values assumed by Machine Learning it is possible to evaluate the classification performance of the three models. The model obtained using the training set, allows to classify a sample of anomalous behavior and a sample that is related to normal behavior. The attack identification is implemented in water tank system, and the identification approach using Machine Learning aims to avoid dangerous states, such as the overflow of a tank. The results are promising, demonstrating its effectiveness.