A hybrid methodology for anomaly detection in Cyber–Physical Systems

Nicholas Jeffrey, Qing Tan, José R. Villar

Research output: Contribution to journalJournal Articlepeer-review

8 Citations (Scopus)

Abstract

The rapid adoption of Industry 4.0 has seen Information Technology (IT) networks increasingly merged with Operational Technology (OT) networks, which have traditionally been isolated on air-gapped and fully trusted networks. This increased attack surface has resulted in compromises of Cyber–Physical Systems (CPS) with significant economic and life safety consequences. This paper proposes a hybrid model of anomaly detection of security threats to CPS by blending the signature-based and threshold-based Intrusion Detection Systems (IDS) commonly used in IT networks, with a Machine Learning (ML) model designed to detect behaviour-based anomalies in OT networks. This hybrid model achieves more rapid detection of known threats through signature-based and threshold-based detection strategies, and more accurate detection of unknown threats via behaviour-based anomaly detection using ML algorithms.

Original languageEnglish
Article number127068
JournalNeurocomputing
Volume568
DOIs
Publication statusPublished - 1 Feb. 2024

Keywords

  • Cyber–Physical Systems
  • Machine learning
  • Security threats

Fingerprint

Dive into the research topics of 'A hybrid methodology for anomaly detection in Cyber–Physical Systems'. Together they form a unique fingerprint.

Cite this