Original scientific paper

https://doi.org/10.17559/TV-20251011003052

Intelligent Chaotic Harris Hawk Optimization-Driven Deep Learning Framework for Intrusion Detection and Prevention in SCADA Systems

T. John Sunder Singh ; Department of Computer Science and Engineering, Puducherry Technological University, Puducherry, India *
J. I. Sheeba ; Department of Computer Science and Engineering, Puducherry Technological University, Puducherry, India
Pradeep Devaneyan ; Department of Mechanical Engineering, Sri Venkateshwara College of Engineering and Technology, Puducherry, India

* Corresponding author.

Abstract

A smart Intrusion Detection and Prevention System (IDS and IPS) framework is proposed to strengthen the cybersecurity of SCADA (Supervisory Control and Data Acquisition) systems by integrating intelligent optimization, advanced deep learning, and modern cryptographic methods. This framework is designed to detect and prevent cyber threats in real-time, ensuring the safe and uninterrupted operation of critical industrial infrastructure. In this study, SCADA/IIoT traffic is normalized using Min-Max scaling to ensure uniform feature ranges before CHHO-based feature selection and model training. Chaotic Harris Hawk Optimization (CHHO)-based feature selection is utilized in the method to select the most relevant features while reducing computational complexity. In classification, the DL system consisting of CNN (Convolutional Neural Network), Bi-LSTM (Bidirectional Long Short-Term Memory), and Transformer is implemented since it efficiently captures spatial and temporal patterns in SCADA traffic data and long-range dependencies, thus enhancing the accuracy of anomalous activity detection. For real-time and secure communication in intrusion prevention, NTRU cryptography, which is post-quantum secure and lightweight, is combined as a part of IPS process. Data preprocessing and normalization methods are applied to provide better input data for learning. Two realistic industrial datasets, WUSTL-IIoT-2018 and WUSTL-IIoT-2021, are used for model training and evaluation, with various attack scenarios and SCADA-type traffic. The proposed framework works well with high accuracy. The robustness, generalizability, and real-time application of the framework for SCADA security protect critical infrastructure against evolving cyber threats.

Keywords

chaotic harris hawk pptimization (CHHO); deep learning (CNN-BiLSTM-Transformer); intrusion detection and prevention system (IDPS); post-quantum NTRU cryptography; SCADA security

Hrčak ID:

353087

URI

https://hrcak.srce.hr/353087

Publication date:

7.11.2026.

Visits: 0 *