Trustworthy And Reliable Cyberattack Detection System for Industrial IoT using Deep Learning

Authors:

Jahanara Begum 1, Ather Hussain 2, Akhilesh Yadav 3, Afsar Saha 4, Ankit Kumar  5, Vamsi Varma  6

1Assistant Professor, Department of Computer Science and Engineering, Hyderabad Institute of Technology and Management, Hyderabad,Telangana, India

2,3,4,5,6 B.Tech Students, Department of Computer Science and Engineering, Hyderabad Institute of Technology and Management, Hyderabad,Telangana, India

Abstract— IIoT is being used more and more to automate and track industrial environments through machines, sensors and controllers that are connected. However, its extensive connectivity with numerous systems introduces significant security challenges for industrial control systems. IIoT environments are prone to more sophisticated cyber threats, because conventional cybersecurity approaches are constrained by old protocols. In this work, we propose a deep learning based method for detection of the cyberattacks in IIoT systems in particular SCADA systems. The proposed method integrates a PRU model with decision trees (DT) through an ensemble learning strategy, achieving enhanced detection accuracy. This integration enables nonlinear feature extraction and strengthens the system by mitigating the impact of inconsistent features. Evaluation on fifteen SCADA datasets demonstrates that our approach outperforms classical and learning-based detection methods. The results demonstrate that the framework can be utilized for enhancing protection, trust and robustness in industrial IIoT systems.

Keywords— Industrial Internet of Things(IIoT), Cyberattack Detection, Deep Learning, Pyramidal Recurrent Units (PRU), Decision Trees (DT).