AI-based Secure Intrusion Detection Framework for Digital Twin-enabled Critical Infrastructure

Authors:

Abstract
This paper discusses the importance of securing modern society’s critical infrastructure in the face of physical and digital threats. It emphasizes the vulnerabilities introduced by programmable logic controllers (PLCs) and supervisory control and data acquisition (SCADA) systems that highlight the potential of digital twins to enhance security. The paper presents a research contribution aimed at improving the integrity of data exchange among PLCs equipped with digital twins in critical infrastructure systems. It leverages artificial intelligence (AI) algorithms to detect and mitigate different security attacks, such as denial of service (DoS), injection attacks, and data tampering attacks. The proposed framework efficiently classifies the malicious and non-malicious data of digital twin-based critical infrastructure. Further, the trained AI model is deployed on a critical infrastructure’s intrusion detection system that continuously monitors network traffic and system logs in real time, identifying unusual patterns or behaviors that may indicate an intrusion. Further, the performance of the proposed framework is evaluated using accuracy, log loss, and validation curves.