A Novel Hybrid Model Detection of Security Vulnerabilities in Industrial Control Systems and IoT Using GCN+LSTM

Abstract

In this study, we address critical security vulnerabilities in Industrial Control Systems (ICS) and the Internet of Things (IoT) by focusing on enhancing collaboration and communication among interconnected devices. Recognizing the inherent risks and the sophisticated nature of cyber threats in such environments, we introduce a novel and complex implementation that leverages the synergistic potential of Graph Convolutional Networks (GCN) and Long Short-Term Memory (LSTM) models. This approach is designed to intelligently predict and detect intrusion attempts by analyzing the dynamic interactions and data flow within networked systems. Our methodology not only differentiates between the operational nuances of various IoT routing mechanisms but also tackles the core design challenges faced by ICS. Through rigorous experimentation, including the deployment of our model in simulated high-risk scenarios, we have demonstrated its efficacy in identifying and mitigating deceptive connectivity disruptions with a remarkable accuracy rate of 99.99%. This performance underscores the models capability to serve as a robust security layer, ensuring the integrity and resilience of ICS networks against sophisticated cyber threats. Our findings contribute a significant advancement in the field of cybersecurity for ICS and IoT, proposing a comprehensive framework that can be centrally integrated with existing security information and incident management systems for enhanced protective measures.

Keywords

• Ad-hoc network
• graph convolutional networks (GCN)
• industrial control system (ICS)
• Internet of Things (IoT)
• intrusion detection system (IDS)
• security vulnerabilities