Anomaly Detection in IIoT Transactions using Machine Learning: A Lightweight Blockchain-based Approach

Abstract

The integration of secure message authentication systems within the Industrial Internet of Things (IIoT) is paramount for safeguarding sensitive transactions. This paper introduces a Lightweight Blockchain-based Message Authentication System, utilizing k-means clustering and isolation forest machine learning techniques. With a focus on the Bitcoin Transaction Network (BTN) as a reference, this study aims to identify anomalies in IIoT transactions and achieve a high level of accuracy. The feature selection coupled with isolation forest achieved a remarkable accuracy of 92.90%. However, the trade-off between precision and recall highlights the ongoing challenge of minimizing false positives while capturing a broad spectrum of potential threats. The system successfully detected 429,713 anomalies, paving the way for deeper exploration into the characteristics of IIoT security threats. The study concludes with a discussion on the limitations and future directions, emphasizing the need for continuous refinement and adaptation to the dynamic landscape of IIoT transactions. The findings contribute to advancing the understanding of securing IIoT environments and provide a foundation for future research in enhancing anomaly detection mechanisms.