Blockchain-based intrusion detection system with artificial intelligence

Soh, Wen Kai (2025) Blockchain-based intrusion detection system with artificial intelligence. Final Year Project, UTAR.

Preview

PDF Download (2932Kb) | Preview

Abstract

This project presents the development of a Blockchain-Based Intrusion Detection System with Artificial Intelligence, designed to address the limitations of traditional intrusion detection frameworks that lack contextual awareness, secure alert storage, and automated response. The system integrates a rule-based signature engine with a large language model to detect both known and previously unseen network threats through real-time traffic analysis. Signature-based detection matches flows against predefined patterns, while the LLM performs context-aware reasoning to identify complex or ambiguous behaviours, producing alerts with human-readable explanations and severity levels. To ensure alert integrity and traceability, all detection events are logged to a private Ethereum blockchain using smart contracts, providing a decentralised and tamper-resistant audit trail. Simultaneously, off-chain logging is enabled to ensure efficient notification of intrusion events. A web-based dashboard offers live monitoring of packet capture, active flows, alert statistics, and blockchain synchronisation. The system was designed for modularity, with configurable components for flow processing, AI analysis, and on-chain logging. It achieved a detection accuracy of 93.95% and a false positive rate of 5.00%, confirming the effectiveness of its hybrid detection approach. This prototype demonstrates the feasibility of combining AI and blockchain technologies to build an IDS that is not only accurate but also transparent, explainable, and resistant to tampering, thus making it a promising foundation for modern, resilient cybersecurity systems.

Actions (login required)