IoT vulnerability detection using featureless LLM CyBert model

Bin Hulayyil, Sarah, Li, Shancang and Saxena, Neetesh ORCID: https://orcid.org/0000-0002-6437-0807 2025. IoT vulnerability detection using featureless LLM CyBert model. Presented at: IEEE International Conference on Trust, Security and Privacy in Computing and Communications, Sanya, China, 17-21 December 2024. Proceedings of the 23rd International Conference on Trust, Security and Privacy in Computing and Communications. IEEE, pp. 2474-2480. 10.1109/TrustCom63139.2024.00343

Full text not available from this repository.

Abstract

This work aims to leverage large language models (LLMs) and a featureless approach to effectively detect vulnerabilities in Internet of Things (IoT) network traffic. By directly learning from the Ripple20 dataset, a featureless LLM model, CyBERT, was designed that can efficiently distinguish between secure and vulnerable IoT devices without relying on handcrafted features. This LLM-based classifier could be instrumental in identifying IoT networks that pose potential threats to other connected devices by uncovering critical vulnerabilities. The experimental results demonstrate the exceptional capabilities of the featureless CyBERT model, which achieves high accuracy, precision, recall, and F1 score in detecting zero-day vulnerabilities. Moreover, the model significantly outperforms traditional methods in terms of detection speed. These results have profound implications for the future of IoT security, paving the way for real-time threat and attack detection.

Item Type:	Conference or Workshop Item (Paper)
Date Type:	Published Online
Status:	Published
Schools:	Schools > Computer Science & Informatics
Publisher:	IEEE
ISBN:	979-8-3315-0621-6
ISSN:	2324-898X
Date of Acceptance:	2 November 2024
Last Modified:	13 May 2025 09:15
URI:	https://orca.cardiff.ac.uk/id/eprint/177494

Actions (repository staff only)

Edit Item