A Stacking Ensemble of Deep Learning Models for IoT Network Intrusion Detection

Lazzarini, Riccardo; Tianfield, Huaglory; Charissis, Vassilis

Abstract

The number of Internet of Things (IoT) devices has increased considerably inthe past few years, which resulted in an exponential growth of cyber attackson IoT infrastructure. As a consequence, the prompt detection of attacks inIoT environments through the use of Intrusion Detection Systems (IDS) hasbecome essential. In this article, we propose a novel IDS approach based ona stacking ensemble of deep learning (DL) models. Our approach is namedDeep Integrated Stacking for the IoT (DIS-IoT) and it combines four differentDL models into a fully connected DL layer, creating a standalone ensemblemodel. We evaluated DIS-IoT on two open-source datasets, namely ToN IoTand CICIDS2017, in binary and multi-class classification and compared resultswith other standard DL methods. We demonstrated that DIS-IoT iscapable of a high level of accuracy with a very low False Positive rate (FPR)in both datasets outperforming all other models. Results from our experimentswere also compared against results available in the literature, whichwere obtained from similar approaches on the ToN IoT dataset, showing thatour model performs better or on par with the others.

Citation

Lazzarini, R., Tianfield, H., & Charissis, V. A Stacking Ensemble of Deep Learning Models for IoT Network Intrusion Detection

• Deposit Date: May 18, 2023
• Publisher: Elsevier
• Keywords: IoT; intrusion detection systems; Deep Learning; ensemble learning; ensemble neural networks; Stacking; Machine learning; networks; simulation
• Publisher URL: https://doi.org/10.2139/ssrn.4412746