Harnessing Machine Learning for Enhanced Internet of Things (IoT) Security and Attack Detection
Md Ahnaf Akif, İsmail Bütün, Imad Mahgoub · 2024
The rapid growth in IoT devices has modernized connectivity across industries. It has also raised severe security concerns regarding IoT networks, making them a prime target of cyber attacks. In this work, we present an in-depth study on the detection of IoT attacks using state-of-the-art machine learning techniques such as Artificial Neural Networks, Random Forests, Extreme Gradient Boosting, and Convolutional Neural Networks. After underlining the careful cleaning and preprocessing of data, we have conducted several simulations on the IoT-23 dataset, which includes many attacks against devices, like Distributed Denial of Service and malware from the Mirai and Okiru botnets. In this paper, we present experimental results that show our approach significantly improves the accuracy and efficiency of attack detection in IoT environments with respect to the literature. Results point out that good data preprocessing lies at the very core of better detection performance; it provides a framework for developing practical, scalable, and reliable security solutions in the future for the IoT.