Network-Based Intrusion Detection in IoT Environments Using Hybrid Machine Learning Techniques

Abstract

The recent proliferation of the Internet of Things (IoT) has emerged many security threats, especially on network-based communication platforms. With the increase of cyber-attacks against critical systems, in particular DDoS, e-fraud type or other, the demand for intelligent, adaptive IDS is clearly important. In this context, this paper proposes a hybrid machine learning model integrating traditional algorithms like Support Vector Machines (SVM) and emerging deep neural networks to improve the precision and robustness of detection in IoT systems. The performance of the proposed framework is tested on various benchmark datasets which includes NSL-KDD, DS2OS, and IoT Botnet using various performance parameters such as Accuracy, Precision, Recall, F1-score. The suggested approach yielded a high detection rate of 96.38%, indicating the capability of our system to pinpoint sophisticated intrusion instances on multi-type IoTNs. Experimental results demonstrate that the hybrid solution has the potential for false positives reduction as well as for enhancing response against the threat in real environment. This study is an attempt to bring cyber-security solutions for the emerging landscape of IoT infrastructures, which are scalable, adaptive networked based and intelligent.