IoT Device Identification Using Supervised Machine Learning

Abstract

Internet of Things (IoT) has been increasingly becoming mainstream and can be considered as the next stage of the internet revolution. The increasing use of IoT-based applications presents several issues to massively connected devices. For example, companies and organizations need to have a fast and reliable way to identify IoT devices on their networks to manage access and prevent vulnerable devices from connecting. On the other hand, machine learning has been widely used for image processing, intrusion detection, and malware classification. However, there are few studies on device identification using machine learning. In this paper, we propose a machine learning-assisted approach for IoT device identification. That includes four essential components: network traffic collection, feature extraction, data labeling, and machine learning. We test and evaluate four machine learning classifiers in a testing network, including multiple IoT devices. The evaluation results indicate a 79% accuracy in identifying the IoT devices in the considered network testbed.