A lightweight IoT device identification using enhanced behavioral-based features

As the Internet of Things (IoT) landscape expands, new devices with various functionalities are continuously being integrated into the IoT ecosystem. When traditional systems, which involve human interaction, are replaced by devices, it becomes crucial to upgrade the conventional authorization and authentication systems. This is essential to establish a new access control system designed to manage accessibility of multiple devices. Traditional device identification approaches often struggle to accommodate the dynamic behaviors exhibited by IoT devices. In response, this paper introduces an innovative approach that leverages enhanced behavioral features to generate a representation of device behavior. This representation is then employed to train machine learning models for classifying devices based on their behaviors. Furthermore, this paper also considers special scenarios where the access management system lacks access to full network traffic data. In such cases, device identification is achieved based on HTTPS features and user agent information. We conducted experimental analyses using real data from state-of-the-art IoT device profiling datasets. The performance results indicate that the extracted behavioral-based features have the capability to identify multiple IoT devices with various functionalities and vendors.