The Identification of IoT Devices Using Neural Networks and Passive DNS Traffic
Kexian Zhou, Zhen Zhou Wu · Advances in transdisciplinary engineering · 2025
The rapid proliferation of Internet of Things (IoT) devices has introduced significant challenges in network visibility, security, and management. Traditional device fingerprinting approaches often rely on active probing or deep packet inspection, both of which are intrusive and difficult to scale in heterogeneous environments. In this paper, we propose a novel passive identification method that leverages DNS traffic features and neural network models to accurately classify IoT device types without requiring payload inspection. We construct a labeled dataset by mapping observed DNS queries to known IoT device categories and extract temporal, lexical, and behavioral features. A multi-layer neural network is trained to infer device identity based solely on passive traffic patterns. Experimental results across multiple device brands and network conditions demonstrate the proposed method achieves high classification accuracy, low false-positive rates, and strong generalizability to unseen devices. This work provides a scalable and privacy-preserving solution for IoT inventory and anomaly detection in large-scale networks.