中圖分類號：TN918;TP309文獻標識碼：ADOI:10.19358/j.issn.2097-1788.2025.02.003
引用格式：李榮，賴怡聰，李樂言. 基于雙模態融合的ZigBee設備識別方法［J］.網絡安全與數據治理，2025，44（2）：17-25.

A dual-modal fusion approach for ZigBee device identification

Abstract： Smart homes bring convenience to daily life but also introduce privacy and security risks. Smart sensors continuously collect information about their surroundings and transmit data remotely via wireless communication. To protect user privacy, this paper proposes a ZigBee device identification method based on dual-mode fusion. Firstly, ZigBee traffic data from smart home devices is passively captured through the air interface. Then, the device traffic is fragmented: in the text modality, features such as timing and length of encrypted traffic are extracted, while in the image modality, high-dimensional features of traffic images are obtained. Finally, the textual and image features of the encrypted traffic are fused to construct a device type recognition model based on bimodal fusion. Experimental results involving 15 devices from 5 manufacturers show that, even with encrypted wireless traffic, this method achieves an identification accuracy of approximately 99% for ZigBee devices, which can effectively identify nearby smart sensors and protect user privacy.

Key words : privacy security; ZigBee; deep learning; smart home