Distinguishing Between Smartphones and IoT Devices Via Network Traffic
Shuodi Hui,Huandong Wang,Dianlei Xu,Jing Wu,Yong Li,Depeng Jin
DOI: https://doi.org/10.1109/jiot.2021.3078879
IF: 10.6
2021-01-01
IEEE Internet of Things Journal
Abstract:Internet of Things (IoT) devices are increasingly growing in mobile networks with the ubiquity of various IoT services. They share the same infrastructure with smartphones while having different requirements for communication resources and security defense mechanisms. Distinguishing IoT devices from smartphones has far-reaching implications on effective network design, resource allocation scheme, pricing scheme, etc. In this article, we distinguish between 12 107 IoT devices and 12 693 smartphones in the real world via characterizing their network traffic. The IoT devices fall into five categories, namely, locating, monitoring, portable, point of sale (POS), and vehicle. We analyze the device behaviors from the network domain, physical domain, and time domain, make comparisons between each kind of IoT devices and smartphones, and design effective features based on the distinguishable network behavior characteristics at packet level, traffic level, and mobility level. Then, we train several classifiers based on our feature set to identify different kinds of mobile devices. Specifically, the accuracy of identifying IoT devices from smartphones achieves 95.86%, and the accuracies of distinguishing IoT devices in each category from smartphones are all over 95%. In the trained classifiers, feature importance verifies the discriminability of different network traffic characteristics observed in our multidomain measurement. Our study reveals the network traffic behavior characteristics for IoT devices, and successfully distinguishes them from smartphones, which paves the way for better network design, resource allocation, pricing scheme, and security defense mechanisms.