A Novel IM Sync Message-Based Cross-Device Tracking. (22nd September 2020)

• Record Type: Journal Article
• Authors: Guo, Naixuan; Luo, Junzhou; Ling, Zhen; Yang, Ming; Wu, Wenjia; Gu, Xiaodan
• Other Names: Xu Xiaolong Academic Editor.
• Is Part Of: Security and communication networks. Volume 2020(2020)
• Publication Date: 2020-09-22
• Subjects: Computer networks -- Security measures -- Periodicals; Computer security -- Periodicals; Cryptography -- Periodicals; 005.805
• DOI: 10.1155/2020/8891664
• Languages: English
• ISSNs: 1939-0114

Abstract

Abstract : Cybercrime is significantly growing as the development of internet technology. To mitigate this issue, the law enforcement adopts network surveillance technology to track a suspect and derive the online profile. However, the traditional network surveillance using the single-device tracking method can only acquire part of a suspect's online activities. With the emergence of different types of devices (e.g., personal computers, mobile phones, and smart wearable devices) in the mobile edge computing (MEC) environment, one suspect can employ multiple devices to launch a cybercrime. In this paper, we investigate a novel cross-device tracking approach which is able to correlate one suspect's different devices so as to help the law enforcement monitor a suspect's online activities more comprehensively. Our approach is based on the network traffic analysis of instant messaging (IM) applications, which are typical commercial service providers (CSPs) in the MEC environment. We notice a new habit of using IM applications, that is, one individual logs in the same account on multiple devices. This habit brings about devices' receiving sync messages, which can be utilized to correlate devices. We choose five popular apps (i.e., WhatsApp, Facebook Messenger, WeChat, QQ, and Skype) to prove our approach's effectiveness. The experimental results show that our approach can identify IM messages with high F 1 -scores (e.g., QQ's PC message is 0.966, and QQ's phone message is 0.924) and achieve an average correlating accuracy of 89.58% of five apps in an 8-people experiment, with the fastest correlation speed achieved in 100 s.