Identify VPN Traffic Under HTTPS Tunnel Using Three-Dimensional Sequence Features
Open Access
- 9 December 2022
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 18-23
Abstract
VPN traffic identification under HTTPS tunnel is an urgent problem in the field of network security and network management. In the research scenario of TLS encrypted VPN traffic, the existing work mainly focuses on the classification of traffic application types, and there is no specific research on VPN traffic identification under HTTPS tunnel, this makes the existing methods in the VPN traffic identification scenario under HTTPS tunnel subject to the impact of different application type traffic patterns and different network conditions, resulting in reduced identification accuracy and robustness of the model. At the same time, the existing work also lacks the analysis of the TPR and FPR indicators of the model, which results in that the practicability of the existing results cannot be guaranteed. To solve these problems, this paper has confirmed the simulation scenario of VPN traffic identification under HTTPS tunnel. In order to better mine the network behavior characteristics of VPN services, we propose three-dimensional sequence features (TSF) from three aspects: changes in the number of packets, changes in the payload length of packets, and changes in the interaction rules of uplink and downlink packets. Then we convert the features into RGB images to build a CNN based VPN traffic identification framework under HTTPS tunnel. Experiments on the ISCX-TOR dataset show that the Accuracy of the TSF method is 100%, the TPR and F1-Score are more than 99%, and the FPR is only 0.03%. This means that the TSF method has excellent identification accuracy in the VPN traffic identification scenario under the HTTPS tunnel, and can meet the requirements of actual use to a certain extent.Keywords
This publication has 5 references indexed in Scilit:
- Detection of Anonymising Proxies Using Machine LearningInternational Journal of Digital Crime and Forensics, 2021
- Detection of Virtual Private Network Traffic Using Machine LearningInternational Journal of Wireless Networks and Broadband Technologies, 2020
- Network traffic identification of several open source secure proxy protocolsInternational Journal of Network Management, 2019
- Deep learning-based real-time VPN encrypted traffic identification methodsJournal of Real-Time Image Processing, 2019
- Characterization of Encrypted and VPN Traffic using Time-related Features2016