基于VPN通道下的加密流量分类算法  

Encrypted Traffic Classification Algorithm Based on VPN Channel

作　　者：魏洁玲 马秀丽[1] 金彦亮[1] 王瑞[1] WEI Jieling;MA Xiuli;JIN Yanliang;WANG Rui(School of Communication and Information Engineering,Shanghai University,Shanghai 200444,China)

机构地区：[1]上海大学通信与信息工程学院,上海200444

出　　处：《应用科学学报》2023年第4期646-656,共11页Journal of Applied Sciences

基　　金：国家自然科学基金(No.61771299)资助。

摘　　要：为了改善网络管理水平、加强网络安全监督,针对虚拟专用网络(virtual private network,VPN)通道下流量加密性强、不透明度高的特点,设计了加密流量数据的新构图方式,提出了基于变体ResNet18网络的加密流量分类算法。为了验证算法有效性,采集真实VPN通道下的热门app流量,成功实现了多VPN通道下的多应用流量分类。所提算法最终在公有数据集与真实采集数据集上的分类准确率分别达到98.1%和96.0%。实验结果表明,该算法具有通用性且具有一定的实际价值。This paper proposes a new encrypted traffc classification algorithm based on a variant ResNet18 network to improve network management and strengthen network se-curity supervision.A three-channel image construction is designed to address the strong encryption and high opacity characteristics of traffc in virtual private network(VPN)channels.The proposed method successfully identifies different apps’traffc in different VPN channels,as validated using popular apps’traffc collected from real VPN channels.The algorithm achieves 98.1%and 96.0%classification accuracy on public and self-collected datasets,respectively.Experimental results demonstrate the algorithm’s universality and practical value.

关 键 词：深度学习 加密流量 虚拟专用网络 残差网络 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]