基于Res-CAN的Tor网站指纹识别模型  

Res-CAN Based Tor Website Fingerprinting Identification Model

作　　者：王曦锐 芦天亮[1] 杨成 于兴崭 WANG Xirui;LU Tianliang;YANG Cheng;YU Xingzhan(School of Information and Cyber Security,People’s Public Security University of China,Beijing 100038,China)

机构地区：[1]中国人民公安大学信息网络安全学院,北京100038

出　　处：《中国人民公安大学学报（自然科学版）》2023年第2期76-84,共9页Journal of People’s Public Security University of China(Science and Technology)

基　　金：北京市社会科学基金(21JCC108);中国人民公安大学2022年基本科研业务费项目(2022JKF02022)。

摘　　要：网站指纹识别技术通过分析流量特征判断用户访问的网站站点,能够有效监管TOR匿名网络的用户行为。现有的识别方法通常需要大规模的数据样本以获得高的识别准确率,且普遍存在概念漂移问题。针对以上问题,本文提出一种基于残差和协作对抗网络(Residual network and Collaborative and Adversarial Network,Re s-CAN)的网站指纹识别模型。该模型使用残差网络(Residual network)作为特征提取器以减少网络的优化难度。同时,将协作对抗网络(Collaborative and Adversarial Network,CAN)应用于网站指纹识别问题,使得特征提取器同时学习领域相关和领域无关特征,实现源域与目标域的特征空间对齐。实验结果表明,本文提出的方法在小样本环境下网站指纹识别准确率达到91.2%,优于现有的利用对抗领域自适应网络(Domain-Adversarial Neural Networks,DANN)迁移学习方法,且抗概念漂移能力较高。Website fingerprint identification technology can effectively supervise the user behavior of Tor anonymous network by analyzing traffic characteristics to determine the websites visited by users.Current recognition methods usually need large-scale data samples to obtain high recognition accuracy,and there is a widespread problem of concept drift.In view of the above problems,a website fingerprint identification model is proposed based on residual network and Collaborative and Adversarial Network.Residual network is used as feature extractor to reduce the difficulty of network optimization.At the same time,the collaborative and adversarial network is applied to website fingerprint identification,so that domain informative and domain uninformative features can be learned by the feature extractor,realizing the feature space alignment of source domain and target domain.The experimental results show that the accuracy for website fingerprint identification of the method proposed in this paper can reach 91.2%in a small sample environment,which is better than the current transfer learning methods using domain-adversarial neural networks.Furthermore,the ability to resist concept drift is high.

关 键 词：网站指纹 匿名网络 残差网络 领域自适应 迁移学习 

分 类 号：D918.91[政治法律—法学]