摘要
为进一步提升Tor(the onion router)网络流关联技术的准确率,减少时空开销以及增强容错性,提出一种基于孪生神经网络的Tor网络流关联方法。提取Tor网络流量的包间隔与包大小作为原始流量特征,利用孪生神经网络对特征进行关联性分析。通过孪生神经网络提取入口流与出口流的特征向量并进行相似度计算,根据阈值选择函数选择关联阈值判断流量是否关联。实验结果表明,所提方法关联准确率达到96.21%,误报率仅为0.1%,较现有方法准确率提升2.05%,误报率显著降低,进一步降低了关联成本。
To further improve the accuracy of flow correlation technology on Tor(the onion router),reduce spatiotemporal overhead,and enhance fault tolerance,a flow correlation method based on Siamese neural networks on Tor was proposed.The packet interval and packet size of Tor traffic were extracted as raw traffic features,and Siamese neural networks were used to perform correlation analysis on the features.The feature vectors of the entrance flow and exit flow were extracted through the Siamese neural network and the similarity was calculated.According to the threshold choice function,the correlation threshold was selected to determine whether the flow was related.Experimental results indicate that the proposed method achieves a true positive rate of 96.21%and a false positive rate of only 0.1%.Compared to the existing methods,the true positive rate is improved by 2.05%,and the false positive rate is significantly reduced,while the correlation cost is further reduced.
作者
孟玉飞
翟江涛
刘光杰
MENG Yu-fei;ZHAI Jiang-tao+;LIU Guang-jie(School of Electronic and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China)
出处
《计算机工程与设计》
北大核心
2024年第5期1321-1328,共8页
Computer Engineering and Design
基金
国家自然科学基金项目(61931004、62072250)
国家重点研发计划基金项目(2021QY0700)。
关键词
匿名通信
洋葱路由
流关联分析
孪生网络
深度学习
卷积神经网络
门控循环神经网络
anonymous communication
onion router
flow correlation analysis
Siamese network
deep learning
convolution neural network
gated recurrent neural network
