Relation-CNN: Enhancing website fingerprinting attack with relation features and NFS-CNN

Hongcheng Zou, Ziling Wei, Jinshu Su, Shuhui Chen, Zheng Qin

Abstract: Highlights•We first propose relation features from Tor traffic.•We design NFS-CNN, a deep learning model, which can reduce the information loss.•Based on NFS-CNN, we propose Relation-CNN, a new website fingerprinting attack.•Results show that relation features reveal unique discriminative information.•We also validate the superiority of Relation-CNN via comprehensive evaluations.