SCSM: A Traffic-Native Foundation Model for Transferable Website Fingerprinting

2026-10-06

AI Summary

This study addresses the performance degradation of website fingerprinting under environmental shifts and the inability of existing methods to capture native traffic semantics. We pioneer a traffic-native pretraining paradigm by proposing a transferable foundation model. Rather than relying on manual perturbations or transplanted language architectures, our approach takes windowed traffic count matrices as input, constructs pretraining views through segment composition, and trains a state-space encoder via unsupervised contrastive learning. Combined with multi-timescale inference aggregation, this method effectively preserves authentic traffic dynamics. Experimental results demonstrate that the proposed model outperforms the strongest baseline on the GTT23 dataset by 16.4% in average Top-3 accuracy and achieves a 13.8% improvement in cross-domain Macro-F1 score.

Abstract

Website fingerprinting infers the websites visited by users from encrypted traffic metadata. However, models trained under fixed collection conditions often degrade as website sets, collection times, network paths, browsers, or defenses change. Existing transferable attacks either rely on handcrafted perturbations of individual traces or adapt language-oriented architectures to traffic, limiting their ability to capture traffic-native semantics. To address these limitations, we propose SCSM, a traffic-native foundation model for transferable website fingerprinting. Specifically, SCSM constructs pairs of pretraining views from the same group of unlabeled traces through Segmentation, Combination, Scaling, and Masking. These operations produce diverse observable patterns while preserving the underlying packet events and local traffic dynamics of real trace fragments. The corresponding windowed traffic counting matrices serve as inputs for contrastive pretraining of a state-space encoder without website annotations. The pretrained encoder is then fine-tuned on a small labeled support set, and predictions are aggregated across temporal scales at inference. Experimental results demonstrate that SCSM surpasses the strongest baselines by 16.4\% in average top-3 accuracy over six temporal-drift tasks on GTT23 and by 13.8\% in average macro-F1 over four cross-domain datasets. The code and datasets will be made available at https://github.com/SJTU-dxw/WF-SCSM.

Problem

• Website Fingerprinting
• Transferable Attack
• Encrypted Traffic
• Traffic-Native Semantics
• Domain Adaptation

Innovation

• Website Fingerprinting
• Foundation Model
• Contrastive Pretraining
• State-Space Model
• Transfer Learning