Reproduce “STAR: Semantic-Traffic Alignment and Retrieval for Zero-Shot HTTPS Website Fingerprinting” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yifei Cheng, Yujia Zhu, Baiyang Li, Xinhao Deng, Yitong Cai, Yaochen Ren, Qingyun Liu
STAR introduces a zero-shot HTTPS website fingerprinting method that reformulates the problem as a cross-modal retrieval task, aligning encrypted traffic patterns with crawl-time semantic website profiles. The system achieved 87.87% top-1 accuracy for identifying 1,600 unseen websites without prior labeled traffic and demonstrated robust open-world detection with an AUC of 0.963.
Key insights
- "Semantic leakage" through encrypted traffic patterns, specifically the structural characteristics of website resource loading, remains a dominant privacy vulnerability despite advanced HTTPS mechanisms.
- Encrypted traffic and semantic website profiles exhibit robust "alignment anchors," such as correlations between request packet lengths and URI sizes or cumulative response sizes and resource content volumes.
- Website fingerprinting can be successfully reframed as a zero-shot cross-modal retrieval task, enabling the identification of websites without any prior labeled traffic data for those specific sites.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.