LASERBEAK: Evolving Website Fingerprinting Attacks with Attention and Multi-Channel Feature Representation
Published in IEEE Transactions on Information Forensics and Security (TIFS), 2024
In this paper, we present LASERBEAK, a state-of-the-art website fingerprinting attack targeting Tor traffic. LASERBEAK achieves nearly 96% accuracy against traffic obfuscated by the FRONT defense by combining multi-channel traffic representations and advanced attention mechanisms inspired by computer vision models. The attack significantly outperforms prior attacks, demonstrating a 36.2% improvement against heavily defended traffic. Furthermore, LASERBEAK maintains high precision and recall even in a large open-world scenario, achieving over 80% recall at 99% precision on obfuscated traffic. This work underscores the critical need for stronger traffic analysis defenses in anonymity networks like Tor.
Recommended citation: Nate Mathews, James K. Holland, Nicholas Hopper, Matthew Wright. (2024). "LASERBEAK: Evolving Website Fingerprinting Attacks with Attention and Multi-Channel Feature Representation." IEEE Transactions on Information Forensics and Security (TIFS), 2024. DOI: 10.1109/TIFS.2024.3468171.
Download Paper