K-Fingerprinting - A Robust Scalable Website Fingerprinting Technique

Explore a conference talk on k-fingerprinting, a robust and scalable website fingerprinting technique presented at USENIX Security '16. Delve into the research by Jamie Hayes and George Danezis from University College London, which introduces a novel approach based on random decision forests. Learn how this technique outperforms current state-of-the-art attacks, even against website fingerprinting defenses, and its effectiveness in handling large amounts of noisy data. Discover the impressive accuracy rates achieved in identifying monitored hidden services and understand the varying vulnerability of different web resources to this attack. Gain insights into the methodology, data collection process, and limitations of k-fingerprinting, as well as its implications for encrypted and anonymized network connections.

Introduction
Contributions
Features
How does it work
Base Rate
Accuracy Metrics
How kfingerprinting works
Data collection
Accuracy
Alexa
Hidden Service
Limitations
Conclusion
Interview