Website Fingerprinting Attacks and Defense Techniques: A Survey

Abstract

Anonymity networks like Tor protect the end users privacy by hiding the browsing activity. However, this protection is often abused for online activities which are not legal. We find works in the literature, which tend to reveal the identities of users with advanced traffic analysis. Contrary to these, there are also works which thwart such traffic analysis to protect users’ identities. The first class of work is known as website fingerprinting (WF) and mainly rely on machine learning and deep learning algorithms to analyze encrypted traffic. The second class of work has several defense mechanisms to counter website fingerprinting attacks. In this article, we provide an in-depth analysis of both website fingerprinting attacks and defenses covering recent advancements in the domain. First, we look at WF attacks by dividing them into two groups: those using traditional machine learning techniques, and the others using deep learning models. Next, we provide detailed coverage of defense mechanisms. We also cover details of publicly available datasets, commonly used evaluation metrics for assessing the robustness of the WF, and experimental tools used for traffic analysis. Finally, we highlight some important research gaps that need to be filled to make progress towards designing robust attack frameworks. © 2026 Copyright held by the owner/author(s).