It Leaks More Than You Think: Fingerprinting Users from Web Traffic Analysis

Acta Informatica Pragensia

Xujing Huang¹ 

[1] School of Electronic Engineering and Computer Science,Queen Mary University ofLondon;

关键词: Side-channel leakages;    User identities;    Web applications;    Google accounts;   

DOI  :  10.18267/j.aip.70

来源: DOAJ

【 摘 要 】

We show how, in real-world web applications, confidential information about user identities can be leaked through “non-intuitive communications”, in particular web traffic which appear to be not related to the user information. In fact, our experiments on Google users demonstrate that even Google accounts are vulnerable on traffic attacks against user identities, using packet sizes and directions. And this work shows this kind of non-intuitive communication can leak even more information about user identities than the traffic explicitly using confidential information. Our work highlights possible side-channel leakage through cookies and more generally discovers fingerprints in web traffic which can improve the probability of correctly guessing a user identity. Our analysis is motivated by Hidden Markov Model, distance metric and guessing probability to analyse and evaluate these side-channel vulnerabilities.

【 授权许可】

Unknown