Performance of side-channel attacks on encrypted 802.11 web traffic and impact of network-level noise

Abstract

In today's world, Wi-Fi networks are widespread and intertwined with our daily lives: we access them through our smartphones, our tablets and our laptops. This means that inevitably a lot of sensitive data is transferred through these networks. Encryption, which is present in virtually every Wi-Fi network, helps protect this data. Nonetheless, through side-channel attacks it is possible to obtain information and potentially breach a user's privacy. This thesis has the objective of devising such an attack, in order to identify the website a given user is accessing. A solid dataset, consisting of numerous captures pertaining to a specific website, provides the means for an algorithm based on machine learning to classify a traffic capture. Another objective is to study the effect of network-level noise on the accuracy of the algorithm.