Application of Data Analytics: Case Studie

Data analytics is the technique of finding knowledge by examining raw data. It is an important tool for researchers to verify existing knowledge or infer new knowledge. In this dissertation, we focus on anonymous traffic and privacy-aware systems. Our research is divided into three data analytics case studies. We use data analytics to learn from and improve existing systems. Tor, an anonymous network, is designed to protect Internet users from traffic analysis attacks. Researchers have shown that traffic analysis like timing attack and website fingerprinting attack are still realistic and can be used to deanonymize Tor users. We first analyze the anonymity of Tor itself; we show that a timing attack can be used to bypass the anonymity provided by Tor. We also propose a schema to identify this type of timing attack. Our second case study is about website fingerprinting. We propose a new realistic cover traffic algorithm to mitigate website fingerprinting attacks. Our algorithm reduces the accuracy of website fingerprinting attacks to 14% with zero latency overhead and 20% bandwidth overhead. Our third case study is about webbrowser fingerprinting in anonymous communications. We analyze the network traffic generated by web browsers and show that features of webbrowsers can be inferred with high probability.