Abstract

The wide adoption of Web applications in various sectors of our society, such as government, finance, education, health care, media, etc., has implicitly introduced new security
challenges. Among such challenges are side channel attacks that may disclose private user inputs from encrypted raffic. Such attacks might have a serious impact upon user privacy in such applications. In this thesis, we propose a new concept and algorithms that can preserve user privacy in Web applications. In order to achieve this, we define a new privacy
model based on a well known concept, namely, differential privacy. The intent is to make
padded traffic differentially private such that adversaries cannot infer private user inputs
even when they possess prior knowlege about such inputs. At the same time, we intent
to achieve a balance bewteen privacy and the incurred communication overhead. In order
to demonstrate the usefulness of our model, we implement the proposed algorithms and
conduct experiments based on data collected from well known Web applications.