Intersection Attacks on Discrete Epochs

Abstract

Anonymous messaging systems with churn in the set of online users are vulnerable to intersection attacks. Researchers have evaluated the success of the state of the art intersection attack using a model of user messaging simulated from a generated social graph. This thesis compares the success of the state of the art intersection attack using a model simulated from a generated social graph versus models simulated from real social graphs, such as those of Twitter and Google+. We find that users lose anonymity at a slower rate if the model uses a real social graph rather than a generated social graph.