Dummy traffic against long term intersection attacks

Abstract

In this paper we propose a method to prevent so called "intersection attacks" on anonymity services. Intersection attacks are possible if not all users of such a service are active all the time and part of the transfered messages are linkable. Especially in real systems, the group of users (anonymity set) will change over time due to online and off-line periods. Our proposed solution is to send pregenerated dummy messages to the communication partner (e.g. the web server), during the user’s off-line periods. For a detailed description of our method we assume a cascade of Chaumian MIXes as anonymity service and respect and fulfill the MIX attacker model.