Unobservable surfing on the world wide web: Is private information retrieval an alternative to the MIX based approach?

Abstract

The technique Private Information Retrieval (PIR) perfectly protects a user’s access pattern to a database. An attacker cannot observe (or determine) which data element is requested by a user and so cannot deduce the interest of the user. We discuss the application of PIR on the World Wide Web and compare it to the MIX approach. We demonstrate particularly that in this context the method does not provide perfect security, and we give a mathematical model for the amount of information an attacker could obtain. We provide an extension of the method under which perfect security can still be achieved.