Collaborative Research: Anonymous Protocols

Purdue University, West Lafayette IN

Investigators

Abstract

The purpose of this work is to support the understanding and development of protocols that can provide personal privacy over the Internet. While there have been some protocols proposed and implemented that hide the IP address of network communicators, there has been no systematic study of such protocols, nor have methods of analyzing the anonymity and security of those protocols been developed. A thorough understanding of anonymity is important not only for the development, analysis, and use of anonymous protocols for personal privacy, but also for understanding how network attackers can hide their location and for developing methods to track those attackers. Our eventual goal is to develop new techniques and protocols for anonymity, including a \logic of anonymity" similar to those developed for authentication protocols that can be used to describe and analyze anonymous protocols. This proposal considers a number of contributions on this research area. Our first contri- bution will be to take a more general view of anonymous communication than previous work by considering protocols and techniques for hiding the identity of a responder from all other parties, and for providing completely anonymous communication in which both the initiator and responder are anonymous to each other and all other parties. The former technique is useful for anonymous web sites (rather than just anonymous web browsing); the later technique is useful for true two-party privacy over the Internet. Our second class of contributions will be to consider generalized protocol construction. This extends our past work, which de-coupled the forward and reverse paths of anonymous communication, giving rise to a large family of anony- mous protocols. We will continue investigation of those protocols, and examine what network support is available for anonymous communication and what support might best be provided in the future. Our third contribution will be to provide a formalism for anonymous protocols such that beliefs regarding entity identities can be reasoned by stating assumptions and deriving log- ical conclusions from following steps of defined protocols. Fourth, we will consider the network performance of anonymous routing protocols in an in-depth fashion, including not only latency and traffic, but also requirements of streaming media applications and traditional TCP-based bulk-data transfer applications. Finally, we will consider heretofore ignored security concerns of existing protocols, initially considering path analysis attacks and attacks that cause some participants to be unfairly accused by third parties. The formalism provided by our proposed logic of anonymity will be a foundational component of our security analysis work. Collaboration between these our two institutions will allow us to leverage the existing strength in networking at the University of Massachusetts and the existing strength in secu- rity at Purdue. 1