PACOM: Parasitic anonymous communication in the BitTorrent network

Abstract

Existing anonymous communication systems mask the identities of users by adopting intermediary nodes to transform message flows. However, some recently presented traffic analysis algorithms are still able to undermine the anonymity of these systems. The traditional flow transformation strategies fail to completely eliminate the traffic correlation between adjacent communication links to prevent such attacks. To address this problem, we propose a novel parasitic anonymous communication system, named PACOM. Each PACOM client is parasitic in the BitTorrent network which is the most popular Peer-to-Peer file sharing network, and conceals the communication path in the request driven traffic compatible with the BitTorrent protocol. The traffic patterns of adjacent communication links can be proved to be statistically independent, which effectively resists the traffic analysis attacks. Meanwhile, the “effective anonymity set size” of the system can be extended enormously by mixing the PACOM clients with other millions of BitTorrent clients in the Internet. To validate the PACOM solution, we analyse the anonymity of PACOM theoretically and conduct comprehensive simulations and emulations to test the scalability and effectiveness of PACOM against various attacks.