Achieving Sender Anonymity in Tor against the Global Passive Adversary

Abstract

Tor is the de facto standard used for anonymous communication over the Internet. Despite its wide usage, Tor does not guarantee sender anonymity, even in a threat model in which the attacker passively observes the traffic at the first Tor router. In a more severe threat model, in which the adversary can perform traffic analysis on the first and last Tor routers, relationship anonymity is also broken. In this paper, we propose a new protocol extending Tor to achieve sender anonymity (and then relationship anonymity) in the most severe threat model, allowing a global passive adversary to monitor all of the traffic in the network. We compare our proposal with Tor through the lens of security in an incremental threat model. The experimental validation shows that the price we have to pay in terms of network performance is tolerable.