Byzantine attacks on anonymity systems

In the study of anonymous systems, research into new designs for anonymity has been balanced by analysis of attack strategies for violating anonymity. Such strategies shed light on how well systems will defend users’ privacy and suggest new directions for research and development. Any analysis of attacks, however, must be parameterized by a model of adversary capabilities. The choice of model has been evolving over the years; for example, as the scale of anonymous systems has grown, the popular global passive adversary becomes less realistic and a more limited adversary may be more appropriate [36]. At the same time, with more loose dynamics connecting the participants of today’s anonymous networks, the likelihood that a large number of participants may be compromised and colluding together is perhaps higher than it used to be. This motivates deeper study of attacks that such participating adversaries can pose.