Preventing denial-of-request inference attacks in location-sharing services

Location-sharing services (LSSs), such as Google Latitude, have been popular recently. However, location information is sensitive and access to it must be controlled carefully. We previously study an inference problem against an adversary who performs inference based on a Markov model that represents a user´s mobility patterns. However, the Markov model does not capture the fact that a denial of a request enforced by the LSS itself implies that a target user is visiting some private location. In this paper, we develop an algorithmic model for representing this new class of inference attacks and conduct experiments with a real location dataset to show that threats posed by the denial-of-request inference attacks are significantly real.