Give Me Two Points and I'll Tell You Who You Are

Vehicular mobility traces are datasets of vehicle location records in a region with high spatio-temporal precision. The access to this sensitive information can threaten the safety and privacy of drivers, given that the analysis of this data makes it possible to discover other contextual and latent information, such as daily home routes or workplace's address. To prevent this, many obfuscation and anonymization techniques have been proposed to mitigate the problem of user location privacy. In this work, we analyze an anonymization technique called mix-zone, which selects urban regions that promote the simultaneous anonymization of vehicles by changing their current pseudonym. We show how information about drivers' behavior in a city, such as their road preferences, can be used to reidentify their trajectories. To do this, we present a simple and efficient re-identification technique that uses only two geo-referenced points as input data. We validate our technique with a real dataset of taxicabs, being able to re-identify up to 100% of the anonymized trajectories.