Carmela Troncoso is an assistant professor at the École Polytechnique Fédérale De Lausanne EPFL
Information about whereabouts is often used to support smart services and applications, e.g., generating live traffic maps or predicting visits to businesses. In this setting, rather than collecting or sharing raw data, entities often use aggregation as a privacy protection mechanism, aiming to hide individuals’ location traces.
In this talk, I will present our work evaluating the use of aggregation as a privacy-preserving mechanism for location time series. I will focus mainly on three attacks: profiling (how much aggregates reveal about users’ movement patterns), localization (how much aggregates reveal about users’ punctual locations), and membership (how much aggregates reveal about which users contribute to the published statistics). I will also show that using typical differential privacy mechanisms to enhance privacy is ineffective at protecting users against these attacks.