Activity tracking: A new attack on location privacy

The exposure of location information in location-based services (LBS) raises users' privacy concerns. Recent research reveals that in LBSs users concern more about the activities that they have performed than the places that they have visited. In this paper, we propose a new attack with which the adversary can accurately infer users' activities. Compared to existing attacks, our attack provides the adversary not only with the places where users perform activities but also with the information when they stay at each of these places. To achieve this objective, we propose a new model to capture users' mobility and their LBS requests in continuous time, which naturally expresses users' behaviour in LBSs.We then formally implement our attack by extending an existing framework for quantifying location privacy. Through experiments on a real-life dataset, we show the effectiveness of our new tracking attack.