Protecting location privacy in online and offline contexts

点击翻译按钮获取中文摘要

The advent of Location-Based Services (LBSs), which personalize the information provided according to the position of their users (e.g., geolocated search), has been accompanied by the large-scale collection of their mobility data. On the one hand, these mobility datasets have a high scientific, societal and economical value. On the other hand, learning the location of an individual is one of the greatest threats against his/her privacy due to its strong inference potential and the possibility of deriving a wealth of personal information. In particular in the past, I have designed inference attacks that use the location data of a user to deduce other personal information (such as the points of interests characterizing his/her mobility), to predict his/her future movements or even to perform a de-anonymization attack. The scope of my research program covers two different contexts in which the location privacy of a user should be protected. The first context corresponds to the situation in which the user is online (i.e., when he/she benefits from a location-based service in real-time). In this setting, I propose to investigate two different approaches whose objective is to enable privacy-preserving LBSs to operate while minimizing the trust assumptions: the local computation approach and the cooperative one. The second context considered is the offline setting, in which the location data of thousands of users has been collected and has to be sanitized before it is released (e.g., before opening or sharing this data). More precisely, during my discovery grant I propose to work on the design of sanitization methods for mobility mining, whose objective is to produce a data structure that can be used to derive generic mobility patterns of the population while hiding individual movements. Finally at the fundamental level, I am deeply interested in how to model and quantify location privacy in a manner that is both meaningful and useful for practitioners who need to assess the privacy risks of processing, sharing and collecting location data. Thus I propose to study how to integrate the semantic dimension in the currently existing location privacy models. The societal impact of my research program’s outcomes can be important, as they have the potential to improve significantly the privacy situation of users of LBSs. In addition, the solutions developed will act as enablers by helping Canadian companies to implement privacy-preserving LBS. In particular, a major social and economic challenge is to foster the development of LBS while providing sufficient privacy guarantees. Thus, privacy-preserving LBS have to be developed to avoid the transformation of Big Data into Big Brother, and the results of my research program will directly contribute to this. Finally, the research conducted will be done in cooperation with and contribute to the formation of HQP (i.e., PhD and master students).