Section: New Results
Data anonymization
Participants : Claude Castelluccia, Gergely Acs.
With billions of handsets in use worldwide, the quantity of mobility data is gigantic. When aggregated they can help understand complex processes, such as the spread viruses, and built better transportation systems, prevent traffic congestion. While the benefits provided by these datasets are indisputable, they unfortunately pose a considerable threat to location privacy. At KDD 2014 [9] , we present a new anonymization scheme to release the spatio-temporal density of Paris, in France, i.e., the number of individuals in 989 different areas of the city released every hour over a whole week. The density is computed from a call-data-record (CDR) dataset, provided by the French Telecom operator Orange, containing the CDR of roughly 2 million users over one week. Our scheme is differential private, and hence, provides provable privacy guarantee to each individual in the dataset. Our main goal with this case study is to show that, even with large dimensional sensitive data, differential privacy can provide practical utility with meaningful privacy guarantee, if the anonymization scheme is carefully designed. This work is part of the national project XData (http://xdata.fr ) that aims at combining large (anonymized) datasets provided by different service providers (telecom, electricity, water management, postal service, etc.).