GridPerturbator: A Proļ¬le-aware Location Privacy Preserving Platform

Introduction


With the explosive growth in the sales of Android Smart- phones, people's daily life is firmly connected with this device. When using the Smart-phones, people's location information can be easily leaked out. People's location can be also passively retrieved by the ad libraries which can not be trusted. Since human's mobility trace is regular in both temporal and spatial domain, people can be reidentified if their location information, especially their "top N" locations are leaked out. There need a good mechanism in Android system which can preserve user's location-privacy and at the same time meet their needs of location-based service.

In our work, we add a layer between the application and the location requesting API in Android system. Our platform can provide a overall management of location requests made by the Android application. By automatically learning the mobility trace of the user, our platform can hide the user's top location by reducing the granularity in both spatial and temporal domain.

Demo


Learning the Top Location

User's "Top Location" informaiton is very sensitive since human's mobility trace is regular in both temporal and spatial domain, people can be reidentified if their location information, especially their "top N" locations are leaked out. Our platform can automatically learn the user's mobility pattern and then give obfuscation in both spatial and temporal domain, which can protect the user's privacy.

Fig.1 User's Mobility Trace

Fig.2 Top Location Learned by GridPerturbator

Fig.1 shows a user's mobility trace. From the mobility trace, we can easily get this user's Top Location by clustering. Fig.2 shows the result of the cluster algorithm . User's top 3 location in Fig.2 gives us many useful information about the user. In this example, the top 1 location stands for the working place of the user. The top 2 location stands for the living place of the user. The top 3 location is a library, which can reveal the usr's lifestyle. Combining the top location information with some other side information, the attacker can reidentify the user and continue to track the user to acquire more sensitive personal information.

GridPerturbator can reduce the granularity in both spatial and temporal domain so as to obfuscate user's mobility trace and protect user's privacy

This is an on-going project, more information to be continued...