| Description |
Many networked users and devices are interested in participating in distributed sensing and data collection for the purpose of the betterment of human society or for earning monetary rewards. Preservation of their privacy is an important requirement for users participating and contributing to the data collection. In this thesis, we develop a novel privacy preserving approach for collecting noise-free data from vehicular users. Collection of noise-free "pure" data enhances utility of applications that use it. We collect the data in a noise-free form by introducing temporal and spatial variations using random delays and indirections. We assume that the node collecting the user data, which we call the central controller, is adversarial and wants to localize the users/vehicles providing the data. We formulate an adversary attack and show that the adversary can localize a casual user based on multiple uploads made by it. Next, we propose steps that the participating users should take so that the adversary is unable to localize them. We run simulations on a custom toy map and a real traffic scenario in the city of Luxembourg to evaluate our approach. Our results show that for a particular set of indirection probability and random delays, the adversary can localize a few users within 500 meters. However, if the participating users only make uploads to the server following our approach, then we can significantly increase the adversary's error in localizing the participating users. |
