Preserving Location Privacy on the Release of Large-scale Mobility Data

Mobility models play an important role in wireless network simulation. While being widely used due to simplicity, synthetic models usually suffer from the inadequate semantics to characterize real-world movements. In contrast, traces are highly desirable for simulation since they are extracted from realistic movements. However, even releasing anonymized traces could potentially cause privacy exposure. To tackle this dilemma, our paper proposes a novel approach to produce mobility traces while still preserving location privacy. Our algorithm depends only on certain wireless relationships observed in a large-scale mobile dataset collected in campus, instead of using any of the actual location information for trace generation. We argue that wireless relationships rather than geo-locations are the critical aspects to preserve in mobility patterns. A set of metrics are applied to evaluate the performance of the proposed approach in terms of preserving the original wireless relationships in the output traces, demonstrating promising initial results.