Tracking Groups in Mobile Network Traces

Abstract

Detecting and tracking groups in mobility network traces is critical for developing accurate mobility models, which in turn are needed for mobile/wireless network design. One approach is to represent mobility traces as a temporal network and apply group (community) detection algorithms to it. However, observing detailed changes in a group over time requires analyzing group dynamics at small time scales and introduces two challenges: (a) group connectivity may be too sparse for group detection; and (b) tracking evolving groups and their lifetimes is difficult. We proposes a group detection framework to address these time scale challenges. For the time-dependent aspect of the groups, we propose a time series segmentation algorithm to detect their formations, dissolutions, and lifetimes. We generate synthetic datasets for mobile networks and use real-world datasets to test our method against state-of-the-art. The results show that our proposed approach achieves more accurate fine-grained group detection than competing methods.