Reality Mining with Mobile Data: Understanding the Impact of Network Structure on Propagation Dynamics

Recent studies have increasingly turned to graph theory to model Realistic Contact Networks (RCNs) for characterizing propagation dynamics. Several of these studies have demonstrated that RCNs are best described as having exponential degree distributions. In this article, based on the mobile data gathered from in-vehicle wireless devices, we show that RCNs do not always have exponential degree distributions, especially in dynamic environments. On this basis, a model is designed to recognize the structure of networks. Based on the model, we investigate the impacts of network structure on disease dynamics that is an important empirical study to the propagation dynamics. The time-varying infected number R is the important parameter that is used to quantify the disease dynamics. In this study, the prediction accuracy for R is improved by utilizing realistic structural knowledge mined by our recognition model.