Predicting Temporal Social Contact Patterns for Data Forwarding in Opportunistic Mobile Networks

In this paper, we predict nodes' social contact patterns from the temporal perspective, and propose a novel approach to improve the performance of data forwarding in opportunistic mobile networks (OMNs). Specifically, considering both the average separating time and the variance of the separating time, we first introduce the definition of temporal closeness and temporal centrality. Then, several intuitive prediction methods are designed to predict nodes' future temporal social contact patterns based on the observations from extensive real trace-driven simulation results. Afterward, based on the predicted temporal social contact patterns, we propose an efficient temporal closeness and centrality-based data forwarding strategy named TCCB for OMNs. The core idea of TCCB is to capture and utilize the temporal correlations to infer the future temporal social contact patterns in the remaining valid time of the data. Finally, extensive real trace-driven simulations are conducted to evaluate the performance of TCCB. The results show that TCCB is close to Epidemic in terms of delivery ratio but with significantly reduced delivery cost. Furthermore, TCCB performs much better than Bubble Rap and Prophet in terms of delivery ratio, but the delivery cost of TCCB is very close to that of Bubble Rap.