Reducing the Overhead of Multicast Using Social Features in Mobile Opportunistic Networks

Abstract

Through utilizing user mobility and short-range device-to-device communication techniques, mobile opportunistic networks (MONs) enable end-to-end message delivery without the dependence on reliable network infrastructures. Examples of these networks include mobile social networks and vehicular networks. Multicast in MONs is used to disseminate information to a group of mobile nodes, which have attracted considerable attention due to high resource utilization. In this paper, we focus on reducing the overhead of multicast in MONs without compromising the delivery performance, through utilizing static social features of nodes and time-varying social behaviors. We first conduct a trace data analysis using the information entropy theory to identify the most important and representative social features in a popular trace, the Infocom06 trace. Based on these static social features, we propose a Social Profile-based Multicast (SPM) routing scheme, that supports efficient multicast message dissemination with a small maintenance overhead, i.e., little cost on maintaining the historical records. Furthermore, by exploring the time-varying social behaviors during daytime and nighttime, we verify that a small number of forwardings during that daytime is sufficient to achieve a desirable delivery ratio. We thus propose an improved overhead-reducing scheme Social Profile-based Multicast-Overhead Reducing (SPMOR) that restricts the number of forwardings during the daytime. The extensive trace-driven simulations show that SPM achieves desirable delivery performance with small maintenance and transmission cost, and SPMOR can further reduce the transmission overhead under diurnal user behaviors. At last, we conduct a similar study on a campus-based MON trace, SocialBlueConn. We find that the main conclusions and performance results on SocialBlueConn are consistent with the Infocom06 trace, which verifies our methodology is scalable.