Distribution of inter-contact time: An analysis-based on social relationships

Communication in delay tolerant networks (DTNs) relies on message transport by mobile nodes, and a correct understanding of the node mobility characteristics is therefore crucial to the design of an efficient DTN routing protocol. However, previous work has mainly focused on uncovering all behaviors of node movement, which is not conducive to accurately detecting the specific movement characteristics of a different node. In this paper, we seek to address this problem based on a consideration of social relationships. We first consider social ties from both static and dynamic perspectives. For a static perspective, in addition to certain accidental events, social relations are considered for a long time granularity and tend to be stable over time. For a dynamic perspective, social relations are analyzed in a relatively short time granularity and are likely to change over time. Based on these perspectives, we adopted different efficient approaches to dividing node pairs into two classes, i.e., familiar and unfamiliar pairs. A threshold approach is used for static social ties whereas a density-based aggregation method is used for dynamic social relationships. Extensive experimental results show that both familiar and unfamiliar node pairs have the same inter-contact time distribution, which closely follows a power-law decay up to a certain point, beyond which it begins to exponentially decay. The results also demonstrate that the inter-contact time distribution of familiar pairs decays faster than that of unfamiliar pairs, whether from a static or dynamic perspective. In addition, we also analyze the reason for the difference between the inter-contact time distributions of both unfamiliar and familiar pairs.