Reaserch On Data Transmission Mechanism In Mobile Social Network

Recently, with the rapid development of smart portable devices (e.g., smartphones, tablets, wearable devices and so on), and the improvement of short-distance communication capability, mobile social networks (MSNs) has caused extensive concern. Without the constraint of real-time connectivity and additional extern communication infrastructure, MSNs can form communication networks at any time and place. Therefore, they can be widely applied in handheld device network forming, vehicular networks, and providing network services in undeveloped areas. And MSNs have far-reaching impacts ont the realization of new type communication network in the future.Although previous researchers have done a lot of work in MSNs, most of them use the ag-gregated graph model and ignore the dynamic change of MSNs. To overcome this limitation, this thesis introduce the time-ordered graph to model the MSNs, studying the network change of MSNs in a dynamic view and mining the changing pattern for the better application of MSNs. The main work and contribution are summarized as follows:The time-ordered graph is introduced in this thesis to model the MSNs and real trace data is used to analysis the network. The MSNs is found to have the characteristics of persistence and periodicity, after analyzing the network change of different time in one day or one week, from perspectives of network structure similarity, community similarity and vertex similarity.The temporal contact pattern between nodes in MSNs is explored. And two distributed met-rics, temporal closeness and temporal centrality, are first proposed to characterize the temporal importance of nodes in the network. Based on the dynamic characteristics of MSNs, a prediction mechanism is designed and evaluated to estimate the node importance in the future time. And final-ly, a new data forwarding strategy based on the temporal contact pattern is put forward. Evaluation results ensure the efficiency of the proposed strategy.The quickest diffusion problem is also considered in MSNs. By using the characteristic of network community structure that nodes inside the community contact each other more frequent-ly than nodes belong to different communities, a community based diffusion mechanism is first proposed. Meanwhile, considering the dynamic instinct of the MSNs, the community changing pattern is explored. Then at last, the thesis put forward a more accurate diffusion mechanism based on temporal community. Evaluation results ensure the efficiency of the proposed diffusion mechanism.