Interests And Behavior Prediction-based Dynamic Resource Discovery Mechanism For Mobile Social Networking

With the popularization of portable sets(mobile phone, PAD, laptop) and full-blown application of short-range communication technology, more and more people spontaneously organize themselves into a disconnected and delay-tolerant Ad Hoc network which is called Mobile Social Networks(MSN). In this network, the users with similar interest assemble together to form a community. Meanwhile, the nodes do not depend on external infrastructure to communicate with each other, but share information in the form of end-to-end.However, the network property of MSN leads the logical relationship and the underlying physical connection between different nodes to be the critical elements which influence the resource discovery efficiency. Combined with the interest feature and behavior regularity between different nodes, the resource explore process dominated by interest and the guarantee of the underlying physical connection between different nodes is of great significance to further increase the resource discovery efficiency.Therefore, this thesis focuses on the behavior rules of time, space and social relation between different nodes according to resource discovery in disconnected delay-tolerant mobile social networks. Moreover, we proposed an Interests and Behavior prediction-based dynamic Resource Discovery mechanism(IBRD). The main contents of this research can be concluded as follows.(1) Making a conclusion about different research discovery methods under the current social network, including method principle, merits and demerits; specifically discussing the critical technology in these methods.(2) Analyzing on time, space and social relation according to the mobile social network data set to acquire user behavior properties and to deeply investigate the influences on research discovery efficiency generated by users’ behavior in mobile social network.(3) Utilizing the user behavior properties to set up a valid recessive Markov model and a construction method based on temporal and spatial association of community to cluster nodes.(4) Designing two types resource explore methods to ensure the searching efficiency the average delay and the communication cost according to the two kinds of nodes about whether shares interests.(5) Building up the prediction results of the Markov model according to the users behavior properties; realizing the assistant searching strategies and the dynamic maintenances of the virtual interest communities; further raising the resource explore efficiency and decreasing the average delay and the communication cost.The opportunistic network environment(ONE) has been used as the simulation platform in order to evaluate the performances of our scheme. The experiment is conducted from simulated scene and Real data set, and is assessed by the research discovery efficiency, the transmission delay, and the communication cost. Simulation results show that our proposed scheme consistently outperforms the state-of-the-art resource discovery schemes in terms of the searching efficiency, the average delay and the communication cost.