Research On Analysis Method Of Social Characteristics In Opportunistic Networks Based-on Map Reduce

In recent years, with the popularity of a large number low cost mobile devices having the capability of short distance wireless communication, the development of opportunistic networks,which data exchange are characterized by the encounter opportunities through the human carrier of mobile devices, has a significant impact for the future pervasive computing. Therefore, the study for massive node data attributes in opportunistic networks has become more and more important for enterprises and users.Currently, the study of the social characteristics of node data in social networks is the focus of the research areas in the world. However, the study for the social characteristics of opportunistic networks is rare. The social characteristics analysis for mass node data of networking opportunities exist the following problems: the problem of too large data set and low efficiency for analysis and process massive node data in opportunistic networks; the problem of the non-standard format and redundant data in using specific calculation model for dealing with massive data sets; and the traditional serial computation method of social analysis algorithm is inefficient, and unable to meet the demand on the performance of the algorithm.To quickly and efficiently analyze the social characteristics of massive node data in opportunistic networks, in this thesis, we research the above three problems and put forward the solutions: For the performance problems of analysis in opportunistic networks caused by the explosive growth of the massive nodes data, our proposal is to analyze the massive data using the MapReduce parallel computing technology, where the opportunistic network will be abstracted into a graph, nodes and the relationship of the graph were standardized modeling. Using the MapReduce framework for graph data processing can effectively enhance the operation performance and efficiency; For the problem of the specification of data format in social characteristic analysis in opportunistic networks based on MapReduce, we design and implementation the data pre processing algorithm based on MapReduce, which can extract the needed data information from the original data of opportunistic networks and perform format conversion and redundancy removement to provide a standard format of the data sets for social characteristics analysis; For the problem of seriously low performance in the traditional serial computation method of social characteristics analysis algorithm, we present two improved parallel connected component analysis algorithms: the Label Propagation Algorithm based on MapReduce(LPA-MR) for non directed graph connectedcomponents and the Bidirectional and Synchronous Label Propagation algorithm based on MapReduce(BSLPA-MR) for strongly connected components, and the experimental tests show that the two algorithms reduces the computational time and improve the computational efficiency through the parallel design, thus can provide the reliable algorithm in society characteristics analysis of the mass of node information in opportunistic networks.Finally, a analysis system prototype of social characteristics of opportunity network based on MapReduce is designed and implemented, and the model and algorithm mentioned above are applied in the system. The experimental results show that the introduction of the MapReduce computation model and performance optimization of the social characteristics analysis algorithm are not only greatly improves the processing efficiency of the whole system, but also provides a reliable and effective method for the social characteristics analysis in the opportunistic networks.