移动机会网络中的大数据路由理论和方法及应用基础研究

本项目旨在研究移动机会网络中的大数据路由问题。我们将突破用静态图理论研究网络数据路由问题的传统思路，把移动机会网络拓扑结构动态演化建模作为研究切入点，从时间和空间统一建模的观点,建立移动机会网络拓扑结构动态演化的时空图模型，用来准确刻画移动机会网络的"存储-携带-转发"数据路由新模式；提出移动机会网络中的大数据分块路由方法,以提高移动机会网络大数据路由效率；研究移动机会网络数据路由的不确定性和时间序列单跳决策特点，运用马尔科夫决策理论, 建立移动机会网络的数据路由路径选择不确定决策模型；综合考虑节点移动性、剩余能量和通信链路稳定性，运用无监督强化学习方法，设计分布式、自适应、高效数据路由路径选择决策算法；以移动学习机会网络中的大数据路由为背景，开展应用研究。最终为移动机会网络中的大数据路由建立新的理论和方法基础，其研究具有重要的理论意义和广阔的应用前景。

This project aims to study the issue of big data routing in mobile opportunistic networks. Firstly,we will break through the traditional idea of using static graph theory to study the problem of data routing in networks, with the modeling of the topology structure dynamic evolution of mobile opportunistic network as our study starting point, we propose a spacial-temporal graph model of the topology structure dynamic evolution of mobile opportunistic networks from the unified modeling perspective of time and space, and our purpose is to use the spacial-temporal graph model to precisely describe the new data routing scheme of storage-carry-forward in mobile opportunistic networks. Secondly，we will propose a new big data block routing method to improve the big data routing efficiency in mobile opportunistic networks. Thirdly, we will study the uncertainty and the time series characteristics of single hop decision of data routing in mobile opportunistic networks,and propose the uncertain decision model of data routing path selection for mobile opportunistic networks based on the Markov decision theory. Fourthly, considering the movement, residual energy of nodes and the reliability of wireless links between nodes, and using the unsupervised reinforcement learning method,we will design a distributed adaptive efficient decision algorithm for data routing path selection in mobile opportunistic networks. Finally, with the big data routing in mobile learning opportunistic networks as the background, we will carry out the application research. Our ultimate goal is to propose the new theory and method for the big data routing in mobile opportunistic networks. Thus, this project has important theoretical significance and wide application prospect.