基于图谱和图熵的大规模网络虚拟网络嵌入及负载平衡研究

Network virtualization (NV) is a promising solution to overcome the resistance of the current Internet in aspects of architecture change, and virtual network embedding (VNE) is recognized as one of the key technologies in NV. In this program, the theory, methods and technologies for embedding virtual network into a large substrate physical network are investigated. This research models the ordinary VNE problem with a way of applying algebra graph theory, and further explores the theory and methods to solve it by analyzing the algebra properties of the virtual and physical network graphs. .By introducing an intermediate structure called agent virtual network (AVN) with regularly clustered substructure between VN and SPN, and applying the theory of their graph spectra and entropy, the abstract graph model of the VNE problem is built. Based on this model, the efficient strategies to find the solutions of the virtual network embedding into large scale network, and to balance the load in nodes and links of the SPN network, are proposed. All these studies in this program are mainly aimed to build efficient scheme for the large SPN to solve the VNE, and the load balancing problem of nodes and links caused in the process of dynamical VN operation. . The main distinguishing aspect of this project is to utilize the good algebra properties of regularly clustered graph to simplify the VNE problem on large networks. This research is significant in developing new methods for enabling network virtualization technology to obtain the practical deployment in the future Internet architecture.

网络虚拟化(NV)是解决当前网络僵化问题的基本途径，而虚拟网络嵌入(VNE)是网络虚拟化的关键技术。本项目主要研究大规模网络虚拟网络嵌入的理论、方法和技术。研究使用代数图论的方法建模一般VNE问题，并分析虚拟网络和物理网络图模型的代数特征以探索求解理论方法。.通过在虚拟网络(VN)和底层物理网络(SPN)之间增加具有规则分簇子结构的代理虚拟网络AVN，应用它们通信拓扑图的谱和熵理论，对大规模网络的VNE问题建立了抽象图模型；并研究基于该模型的大规模网络VNE问题求解方法，以及相应的节点和链路负载平衡算法。目标是为大规模网络构建高效的VNE求解方案，并解决VN在动态运行过程中对SPN节点和链路产生的负载平衡问题。.本项目研究主要的特色是利用规则分簇图结构的良好代数特征，简化大规模网络的VNE问题，其成果将对发展网络虚拟化技术，使其在未来互联网络架构中获得实际部署具有重要意义。

本项目主要研究了大规模网络虚拟网络嵌入的理论、方法和技术。研究使用代数图论的方法建模一般VNE问题，并分析虚拟网络和物理网络图模型的代数特征以探索求解理论方法。项目通过利用规则分簇图结构的良好代数特征简化大规模网络的VNE问题，应用它们通信拓扑图的谱和熵理论，对大规模网络的VNE问题建立了抽象图模型；研究了基于大规模网络VNE问题求解方法，以及相应的节点和链路嵌入策略。.项目的研究设计了大规模网络构建高效的VNE求解方案，并且通过实验证明了这些方法较存在的方法具有更高的效率和嵌入质量，尤其针对大规模物理网络的情形。本项目研究其成果将对发展网络虚拟化技术，使其在未来互联网络架构中获得实际部署具有重要意义。