跨多域虚拟网络嵌入关键技术研究

发布时间：2018-05-16 04:16

本文选题：虚拟网络 + 多域　；参考：《电子科技大学》2014年硕士论文

【摘要】：跨入新世纪,国际化浪潮以及市场经济的冲击迎面而来,互联网新型应用层出不穷,现有的互联网架构很难满足用户对底层网络安全性、可扩展性和服务质量等方面的要求,在一定程度上呈现出僵化现象。网络虚拟化是一种有效的解决方案,网络虚拟化的多态性能确保底层网络资源共享、在同一底层网络上异构网络共存和网络本身的多样性。在实际应用中,底层网络属于多个基础设施供应商,如何把虚拟网络嵌入到由多个基础设施供应商共同管理的底层网络是一个重要挑战。本文将从资源匹配、虚拟网络请求分割、虚拟网络嵌入和资源绑定四个方面介绍跨多域虚拟网络配置过程。本文第一章简单介绍网络虚拟化以后,重点给出跨多域虚拟网络嵌入的定义和面临的挑战。第二章首先对跨多域虚拟网络配置问题进行数学抽象,然后介绍了资源匹配的概念,并提出用概念聚类技术对网络资源进行整合和分类并生成系统树图,然后使用基于相似性的匹配算法进行资源匹配。本章的重点内容是在降低分割成本的同时跨多域分割虚拟网络请求。本章提出了跨多域虚拟网络分割的启发式算法(递归调用最大流最小割算法)和精确算法,通过仿真得出精确算法更适合小型网络,启发式算法凭借它的鲁棒性和大型网络线性退化性优势更适合动态变化的大型网络。第三章主要介绍了虚拟网络的嵌入过程,提出用精确嵌入算法使虚拟节点和链路同时进行映射。该算法在最小化虚拟网络嵌入成本的同时提高了请求接受率。分别用并行请求处理和顺序请求处理两种方式对虚拟网络嵌入过程进行仿真,通过仿真得出该精确嵌入算法更适合小型网络,而且并行请求处理比顺序请求处理有更低的嵌入成本和更高的请求接受率,更有利于底层网络的充分利用。当虚拟网络子图嵌入到底层网络以后,提出用最短路径启发式算法和多商品流算法把虚拟网络子图之间的虚拟链路嵌入到底层网络的域间链路。仿真显示多商品流算法有更高的请求接受率和更低的嵌入成本。全部嵌入以后,对相关资源进行绑定和对底层网络资源进行更新。第四章介绍了本文中用到的仿真平台。第五章对全文的工作进行了总结,并对跨多域虚拟网络嵌入的未来工作进行了展望。
[Abstract]:In the new century, with the impact of internationalization and market economy, new applications of the Internet emerge in endlessly. The existing Internet architecture is difficult to meet the requirements of users for the security, scalability and quality of service of the underlying network. To a certain extent, there is a phenomenon of ossification. Network virtualization is an effective solution. The polymorphism of network virtualization can ensure the sharing of the underlying network resources, the coexistence of heterogeneous networks and the diversity of the network itself in the same underlying network. In practical applications, the underlying network belongs to many infrastructure providers. How to embed virtual network into the underlying network managed by multiple infrastructure providers is an important challenge. This paper introduces the configuration process of cross-domain virtual network from four aspects: resource matching, virtual network request segmentation, virtual network embedding and resource binding. In the first chapter, the definition and challenges of virtual network embedding across multiple domains are given. The second chapter firstly abstracts the problem of multi-domain virtual network configuration, then introduces the concept of resource matching, and proposes to integrate and classify network resources with concept clustering technology and generate system tree graph. Then the similarity-based matching algorithm is used for resource matching. The focus of this chapter is to split virtual network requests across multiple domains while reducing the cost of segmentation. In this chapter, a heuristic algorithm (recursive calling maximum flow minimum cut algorithm) and an accurate algorithm are proposed to segment virtual networks across multiple domains. The simulation results show that the exact algorithm is more suitable for small networks. Heuristic algorithm is more suitable for large dynamic networks by virtue of its robustness and linear degradation. In chapter 3, the embedding process of virtual network is introduced, and a precise embedding algorithm is proposed to map the virtual node and link simultaneously. The algorithm not only minimizes the embedding cost of the virtual network, but also improves the request acceptance rate. Parallel request processing and sequential request processing are used to simulate the embedding process of virtual network, and the simulation results show that the accurate embedding algorithm is more suitable for small network. Moreover, parallel request processing has lower embedding cost and higher request acceptance rate than sequential request processing, which is more conducive to the full utilization of the underlying network. When the virtual network subgraph is embedded into the underlying network, the shortest path heuristic algorithm and the multi-commodity flow algorithm are proposed to embed the virtual link between the subgraph of the virtual network into the inter-domain link of the underlying network. Simulation results show that the algorithm has higher request acceptance rate and lower embedding cost. After embedding, bind the related resources and update the underlying network resources. The fourth chapter introduces the simulation platform used in this paper. The fifth chapter summarizes the work of this paper and looks forward to the future work of cross-domain virtual network embedding.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.01

【相似文献】