用户优先级虚拟网络映射算法的实现

发布时间：2018-12-27 10:51

【摘要】：近些年来，随着网络中越来越多的分布式应用和新兴网络技术的出现，互联网体系结构“僵化”问题已经成为其发展的重要桎梏。网络虚拟化技术允许多个异构网络同时在同一个基础物理网络架构上存在，并且相互独立，互不干涉，其在一定程度上缓解了互联网体系结构“僵化”问题给网络技术发展带来的阻碍。而网络虚拟化技术实现的最大挑战就是,在最大化利用底层物理资源和提高基础架构提供商利益的条件下，如何更加有效地为尽可能多的虚拟网络分配合适的底层物理资源，即虚拟网络映射问题。为了解决虚拟网络映射问题，本文以底层物理网络资源利用率最大化为目标，建立了在控制转发分离网络构架下基于用户优先级的虚拟网络映射整数线性规划模型，并提出了一种基于离散空间的改进粒子群算法。该算法以底层物理网络剩余资源为适度函数，对于粒子群算法中粒子的位置、速度和相关操作进行了重新定义，使得其在迭代的过程中，粒子进化更具方向性，同时引进不同粒子位置互斥性用以解决粒子群算法易早熟陷入局部最优解的缺陷。通过仿真实验的数据，我们可以分析得出改进粒子群算法在解决虚拟网络问题的可行性和高效性。为了将虚拟网络映射功能真正应用于实际，，本文针对以Openflow交换机搭建的底层物理网络，设计并开发了虚拟网络部署系统。该系统通过Web Service技术将虚拟网络映射算法封装成Web服务，满足不同用户使用需求。系统通过获取用户应用需求，自动生成虚拟网络并完成映射，将映射结果返回给用户，并且转化为交换机配置信息交给底层网络，真正将虚拟网络成功部署到底层网络。
[Abstract]:In recent years, with more and more distributed applications and emerging network technologies in the network, the problem of "rigid" Internet architecture has become an important shackles of its development. Network virtualization technology allows multiple heterogeneous networks to exist simultaneously on the same physical network architecture, independent of each other and non-interference with each other. To some extent, it alleviates the hindrance to the development of network technology caused by the "rigidity" of Internet architecture. The biggest challenge of network virtualization is how to allocate appropriate physical resources to as many virtual networks as possible under the conditions of maximizing the use of underlying physical resources and improving the benefits of infrastructure providers. That is, virtual network mapping problem. In order to solve the problem of virtual network mapping, an integer linear programming model of virtual network mapping based on user priority is established in order to maximize the utilization of the underlying physical network resources. An improved particle swarm optimization algorithm based on discrete space is proposed. The algorithm takes the remaining resources of the underlying physical network as the appropriate function, and redefines the position, velocity and related operations of the particles in the particle swarm optimization algorithm, which makes the evolution of particles more directional in the iterative process. At the same time, the mutual exclusion of different particle positions is introduced to solve the problem that particle swarm optimization algorithm is prone to premature falling into local optimal solution. The feasibility and efficiency of the improved particle swarm optimization (PSO) algorithm in solving the virtual network problem can be analyzed by the data of simulation experiment. In order to truly apply the virtual network mapping function to practice, this paper designs and develops a virtual network deployment system for the underlying physical network based on Openflow switch. The system encapsulates the virtual network mapping algorithm into Web service through Web Service technology to meet the needs of different users. By obtaining the user's application requirements, the system automatically generates the virtual network and completes the mapping, returns the mapping result to the user, and converts the configuration information of the switch to the underlying network, and truly deploys the virtual network to the underlying network successfully.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.01

【参考文献】