无线mesh网中基于Hopfield神经网络的虚拟子网映射算法研究

发布时间：2018-05-25 00:34

本文选题：无线网状网网 + 网络虚拟化　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：无线网状网技术(wireless mesh network,WMN)是一种与传统无线网络完全不同的新型无线网络技术。它具有快速部署、易于安装、成本低廉、覆盖范围广和高网络带宽等优点,有极为广阔的应用前景。但无线网状网网络在实际部署中也面临着很多挑战性的问题,如难以保证用户的接入带宽、丢包率较高、移动用户通信中断等问题。对于这些问题,可以采用无线网状网网络虚拟化的方案来解决。本文主要研究的是无线网状网网络虚拟化。文中首先介绍无线网状网及网络虚拟化的相关研究,还介绍了无线网状网网络虚拟化的必要性和应用。其次是对无线mesh网中的资源调度算法进行分类总结,本文中无线网状网调度的对象不再是用户或链路,而是虚拟子网,并对现有的虚拟网络映射算法进行总结。最后,采用Hopfield神经网络算法求解资源分配和虚拟网络映射(virtual network embedding,VNE)。网络虚拟化中涉及到的一个最大挑战就是VNE问题。本文主要研究的就是VNE问题。对于无线网状网网络虚拟化环境中的VNE包括底层物理网络的资源调度、虚拟节点映射和虚拟链路映射,是一个联合优化问题。文中对这个非常复杂的NP难问题,进行拆分,分解为资源调度和虚拟子网映射两个优化问题。本文采用的是OFDMA mesh系统,调度的资源是子信道。对于虚拟子网映射中的节点映射,与有线网络不同,无线环境中可以采用位置确定,我们要优化的就是虚拟链路映射了。信道资源调度和虚拟链路映射两个优化问题都是组合优化问题,我们采用现代优化算法中的神经网络算法求解,利用Hopfield神经网络并行计算、快速收敛的特性求解这两个优化问题。仿真结果表明利用神经网络算法能有效的求解出信道资源的分配和虚拟链路的映射路径。还有文中对虚拟链路的映射采用多径映射,仿真结果表明多径映射能明显提高虚拟子网的成功接入率和物理网络的资源利用率。最后对于同时到达的虚拟子网的映射,采用优先级的方式进行映射并给出了仿真结果。
[Abstract]:Wireless mesh network (WMN) is a new wireless network technology which is completely different from the traditional wireless network. It has the advantages of rapid deployment, easy installation, low cost, wide coverage and high network bandwidth, and has a very broad application prospect. However, the wireless mesh network is also faced with the actual deployment of the network. Many challenging problems, such as the difficulty of guaranteeing the user's access bandwidth, the high packet loss rate and the communication interruption of the mobile users. For these problems, the wireless mesh network virtualization scheme can be used to solve these problems. The main research is the wireless mesh network virtualization. First, the wireless mesh network and the network virtualization are introduced. It also introduces the necessity and application of wireless mesh network virtualization. Secondly, it classifies the resource scheduling algorithms in wireless mesh network. In this paper, the object of the wireless mesh network scheduling is no longer the user or link, but a virtual subnet, and summarizes the existing virtual network mapping algorithm. Finally, Hopfie is used. The LD neural network algorithm solves the resource allocation and the virtual network mapping (virtual network embedding, VNE). One of the biggest challenges involved in the network virtualization is the VNE problem. This paper mainly focuses on the VNE problem. For the wireless mesh network virtual environment, the VNE includes the resource scheduling of the underlying physical network and the virtual node projection. Mapping and virtual link mapping is a joint optimization problem. In this paper, this very complex NP difficult problem is divided into two optimization problems: resource scheduling and virtual subnet mapping. The OFDMA mesh system is used in this paper. The resource of the scheduling is subchannel. The mapping of nodes in virtual subnet mapping is different from that of wired network. Location determination can be adopted in wireless environment. We want to optimize the virtual link mapping. The two optimization problems of channel resource scheduling and virtual link mapping are combinatorial optimization problems. We use the neural network algorithm in the modern optimization algorithm to solve this problem, and use the Hopfield neural network to compute parallel computing and fast convergence characteristics to solve this problem. Two optimization problems. The simulation results show that the allocation of channel resources and the mapping path of virtual links can be effectively solved by using neural network algorithm. Multipath mapping is used in the mapping of virtual links in the text. The simulation results show that the multipath mapping can obviously improve the power access rate of virtual subnet and the resource utilization of the physical network. Finally, the mapping of the virtual subnet arriving at the same time is carried out by the priority method and the simulation results are given.
【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】：TN929.5

【相似文献】