基于GAlib和聚类技术的无线mesh网关部署优化研究-模式识别与智能系统专业论文.docxVIP

摘 要 随着无线通信技术的发展，无线网络的广阔运用正在逐步改变着人们的日常生 活。无线网络能够实现高容量和高速率的 Internet 接入，逐渐替代有线网络。但是由 于传统单跳无线接入网传输距离有限的问题，已经越来越无法满足人们的需要。无线 mesh 网络以其组网简单快速、网络结构灵活、系统鲁棒性优秀等特点，拥有非常广 阔的应用前景。成为了人们关注的新热点。 本文概述了无线 mesh 网络这种新型的下一代无线网络，介绍其结构，分析其所 具备的特点，与移动 Ad hoc 之间的区别，以及其应用状况。在无线 mesh 网络中骨干 网络的设计中，网关部署是设计的要点。本文分析了网关部署优化需要的各种关键问 题，分析了现有方案的缺陷，并提出一种网关部署算法。 首先将网关部署问题化为数学模型，用集合覆盖问题求解多目标优化问题。其次 提出了一种基于遗传算法的网关部署方法，使用 GAlib 编程实现后制作出了一个界面 友好的可执行文件。可以根据经验和要求输入各种约束值，并得到直观的网关选择点 及分簇方案。最后在上一步的基础上，提出了一种聚类优化方法，使得前一步的部署 方案在网关的选取、簇内平均跳数及负载平衡的要求得到进一步的均衡完善。 同时用 opnet 仿真分析后证明，经过聚类优化后，网关部署方案在网络通信时延 和平均负载中得到改善，在与前面研究者提出方案对比后，证明本方案在网关数目， 平均跳数优化这两方面成效显著。 关键词：无线 mesh 网络；网关部署；遗传算法；聚类 Abstract With the development of wireless communication technology, wide application of wireless internet has been changing people’s lives with each passing day. Since wireless Internet can actualize to satisfy the high capacity and speed of internet access, cable internet has increasingly been substituted. However, due to the distance constraint of the single hop of wireless internet, it cannot satisfy for users’ demand. Wireless mesh internet has a bright/perspective future as a result of its own advantages which includes simplified component speed, the flexibility of internet structure, and the excellence robustness of the system. This thesis aims to briefly explain the next generation of wireless meshes internet and introduce its structure，inherent special feature and distinguish between Ad hoc as well as its application status. The essential points are designed in the wireless mesh internet. This essay also analyzes different aspects of the gateway deployment optimization and the existing detection to bring out an algorithm how to deploy gateways. Firstly, the deployment of the gateway into the mathematical model, set covering problem with multi-objective optimization problem. Secondly, we propose a genetics based gateway deployment method, product a EXE file with friendly interface by GAlib after programmer. According to experien