【作者】 黄昭文；

【导师】 冯穗力；

【作者基本信息】 华南理工大学 ， 通信与信息系统， 2010， 博士

【摘要】 Wireless Mesh Network (WMN)是无线Ad Hoc网络在组网结构、通信机制等方面的扩展。以往对Ad Hoc网络及IP网络的研究成果可以作为参考,但不能完全适用于WMN。WMN在网络结构、网络容量、物理层、MAC层、资源调度、路由协议、QoS(Quality of Service)保障等方面出现了许多新课题,而这些课题是WMN实现大规模应用的关键环节。WiMAX Mesh网络作为主流WMN网络,在资源调度与QoS保障方面仍存在一些问题需要解决。IEEE 802.16为WiMAX Mesh模式定义了三种网络资源调度机制,即协同集中式调度、协同分布式调度、非协同分布式调度,但是规范中对各种调度机制并没有给出完整的实现方案,对于实际组网所关心的网络稳定性、容灾性问题,以及对网络新技术如OFDMA等的引入等都没有给出明确的定义。在QoS保障方面,虽然IEEE 802.16标准为点到多点PMP模式在MAC层定义了若干QoS级别和保障方案,但是对WiMAX Mesh模式,在规范中却没有定义类似于PMP模式的QoS级别和保障机制。由于Mesh网络固有的多跳属性给相关机制的设计和实现造成一定困难,使得WiMAX Mesh难以支持端到端的QoS,从而对WiMAX Mesh技术投入实际应用造成了障碍。本文以WiMAX Mesh为研究重点,分析了业界在WMN资源调度和QoS保障方面的研究现状,结合WiMAX Mesh网络技术特点,以提高资源利用率,实现端到端QoS保障为目标,提出了一系列优化算法和保障机制,并通过理论分析和实验仿真进行了验证。本论文的主要成果和创新点包括以下6点:(1)系统分析主要WMN网络(包括WiFi Mesh和WiMAX Mesh)的技术原理,总结网络容量、资源调度和QoS保障的研究现状和存在问题。本文对WMN网络的物理层、MAC层、组网方案、网络接入、拓扑控制、调度算法、QoS保障机制等方面的工作原理进行了深入调研,这是开展WMN网络资源调度和QoS保障研究工作的理论基础。(2)根据WiMAX Mesh网络流量特性,首次提出一种网络流量预测新方法(WARPA)。在对WiMAX Mesh网络流量特性进行分析的基础上,本文提出了一种基于小波变换和AAR模型的WiMAX Mesh网络流量预测新方法。仿真实验的结果表明,本方法的预测精度较高,运算量适中,适合于对网络流量进行准确预测。(3)在(1)的基础上,提出一种多出口WiMAX Mesh组网方案和资源调度算法(MEMSA)。本文分析了WiMAX Mesh网络的调度机制和单出口WiMAX Mesh网络在吞吐量和容灾性方面存在的问题,提出了一种多出口组网方案,并在该方案基础上提出了一种全局最优的资源调度算法,以及把该算法在WiMAX Mesh网络实施的详细步骤。本文在具有20和50个SS节点的Mesh网络中对该算法进行仿真,本算法可以把调度效率提高53%以上。(4)提出基于OFDMA的WiMAX Mesh系统模型,并提出一种基于蚁群算法的资源调度算法(AWRAA)。为了实现基于OFDMA的WiMAX Mesh网络资源的最优化调度,需要综合考虑网络节点和信道的数量、调制方式、误比特率、调度模式、QoS需求等多种因素,同时使网络吞吐量达到最大化,所以问题的求解难度极高。为此,本文建立了基于OFDMA的WiMAX Mesh网络系统优化模型。模型的优化目标是在充分考虑功率、调度和吞吐量约束条件下最大化网络吞吐量。为了对系统模型进行求解,本文结合WMN网络特点,提出了一种较为完善的基于蚁群算法的求解算法。该算法具有容易实现并且灵活支持不同的需求和约束的特点。这些特点使得本方法可方便地应用于实际网络环境。仿真结果表明,本算法有效解决了OFDMA WiMAX Mesh的资源调度难题。(5)首次提出一种基于全局优化的WiMAX Mesh网络QoS保障算法(WMNQGA)。本文在分析现有WiMAX Mesh QoS保障机制的基础上,提出了一种有效、动态的QoS保障算法。该算法把WiMAX Mesh网络视为一个整体,定义了各节点在进行数据包处理时应该遵循的约束条件,并给出了各节点对网络流量的处理方式。此算法是在WiMAX Mesh网络的QoS保障方面一个新的探索。仿真结果表明,本算法既提高了全网吞吐量,同时在不同的拓扑结构下仍能确保各节点对带宽使用的比例公平性。(6)首次在WiMAX Mesh网络中提出智能标签技术,并提出一种基于智能标签的WiMAX Mesh网络QoS保障算法(ITBQoS)。本文在分析WiMAX Mesh网络QoS保障机制研究现状的基础上,首次提出了一种称之为“智能标签”的概念和算法,接着定义了与数据流相关的智能标签的结构和属性,各网络节点对智能标签及数据流的约束条件和处理算法,制定了不同优化目标下的系统模型,并给出了相应启发式算法。仿真结果表明,智能标签技术在不同带宽和故障概率下均取得较好性能,既可提升WiMAX Mesh网络吞吐量,也可满足各节点的QoS需求,具有良好的应用前景。更多还原

【Abstract】 Wireless Mesh Network, WMN, is an extension to the mobile Ad Hoc network, both in network architecture and communication mechanism. The current achievements in the research of Ad Hoc Network and IP Network can not be used in WMN. There are some open issues in network architecture, capacity, physical layer, MAC layer, resource scheduling, routing protocol and QoS guaranteeing mechanism. And these issues are critical to the deployment of WMN at large scale.There are some problem in the field of resource scheduling and QoS guaranteeing for the mainstream of WMN, WiMAX Mesh Network. The IEEE 802.16 mesh mode has defined three different scheduling schemes to manage the minislots in the data subframe, i.e., centralized scheduling, uncoordinated distributed scheduling, and coordinated distributed scheduling, but the IEEE 802.16 neither explain them in detail, nor give any guideline for the stability and robustness of network. And it does not cover the problem of deploying new network technology, for example OFDMA, in WiMAX Mesh Mode. As to QoS guaranteeing, though the IEEE 802.16 defines some QoS Class and guaranteeing schemes for WiMAX Point to Multipoint Mode, there is no counterpart for WiMAX Mesh Mode. Since its character of multi-hop, it is difficult to support end-to-end QoS in WiMAX Mesh Network, and it has turned out to be an obstacle for the deployment.In this dissertation, we firstly analyze the current achievements in resource scheduling and QoS guaranteeing for WMN, then advance a series of algorithm and guaranteeing mechanism to improve the wireless resource utilization and end-to-end QoS guaranteeing, based on the character of WiMAX Mesh Network. The main research work and achievement of this dissertation includes:(1) The technical principle of mainstream of WMN, including WiFi Mesh and WiMAX Mesh, was studied systematically. We made thorough investigations in the field of network architecture, capacity, physical layer, MAC layer, resource scheduling, routing protocol and QoS guaranteeing mechanism. And this is the theoretical basis of our research.(2) The character of traffic in WiMAX MESH Network is analyzed, and a predicting method based on wavelet transformation and AAR model, WARPA, is proposed. To allocate bandwidth in WiMAX Mesh Network dynamically and efficiently, we need to predict the real-time traffic volume correctly, which will be sent to the Mesh scheduler. The Mesh scheduler will grant bandwidth to each Mesh node according to the value. In our study, the current predicting method, including ARMA, is found to be not so proper for using in WiMAX Mesh Network. With our predicting method, the noise of traffic volume with wavelet transformation is decreased, and the result is fed into the AAR model, which will adjust the model argument automatically, to do the predicting job. To verify the algorithm, we used the traffic volume data captured in Auckland college into the model. It is found that the algorithm is about 2 percent more accurate than the AARMA model. The simulation result showed that our method has more precise prediction and less computation time, and is more suitable for WiMAX Mesh Network.(3) Based on achievement (1), we consider the scheduling efficiency, throughput and robustness for WiMAX Mesh Network. We analyze the scheduling mechanism of WiMAX Mesh Network and point out problem of current backhaul network structure. We proposed a new WiMAX Mesh network topology with multiple BS as egress, and propose an efficient optimal scheduling algorithm, MEMSA, for this new network topology. The detailed procedure for deploying this algorithm is described as well. We developed a simulation platform to verify this algorithm in WiMAX Mesh network with 20 and 50 MESH SS nodes respectively. The simulation result shows that our algorithm improves the efficiency more than 53 percent. We also illustrate that the bottleneck problem of the backhaul network can be solved by our scheduling algorithm.(4) Although OFDMA is a candidate physical layer protocol for 802.16, there are few solutions for the wireless resource-allocating problem in OFDMA mesh network. We study the current resource allocation methods in OFDM and OFDMA network, and introduce the architecture of OFDMA mesh network. We build a mathematic system model for the OFDMA mesh network, and propose a sophisticated ACO-based algorithm for wireless resource allocation, AWRAA. With our algorithm, we get acceptable solution, while fulfilling power and QoS constraints. Since our algorithm is easy to adapt to more realistic problems, it turns out to be a feasible method. The simulation result showed our algorithm is able to find a sub-optimal answer for resource allocation in OFDMA mesh network.(5) Since the IEEE 802.16 Specification has not defined how to guarantee QoS in WiMAX Mesh Network, it became a critical problem for deployment. We analyzed several existing WMN QoS guaranteeing mechanisms, and found they have some limitation and could not guarantee QoS in WMN very well. We advanced an effective and dynamic QoS-Guaranteeing algorithm, WMNQGA, for WMN, which described the traffic handling method of each node in view of integrity. It calculated the global optimized traffic-handling pattern for each node, and gave some constraints for the WiMAX Mesh nodes. Our algorithm can improve the network throughput, and ensure fairness. To verify our algorithm, we developed a WiMAX Mesh simulating platform and ran our simulation on it. The simulation result showed that our algorithm is effective and feasible.(6) Though it is important for the application of WiMAX Mesh Network, the IEEE 802.16 Specification has not defined how to guarantee QoS in WiMAX Mesh Network. And most of the current related research on WMN QoS only focuses on routing Algorithm or Hop-by-Hop QoS guaranteeing. By analyzing the current status of QoS guaranteeing mechanism, we advanced a new technology named“Intelligent Tag”. Based on it, we proposed a QoS-Guaranteeing algorithm for WiMAX Mesh Network, ITBQoS, which defined the ITtag handling method of each node in view of the whole network, and then we give the detailed procedure of handling the ITtag in each node. We also give some constraint on the WiMAX Mesh nodes in processing the packets. To verify our algorithm, we developed a WiMAX Mesh simulating platform and run our simulation on it. The simulation result showed that our algorithm is effective and feasible.更多还原