【作者】 韩冰青；

【导师】 张宏；

【作者基本信息】 南京理工大学 ， 计算机应用技术， 2010， 博士

【摘要】 移动Ad Hoc网络代表一种无基础设施的、完全分布式的、多跳的自治系统。人们有着很大的兴趣来设计和实现这种网络,以提供多样化的应用服务,如救灾和临时会议等。近年来,随着多媒体应用的不断普及,在移动Ad Hoc网络中提供服务质量支持已成为一个重要而又具有挑战性的任务。但是,由于无线网络带宽有限,有限的无线资源与多媒体业务的服务质量需求的矛盾日益突出,如何设计合理有效的资源分配方案成为保证服务质量的关键。本文研究的主要目标是在有限带宽条件下,为移动Ad Hoc网络建立一种有效的资源分配方案,以保证多媒体应用程序的端到端的QoS (Quality of Service)需求,并满足它们对时延、带宽或丢包率等方面的要求,同时能提高网络资源利用率。本文首先研究了移动Ad Hoc网络的资源分配模型和算法,接着研究了网络层资源分配技术,然后研究了MAC (Media Access Control)层资源分配技术,最后对联合网络层和MAC层的跨层资源分配技术进行了研究。本文的主要研究内容和成果如下：(1)在系统地分析Ad Hoc网络的信息流竞争特点的基础上,提出了新的基于价格的移动Ad Hoc网络资源分配算法。首先,构建移动Ad Hoc网络资源分配模型,将资源优化问题转换为拉格朗日对偶问题；接着,以价格作为资源分配的度量指标,通过路由发现时的价格查询以及路由响应时的价格反馈,消除分布式预测带来的不一致性；最后,利用小波变换将自相似性业务的长相关性转化为短相关性,对网络流量进行预测,提出一种改进的Ad Hoc网络资源分配算法。仿真结果表明：所提出的算法具有良好的收敛性,与现有的资源分配算法相比提高了资源利用率。(2)在网络层资源分配方面,提出了新的路由算法。在移动Ad Hoc网中,判断路径优劣的度量参数主要有跳数、延迟、能量等,常常导致过度使用中心节点,引起网络拥塞。针对这一问题,首先提出了一种基于价格感知的多目标优化按需路由协议。该协议以价格为路由度量指标,采用分层的体系结构以获得全网一致的拓扑视图。仿真结果表明：该协议提高了分组投递率并降低了端到端平均时延,在动态移动环境下具有较好的稳定性。接着,在分析DYMO (Dynamic MANET On-Demand)路由协议优缺点的基础上,将DYMO协议和服务质量及多径路由结合起来,提出一种新的QoS感知的多径DYMO路由协议QA-DYMO (QoS-Aware Multi-path Dynamic MANET On-Demand)。该协议以多径路由为基础,利用多条链路不相交路径来发送数据,能够适应Ad Hoc网络的动态变化和较好的支持QoS。(3)在MAC层资源分配方面,提出新的冲突退避算法和分析模型。首先,在分析传统的IEEE 802.11 DCF协议退避算法的基础上,设计了改进的多优先级退避算法和区分优先级的帧间隔,提出了一种支持区分服务的改进方案D-DCF (Differentiated Distributed Coordination Function)。接着,提出一种四维马尔可夫链模型对D-DCF进行建模,该模型能有效地评估饱和及非饱和状态下的网络性能。理论和仿真结果表明：D-DCF在吞吐量和平均分组延迟上的性能都优于802.11 DCF,并且能够支持3种优先级别的区分服务。(4)在跨层资源分配方面,提出一种联合网络层和MAC层的跨层资源分配算法CL-QARA (Cross-Layer QoS Aware Resource Allocation)。单纯的基于MAC层或者网络层的资源分配方案无法满足动态变化的Ad Hoc网络环境下的服务质量要求,而将跨层技术运用于资源分配中,可以提供多种服务质量的应用。CL-QARA的主要思想是引入价格作为资源分配的度量指标,以QoS带宽需求为参数,将网络层的动态资源分配信息与MAC层CSMA/CA接入机制相结合,以改进MAC层的冲突退避算法。并设计了改进的退避算法和呼叫接入控制算法,以实现MAC层与网络层的跨层技术。通过QoS感知的资源分配算法和跨层技术协同工作,为QoS服务提供了业务保障。仿真结果表明,CL-QARA算法具有良好的收敛性和稳定性,与其它算法相比,CL-QARA能有效地提供QoS保证,提高了网络的效用和性能。更多还原

【Abstract】 Mobile ad hoc networks represent autonomous distributed systems that are infrastructure-less, fully distributed, and multi-hop in nature. There is considerable interest in designing and implementing such networks to provide services in a variety of diverse applications, e.g., disaster relief and temporary meetings. Recently, due to increasing popularity of multimedia applications, QoS support in mobile ad hoc networks has become an important yet challenging objective. However, due to the limited bandwidth, the conflict between the limited wireless resource and QoS requirements of multimedia service grows increasingly. How to design an effective resource allocation is the key to support QoS.The main objective of this dissertation is to develop an effective resource allocation scheme for mobile ad hoc networks under limited bandwidth conditions. It can guarantee satisfactory end-to-end QoS to multimedia applications according to certain QoS measures such as delay, bandwidth or packet loss, while achieving efficient network resource utilization. In this dissertation, resource allocation model and algorithm in mobile ad hoc network is firstly studied. Secondly, the network-layer resource allocation technique is studied. Thirdly, the MAC-layer resource allocation technique is studied. Finally, the cross-layer resource allocation with joint design of network layer and MAC layer is studied.The main research contributions of this dissertation are as follows:(1) With detailed analysis of the contest characteristics of ad hoc network’s information flow, a new price-based resource allocation algorithm of mobile ad hoc network is proposed. Firstly, a resource allocation model of mobile ad hoc network is presented by converting the optimization problem to the Lagrangian dual problem. Secondly, price is introduced to measure resource allocation. Price query is done with the route discovery and price feedback is done with the route reply. It can eliminate the inconsistency of distributed prediction. Finally, long-range dependence can be converted into short-range dependence and the network traffic can be predicted with wavelet. An improved resource allocation algorithm is proposed. The simulation results show that the resource allocation algorithm has good convergence and it improves the resource utility compared to other algorithms.(2) In the network-layer resource allocation, new routing algorithms are proposed. In mobile ad hoc networks main routing metrics are hops, delay and energy etc, which lead to overload of central nodes and can cause network congestion. An on-demand routing protocol based on price awareness is proposed to deal with the problem. This protocol chooses price as routing metric and uses layered architecture. The method can obtain the same view of the whole network. The simulation results show that it improves the packet delivery fraction and reduces the average end-to-end delay and has good stability. Then, based on the analysis of the dynamic MANET on-demand (DYMO) routing protocol, a novel QoS-aware multi-path DYMO (QA-DYMO) protocol is proposed to provide QoS. QA-DYMO establishes and utilizes multiple routes of link-disjoint paths to send data packets concurrently. It can adapt to the dynamic changes of ad hoc network and support QoS better.(3) In the MAC-layer resource allocation, new backoff algorithm and analytical model are proposed. Firstly, based on the analysis of the traditional IEEE 802.11 DCF protocol, multi-priority backoff algorithm and inter-frame space are designed. A differentiated distributed coordination function (D-DCF) is proposed to support differentiated service. Then, a four-dimensional Markov chain model is proposed to modeling D-DCF. The model can effectively evaluate the network performance of saturated and non-saturated conditions. Theory and simulation results show that D-DCF performs better than 802.11 DCF considering throughput and average packet delay and it can support three kinds of priority of differentiated service.(4) With the cross-layer technique, a cross-layer resource allocation algorithm CL-QARA is proposed, which is jointly designed between network layer and MAC layer. Resource allocation schemes which are purely based on MAC layer or network layer can not meet QoS requirements of the dynamic ad hoc networks. But if the cross-layer technique is used in resource allocation, it can provide a variety of QoS applications. The key idea of CL-QARA algorithm is to introduce price and QoS bandwidth to measure resource allocation. The dynamic resource allocation information in network layer is combined with the CSMA/CA admission control in MAC layer to improve the backoff algorithm. A new backoff algorithm and call admission control algorithm is designed to implement the cross-layer technique between MAC layer and network layer. The QoS-aware resource allocation algorithm is cooperated with the cross-layer technique to provide QoS guarantee. Simulation results show that CL-QARA has good convergence and stability. Compared to other algorithms. CL-QARA provides better QoS guarantee and improves the network utility and performance.更多还原