基于演化博弈的跨时间的公平带宽分配算法

发布时间：2018-07-05 08:46

本文选题：数据中心网络 + 带宽分配　；参考：《浙江工商大学》2017年硕士论文

【摘要】：作为能够通过网络随时随地获取高可用计算资源的模式,云计算(Cloud Computing)得到了迅速发展。但是在发展的同时,随着不断壮大的互联网规模以及日益增多的网络应用类型,致使网络流量的需求也不断的激增,网络带宽逐渐成为影响云计算技术发展的一个重要因素。因此,如何协调网络带宽分配、减少网络延迟、提高网络吞吐量等因素,制定出适合运营商和租户间双赢的网络带宽分配模型,是云计算企业需要在竞争严重的市场中急于解决的重要问题之一。公平带宽分配能够有效解决网络带宽的优化配置,在此背景下,基于新的网络环境下对于宽带的需求以及企业在现实运营中遇到的问题,租户在现实工作生活中使用网络流量所遇到的问题,建立和设计更加贴近实际应用的带宽分配的博弈模型和相应公平分配带宽的算法,这无论从理论研究还是现实运用方面都具有重要研究意义。本文首先阐述公平带宽分配的研究背景与意义,再对目前带宽分配的国内外研究现状做了详细的说明,分析且指出其不足之处。其次本文结合实际的网络现状,研究了基于跨时间的公平带宽分配问题。文章首先提出了两类新型的基于链路比例的网络带宽分配模型,静态链路比例模型(SLP)以及适应性链路比例模型(ALP)。在第一类模型中网络的高可用率没有得到保障,而在第二类模型中虽然网络的高可用率得到了保障,但是忽略了对历史带宽占用的考虑。因此我们针对以上两类模型进行分析,取其精华去其糟粕,创新性地引出了跨时间的公平带宽分配策略(FOT),并在此基础上提出了一个演化博弈模型。最后通过设计演化规则,说明演化会趋向稳定,也即,本文所提出的演化博弈下的带宽分配方法是可行的。然后通过比较分析多种带宽分配策略,证明所提出的带宽分配方法是优于其他带宽分配方法。
[Abstract]:Cloud Computing (Cloud Computing) has developed rapidly as a model that can access high available computing resources anytime and anywhere through the network. But at the same time, with the growing scale of the Internet and the increasing types of network applications, the demand for network traffic is also increasing. Network bandwidth has gradually become an important factor affecting the development of cloud computing technology. Therefore, how to coordinate network bandwidth allocation, reduce network delay, improve network throughput and other factors, establish a win-win network bandwidth allocation model between operators and tenants. Cloud computing enterprises need to solve one of the most important problems in the highly competitive market. Fair bandwidth allocation can effectively solve the optimal allocation of network bandwidth. In this context, based on the demand for broadband in the new network environment and the problems encountered by enterprises in real operation, The problems encountered by tenants in using network traffic in their real work life, the game model of bandwidth allocation which is closer to the actual application and the corresponding algorithm of fair bandwidth allocation are established and designed. It is of great significance in both theoretical research and practical application. This paper first describes the background and significance of the research on fair bandwidth allocation, then gives a detailed description of the current domestic and foreign research on bandwidth allocation, and points out its shortcomings. Secondly, this paper studies the problem of fair bandwidth allocation based on cross-time. In this paper, two new network bandwidth allocation models based on link ratio, static link proportion model (SLP) and adaptive link proportion model (ALP) are proposed. In the first model, the high availability rate of the network is not guaranteed, while in the second model, the high availability rate of the network is guaranteed, but the consideration of the historical bandwidth occupancy is neglected. Therefore, we analyze the above two kinds of models, take the essence of the model to get rid of the dross, and creatively lead to a cross-time fair bandwidth allocation strategy (FOT), and then put forward an evolutionary game model. Finally, the evolution rules are designed to show that the evolution tends to be stable, that is, the bandwidth allocation method proposed in this paper is feasible. Then, by comparing and analyzing various bandwidth allocation strategies, it is proved that the proposed bandwidth allocation method is superior to other bandwidth allocation methods.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O225;TP393.09

【参考文献】