基于博弈论的云计算资源分配机制研究
发布时间：2018-07-26 10:08
【摘要】：作为一种新型的计算模式，云计算正逐渐成为信息产业的最新发展趋势。云计算以其易扩展、按需使用、按使用付费的特点迅速吸引了大量的用户。通过使用云计算，用户不仅降低了在IT设备上的成本投入，同时还能专注于自身核心业务的研发而无须担心运营环境是否支持业务的拓展。目前国内外各大IT公司包括谷歌、亚马逊、微软、百度等都相继推出了自己的云计算服务平台。资源分配是云计算的核心问题之一，高效的资源分配方案不仅能提高资源的利用率，还能提高用户的满意度。传统的资源分配方法以系统整体性能作为衡量标准，所采用的优化方法并未考虑到用户的个性化需求。由于云用户需求日益个性化以及云计算更多的应用于商业领域，因此在资源分配时须以用户为中心，提高用户的满意度。本文采用博弈论与机制设计理论对云计算资源分配问题展开研究。本文的工作内容与创新点如下： （1）针对用户可以根据任务的起止期限而动态达到，采用机制设计理论设计了一种基于预分配资源的云计算资源在线拍卖机制，允许用户在任务的时间窗口内进行多次竞拍，并证明了该机制满足个体理性与激励兼容特性。 （2）从提高社会价值的角度出发，改进了上述基于预分配资源的在线拍卖机制，提出了一种基于高价值优先的资源在线拍卖机制。并通过实验对两种机制的性能进行了分析与比较。 （3）在云联盟结构中，引入工作分解模型，用户可以将任务分解成多个可以同时执行的子任务，一部分提交到云联盟中执行，另一部分在本地执行。将多用户多服务提供商的资源匹配问题建模为一次博弈，提出了基于云联盟的工作分解博弈模型，研究该博弈纳什均衡的存在性并分析均衡下用户的行为模式。 在目前我国对以用户为中心的云计算资源分配问题缺乏研究的背景下，本文使用博弈论与机制设计理论对资源分配问题所做的研究成果对我国云计算的研究与应用具有一定的参考价值。
[Abstract]:As a new computing model, cloud computing is becoming the latest trend of information industry. Cloud computing has attracted a large number of users because of its easy to expand, on demand, and pay for the use of the characteristics of quickly attracted a large number of users. By using cloud computing, users not only reduce the cost of IT devices, but also focus on the development of their core business without worrying whether the operating environment supports business expansion. At present, the major IT companies at home and abroad, including Google, Amazon, Microsoft, Baidu and so on, have launched their own cloud computing service platform. Resource allocation is one of the core problems in cloud computing. Efficient resource allocation can not only improve the utilization of resources, but also improve the satisfaction of users. The traditional method of resource allocation is based on the overall performance of the system, and the optimization method does not take into account the individual needs of users. Due to the increasingly personalized needs of cloud users and more applications of cloud computing in the business field, it is necessary to focus on users in resource allocation to improve their satisfaction. In this paper, game theory and mechanism design theory are used to study the resource allocation of cloud computing. The contents and innovations of this paper are as follows: (1) aiming at the dynamic realization of users according to the starting and ending period of the task, an online auction mechanism of cloud computing resources based on pre-allocated resources is designed based on mechanism design theory. It allows users to bid several times in the time window of the task, and proves that the mechanism meets the characteristics of individual rationality and incentive compatibility. (2) from the point of view of enhancing social value, The online auction mechanism based on preallocated resources is improved, and a resource online auction mechanism based on high value priority is proposed. The performance of the two mechanisms is analyzed and compared through experiments. (3) in the cloud alliance structure, the work decomposition model is introduced, the user can decompose the task into several sub-tasks that can be executed simultaneously. Some are submitted to the cloud alliance for execution, and the other part is executed locally. The resource matching problem of multi-user and multi-service provider is modeled as a single game. A work decomposition game model based on cloud alliance is proposed. The existence of Nash equilibrium is studied and the user behavior pattern under equilibrium is analyzed. In the context of the lack of research on user-centric resource allocation in cloud computing in China, In this paper, the research results of resource allocation based on game theory and mechanism design theory have a certain reference value for the research and application of cloud computing in China.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.07
本文编号：2145658
[Abstract]:As a new computing model, cloud computing is becoming the latest trend of information industry. Cloud computing has attracted a large number of users because of its easy to expand, on demand, and pay for the use of the characteristics of quickly attracted a large number of users. By using cloud computing, users not only reduce the cost of IT devices, but also focus on the development of their core business without worrying whether the operating environment supports business expansion. At present, the major IT companies at home and abroad, including Google, Amazon, Microsoft, Baidu and so on, have launched their own cloud computing service platform. Resource allocation is one of the core problems in cloud computing. Efficient resource allocation can not only improve the utilization of resources, but also improve the satisfaction of users. The traditional method of resource allocation is based on the overall performance of the system, and the optimization method does not take into account the individual needs of users. Due to the increasingly personalized needs of cloud users and more applications of cloud computing in the business field, it is necessary to focus on users in resource allocation to improve their satisfaction. In this paper, game theory and mechanism design theory are used to study the resource allocation of cloud computing. The contents and innovations of this paper are as follows: (1) aiming at the dynamic realization of users according to the starting and ending period of the task, an online auction mechanism of cloud computing resources based on pre-allocated resources is designed based on mechanism design theory. It allows users to bid several times in the time window of the task, and proves that the mechanism meets the characteristics of individual rationality and incentive compatibility. (2) from the point of view of enhancing social value, The online auction mechanism based on preallocated resources is improved, and a resource online auction mechanism based on high value priority is proposed. The performance of the two mechanisms is analyzed and compared through experiments. (3) in the cloud alliance structure, the work decomposition model is introduced, the user can decompose the task into several sub-tasks that can be executed simultaneously. Some are submitted to the cloud alliance for execution, and the other part is executed locally. The resource matching problem of multi-user and multi-service provider is modeled as a single game. A work decomposition game model based on cloud alliance is proposed. The existence of Nash equilibrium is studied and the user behavior pattern under equilibrium is analyzed. In the context of the lack of research on user-centric resource allocation in cloud computing in China, In this paper, the research results of resource allocation based on game theory and mechanism design theory have a certain reference value for the research and application of cloud computing in China.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.07
【参考文献】
相关期刊论文 前2条
1 孙瑞锋;赵政文;;基于云计算的资源调度策略[J];航空计算技术;2010年03期
2 张煜;林莉;怀进鹏;李先贤;钟亮;;网格环境中信任-激励相容的资源分配机制[J];软件学报;2006年11期
，本文编号：2145658
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2145658.html
