面向移动云计算的虚拟化资源管理
发布时间：2018-11-03 13:55
【摘要】：近年来，移动互联网业务迅猛发展。与PC和传统互联网业务相比，，移动云计算有着瘦终端、网络带宽重载和高时延等特点，这就决定了有效的资源管理成为了移动云计算中的热点与难点。 本文首先在深入研究现有云计算和移动互联网技术的基础上，分析针对移动云计算的资源管理需求。其次，为虚拟化资源管理平台设计了一个多层映射的管理模型。然后，提出了基于元胞粒子群的资源调度算法，基于小波分解和支持向量机的资源分配预测算法。最后，分别通过实验验证了上述两种算法的正确性和优越性，检验了虚拟化资源管理模型的效果。 本文的主要创新点如下： (1)提出层次化的虚拟化资源管理模型。根据服务调度粒度和灵活性，提出了服务-资源映射模型，同时给出资源分配的方案，实现了细粒度化用户请求。 (2)提出了基于元胞粒子群的虚拟资源调度算法。通过分析虚拟化资源调度问题的特点，构建合理的多目标优化模型，结合微粒群优化算法和元胞自动机，设计了合理的、高效的资源调度算法。 (3)提出了基于小波分解和支持向量机的资源分配预测算法。根据DB2小波分解和SVM提出了以小波分解-SVM建模-小波重构为过程的预测算法，对资源使用的历史数据进行训练学习，能够准确实现下个时间节点的资源分配预测。
[Abstract]:In recent years, mobile Internet services are developing rapidly. Compared with PC and traditional Internet services, mobile cloud computing has the characteristics of thin terminal, network bandwidth overload and high delay, which determines that effective resource management has become a hot and difficult point in mobile cloud computing. This paper firstly analyzes the resource management requirements of mobile cloud computing on the basis of in-depth research on the existing cloud computing and mobile Internet technologies. Secondly, a multi-layer mapping management model is designed for the virtualized resource management platform. Then, a resource scheduling algorithm based on cellular particle swarm optimization and a resource allocation prediction algorithm based on wavelet decomposition and support vector machine are proposed. Finally, the correctness and superiority of the two algorithms are verified by experiments, and the effectiveness of the virtual resource management model is tested. The main innovations of this paper are as follows: (1) A hierarchical virtual resource management model is proposed. According to the granularity and flexibility of service scheduling, the service-resource mapping model is proposed, and the scheme of resource allocation is given, which realizes the fine-grained user request. (2) A virtual resource scheduling algorithm based on Cellular Particle Swarm Optimization (CPSO) is proposed. By analyzing the characteristics of the virtual resource scheduling problem, a reasonable multi-objective optimization model is constructed, and a reasonable and efficient resource scheduling algorithm is designed by combining the particle swarm optimization algorithm and cellular automata. (3) Resource allocation prediction algorithm based on wavelet decomposition and support vector machine is proposed. According to DB2 wavelet decomposition and SVM, a prediction algorithm based on wavelet decomposition-SVM modeling and wavelet reconstruction is proposed. The historical data of resource utilization can be trained and studied, and the prediction of resource allocation at the next time node can be realized accurately.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP3;TN929.5
本文编号：2307990
[Abstract]:In recent years, mobile Internet services are developing rapidly. Compared with PC and traditional Internet services, mobile cloud computing has the characteristics of thin terminal, network bandwidth overload and high delay, which determines that effective resource management has become a hot and difficult point in mobile cloud computing. This paper firstly analyzes the resource management requirements of mobile cloud computing on the basis of in-depth research on the existing cloud computing and mobile Internet technologies. Secondly, a multi-layer mapping management model is designed for the virtualized resource management platform. Then, a resource scheduling algorithm based on cellular particle swarm optimization and a resource allocation prediction algorithm based on wavelet decomposition and support vector machine are proposed. Finally, the correctness and superiority of the two algorithms are verified by experiments, and the effectiveness of the virtual resource management model is tested. The main innovations of this paper are as follows: (1) A hierarchical virtual resource management model is proposed. According to the granularity and flexibility of service scheduling, the service-resource mapping model is proposed, and the scheme of resource allocation is given, which realizes the fine-grained user request. (2) A virtual resource scheduling algorithm based on Cellular Particle Swarm Optimization (CPSO) is proposed. By analyzing the characteristics of the virtual resource scheduling problem, a reasonable multi-objective optimization model is constructed, and a reasonable and efficient resource scheduling algorithm is designed by combining the particle swarm optimization algorithm and cellular automata. (3) Resource allocation prediction algorithm based on wavelet decomposition and support vector machine is proposed. According to DB2 wavelet decomposition and SVM, a prediction algorithm based on wavelet decomposition-SVM modeling and wavelet reconstruction is proposed. The historical data of resource utilization can be trained and studied, and the prediction of resource allocation at the next time node can be realized accurately.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP3;TN929.5
【参考文献】
相关期刊论文 前6条
1 孙健;贾晓菁;;Google云计算平台的技术架构及对其成本的影响研究[J];电信科学;2010年01期
2 陈国宏;蔡彬清;李美娟;;元胞自动机:一种探索管理系统复杂性的有效工具[J];中国工程科学;2007年01期
3 张荣建;张志强;祖述勋;;钢管混凝土拱桥安全性评价的SVM机器模型[J];混凝土;2011年11期
4 刘勇;马良;;元胞微粒群算法及其在多维背包问题中的应用[J];管理科学学报;2011年01期
5 林立;邹昌伟;;基于Android平台的云计算研究[J];软件导刊;2010年11期
6 刘勇;马良;;非线性0-1规划的元胞蚁群算法[J];系统管理学报;2010年03期
相关硕士学位论文 前1条
1 王华军;基于小波分解和ARIMA的网络流量模型[D];山东大学;2011年
本文编号：2307990
本文链接：https://www.wllwen.com/kejilunwen/jisuanjikexuelunwen/2307990.html
