基于云计算的视频监控系统能耗优化研究

项目名称： 基于云计算的视频监控系统能耗优化研究

项目编号： No.61202340

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 熊永华

作者单位： 中国地质大学（武汉）

项目金额： 23万元

中文摘要： 针对基于云计算的视频监控系统，从监控节点、物理节点和存储节点三个层面研究其能耗优化方法。通过研究基于数据驱动的视频质量和综合能耗建模方法，建立兼顾视频质量和综合能耗的多目标优化模型，设计监控节点低能耗部署策略，在保证视频质量的同时降低监控节点能耗；通过分析监控任务对虚拟机的性能要求，建立基于专家知识的虚拟机接入能力智能评价模型，提出基于虚拟机超载系数的模型动态更新方法，设计面向负载均衡的虚拟机接入调度策略，在兼顾服务质量的同时减少虚拟机数量，从而减少所需物理节点数量，降低能耗；通过设计基于SLA中访问时间约束的静态分类方法和基于访问时间特征聚类的动态迁移机制，提出融合静态分类与动态迁移的存储管理方法，建立基于运行计划表的存储节点运行策略，在符合用户访问需求的同时，减少存储节点能耗。通过理论研究与实验验证，形成对云视频监控系统能耗优化的一套有效的和可行的新方法，具有重要的理论意义和应用价值。

中文关键词： 视频监控；云计算；能耗；优化；

英文摘要： This project studies the energy consumption optimization methods for cloud-computing-based video surveillance systems from three levels, including monitoring nodes, physical nodes and storage nodes. By studying data-driven approaches of modeling the video quality and overall energy consumption of monitoring node, a multi-objective optimization model considering both video quality and energy consumption is established. A low-energy-consumption deployment strategy of monitoring nodes is designed to reduce the energy consumption of the monitoring node, and to ensure the video quality. Through analyzing the performance requirements of virtual machine coming from monitoring tasks, an expert-knowledge-based intelligent evaluation model of virtual machine accessibility is built, which can be updated by a dynamic update method based on the virtual machine overload factor. A virtual machine access scheduling strategy oriented to load balance is devised to cut down the number of virtual machines, with the quality of service guaranteed. Thus both the required number of physical nodes and energy consumption will decrease. By designing static classification methods based on access time constraints in Service Level Agreement (SLA), as well as developing dynamic migration mechanism based on access time feature clustering, a st

英文关键词： video surveillance；cloud computing；energy consumption；optimization；