带能量供给与变速延迟约束的电压调整节能调度算法研究

Mobile applications and smart phones are widespread nowadays and hence the battery life of the devices has become one of the greatest concern for the consumers. On the other hand, the big data processing and cloud computing has been emergency demand for the enterprises, but the data center needs much energy costs. The voltage scaling based energy-saving technique can effectively reduce the energy for the processors, which is one of the most important topics in the theoretical research. In the literature of dynamic voltage scheduling(DVS), few works have simultaneously considered considering the energy supply and speed-adjusting delay in the processors. This project intends to focus on such constraints. While satisfying the time constraints of the job execution, the objective is to optimize the energy consumption. We will model the physical characteristics of the speed-adjusting delay and energy supply. By adopting the techniques of structure analysis and algorithm design/analysis we will study the structural properties of the min-energy scheduling. The output is to design the optimal min-energy scheduling algorithms and examine the optimality of the algorithm. The project will extend the research of voltage scaling scheduling to the actual physical-constrained environment, efficiently enlarge the battery life of the devices and reduce the costs for data centers. In summery, the research will render several effective and valuable solutions and enrich the dynamic voltage scheduling.

当前无线设备电池供电时间有限、大规模数据计算的能耗成本高昂，基于电压调整的节能调度研究能有效地从处理器运行层面就开始优化能量使用。但目前的电压调整节能调度算法研究，尚未深入研究处理器中动态能量供给与变速延迟约束所带来的影响。本项目拟通过对动态能量供给与变速延迟约束进行建模，真实反映处理器所处的实际物理约束条件。以满足所有工作在截止时间前完成为前提，在动态能量供给与变速延迟约束下，最优化处理器工作执行所带来的能耗。采用结构分析、算法设计与分析的研究方法，分析最优调度的结构特点，设计算法以计算最优调度，证明算法的最优性。本项目的研究将拓展电压调整节能调度研究到更为贴近实际物理约束的情况下，为移动设备、数据中心节点的能耗优化提供切实有效的理论参考。

当前移动应用已广泛普及，智能手机、平板电脑等移动设备的电池使用时间成为消费者最关注的问题之一。基于速率调节的调度策略优化可以有效降低无线设备在任务执行和数据传输中的能耗，是理论研究的重要课题之一。本研究针对以上问题，从能量采集、数据共享等特征建模开始入手，在满足工作集时间约束的前提下，最优化能量使用效率。本研究证明了能耗最优调度的最优结构，提出最优算法和近似算法以逼近最优解，从理论和实验角度证实所提出算法的有效性。本研究的研究拓展节能调度研究到能量采集、延迟、共享环境等实际物理约束环境下。