单处理器多外设实时系统全局能耗优化算法研究

System-wide Energy Minimization for Uni-processor Real-time Systems with Multiple Devices

在电池供电的实时嵌入式系统中,能耗是系统设计的一个重要研究问题.动态电压调度和动态电源管理是两种重要的节能技术.前者是动态改变处理器电压/频率,降低处理器能耗;而后者是动态调整片外设备的工作模式,减少片外设备能耗.目前只有少量研究把这两种技术综合在一起.本文研究支持这两种技术的嵌入式全系统实时任务节能调度问题.针对连续和离散处理器频率模型,论文分别提出高效的算法,通过计算系统运行的能耗最小处理器最优频率和设备最优空闲时间,来实现全系统综合节能的目的.实验模拟表明本文算法大大优于其他算法.

Energy-efficiency has become one of the most important design issues of modem battery-powered real-time embedded sys- tems. Dynamic Voltage Scaling （DVS） and Dynamic Power Management （ DPM } are two efficient techniques for reducing the system-level energy consumption. DVS aims at reducing the processor energy by adjusting the supply voltage/frequency dynamically, while DPM involves putting the off-chip devices to low-power states at run time. In this paper, we investigate the problem of systemwide energy-efficient scheduling for frame-based real-time tasks running on a DVS capable CPU with potentially multiple devices. Several efficient algorithms are proposed to determine the optimal frequency of the processor and idle time of devices resulting in minimum system-wide energy consumption for the continuous and discrete frequency model respectively. Extensive experiments are conducted and the results show that our approach significantly outperforms existing approaches in terms of system-wide energy savings.