On-Line Dynamic Voltage Scaling for Hard Real-Time Systems Using the EDF Algorithm

Abstract

Recently, there has been a rapid and wide spread of non-traditional computing platforms, especially mobile and portable computing devices. As applications become sophisticated and computation power increases, the most serious limitation on these devices is the available battery life. Dynamic voltage scaling (DVS) has been a key technique to exploit the hardware characteristics of processors to reduce energy dissipation by lowering the supply voltage and operating frequency. This paper presents a novel on-line DVS algorithm called OLDVS that, when coupled with the underlying OS task management mechanism and real-time scheduler, can make significant energy savings, while preserving timeliness guarantees made by the underlying real-time scheduling algorithm. While most existing DVS algorithms are confined to periodic tasks only, OLDVS does not assume task periodicity, nor does it require any a priori information on the task set to be scheduled. OLDVS requires only O(1) computation on each task context switch, thus making it fairly easy to be incorporated into a real-time operating system. The OLDVS algorithm considers a general task model which is very difficult, if not impossible, for the existing DVS algorithms to handle. Our simulation results show that OLDVS achieves great energy savings and outperforms the existing DVS algorithms when the ratio of the computation requirement of aperiodic tasks to the total computation requirement is higher than 40%. The performance advantage becomes much larger as the ratio increases.