Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2008
Session Number:P1
Session:
Number:P1-45
Energy-Aware Task Scheduling for Multi-Processor Systems considering Battery Lifetime
Il-jong Jung, Jong-wha Chong,
pp.-
Publication Date:2008/7/7
Online ISSN:2188-5079
Summary:
Increasing needs of large-scale tasks, using single processor in a system is inefficient because it causes waste of energy consumption due to high operating frequency. To overcome the defect of a single processor, multi-processor systems which can operate tasks on low frequency are required to minimize energy consumption. Since the multi-processor systems are expanded to portable devices such like PDAs, laptop computers and cellular phones, it is necessary to not only minimize energy consumption but also optimize battery lifetime because these devices are powered by batteries. This paper presents two solutions to maximize the portable device lifetime. First, the tasks are partitioned to subtasks have the different operating frequencies, which are higher than the previous subtask in a task. The operating frequencies and the subtasksy?execution cycles are decided by the formulation to guarantee the deadlines and the energy efficiency. This method maximizes the energy reduction because the latter subtasks operated by the higher frequency are not executed if tasksy?execution cycles are decreased. To minimize the average energy consumption, the tasksy?execution times are extended to the deadlines to lower the operating frequencies. Second, the frequencies of tasks scheduled in processors are adjusted by Dynamic Voltage Scaling (DVS) to avoid that the maximum frequency of each task overlaps. This process optimizes the battery lifetime, satisfying recovery effect. The simulation results show that the proposed method saves 15%~50% more energy than the existing method.