Abstract
Dynamic voltage scaling (DVS) has become the most efficient and im-portant technique for reducing energy consumption of processors, es-pecially in embedded hard real-time systems. A number of approaches have been proposed by different research groups to utilize DVS on different task models and system configurations. In this paper, we provide a comprehensive comparison of DVS-capable low-power real-time scheduling algorithms for a periodic fixed-priority task model. Our analysis focuses on their energy efficiency and performance com-parison. This paper intends to serve as a reference guild in the design of a low-power real-time system. In addition to qualitative analysis, we also conducted experiments to quantitatively compare the perfor-mance of these algorithms on two practical scenarios. Finally, we give the definition of a task set suitable for each algorithm.