A Practical Dynamic Frequency Scaling Scheduling Algorithm for General Purpose Embedded Operating System

Dynamic frequency scaling (DFS) techniques for real-time embedded systems have been widely studied. However, most of the scheduling algorithms so far concern only special purpose real-time systems. In this paper we devise a power-aware parameter task model for time-sharing real-time systems of general purpose, as well as a scheduling algorithm that is based on the task model. We implement the model and corresponding algorithm on embedded Linux. With the energy consumption measurement method presented in this paper, the experiment shows that for real-time tasks in Linux, 31.7% power consumption can be saved compared with the circumstance under which the processor runs at highest frequency.