A car racing based strategy for the Dynamic Voltage and Frequency Scaling technique

This work proposes the application of a car racing analogy to develop real-time applications with low energy consumption using the Dynamic Voltage and Frequency Scaling technique. Although several previous works had dealt with the DVFS technique, our proposed method offers two main improvements. First, it can be applied to multiple preemptable real-time tasks. Second, in our method, energy consumption optimization may be carried out even when the worst case execution path is followed. The proposed method takes into account the limited set of available frequencies, and consider both the timing and energy overheads for changing the frequency/voltage. Our simulation results indicate clearly the practical usability, efficiency, and the advantages of the proposed method.