A Dynamic Voltage Scaling Algorithm for Sporadic Tasks

  • Authors:
  • Ala' Qadi;Steve Goddard;Shane Farritor

  • Affiliations:
  • -;-;-

  • Venue:
  • RTSS '03 Proceedings of the 24th IEEE International Real-Time Systems Symposium
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Dynamic voltage scaling (DVS) algorithms save energyby scaling down the processor frequency when the processoris not fully loaded. Many algorithms have been proposedfor periodic and aperiodic task models but none support thecanonical sporadic task model. A DVS algorithm, calledDVSST, is presented that can be used with sporadic tasksin conjunction with preemptive EDF scheduling. The algorithmis proven to guarantee each task meets its deadlinewhile saving the maximum amount of energy possible withprocessor frequency scaling.DVSST was implemented in the µC/OS-II real-time operatingsystem for embedded systems and its overheadwas measured using a stand-alone Rabbit 2000 test board.Though theoretically optimal, the actual power savings realizedwith DVSST is a function of the sporadic task set andthe processor's DVS support. It is shown that the DVSST algorithmachieves 83% of the theoretical power savings for aRobotic Highway Safety Marker real-time application. Thedifference between the theoretical power savings and the actualpower savings is due to the limited number of frequencylevels the Rabbit 2000 processor supports.