Affiliations: Electronics and Telecommunications Research Institute,
Daejeon, 305-720, South Korea | Information and Communications University, Daejeon,
305-714, South Korea | Department of Computer Science Engineering, Arizona
State University, Tempe, AZ 85287-5406, USA | Electrical and Computer Engineering Department
University of Massachusetts, Amherst, MA 01003, USA
Abstract: For energy-constrained real-time embedded systems, the power-delay
tradeoff property of Volta ge-Clock Scaling (VCS) needs to be carefully
considered in scheduling real-time tasks for meeting strict timing
requirements. In addition, non-real-time tasks require minimized response
times. Thus, a bounded energy allocation model should be adopted to solve a
combination of optimization objectives in systems having mixed hard and soft
real-time tasks. In this paper, we propose an energy sharing model that
allocates the energy budget among hard and soft real-time tasks, exploiting the
interplay between utilization and energy consumption in VCS-based
earliest-deadline-first scheduling (VCS-EDF). Also a dynamic scheduling is
presented which is designed to reduce energy consumption by switching between
two scheduling policies and utilizing an explicit pattern of event occurrences
at run-time. Through simulations, we show that this Dual-Policy Dynamic
Scheduling can outperform in the reduction of energy consumption while
introducing small delay in average response time of non-real-time tasks.
Keywords: bounded energy, energy allocation, energy saving, Voltage-Clock Scaling, VCS-EDF scheduling