Energy-modulated quality of service: New scheduling approach

A conventional computational system architecture is formed by a processor which executes tasks and a management unit which controls the task execution routine according to a required strategy. Different strategies (or task scheduling schemes) are developed to maximize system performance. The system performance can be maximized according to the specification of each algorithm, if a task is schedulable. Poor energy condition in energy harvesting systems is one of the factors which cause conventional schedulers to fail. This paper presents a new scheduling approach which helps computation to survive in any energy-scarce environment by sacrificing small amounts of quality of service. Systems operating under the proposed approach can dynamically adapt its computational activity to the energy conditions. The paper also contains simulation results of a special design of FFT processor which works as a load for the proposed scheduling approach.