Energy-efficient scheduling with reliability guarantee in embedded real-time systems

Embedded systems have been widely applied in many areas such as aerospace, intelligent transportation, smart grid, and, medical care. However, limited energy supply and runtime faults are two challenges that hinder the embedded system to be even more pervasively applied. In this paper a greedy energy-efficient (GEE) algorithm is proposed to improve the energy efficiency under the constraint of reliability. As GEE may lead to more preemptions with reduced execution frequency, this paper analyzes the preemption during the execution and proposes a series of constraints to reduce the frequency of preemption. This reduces both the energy consumption and execution time. To further balance the execution frequency, two energy-efficient algorithms are presented based the processor utilization. Finally, considering the fact that the actual execution time (AET) of many tasks may be less than the worst-case execution time (WCET), a global dynamic scheduling algorithm is proposed to maximize the energy efficiency while ensuring the reliability. The experimental results show that the proposed techniques have significant improvements of energy efficiency with guaranteed system reliability.