A Power Consumption Perception and Dynamic Adjustment Model for Embedded Real-Time Systems.
Yingjie Wang,Fengbo Bai,Zumin Wang,Zixu Wang,Zizhong Wang
DOI: https://doi.org/10.1109/hpcc-dss-smartcity-dependsys53884.2021.00317
2021-01-01
Abstract:Embedded real-time systems are usually adopted on platforms with constrained resources in energy, but these systems have been required more and more powerful, of which, the powerful functions would inevitably increase the power consumption. Therefore, reduction of the power consumption for an embedded system becomes as important as the accomplishment of the task execution deadline. That means, the strategy of task scheduling technology and selection of voltage/frequency mechanism should be considered together, so that scheduling could provide an approach to achieve system energy saving. In the practice of this topic for energy consumption optimization, experimental facts have shown that the adoption with combining the Dynamic Voltage and Frequency Scaling (DVFS) technology and scheduling strategies could be a promising method to reduce system energy consumption. In this paper, an energy-aware scheduling model of embedded real-time system adopting the combination with dynamic power management is proposed, to ensure the tasks executed before the deadline and minimize processor energy consumption. Based on both considering aspects of timing control and energy consumption control, a dynamic power management scheduling algorithm is designed to sufficiently explore the adoption of the DVFS technology, to reduce the energy consumption, with fulfilling the execution deadline requirement and energy consumption adjustment, under the condition of multiple preemptive tasks.