Enhanced Energy-Aware Feedback Scheduling of Embedded Control Systems

Enhanced Energy-Aware Feedback Scheduling of Embedded Control Systems Feng Xia1,4, Longhua Ma2, Wenhong Zhao3, Youxian Sun2, Jinxiang Dong1 1College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China Email: f.xia@ieee.org 2State Key Lab of Industrial Control Technology, Zhejiang University, Hangzhou 310027, China Email: lhma@iipc.zju.edu.cn 3College of Mechanical and Electrical Engineering, Zhejiang University of Technology, Hangzhou 310032, China 4Faculty of Information Technology, Queensland University of Technology, Brisbane QLD 4001, Australia Abstract— Dynamic voltage scaling (DVS) is one of the most effective techniques for reducing energy consumption in embedded and real-time systems. However, traditional DVS algorithms have inherent limitations on their capability in energy saving since they rarely take into account the actual application requirements and often exploit fixed timing constraints of real-time tasks. Taking advantage of application adaptation, an enhanced energy-aware feedback scheduling (EEAFS) scheme is proposed, which integrates feedback scheduling with DVS. To achieve further reduction in energy consumption over pure DVS while not jeopardizing the quality of control, the sampling period of each control loop is adapted to its actual control performance, thus exploring flexible timing constraints on control tasks. Extensive simulation results are given to demonstrate the effectiveness of EEAFS under different scenarios. Compared with the optimal pure DVS scheme, EEAFS saves much more energy while yielding comparable control performance. Index Terms— Feedback Scheduling, Embedded Control Systems, Energy Management, Application Adaptation I. INTRODUCTION Power management has become a critical design issue, particularly in battery operated real-time embedded systems. Low power design not only reduces the operational cost but also increases the system reliability, while prolonging the battery’s lifetime [1]. Dynamic voltage scaling (DVS) [2,3] is one of the most effective approaches to power consumption reduction. However, conventional real-time DVS algorithms rarely take into account the resulting performance of target applications when determining the voltage level of the processor. Though much effort has been made on DVS for real-time applications, e.g. [2,4,5], state-of-the-art DVS algorithms usually rely on fixed timing constraints of real-time tasks. They typically derive the processor speed that provides timeliness guarantees during runtime according to pre- specified periods/deadlines of the task set, and these timing attributes will never be intentionally changed in favour of energy savings, e.g., in response to the actual application requirements. In practice, however, the resources that an application demands may vary over time. One representative example is control systems. From the control perspective, smaller sampling periods are beneficial to rapid recovery of steady states. Consequently, the negative effect of perturbations will be alleviated, and the quality-of-control (QoC) will then be improved. When the system is in a steady state, however, an unnecessarily small sampling period implies waste of resources (e.g., CPU time and energy). In this case, the sampling period may be enlarged to some extent without significantly degrading the control performance [6-10]. This feature of real-time control applications makes it possible to dynamically allocate CPU resource to each control task according to their real demands. Improving QoC and reducing energy consumption pose conflicting requirements. The objective of this paper is to develop an approach to reduce CPU energy consumption while preserving QoC guarantees. The effects of sampling periods on energy consumption and QoC will be exposed via motivating examples, respectively. An enhanced energy-aware feedback scheduling (EEAFS) scheme will be proposed, which takes advantage of application adaptation. In particular, the proposed scheme has the following features: x It integrates feedback scheduling with DVS, providing an effective way for managing QoC and energy consumption simultaneously in embedded real-time control systems. DVS provides an enabling technology for feedback schedulers to manipulate the tasks’ execution times, while feedback scheduling enables further energy savings over pure DVS schemes. x By exploiting direct feedback scheduling [11,12], the sampling periods of control loops (in addition to the CPU speed) are adjusted dynamically to make better compromise between application performance and energy consumption. In other words, the proposed scheme utilizes flexible timing constraints of real-time tasks to enhance the performance of DVS in saving energy. This is in contrast to most to appear in Journal of Computers. http://www.academypublisher.com/jcp/index.html previous DVS algorithms that

…(Full text truncated)…

This content is AI-processed based on ArXiv data.