Chinese Journal of Computers   Full Text
Title	A Voltage Allocation Technique with Checkpoint-Based Schedulability Test in Embedded Real-Time Systems
Authors	LI Guo-Hui1) YANG Bing1) HU Fang-Xiao1) XU Hua-Jie2) DU Jian-Qiang3)
Address	1)(School of Computer Science & Technology, Huazhong University of Science & Technology, Wuhan 430074) 2)(School of Computer and Information, Shanghai Second Polytechnic University, Shanghai 201209) 3)(Department of Computer Science, Jiangxi University of Traditional Chinese Medicine, Nanchang 330006)
Year	2009
Issue	No.12(2403—2410)
Abstract &   Background	Abstract Fault-tolerance through software/hardware redundancy as well as power management through frequency and voltage scaling has been well studied separately in the context of embedded real-time systems. However, the real-time characteristic, the high-level reliability and the low power consumption make the combination of fault-tolerance and power management necessarily for embedded real-time systems. This paper proposes a new method for aperiodic task scheduling and voltage allocation called voltage allocation with checkpoint-based schedulability test (CST-VA). CST-VA not only guarantees the timing constrains, also provides higher reliability through checkpointing and minimizes power consumption via optimal voltage allocation technique. Simulation results show that CST-VA has better performance compared with the existing voltage allocation technique and is more appropriate for embedded real-time systems. Keywords embedded real-time systems; aperiodic tasks; fault-tolerance; power management; voltage allocation Background The work is supported in part by the National Natural Science Foundation of China under the grant No.60873030; the National High Technology Research and Development Program (863 Program) of China under the grant No.2007AA01Z309; and Special Funds of Central Colleges for Basic Scientific Research and Operating Expenses. The problem of task scheduling and voltage allocation, which should minimize the total processor energy consumption while guaranteeing timing constraints of tasks, is a representative requirement for embedded real-time systems. At present, researchers are interested in the trade-off between system reliability and energy consumption for real-time systems. The slack time in real-time systems can be used by recovery schemes to increase system reliability as well as by frequency and voltage scaling techniques to save energy. Moreover, the rate of transient faults also depends on system operating frequency and supply voltage. So, it gives us the new challenge for considering the combination of system reliability and energy consumption for the problem of task scheduling and voltage allocation. Fault tolerance through redundancy as well as energy management through frequency and voltage scaling has been well studied separately in the context of real-time embedded systems. However, there are relatively less researches addressing the combination of fault tolerance and energy management. Furthermore, most of the existing methods are concerned about periodic task set only. In many real-time embedded systems, tasks can be classified as periodic tasks and aperiodic tasks. Aperiodic tasks can be processed with general scheduling algorithms. However, in many cases, it is more efficient to process aperiodic tasks with separate algorithms which make use of the knowledge that aperiodic tasks may themselves have a variety of timing requirements. This paper focuses on the variable voltage allocation problem for aperiodic task set in embedded real-time systems based on earliest deadline first (EDF) scheduling strategy. First, the authors present a checkpointing-based schedulability test, and with this test, provide a deterministic guarantee and get the optimal number of checkpoints that a task should be inserted if possible. The optimal number of checkpoints can help the task to guarantee the timing constraints and minimize the worst case execution time in the present of transient faults. Based on the checkpointing-based schedulability test, the authors propose a voltage allocation method through dynamic voltage and frequency scaling (DVFS) technique. The method not only guarantees timing constraints, but also provides higher system reliability through checkpointing and minimizes energy consumption via optimal voltage allocation techniques. This paper contributes on the problem of task scheduling and voltage allocation focusing on aperiodic tasks for real-time embedded systems. The simulation results show that the proposed method has significantly better performance compared with the existing voltage allocation technique and is more appropriate for embedded real-time systems.