计算机集成制造系统 ›› 2018, Vol. 24 ›› Issue (第6): 1445-1454.DOI: 10.13196/j.cims.2018.06.014

• 当期目次 • 上一篇    下一篇

开放式数控系统的低能耗和可靠性调度算法

邓昌义1,2,郭锐锋1,2,段立明1,2,王帅1,2,杜少华3   

  1. 1.中国科学院沈阳计算技术研究所
    2.中国科学院大学
    3.沈阳高精数控智能技术股份有限公司
  • 出版日期:2018-06-30 发布日期:2018-06-30
  • 基金资助:
    国家科技重大专项资助项目(2014ZX04014-021)。

Reliability scheduling algorithm for low power and reliability for open CNC system

  • Online:2018-06-30 Published:2018-06-30
  • Supported by:
    Project supported by the National Science and Technology Major Project,China(No.2014ZX04014-021).

摘要: 为充分利用多核处理器的并行能力,同时在保证数控系统可靠性约束下实现系统低能耗。在任务级粗粒度软件流水线技术基础上,首先提出非依赖有向无环图算法将周期性依赖任务转换为基于重定时的一组独立任务,通过任务并发执行和动态电压调节技术降低系统能耗。其次,在保证系统可靠性方面,算法分为无错和容错两个阶段。在无错阶段,一个任务的多个副本同时执行,通过投票确定任务是否执行正确。当不一致时,任务进入容错模式。在容错模式下,剩余副本独占处理器以最高速度完成容错。实验结果表明,所提出的算法在保证数控系统可靠性前提下,可以降低系统能耗,相比已有算法不仅具有更高的可靠性而且能耗更低。

关键词: 数控系统, 多核处理器, 实时操作系统, 能耗, 可靠性, 调度

Abstract: To make full use of parallel ability of multicore for ensuring the real-time system reliability constraints and low power consumption,a novel non-dependent directed acyclic graph algorithm was given based on coarse-grained software pipelining,which could transform a periodic dependency task graph into a set of independent tasks with retiming technique.The system energy consumption could be reduced by task parallel execution and dynamic voltage scaling.In terms of guaranteeing system reliability,the algorithm was divided into two phases: fault-free and fault-tolerant.In the fault-free phase,multiple versions of task were executed at the same time,and results of multiple versions were used for the majority voting to determine whether the task had been executed successfully.If not,the task would enter the fault-tolerant mode.In the fault-tolerant mode,the remaining copies performed the fault tolerance at the highest speed in an exclusive processor.Experimental results showed that the proposed algorithm could reduce the system energy consumption under premise of guaranteeing the reliability of system,which yielded high-reliability and lower energy consumption than the existing algorithms.

Key words: numerical control system, multicore processor, real-time operating system, energy consumption, reliability, scheduling

中图分类号: