华南理工大学学报(自然科学版) ›› 2015, Vol. 43 ›› Issue (9): 81-87.doi: 10.3969/j.issn.1000-565X.2015.09.013

• 计算机科学与技术 • 上一篇    下一篇

基于 PI 反馈的分布式控制系统动态负载均衡算法

汤峰 张平李方 黄致祥   

  1. 华南理工大学 计算机科学与工程学院,广东 广州 510640
  • 收稿日期:2015-01-22 修回日期:2015-06-19 出版日期:2015-09-25 发布日期:2015-09-07
  • 通信作者: 张平(1964-),男,博士,教授,主要从事机器人控制研究. E-mail: pzhang@scut.edu. cn
  • 作者简介:汤峰(1979-),女,博士生,工程师,主要从事机器人控制研究. E-mail: fengtang@ scut.edu.cn
  • 基金资助:
    广东省科技计划项目(2014B090921007);广州市科技计划项目(20150810068);广州市海珠区科技计划项目
    (2014-cg-02)

PI Feedback-Based Dynamic Load-Balancing Algorithm for Distributed Control System

Tang Feng  Zhang Ping  Li Fang  Huang Zhi-xiang   

  1. School of Computer Science and Technology,South China University of Technology,Guangzhou 510640,Guangdong,China
  • Received:2015-01-22 Revised:2015-06-19 Online:2015-09-25 Published:2015-09-07
  • Contact: 张平(1964-),男,博士,教授,主要从事机器人控制研究. E-mail: pzhang@scut.edu. cn
  • About author:汤峰(1979-),女,博士生,工程师,主要从事机器人控制研究. E-mail: fengtang@ scut.edu.cn
  • Supported by:
    Supported by the Science and Technology Planning Projects of Guangdong Province(2014B090921007)

摘要: 针对分布式控制系统由于负载不均衡、网络通信量大等引起的时延问题,文中设计了基于请求划分的任务分配模型,提出了基于实时动态比例积分(PI) 反馈控制的负载均衡算法. 该算法利用增量 PI 控制的思想,根据服务器节点性能的实时反馈值动态调节服务器节点的分配权值,通过虚拟节点转移算法局部调整虚拟节点的分配,以维护哈希空间的稳定. 仿真实验结果表明,该算法实现了分布式控制系统的动态负载均衡,减小了服务器资源消耗及用于存取远程数据的通信开销,提高了控制系统的实时性,具有良好的扩展性和容错性.

关键词: 分布式控制系统, 负载均衡, 哈希函数, 反馈控制

Abstract: In order to solve the problem of slow response caused by the load imbalance and the communication
overhead in distributed control systems,a task allocation model is constructed on the basis of the request division,and a dynamic load-balancing algorithm is proposed on the basis of the real-time dynamic proportional integral (PI)feedback control. This algorithm adopts the PI control method to dynamically adjust the allocation weight of server nodes according to the real-time feedback values of the performance of the nodes,and then employs the virtual node transfer algorithm to partially adjust the distribution of virtual nodes,so as to maintain the stability of Hash space.Simulation results show that the proposed algorithm realizes the dynamic load balancing of distributed control system,reduces the communication overhead and improves the real-time performance of the control system,and that the proposed algorithm is of a high expansion and an excellent fault tolerance.

Key words: distributed control systems, load balancing, Hash functions, feedback control