仿真网格中一种基于知识的动态任务调度算法

仿真网格是以通用网格技术为基础、面向仿真领域的专用网格,目前国际上对仿真网格的研究尚处于起步阶段.现有的分布式仿真HLA(high level architecture)体系结构中的仿真资源和联邦成员是静态绑定的,网格技术的引入使得仿真资源的动态分配成为可能.根据仿真网格任务调度的特点,在仿真网格中建立了一种任务调度模型,并针对该模型,提出了一种新的基于知识的动态任务调度算法KMO,该算法适用于将N个相互独立的计算需求不同的仿真任务调度到M个随时间动态变化的仿真资源上,它能对若干次调度后的结果进行统计并提炼成"知识"反馈给算法预处理部分,使得该算法在动态多变的环境中能获得比较稳定的性能.实验结果表明,在仿真网格环境中,该算法的性能优于网格中传统的任务调度算法.

A Dynamic Knowledge-Based Task Scheduling Algorithm in Simulation Grid Environment

Simulation grid is a specific grid which is based on general grid technology and oriented towards simulation.At present,researches on simulation grid are still at the fledgling stage.Simulation resources as well as federates in HLA-based distributed simulation are statically bound now.Introduction of grid technology makes the dynamic allocation of the simulation resources come to reality.However,there is little relevant literature discussing task scheduling in simulation grid now.Therefore,it's a new research field of high practical value.Simulation grid provides a flexible and powerful simulation platform to execute large-scale simulation applications.The focus here is on the task scheduling in simulation grid environment.In this paper,a new task scheduling model in simulation grid is built firstly,and a novel knowledge-based dynamic task scheduling algorithm KMO is proposed for scheduling N independent tasks with the different length onto a simulation gird with M resources whose computing power varies over time.In this algorithm,the results of several scheduling are collected and refined to form knowledge,which is then used in pretreatment stage in the algorithm.Finally,the experiment result shows that KMO is better than the traditional task scheduling algorithms in simulation grid environment.