多层次分布实时仿真系统的静态调度策略  

Task Matching and Scheduling for Multi-level Distributed Real-time Simulation System

作　　者：潘玉林[1] 陈勇[1] 姚新宇[1] 黄柯棣[1] 

机构地区：[1]国防科技大学机电工程与自动化学院,长沙410073

出　　处：《系统仿真学报》2011年第1期85-88,94,共5页Journal of System Simulation

摘　　要：针对多层次分布实时仿真系统环境的特点,为了解决任务静态调度问题,提出了一种基于实时性分层的调度策略。该算法结合任务分层和启发式列表调度算法,在映射中同时考虑任务节点计算负载和通信开销映射到系统节点的等效负载,并映射当前最小总等效量到相应的系统节点上,从而获取较小的调度长度。实验结果表明,与交互优先算法相比该算法能提供较优的调度结果;相对于不考虑实时性而言,在调度中存在着一定的"不平衡"。To solve the problems of static tasks scheduling for multi-level distributed real-time simulation system, a novel scheduling strategy based on the real-time hierarchical was proposed. This strategy combines the lamination and heuristics list scheduling algorithm. The equivalent count that the calculate load and the communicate spending of simulation tasks mapped into the system nodes is considered, and loads the minimum task to the system node every time. Compared with the Interaction Priority Algorithm, the proposed strategy provides the less scheduling length, and there exists some ＂unbalance＂ in order to satisfy the requirement of real-time.

关 键 词：多层次 分布实时仿真 分层 单向性 调度长度 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]