基于多Agent的分布式多目标任务调度机制研究  

Research on multi-agent-based distributed task scheduling mechanism with multi-objective

作　　者：王占杰[1] 刘晶晶[1] 

机构地区：[1]大连理工大学计算机科学与技术学院

出　　处：《大连理工大学学报》2011年第5期755-760,共6页Journal of Dalian University of Technology

基　　金：国家自然科学基金资助项目(90718030)

摘　　要：针对分布式计算中任务调度的复杂性、动态性和实时性等特点，构建了一个基于多Agent的分布式多目标任务调度模型．此模型中，设计了一种分布武调度机制，使每个节点都具有独立的调度决策能力，增强了系统的可扩展性．同时为了解决分布式调度所产生的节点间任务通信问题，引入任务与节点相关性等概念，有效地降低了节点间的通信代价．模型对多个评价指标进行了综合衡量，如任务调度开销、任务最早执行时间、负载均衡及节点问的通信量等，并根据任务和系统运行的状态特征，实现多目标任务调度，提高了系统的整体性能．A task scheduling model of distributed multi-objective based on multi-agent is built for the features of task scheduling in distributed computing, such as complexity, dynamicity and immediacy. In the model, a distributed scheduling mechanism is designed to make every node have independent scheduling capability, and the extensibility of the system is enhanced. Meanwhile, in order to solve the problem of communication costs caused by distributed scheduling, a new concept called correlation between tasks and nodes, which indicates communication costs between different nodes, is proposed, thus the communication costs are effectively decreased. This model makes synthesis measurement on several indexes, such as task scheduling costs, earliest task execution time, load balancing, communications volume between different nodes and so on. This model also achieves multi-objective task scheduling based on the execution features of tasks and systems in order to improve the whole performance of the system.

关 键 词：分布式任务调度 多AGENT 任务与节点相关性 多目标调度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]