摘要
大多分布式资源调度策略的目标或为提高分布式系统的运行效率,或为提高用户的满意度.文中提出了一种综合考虑资源可信度与用户满意度的调度策略,以用户的满意度为目标函数,且将资源可信度作为用户满意度组成的一部分,通过遗传算法实现优化调度.继而将此调度策略封装进Agent之中,并提出了一种基于Multi-agent分布式系统架构,该架构由用户、作业和资源3类Agent组成,并且具有良好的分布式群集智能效应和自适应能力.算例结果显示,此调度策略不仅提高了系统运行效率,同时增加了用户满意度.
Most distributed resource scheduling strategies consider either improving the efficiency of the system or enhancing user satisfaction.This paper presents a new scheduling strategy integrating resource credibility and user satisfaction.User satisfaction is set to be the objective function of genetic algorithm and resource credibility is made part of it.By packaging this scheduling policy into the agent,a new multi-agent-based distributed system architecture is proposed,which consists of three types of agents： user,job and resource,and has a good distributed swarm intelligence effect and adaptive ability.The simulation results demonstrate that the architecture can improve the efficiency of the system,while increasing customer satisfaction.
出处
《西南大学学报（自然科学版）》
CAS
CSCD
北大核心
2013年第8期155-160,共6页
Journal of Southwest University(Natural Science Edition)
基金
国家自然科学基金-专项基金项目(11047150)
东莞市高等院校科研机构科技计划重点项目(2011108101010)
广东省高新技术产业化项目-工业攻关项目(2012B010100050)
