网格环境下基于序贯博弈的性能-效率平衡型优化  

Performance-Efficiency Balanced Optimization Based on Sequential Game in Grid Environments

作　　者：郑美光[1] 胡志刚[1] 张凯[2] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083 [2]南京中兴通讯中心研究院3G平台软件四部,江苏南京210012

出　　处：《华南理工大学学报（自然科学版）》2010年第1期92-96,107,共6页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(60673165)

摘　　要：为优化应用执行性能和提高系统资源效率,针对网格环境下的资源映射问题,文中提出了一种基于序贯博弈的优化策略.该策略根据平衡的思想,将资源映射过程分为活动分发和处理器分配两个阶段,通过有限次博弈后获得优化解.文中还给出了衡量应用执行性能的重要参数——信任度的计算方法.仿真实验检验了优化策略的可行性和有效性.结果表明,与Min-Min算法相比,文中的优化策略具有更低的时间复杂度、更优的应用执行性能和更高的资源效率.In order to optimize the executing performance of applications and improve the resource utilization efficiency of a system, this paper deals with the resource mapping in grid environments and proposes a novel optimization strategy based on the sequential game. In this strategy, a balanced idea is used to divide the mapping process into two stages, namely the activity distribution and the processor allocation, and an optimal solution is obtained after a limited stage gaming. Moreover, trust degree, an important parameter to measure the executing performance of applications, is dealt with, and the corresponding calculating method is presented. Experimental results indicate that, as compared with the Min-Min strategy, the proposed optimal strategy is of lower time complexity, better executing performance of applications and higher resource utilization efficiency.

关 键 词：网格计算 性能-效率平衡型优化 序贯博弈 资源映射 信任度 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]