数据操作系统中基于博弈的多资源分配算法.pdfVIP

Vol. 3 No.2 去A赞主期 Mar.2014 网络新媒体技术 数据操作系统中基于博弈的多资源分配算法* 张潇王凯武延军 (中国科学院软件研究所总体部北京 1 ∞190) 摘要:数据操作系统需要对 CPU 和内存等多种系统资源进行管理，为了在公平的前提下，解决不同用户对上述资源的不同需 求问题，本文基于完全信息的动态博弈提出了 ICEEI 算法。该方法通过构建博弈树并优化博弈树的方法解决数据操作系统中 的多资掠分配问题。该算法最符合实际的假设是认为任务是不可分的，即数据操作系统分配给每个用户的资源可完全满足 任务的需求。对该算法的公平性进行了讨论，指出其满足共享激励和 Parelo 有效等性质。通过一系列的仿真实验，证明 ICEEI 可以很好地应对用户对资源需求的动态变化，并且与 DRF 相比在有些情况下具有更高的资源利用率。 关键词:博弈论，多资源分配，敏据操作系统 A Multiple Resource Allocation Algorithm ßased on Game Theory in DataOS ZHANG Xiao , WNAG Kai , WU Yanjun (InSlilute of Software Chinese Academy of 5ciences , ZhongGuanCun South 4 Road 桐， Bèiji吨， 1∞190 ， China) Abstract: We consider the problem of how a system allocales resources to difIerenl agents fairly and efficiently in Data 05 , where the system contains difIerent resource types such as CPU and Memory , and agents may have difIerent demands for the resources. In this pa- per , we pro归se the ICEEI algorithm based on dynamic game theory with complete information which solves the multiple 陀source all0- cation problem. The algorithm assumes 出at the task is indivisible , such that the resources the system allocated to each agent fully sat- isfy the demand of each task. Then we consider the faimess problem of the algorithm and prove that it satisfies Sharing incentive and p缸eto efficiency. In addition , we carηout experiments to verify the proposed algorithm , which show that the ICEEI allocation can effi- ciently adjusts to the dynamic change of agents demand for 陀sources ， and use resources more efficiently