网格资源管理关键技术研究

《网格资源管理关键技术研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、电子科技大学博士学位论文网格资源管理关键技术研究姓名：陈佳申请学位级别：博士专业：计算机软件与理论指导教师：吴跃20080801摘要移动Agent的网格资源管理模型中，结合遗传算法的基本理论，充分利用遗传算法固有的分布式与并行性特征，来解决网格资源调度中具有的很强的分布式特征以及网格作业分配中的并发特性。并且遗传算法具有较强的健壮性，对初始化条件要求不高，运行过程中无须干预等特点，使得遗传算法在复杂的网格环境下更容易控制。因此遗传算法适合于解决网格计算环境下的资源调度问题。进一步地，使用改进的并行遗传算法来提高资源调度效率。关键词：网格资源管理、移动Agent、语义、本体

2、、遗传算法ⅡABSTRACTGridistouseInternettechnologytoconnect，assign，manageandcoordinatethedifferentresourceswhicharegeographicallydispersedindifferentlocations，includingcomputingresource，storageresource，communicationresource,softwareresource,informationresourceandknowledgeresourceetc．Andalltheres

3、ourcescomposeavirtualsupercomputerviathelogicrelations．ItmakestheeffectivemanagementandschedulingofGridresourcemorecomplexthantraditionaldistributednetworkduetothedistribution，heterogeneity,dynamic，autonomyandothercharacteristicsofGridresource．ConsequentlytheGridresourcemanagementbecomesa

4、nimportantresearchdomaininGridresearch．BasedonasystematicalsummaryofrelevantworksonGridresourcemanagement，thisdissertationfocuses0nfourkeytechnologiesinGridresourcemanagement．nlemajorcontributionsofthisdissertationaleasfollows：(1)ⅢsdissertationproposesaGridresourcemanagementmodelbasedonmo

5、bileagent．Itutilizestheautonomyandintelligenceofmobileagenttomanagethedistributed，heterogeneous，autonomousanddynamicGridresource．Itisanewmanagementmodelandafeasiblemanagementmethod．(2)ThisdissertationpresentsahierarchyGridresourceorganizationmodelbasedonsemantics．Thereexistspoorexpansibil

6、ity,overloadbandwidthetc．incurrentGridresourceorganizationmodel．Tosolvetheseproblems，thisdissertationinductsthesemantictreeorganizationmarlrlertoorganizetheGridresourcebasedontheGridresurceclusteringresultinthemobileagent-basedGridresourcemanagementmodel．Themodelprovidesagoodorganizationi

7、nfrastructuretotheGridresourcesearch．111edissertationvalidatesthevalidityofthemodelviaexperiment．(3)TllisdissertationproposesaGridresourcesearchstrategybasedonontology．ThereexistspoorreliabilityandlowefficiencyincurrentGridresourcesearchmethods．乃estrategyinductsonto