A Game-Theoretic Formulation博弈论公式 多Agent资源分配

《A Game-Theoretic Formulation博弈论公式 多Agent资源分配》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、AGame-TheoreticFormulationofMulti-AgentResourceAllocationJonathanBredin†,RajivT.Maheswaran‡,C¸agriImer‡,TamerBas¸ar‡,DavidKotz†,andDanielaRus†ABSTRACTfromhosts.Weproposethatagentsuseasystemthatcre-atesincentivesforhoststoparticipate,providesagentswithThispaperconsidersresourceallocationin

2、anetworkwiththecostsoftheiractions,andallowsagentsofheterogeneousmobileagentscompetingforcomputationalpriority.Wefor-prioritiestooperatesimultaneously.mulatethisproblemasamulti-agentgamewiththeplayersWeconstructaresource-allocationpolicywherehostsbeingagentspurchasingservicefromacommonser

3、ver.WetakebidsfromagentsforprioritizedaccesstocomputationalshowthatthereexistsacomputableNashequilibriumwhenresources(CPUtime).Therateatwhichanagentcomputesagentshaveperfectinformationintothefuture.Fromourisproportionaltoitsbidrelativetothesumofallcurrentgame,webuildamarket-basedCPUalloca

4、tionpolicyandbidsatthehost.Hostscollectrevenuesfromeachagentatastrategywithwhichanagentmayplanitsexpendituresarateequaltotheagent’sbid.foramulti-hopitinerary.WesimulateanetworkofhostsWeapplythispolicytoamobile-agentsystemwithsev-andagentsusingourstrategytoshowthatourresource-eraldifferentt

5、ypesofresourcesdistributedthroughoutaallocationmechanismeffectivelyprioritizesagentsaccord-network.Eachagenthasasequenceofresourcestoconsumeingtotheirendowmentsandthatourplanningalgorithm(anitinerary)andanendowmentofelectroniccurrencytohandlesnetworkdelaygracefully.beusedtopurchaseresource

6、accesstocompleteitsitinerary.Weformulatethehosts’resource-allocationproblemas1.INTRODUCTIONagamewiththeplayersbeingagentscompetingforare-Mobile-agentsystemsallowprogramstoautonomouslysourcefromacommonserver.Weshowhowtocomputerelocatefromonehosttoanother.AnagentmayjumptotheuniquepositiveNa

7、shequilibriumexplicitlyunderper-onesitetofilteradatabase,jumptoanothersitetoaccessafectinformationwhentherearetwoormoreplayerstomin-camera,andtoathirdtoprocesstheresultsofthepreviousimizeanagent’sserialexecutiontime.Startingfromthistwohopsbeforereturningtheresultstot