资源描述:
《maximum likelihood estimation of peers’ performance in p2p networks》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、MaximumLikelihoodEstimationofPeers'PerformanceinP2PNetworksZoranDespotovic,KarlAbererEPFL-SwissFederalInstituteofTechnologyLausanne,Switzerlandemail:fzoran.despotovic,karl.abererg@ep°.chAbstractbemorespeci¯candconsideronlyP2Pnetworks.Wewill¯rstreviewtherelevantliterature(
2、Section3)andTheproblemofencouragingtrustworthybehaviorino®eraviewonhowvariousP2Preputationmanage-P2Ponlinecommunitiesbymanagingpeers'reputa-mentapproachescontributetobuildingtrust.Aswetionshasdrawnalotofattentionrecently.However,willsee,mostofthethemsu®erfromthefollowingt
3、womostoftheproposedsolutionsexhibitthefollowingtwoproblems:hugeimplementationoverheadandunclearproblems:hugeimplementationoverheadanduncleartrustrelatedmodelsemantics.Themaincauseofthetrustrelatedmodelsemantics.Inthispaperweshow¯rstproblemliesinthenecessityofaggregatingth
4、ethatasimpleprobabilistictechnique,maximumlikeli-feedbackaboutallpeersinthenetworkinordertoas-hoodestimationnamely,canreducethesetwoproblemssessthetrustworthinessofasinglepeer,whilethesec-substantiallywhenemployedasthefeedbackaggregationondproblemismainlycausedbythecounte
5、rintuitivestrategy.Thus,nocomplexexplorationofthefeedbackfeedbackaggregationstrategiesresultingintheoutputsisnecessary.Instead,simple,intuitiveande±cientthatarehardtointerpret.Inthispaperweshowthatprobabilisticestimationmethodssu±ce.asimpleprobabilistictechnique,maximumli
6、kelihoodestimationnamely,canreducethesetwoproblemssub-1.Introductionstantiallywhenemployedasthefeedbackaggregationstrategy.Operatingonasmallfractionofthefeed-Recentempiricalstudieshaveshownthatmuchofbackavailableinthenetwork,itlendsitselftoanef-eBay'scommercialsuccesscanb
7、eattributedtoitsrep-¯cientimplementation.Ontheotherhand,itsout-utationmechanism(FeedbackForum)asameansofde-putsareprobabilitiesofspeci¯cbehaviorsofthepeersterringdishonestbehavior.Thus,[16]showsthatrep-andassuchhaveaclearandwellfoundedinterpreta-utationpro¯lesarepredicti
8、veoffutureperformance",tion.Finally,itsabilitytodetectpeers'misbehaviorwhile[9]and[12]comeupwith