学术英语论文.docx

学术英语论文.docx

《学术英语论文.docx》由会员分享，可在线阅读，更多相关《学术英语论文.docx（7页珍藏版）》请在冰点文库上搜索。

学术英语论文

NANCHANGUNIVERSITY

课程名称：

学术英语

题目：

AStudyofEnergyEfficient_

CloudComputingPoweredby

_WirelessEnergyTransfer___

英语班级：

理工1615班

专业/年级：

物联网工程161班

姓名/学号：

（47）

二零一八年六月

AStudyofEnergyEfficientMobileCloudComputingPoweredbyWirelessEnergyTransfer

Abstract

Achievinglongbatterylivesorevenself-sustainabilityhasbeenalongstandingchallengefordesigningmobiledevices.Thisstudypresentsanovelsolutionthatseamlesslyintegratestwotechnologies,mobilecloudcomputingandmicrowavepowertransfer（MPT）,toenablecomputationinpassivelow-complexitydevicessuchassensorsandwearablecomputingdevices.Specifically,consideringasingle-usersystem,abasestation（BS）eithertransferspowertooroffloadscomputationfromamobiletothecloud;themobileusesharvestedenergytocomputegivendataeitherlocallyorbyoffloading.AframeworkforenergyefficientcomputingisproposedthatcomprisesasetofpoliciesforcontrollingCPUcyclesforthemodeoflocalcomputing,timedivisionbetweenMPTandoffloadingfortheothermodeofoffloading,andmodeselection.GiventheCPU-cyclestatisticsinformationandchannelstateinformation（CSI）,thepoliciesaimatmaximizingtheprobabilityofsuccessfullycomputinggivendata,calledcomputingprobability,undertheenergyharvestinganddeadlineconstraints.Furthermore,thisstudyrevealsthatthetwosimplesolutionstoachievetheobjecttosupportcomputationloadallocationovermultiplechannelrealizations,whichfurtherincreasesthecomputingprobability.Last,thetwokindsofmodessuggestthatthefeasibilityofwirelesslypoweredmobilecloudcomputingandthegainofitsoptimalcontrol.Andthefutureaspecttostudyissimplytobeanswer.

Keywords:

wirelesspowertransfer;energyharvestingcommunications;mobilecloudcomputing;energyefficientcomputing

Introduction

Mobilecloudcomputing（MCC）asanemergingcomputingparadigmintegratescloudcomputingandmobilecomputingtoenhancethecomputationperformanceofmobiledevices.TheobjectiveofMCCistoextendpowerfulcomputingcapabilityoftheresource-richcloudstotheresource-constrainedmobiledevices（e.g.,laptop,tabletandsmartphone）soastoreducecomputationtime,conservelocalresources,especiallybattery,andextendstoragecapacity.Toachievethisobjective,MCCneedstotransferresource-intensivecomputationsfrommobiledevicestoclouds,referredtoascomputationoffloading.Thecoreofcomputationoffloadingistodecideonwhichcomputationtasksshouldbeexecutedonthemobiledeviceoronthecloud,andhowtoschedulelocalandcloudresourcetoimplementtaskoffloading.TheexplosivegrowthofInternetofThings（IOT）andmobilecommunicationisleadingtothedeploymentoftensofbillionsofcloud-basedmobilesensorsandwearablecomputingdevicesinnearfuture（Huang&Chae,2010）.Prolongingtheirbatterylivesandenhancingtheircomputingcapabilitiesaretwokeydesignchallenges.Theycanbetackledbytwopromisingtechnologies:

microwavepowertransfer（MPT）forpoweringthemobilescomputation-intensivetasksfromthemobilestothecloudandmobilecomputationoffloading（MCO）.Twotechnologiesareseamlesslyintegratedinthecurrentworktodevelopanoveldesignframeworkforrealizingwirelesslypoweredmobilecloudcomputingunderthecriterionofmaximizingtheprobabilityofsuccessfullycomputinggivendata,calledcomputingprobability.TheframeworkisfeasiblesinceMPThasbeenproveninvariousexperimentsforpoweringsmalldevicessuchassensorsorevensmall-scaleairplanesandhelicopters.Furthermore,sensorsandwearablecomputingdevicestargetedintheframeworkareexpectedtobeconnectedbythecloud-basedIOTinthefuture,providingasuitableplatformforrealizingMCO.

Materials

MCOhasbeenanactiveresearchareaincomputersciencewhereresearchhasfocusedondesigningmobile-cloudsystemsandsoftwarearchitectures,virtualmachinemigrationdesigninthecloudandcodepartitioningtechniquesinthemobilesforreducingtheenergyconsumptionandimprovingthecomputingperformanceofmobiles.Nevertheless,implementationofMCOrequiresdatatransmissionandmessagepassingoverwirelesschannels,incurringtransmissionpowerconsumption.Theexistenceofsuchatradeoffhasmotivatedcross-disciplinaryresearchonjointlydesigningMCOandadaptivetransmissionalgorithmstomaximizethemobileenergysavings.Astochasticcontrolalgorithmwasproposedforadaptingtheoffloadedcomponentsofanapplicationtoatime-varyingwirelesschannel.Furthermore,multiusercomputationoffloadinginamulti-cellsystemwasexploredbyShinohara（2014）,wheretheradioandcomputationalresourceswerejointlyallocatedformaximizingtheenergysavingsunderthelatencyconstraints.

AccordingtoSwan（2012）,thethreshold-basedoffloadingpolicywasderivedforthesystemwithintermittentconnectivitybetweenthemobileandcloud.Lastly,theCPU-cyclefrequenciesarejointlycontrolledwithMCOgivenamoreskilledandincreasinglyappropriate

wirelesschannel.TheframeworkisfurtherdevelopedinthecurrentworktoincludethenewfeatureofMPT（Kostaetal.,2012）.Thisintroducesseveralnewdesignchallenges.Amongothers,thealgorithmicdesignoflocalcomputingandoffloadingbecomesmorecomplexundertheenergyharvestingconstraintduetoMPT,whichpreventsenergyconsumptionfromexceedingtheamountofharvestedenergyateverytimeinstant.AnotherchallengeisthatMPTandoffloadingtimesharethemobileantennaandthetimedivisionhastobeoptimized.

Nowthetechnologyisbeingfurtherdevelopedtopowerwirelesscommunications.Thishasresultedintheemergenceofanactivefieldcalledsimultaneouswirelessinformationandpowertransfer（SWIPT）.TheMPTtechnologyhasbeendevelopedforpoint-to-pointhighpowertransmissioninthepastdecades（Brown,1984）.Furthermore,existingwirelessnetworkssuchascognitiveradioandcellularnetworkshavebeenredesignedtofeatureMPT.MostpriorworkonSWIPTaimsatoptimizingcommunicationtechniquestomaximizetheMPTefficiencyandsystemthroughput.Incontrast,thecurrentworkfocusesonoptimizingthelocalcomputingandoffloadingunderadifferentdesigncriterionofmaximumcomputingprobability（Huang&Lau,2014）.

MethodsandResults

Considerasingle-usersystemcomprisingonemulti-antennabasestation（BS）usingtransmit/receivebeamformingfortransferringpowertoasingle-antennamobileorrelayingoffloadeddatafromthemobiletothecloud.Tocomputeafixedamountofdata,themobileoperatesinoneofthetwoavailablemodes:

Localcomputingandoffloading:

inthemodeoflocalcomputing,MPToccurssimultaneouslyascomputingbasedonthecontrollableCPU-cyclefrequencies.Nevertheless,inthemodeofoffloading,thegivencomputationdurationisadaptivelypartitionedforseparateMPTandoffloadingsincetheysharethemobileantenna（Shinohara,2014）.AssumethatthemobilehastheknowledgeofstatisticsinformationofCPUcyclesandchannelstateinformation（CSI）.Theindividualmodesaswellasmodeselectionareoptimizedformaximizingthecomputingprobabilityundertheenergyharvestinganddeadlineconstraints.Fortractability,themetricistransformedintoequivalentones,namelyaveragemobileenergyconsumptionandmobileenergysavings,forthemodesoflocalcomputingandoffloading,respectively.Comparedwiththepriorwork,thecurrentworkintegratesMPTwiththemobilecloudcomputing,whichintroducesnewtheoreticalchallenges.Inparticular,theenergyharvestingconstraintarisingfromMPTmakestheoptimizationproblemforlocalcomputingnon-convex.Totacklethechallenge,theconvexrelaxationtechniqueisappliedwithoutcompromisingtheoptimalityofthesolution.ItisshowninthesequelthatthelocalcomputingpolicyisaspecialcaseofthecurrentworkwherethetransferredpowerissufficientlyhighbySwan（2012）.Furthermore,thecaseofdynamicchannelformobilecloudcomputingisexplored.Approximationmethodsareusedforderivingthesimpleandclose-to-optimalpolicies.

Mobilemodeselection:

Theaboveresultsarecombinedtoselectthemobilemodeformaximizingthecomputingprobability.Givenfeasiblecomputinginbothmodes,theonlyone

yieldingthelargerenergysavingsispreferredandtheselectioncriterionisderivedintermsofthresholdsontheBStransmissionpoweraswellasthedeadlineforcomputing（Huangetal.,2012）.

Optimaldataallocationforadynamicchannel:

Last,theaboveresultsareextendedtothecaseofadynamicchannel,modeledasindependentandidenticallydistributed.blockfading,andnon-causalCSIatthemobile（acquiredfrome.g.,channelprediction）.Theproblemofoptimizinganindividualmobilemode（localcomputingoroffloading）isformulatedbasedonthemaster-and-slavemodelusingthesamemetricasthefixed-channelcounterpart（Kumar&Liu,2013）.

Conclusion

Wirelessandmobilecomputingtechnologiesprovidemorepossibilitiesforaccessingservicesconveniently.Mobiledeviceswillbeimprovedintermsofpower,CPU,andstorage.Mobilecloudcomputinghasemergedasanewparadigmandextensionofcloudcomputing.

Bytwokindsofavailablemodes,wecanpurelyknowoftheEnergyEfficientMobileCloudComputing.ThroughmystudyfortheMobileCloudComputing,wearehereexposingtwosimplesolutionstosolvethisproblem.Althoughmyresearchisprettybasic,itstillbenefittheprocessofthedevelopmentformobilecloudcomputingandhowtomakeitenergyefficient.Webelievethatexploringotheralternatives,suchasintroducingamiddlewarebasedarchitectureusinganoptimizingoffloadingalgorithm