简介:Motivatedbytheprojectsconstrainedbyspacecapacityandresourcetransportingtime,aprojectschedulingproblemwithcapacityconstraintwasmodeled.Ahybridalgorithmisproposed,whichusestheideasofbi-levelschedulingandprojectdecompositiontechnology,andthegeneticalgorithmandtabusearchiscombined.Topologicalreorderingtechnologyisusedtoimprovetheeffciencyofevaluation.Simulationresultsshowtheproposedalgorithmcanobtainsatisfiedschedulingresultsinacceptabletime.
简介:Deadlockmustbeavoidedinamanufacturingsystem.Inthispaper,anefficientalgorithmforfindinganoptimaldeadlock-freeschedulesinamanufacturingsystemwithverylimitedbufferispresented.Thisalgorithmisbasedonheeffectivegeneticalgorithm(GA)searchmethod,andaformalPetrinetstructureisintroducedtodetectthetokenplayerassuringdeadlock-free.InordertomaketheschedulingstrategygeneratedbyGAmeettherequiredconstraintofdeadlock-free,someresultsofthestruetureanalysisofPetrinetareinvolvedasacriteriontoselectdeadlock-freeschedulefromthepopulationgeneratedbyGA.Theeffectivenessandefficiencyoftheproposedapproachisillustratedbyusinganexample.
简介:ThispaperproposesanobjectorientedmodelschedulingforparallelcomputinginmediaMultiProcessorsSystemonChip(MPSoC).Firstly,theCoarseGrainDataFlowGraph(CGDFG)parallelprogrammingmodelisusedinthisapproach.Secondly,thisapproachhasthefeatureofunifiedabstractionforsoftwareobjectsimplementinginprocessorandhardwareobjectsimplementinginASICs,easyformappingCGDFGprogrammingonMPSoC.Thisapproachcutsdownthekerneloverheadandreducesthecodesizeeffectively.Theprincipleoftheorientedobjectmodel,themethodofscheduling,andhowtomapaparallelprogrammingthroughCGDFGtotheMPSoCareanalyzedinthisapproach.Thisapproachalsocomparesthecodesizeandexecutioncycleswithconventionalcontrolflowscheduling,andpresentsrespectivemanagementoverheadforoneapplicationinme-dia-SoC.
简介:Theplatformschedulingprobleminbattlefieldisoneoftheimportantproblemsinmilitaryoperationalresearch.Itneedstominimizemissioncompletingtimeandmeanwhilemaximizethemissioncompletingaccuracywithalimitednumberofplatforms.Thoughthetraditionalcertainmodelsobtainsomegoodresults,uncertainmodelisstillneededtobeintroducedsincethebattlefieldenvironmentiscomplexandunstable.Anuncertainmodelisprposedfortheplatformschedulingproblem.Relatedparametersinthismodelaresettobefuzzyorstochastic.Duetotheinherentdisadvantageofthesolvingmethodsfortraditionalmodels,anewmethodisproposedtosolvetheuncertainmodel.Finally,thepracticabilityandavailabilityoftheproposedmethodaredemonstratedwithacaseofjointcampaign.
简介:Howtodealwiththecollaborationbetweentaskdecompositionandtaskschedulingisthekeyproblemoftheintegratedmanufacturingsystemforcomplexproducts.Withthedevelopmentofmanufacturingtechnology,wecanprobeanewwaytosolvethisproblem.Firstly,anewmethodfortaskgranularityquantitativeanalysisisputforward,whichcanpreciselyevaluatethetaskgranularityofcomplexproductcooperationworkflowintheintegratedmanufacturingsystem,ontheabovebasis;thismethodisusedtoguidethecoarse-grainedtaskdecompositionandrecombinethesubtaskswithlowcohesioncoefficient.Then,amulti-objectiveoptimieationmodelandanalgorithmaresetupfortheschedulingoptimizationoftaskscheduling.Finally,theapplicationfeasibilityofthemodelandalgorithmisultimatelyvalidatedthroughanapplicationcasestudy.
简介:Inthispaper,Petrinetstechniqueisintroducedintomobileadhocnetworks(MANET)andapacket-flowparallelschedulingschemeispresentedusingStochasticPetriNets(SPN).Theflowingoftokensisusedingraphicsmodetocharacterizedynamicalfeaturesofsharingasinglewirelesschannel.ThroughSPNreachabilityanalysisandisomorphiccontinuoustimeMarkovprocessequations,somenetworkparameters,suchaschannelefficiencyand,one-hoptransmissiondelay,canbeobtained.ComparedwithThoseofTheconventionalperformanceevaluationmethods,theaboveparametersaremathematicalexpressionsinsteadoftestresultsfromsimulator.
简介:Apowerallocationschemeformulti-usermultiple-inputmultiple-outputorthogonalfrequencydivisionmultiplexing(MIMO-OFDM)systemswithchannelstateinformation(CSI)ontransmitterandreceiverispressed.Multi-userpowerallocationcanbedecoupledintosingleuserpowerallocationthroughoutnullspacemappingofmulti-userchannelandpowerallocationcanbeperformedthroughoutspatial-spectralwater-fillingforperuser.Todealwithmoreusersinsystemandfadingcorrelation,schedulingisperformedtomaintainthegainofpowerallocation.Theproposedschemecansubstantiallyimprovesystem'sspectralefficiencywithlowomplexity.Simulationresultsvalidatetheaccuracyoftheoreticanalyses.
简介:Amemeticalgorithm(MA)foramulti-moderesourceconstrainedprojectschedulingproblem(MRCPSP)isproposed.WeuseanewfitnessfunctionandtwoveryeffectivelocalsearchproceduresintheproposedMA.Thefitnessfunctionmakesuseofamechanismcalled'strategicoscillation'tomakethesearchprocesshaveahigherprobabilitytovisitsolutionsarounda'feasibleboundary'.Oneofthelocalsearchproceduresaimsatimprovingthelowerboundofprojectmakespantobelessthanaknownupperbound,andanotheraimsatimprovingasolutionofanMRCPSPinstanceacceptinginfeasiblesolutionsbasedonthenewfitnessfunctioninthesearchprocess.AdetailedcomputationalexperimentissetupusinginstancesfromtheprobleminstancelibraryPSPLIB.ComputationalresultsshowthattheproposedMAisverycompetitivewiththestate-of-the-artalgorithms.TheMAobtainsimprovedsolutionsforoneinstanceofsetJ30.
简介:Akindofnetworkedcontrolsystemisstudied;thenetworkedcontrolsystemwithnoisedisturbanceismodeledbasedoninformationschedulingandcontrolco-design.Augmentedstatematrixanalysismethodisintroduced,androbustfault-tolerantcontrolproblemofnetworkedcontrolsystemswithnoisedisturbanceunderactuatorfailuresisstudied.Theparametricexpressionofthecontrollerunderactuatorfailuresisgiven.Furthermore,theresultisanalyzedbysimulationtests,whichnotonlysatisfiesthenetworkedcontrolsystemsstability,butalsodecreasesthedatainformationnumberinnetworkchannelandmakesfulluseofthenetworkresources.
简介:Aself-adaptivelargeneighborhoodsearchmethodforschedulingnjobsonmnon-identicalparallelmachineswithmultipletimewindowsispresented.Theproblems'anotherfeatureliesinoversubscription,namelynotalljobscanbescheduledwithinspecifiedschedulinghorizonsduetothelimitedmachinecapacity.Theobjectiveisthustomaximizetheoverallprofitsofprocessedjobswhilerespectingmachineconstraints.Afirst-infirst-outheuristicisappliedtofindaninitialsolution,andthenalargeneighborhoodsearchprocedureisemployedtorelaxandreoptimizecumbersomesolutions.Amachinelearningmechanismisalsointroducedtoconvergeonthemostefficientneighborhoodsfortheproblem.Extensivecomputationalresultsarepresentedbasedondatafromanapplicationinvolvingthedailyobservationschedulingofafleetofearthobservingsatellites.Themethodrapidlysolvesmostprobleminstancestooptimalornearoptimalandshowsarobustperformanceinsensitiveanalysis.
简介:Inordertolowerthepowerconsumptionandimprovethecoefficientofresourceutilizationofcurrentcloudcomputingsystems,thispaperproposestworesourcepre-allocationalgorithmsbasedonthe'shutdowntheredundant,turnonthedemanded'strategyhere.Firstly,agreencloudcomputingmodelispresented,abstractingthetaskschedulingproblemtothevirtualmachinedeploymentissuewiththevirtualizationtechnology.Secondly,thefutureworkloadsofsystemneedtobepredicted:acubicexponentialsmoothingalgorithmbasedontheconservativecontrol(CESCC)strategyisproposed,combiningwiththecurrentstateandresourcedistributionofsystem,inordertocalculatethedemandofresourcesforthenextperiodoftaskrequests.Then,amulti-objectiveconstrainedoptimizationmodelofpowerconsumptionandalow-energyresourceallocationalgorithmbasedonprobabilisticmatching(RA-PM)areproposed.Inordertoreducethepowerconsumptionfurther,theresourceallocationalgorithmbasedontheimprovedsimulatedannealing(RA-ISA)isdesignedwiththeimprovedsimulatedannealingalgorithm.Experimentalresultsshowthatthepredictionandconservativecontrolstrategymakeresourcepre-allocationcatchupwithdemands,andimprovetheefficiencyofreal-timeresponseandthestabilityofthesystem.BothRA-PMandRA-ISAcanactivatefewerhosts,achievebetterloadbalanceamongthesetofhighapplicablehosts,maximizetheutilizationofresources,andgreatlyreducethepowerconsumptionofcloudcomputingsystems.
简介:Thispaperproposesanewqueuingmodelandadaptiveschedulingschemewhichrealizesmulti-classQoSmechanismunderDiffServarchitecture.Thequeuingmodeliscomposedoftwoparalleloutputsubqueues,eachoutputsubqueueadoptsrandomdropalgorithmbysettingdifferentbufferthresholdfordifferentclasstraffic,soitcanprovidemulticlassQoS.Thenewproposedschedulingschemewhichadaptivelychangestheparameterλcanguaranteetheperformancetargetofhighclasstraffic,inthemeantime,improvetheQoSoflowclassestraffic.
简介:Howtoeffectivelyreducetheenergyconsumptionoflarge-scaledatacentersisakeyissueincloudcomputing.Thispaperpresentsanovellow-powertaskschedulingalgorithm(LTSA)forlarge-scaleclouddatacenters.Thewinnertreeisintroducedtomakethedatanodesastheleafnodesofthetreeandthefinalwinneronthepurposeofreducingenergyconsumptionisselected.Thecomplexityoflarge-scaleclouddatacentersisfullyconsider,andthetaskcomparsoncoefficientisdefinedtomaketaskschedulingstrategymorereasonable.Experimentsandperformanceanalysisshowthattheproposedalgorithmcaneffectivelyimprovethenodeutilization,andreducetheoverallpowerconsumptionoftheclouddatacenter.
简介:Inthispaper,anextendedKendallmodelforthepriorityschedulinginput-linegroupoutputwithmulti-channelinAsynchronousTransferMode(ATM)exchangesystemisproposedandthenthemeanmethodisusedtomodelmathematicallythenon-typicalnon-anticipativePRiorityservice(PR)model.Comparedwiththetypicalandnon-anticipativePRmodel,itexpressesthecharacteristicsofthepriorityschedulinginput-linegroupoutputwithmulti-channelinATMexchangesystem.ThesimulationexperimentshowsthatthismodelcanimprovetheHOLblockandtheperfonnanceofinput-queuedATMswitchnetworkdramatically.ThismodelhasabetterdevelopingprospectinATMexchangesystem.
简介:Amodifiedbottleneck-based(MB)heuristicforlarge-scalejob-shopschedulingproblemswithawell-definedbottleneckissuggested,whichissimplerbutmoretailoredthantheshiftingbottleneck(SB)procedure.Inthisalgorithm,thebottleneckisfirstscheduledoptimallywhilethenon-bottleneckmachinesaresubordinatedaroundthesolutionsofthebottleneckschedulebysomeeffectivedispatchingrules.ComputationalresultsindicatethattheMBheuristiccanachieveabettertradeoffbetweensolutionqualityandcomputationaltimecomparedtoSBprocedureformedium-sizeproblems.Furthermore,itcanobtainagoodsolutioninashorttimeforlarge-scalejob-shopschedulingproblems.