简介:Athresholdsettingschemeisproposedbasedontheresourcemanagementandlimitedfeedbacktheoryinmultiuserorthogonalfrequencydivisionmultiplexing(OFDM)systems.Inadaptiveresourceallocation,thefactorsdenotingthequalityofservice(QoS)andfairnessarebothconsideredastheuserweight.Fromtheaspectoffeedbackoutageprobability,theproposedalgorithmsetsthethresholdforeachuserrelatedtoitsweight,whichbringsenoughfeedbacktotheuserwithgreaterweight.Analysisandsimulationresultsshowthat,comparedwiththethresholdignoringweights,theproposedschemehasmuchlowerfeedbackloadwhilewithbetterQoS.
简介:Inthispaper,thefeedbackloadreductionprobleminwirelesssystemsbasedonorthogonalfrequencydivisionmultiplexing(OFDM)isinvestigatedandanopportunisticfeedbackscheme(OFS)isproposed.ThekeyideabehindOFSisthatonlythekeychannelgainswhichcansignificantlyaffectthesystemthroughputarefedbacktotheBS.Firstly,thekeychannelgainsareprovedtobelongtoachannelgaininterval.Secondly,astatisticalmethodisproposedtoestimatethechannelgaininterval.Thirdly,theopportunisticfeedbackschemeisformulatedandthefeedbackloadofOFSisanalyzed.TheadvantageofOFSisthreefold:thefirstisOFScanworkinbothOFDM-basedmulticastsystemandOFDM-basedunicastsystem.ThesecondisthechannelfadingtypeoftheBS-userlinkisnotrequired,whichismorerealistic.ThethirdisOFScangetbetterfeedbackloadperformancecomparedwithotherschemes,whileachievingalmostthesamethroughputperformancecomparedwiththatoffullfeedbackscheme.
简介:Bydeducingthedistributionofthenormalizedchannelcovariancematrix,anovellimitedfeedbackschemeisproposedundermultipleusers(MU)multiple-inputmultiple-output(MIMO)broadcastchannel(BC)system.Theproposedschemehasadvantagesinthreeaspects.First,ithasnoconstraintsonthenumberofusersorantennas.Second,eachuser'sfeedbackbitsareindependentofthenumberofreceivingantennas.Third,theproposedschemeavoidsthestorageoflarge-sizecodebookonthetransceivers.Simulationresultsshowthattheperformanceoftheproposedschemeisclosetotheperfectchannelstateinformation(CSI)caseanditjustneedsasmallnumberoffeedbackbits.
简介:Themainchallengesofdatastreamsclassificationincludeinfinitelength,concept-drifting,arrivalofnovelclassesandlackoflabeledinstances.Mostexistingtechniquesaddressonlysomeofthemandignoreothers.Soanensembleclassificationmodelbasedondecision-feedback(ECM-BDF)ispresentedinthispapertoaddressallthesechallenges.Firstly,adatastreamisdividedintosequentialchunksandaclassificationmodelistrainedfromeachlabeleddatachunk.Toaddresstheinfinitelengthandconcept-driftingproblem,afixednumberofsuchmodelsconstituteanensemblemodelEandsubsequentlabeledchunksareusedtoupdateE.Todealwiththeappearanceofnovelclassesandlimitedlabeledinstancesproblem,themodelincorporatesanovelclassdetectionmechanismtodetectthearrivalofanovelclasswithouttrainingEwithlabeledinstancesofthatclass.Meanwhile,unsupervisedmodelsaretrainedfromunlabeledinstancestoprovideusefulconstraintsforE.AnextendedensemblemodelExcanbeacquiredwiththeconstraintsasfeedbackinformation,andthenunlabeledinstancescanbeclassifiedmoreaccuratelybysatisfyingthemaximumconsensusofEx.ExperimentalresultsdemonstratethattheproposedECM-BDFoutperformstraditionaltechniquesinclassifyingdatastreamswithlimitedlabeleddata.
简介:Thepapermainlystudiesthesum-rateperformanceoflimitedfeedback(LFB)block-diagonalization(BD)inmulti-userdistributedantennasystem(DAS).Asthechannelstateinformation(CSI)fedbacktobasestation(BS)islimited,multi-userinterference(MUI)iscausedinevitablybecauseofthequantizationerror.ConsideringtheinfluenceoftheMUItothecapacityofDAS,weproposeaparameterofeffectivesum-rateratio(ESRR)todenotethecapacityoffsetundertheconditionoftheBScangetperfectCSIandlimitedCSIfirst,thenweconfirmthattheapproximatedESRRisveryclosetoactualESRRgotthroughsimulations.Afterthat,basedontheapproximatedESRR,anadaptiveminimumbitfeedbackschemewhichcaneffectivelyreducetheoverheadoffeedbackchannelandthecomplexityofthesystemisproposed.Simulationresultsshowtheeffectivenessoftheproposedscheme.