简介:Basedonuser’sin-degreedistribution,traditionalrankingalgorithmsofuser’sweightusuallyneglecttheconsiderationsofthedifferencesamonguser’sfollowersandthefeaturesofuser’stweets.Inordertoanalyzethefactorswhichimpactonuser’sweight,undertheanalysisofthedatacollectedfromSINAMicroblognetwork,thispaperdiscoversthatuserinfluenceandactivedegreesarethedominantfactorsforthisissue.Theproposedalgorithmevaluatesuserinfluencebyuser’sfollowernumber,theinfluenceofuser’sfollowersandthereciprocitybetweenusers.User’sactivedegreeismodeledbyuser’sparticipationandthequalityofuser’stweets.Themodelsaretestedbydifferentdatagroupstoconfirmtheparametersforthefinalcalculation.Eventually,thispapercomparesthecomputationalresultswiththeuser’srankingordergivenbytheSINAofficialapplication.Theperformanceofthisalgorithmpresentsastrongerstabilityonthefluctuantrangeofthevalueofuser’sweight.
简介:Multi-cellmulti-usermultiple-inputmultiple-output(MC-MU-MIMO)isapromisingtechniquetoeliminateinter-userinterferenceandinter-cellcochannelinterferenceinwirelesstelecommunicationsystems.AsthelargenumberofusersinthesystemandthelimitednumberofsimultaneouslysupportableuserswithMC-MU-MIMO,itisnecessarytoselectasubsetofuserstomaximizethetotalthroughput.However,thefullycentralizeduserselectionalgorithmsusedinsinglecellsystem,whichwillincurhighcomplexityandbackhaulloadinmulti-cellcooperativeprocessing(MCP)systems,arenotsuitabletoMC-MU-MIMOsystems.ThisarticlepresentsatwocascadeduserselectionmethodforMCPsystemswithmulti-cellblockdiagonalization.Inthispaper,alocaloptimalsubsetofusers,whichcanmaximizethelocalsumcapacity,isfirstchosenbythegreedymethodineverycooperativebasestationinparallel.Then,allthecooperativebasestationsreporttheirlocaloptimaluserstothecentralunit(CU).Finally,theglobaloptimalusers,whichcanmaximizetheglobalsumcapacityofMCPsystems,areselectedfromtheaggregatedlocaloptimalusersattheCU.Thesimulationresultsshowthattheproposedmethodperformscloselytotheoptimalandcentralizedalgorithm.Meanwhile,thecomplexityandbackhaulloadarereduceddramatically.
简介:Duetothecapabilityofreflectingsocialperceptiononsemanticofresources,folksonomyhasbeenproposedtoimprovethesociallearningforeducationandscholarresearching.However,itsactualimpactissignificantlyinfluencedbythesemanticambiguityproblemoftags.So,inthispaper,weproposedanovelwayofdetectinghomonyms,oneofthemainsourcesoftag’ssemanticambiguityproblem,innoisyfolksonomies.Thestudyisbasedontwohypotheses:1)Usershavingdifferentintereststendtohavedifferentunderstandingofthesametag.2)Usershavingsimilarinteresttendtohavecommonunderstandingofthesametag.Therefore,wefirstlydiscoverusercommunitiesaccordingtousers’interests.Then,tagcontextsarediscoveredinsubsetsoffolksonomyonthebasisofusercommunities.Theexperimentalresultsshowthatourmethodiseffectiveandoutperformthemethodfindingtagcontextsusingalltagsinfolksonomywithoverlappingclusteringalgorithmespeciallywhenvarioususershavingdifferentinterestsarecontainedbythefolksonomy.
简介:Toavoidtheinaccurateestimationoftheactiveuser'snumberandthecorrespondingperformancedegradation,anovelPOR-basedUserIdentificationDetector(UID)isproposedfortheCodeDivisionMultipleAccess(CDMA)systems.ThenewdetectoradoptsthePowerofR(POR)techniqueandtheMultipleSignalClassification(MUSIC)method,whichdoesnotrequiretheestimationofactiveusers'number,andobtainslowerfalsealarmprobabilitythanthesubspace-basedUIDinthemultipathchannels.However,fromouranalysis,increasingtheordermdoesnotimprovetheperformance.Therefore,whenmisone,theperformanceofthenewdetectorismaximal.
简介:ToimproveuserexperienceofcompositeWebservices,auser-awarequalityofservice(QoS)basedWebservicescompositionmodelisproposed.Undersuchmodel,aWebservicesselectionmethodbasedonquantumgeneticalgorithmisproposed.Thisalgorithmusesquantumbitencoding,dynamicstep-lengthquantumgateangleadjustment,neighborhoodservicesearchanddynamicpunishmentstrategytoexpandsearchscopeandspeedupconvergence.SimulationexperimentshowsthatthisalgorithmismoreefficientthanotherexistingalgorithmsinWebservicesselection.
简介:SocialtaggingisoneofthemostimportantcharacteristicsofWeb2.0services,andsocialtaggingsystems(STS)arebecomingmoreandmorepopularforuserstoannotate,organizeandshareitemsontheWeb.Moreover,onlinesocialnetworkhasbeenincorporatedintosocialtaggingsystems.AsmoreandmoreuserstendtointeractwithrealfriendsontheWeb,personalizeduserrecommendationserviceprovidedinsocialtaggingsystemsisveryappealing.Inthispaper,weproposeapersonalizeduserrecommendationmethod,andourmethodhandlesnotonlytheusers'interestnetworks,butalsothesocialnetworkinformation.Weempiricallyshowthatourmethodoutperformsastate-of-the-artmethodonrealdatasetfromLast.fmdatasetandDouban.
简介:在异例察觉,挑战是怎么为一个用户的动态行为建模。许多以前的工作基于固定长度模型代表用户行为。克服他们的缺点,我们与可变长度的序列的状态基于分离时间的Markov链(DTMC)建议一个新奇方法。方法第一产生不同长度的多重壳命令溪流并且把他们合为一般序列的图书馆。然后,状态根据一个有效用户的可变长度的行为的模式被定义,它改进用户介绍的精确和适应性。随后,转变概率矩阵被创造。以便减少计算复杂性,分类价值仅仅由转变可能性被决定,然后与滑动窗户变光滑,并且最后过去常在正常和反常行为之间区别。来自Purdue大学和AT&T香农实验室的数据集上的二实验评估证明建议方法能完成更高的察觉精确性并且比另外的传统的方法要求更少的存储器。
简介:Nowadays,thepassword-basedremoteuserauthenticationmechanismusingsmartcardisoneofthesimplestandconvenientauthenticationwaystoensuresecurecommunicationsoverthepublicnetworkenvironments.Recently,Liuetal.proposedanefficientandsecuresmartcardbasedpasswordauthenticationscheme.However,wefindthatLiuetal.’sschemeisvulnerabletotheoff-linepasswordguessingattackanduserimpersonationattack.Furthermore,italsocannotprovideuseranonymity.Inthispaper,wecryptanalyzeLiuetal.’sschemeandproposeasecurityenhanceduserauthenticationschemetoovercometheaforementionedproblems.Especially,inordertopreservetheuseranonymityandpreventtheguessingattack,weusethedynamicidentitytechnique.Theanalysisshowsthattheproposedschemeismoresecureandefficientthanotherrelatedauthenticationschemes.
简介:Apowerallocationschemeformulti-usermultiple-inputmultiple-outputorthogonalfrequencydivisionmultiplexing(MIMO-OFDM)systemswithchannelstateinformation(CSI)ontransmitterandreceiverispressed.Multi-userpowerallocationcanbedecoupledintosingleuserpowerallocationthroughoutnullspacemappingofmulti-userchannelandpowerallocationcanbeperformedthroughoutspatial-spectralwater-fillingforperuser.Todealwithmoreusersinsystemandfadingcorrelation,schedulingisperformedtomaintainthegainofpowerallocation.Theproposedschemecansubstantiallyimprovesystem'sspectralefficiencywithlowomplexity.Simulationresultsvalidatetheaccuracyoftheoreticanalyses.
简介:Inordertominimizethetransmittedpowerinthemulti-userorthogonalfrequencydivisionmultiplexing(OFDM)system,aschemecombiningtheimprovedparticleswarmoptimization(POS)algorithmwithgeneticalgorithm(GA)isproposedtooptimizethesub-carriersandbitsallocation.Inthealgorithm,arandomvelocitybetweenthemaximumandminimumparticlevelocityisusedastheupdatingvelocityinsteadofmaximumorminimumvelocitywhentheupdatedparticlevelocityishigherthanthemaximumparticlevelocityorlowerthantheminimumparticlevelocity.Then,theconvergencepopulationisusedastheinitialpopulationofthegeneticalgorithmtooptimizethesubcarriersandbitsallocationfurther.Simulationresultsshowthatthetransmittedpoweroftheproposedalgorithmisabout2dBto10dBlowerthanthatofthegeneticalgorithm,particleswarmoptimizationalgorithm,andZhang’salgorithm.
简介:Three-layerAdaptiveBack-PropagationNeuralNetworks(TABPNN)areemployedforthedemodulationofspreadspectrumsignalsinamultiple-accessenvironment.Aconfigurationemployingthree-layeradaptiveBack-propagationneuralnetworksisputforwardforthedemodulationofspread-spectrumsignalsinasynchronousGaussianchannels.Thetheoreticalargumentsandpracticalperformancebasedontheneuralnetworksareanalyzed.Theresultsshowthatwhethertheresistancetothemultipleaccessinterferenceortherobusttonear-fareffects,theproposeddetectorsignificantlyoutperformsnotonlytheconventionaldetectorbutalsotheBPneuralnetworksdetectorandiscomparabletotheoptimumdetector.
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.
简介:Two-dimensional(2D)multiple-inputmultiple-output(MIMO)iscurrentlyconcentratedonpropagationinhorizontalplane,buttheimpactofelevationangleisnotconsidered.However,duetothethree-dimensional(3D)characteroftherealMIMOchannel,2DMIMOcannotachievetheoptimalsystemthroughput.Amultiple-userMIMO(MU-MIMO)userpairingschemewasproposed,inwhichtheverticaldimensionwastakenintoconsideration.Intheproposedscheme,a3DcodebookbasedonfulldimensionMIMOchannelwasdesigned;thentwo3DMU-MIMOuser'spairingschemesareproposedcombiningtheproposedjointandseparate3Dcodebook.Simulationevaluatestheproposed3Dcodebookaideduserpairingschemeandcompareswiththeprevious2DMU-MIMOuserpairingtechnology.Owingtotheadditionalspatialdegreeoffreedominverticaldimension,theproposed3DMU-MIMOuserpairingschemescaneffectivelyimprovetheoverallsystemperformance.
简介:Inwirelesscellularnetworks,theinterferencealignment(IA)isapromisingtechniqueforinterferencemanagement.AnewIAschemefordownlinkcellularnetworkwithmulti-cellandmulti-userwasproposed.Intheproposedscheme,theinterferenceinthenetworksisdividedintointer-cellinterference(ICI)amongcellsandinter-userinterference(IUI)ineachcell.TheICIisalignedontoamulti-dimensionalsubspacebymultiplyingtheICIalignmentprecodingmatrixwhichisdesignedbythesingularvaluedecomposition(SVD)schemeatthebasestation(BS)side.ThealignedICIiseliminatedbytimingtheinterferencesuppressionmatrixwhichisdesignedbyzero-forcing(ZF)schemeattheuserequipment(UE)side.Meanwhile,theIUIisalignedbymultiplyingtheIUIalignmentprecodingmatrixwhichisdesignedbasedonNashbargainingsolution(NBS)ingametheory.TheNBSissolvedbytheparticleswarmoptimization(PSO)method.Simulationsshowthat,comparedwiththetraditionalZFIAscheme,theproposedschemecanobtainhigherdatarateandguaranteethedataratefairnessofUEswithlittleadditionalcomplexity.
简介:Orthogonalfrequencydivisionmultiplexing(OFDM)whichhasbeenadoptedinthelong-termevolution(LTE)systemcanimprovethesystemcapacityobviously.However,italsobringsaboutsevereinter-cellinterference(ICI)forcell-edgeusers(CEUs).Totacklethisproblem,multi-userselectionandpowercontrol(MuS-PC)isproposedasanefficientschemeinuplinkcoordinatedmulti-pointmulti-usermulti-inputmulti-output(CoMP-MU-MIMO)transmission/reception.Thispaperjointlyconsidersuser’ssignaltointerferenceplusnoiseratio(SINR)andproportionalfairness(PF)tomaximizethetotalchannelcapacityinmulti-userselectionbyformulatingapenaltyfunction.Tosimplifythepenaltyfunction’scomputation,particleswarmoptimization(PSO)algorithmisintroduced.Inaddition,powercontrolisadoptedtomaximizeoverallenergyefficiency.SimulationresultsdemonstratethattheMuS-PCschemecannotonlyobtaintheoptimaltotalchannelcapacitywhileguaranteeeachuser’squalityofservice(QoS)andPF,butalsolargelyreducecomputationalcomplexityandimproveenergyefficiency.Asaresult,thepoorcommunicationqualityofCEUscanbeenhanced.
简介:Coordinatedmulti-pointtransmission/reception(CoMP)wasproposedcurrentlyasaneffectivetechnologytoimprovecell-edgethroughputinnext-generationwirelesssystems.Mostoftheexistingworkdiscussedclusteringmethodsmostlytomaximizetheedgeuserthroughputwhileneglectingtheproblemofenergyefficiency,suchasthosealgorithmclusteringbasestations(BSs)ofbetterchannelconditionandBSsofworsechannelconditiontogether.Inaddition,BSsusuallyincreasethetransmitpowertoachievehigherthroughputwithoutanyconsideringofinterferencecausedtootherusers,thatmayresultinenergywaste.TheauthorsfocusonthethroughputmaximizingproblemwhilefullyconsideringenergysavingprobleminCoMPsystems.Acoefficientisdefinedtodescribethefitnessofclusters.Thenasub-carrierallocationalgorithmwithclusteringmethodisputforwardforCoMPdownlink,whichcansavethetransmitpowerofBSandincreasethethroughput.Furthermoreapowerallocationschemeisproposedbasedonnon-cooperationgame;inwhichthetransmitpowerisdecreasedbyBSsgenerallytoreachtheNashequation(NE).Simulationshowsthattheproposedsub-carrierallocationschemeandpowerallocationalgorithmarebetterthantheexistingonesonusers’throughputwhileconsumesmuchlessenergy.