简介:(VNE)虚拟网络嵌入是网络虚拟化的必要部分,它为未来网络被看作最有希望的方法之一。它的主要对象是高效地分配一个虚拟网络(VN)的节点和到一个分享的底层网络(SN)的连接。NP难、退出的研究提出了几个启发式的算法。然而,大多数算法仅仅考虑节点的本地资源,例如中央处理器和带宽(BW),到决定嵌入,并且忽略网络属性的重要影响。基于全部网络的属性,在每个节点之间的连接的一个模型被提出测量节点,和一个新二阶段的嵌入算法评价的资源被建议。此后,印射的节点和印射的连接能联合被考虑。当减少运行时刻时,广泛的模拟证明建议算法由增加VN请求的收入/费用比率和接受比率改进VNE的性能。
简介:Networktrafficclassificationaimsatidentifyingtheapplicationtypesofnetworkpackets.ItisimportantforInternetserviceproviders(ISPs)tomanagebandwidthresourcesandensurethequalityofservicefordifferentnetworkapplications.However,mostclassificationtechniquesusingmachinelearningonlyfocusonhighflowaccuracyandignorebyteaccuracy.TheclassifierwouldobtainlowclassificationperformanceforelephantflowsastheimbalancebetweenelephantflowsandmiceflowsonInternet.Theelephantflows,however,consumemuchmorebandwidththanmiceflows.Whentheclassifierisdeployedfortrafficpolicing,thenetworkmanagementsystemcannotpenalizeelephantflowsandavoidnetworkcongestioneffectively.Thisarticleexploresthefactorsrelatedtolowbyteaccuracy,andsecondly,itpresentsanewtrafficclassificationmethodtoimprovebyteaccuracyattheaidofdatacleaning.Experimentsarecarriedoutonthreegroupsofreal-worldtrafficdatasets,andthemethodiscomparedwithexistingworkontheperformanceofimprovingbyteaccuracy.Experimentshowsthatbyteaccuracyincreasedbyabout22.31%onaverage.Themethodoutperformstheexistingoneinmostcases.
简介:Toenhancethesecurityofnetworksystems,putsforwardakindofsoftwareagentisputforward,whichhastheinductionabilityofnetworkframeworksandtheabilityofbehaviorindependence.Itismobilescanningagent.Moreattentionsispaidtoexpoundhowtodesignandrealizemobilescanningagent.Besides,itisalsoexplainedtheprogramsofmobilescanningagentsystem.Intheend,itexpectsmobilescanningagent.
简介:Astherearelotsofnon-linearsystemsintherealengineering,itisveryimportanttodomoreresearchesonthemodelingandpredictionofnon-linearsystems.Basedonthemulti-resolutionanalysis(MRA)ofwavelettheory,thispapercombinedthewavelettheorywithneuralnetworkandestablishedaMRAwaveletnetworkwiththescalingfunctionandwaveletfunctionasitsneurons.Fromtheanalysisinthefrequencydomain,theresultsindicatedthatMRAwaveletnetworkwasbetterthanotherwaveletnetworksintheabilityofapproachingtothesignals.AnessentialresearchwascarriedoutonmodelingandpredictionwithMRAwaveletnetworkinthenon-linearsystem.Usingthelengthwiseswaydatareceivedfromtheexperimentofshipmodel,amodelofofflinepredictionwasestablishedandwasappliedtotheshort-timepredictionofshipmotion.Thesimulationresultsindicatedthattheforecastingmodelimprovedthepredictionprecisioneffectively,lengthenedtheforecastingtimeandhadabetterpredictionresultsthanthatofARlinearmodel.TheresearchindicatesthatitisfeasibletousetheMRAwaveletnetworkintheshort-timepredictionofshipmotion.
简介:Inpracticalparameterestimation,wehavealwayschoseneitherLeastSquaresEstimation(LSE)orRobustEstimation.Sincethedistributionofobservationsisunknown,toselectacorrectestimationmethodisverydifficult.Itiswellknownthatifobservationsincludegrosserrors,theresultofLSEwillbebadlycontaininated.Ontheotherhand,ifobservationsdonotincludeanygrosserrors,theresultofrobustestimationisnotasgoodasthatofLSE.Tosolvethisproblem,Wang(1999)developedanestimationmethodcalledInformationSpreadEstimation(ISE)basedontheinformationspreadprinciple.TheISEisaverygoodmethodforestimatingoneparameterwhichisveryrobust.However,mostofinstancesinsurveyingdataprocessingaremulti-parameters'estimation,owingtotheinherentrestrictionsofISE,itcannotbeappliedtothesurveyingdataprocessingdirectly.Toapplythegoodmethodtothefieldofsurveyingdataprocessingwidely,theauthorhasdonetheresearchdeeply.ThispaperappliesISEsuccessfullytotheadjustmentoflevelingnetworkbyusingthespecialtiesofleveling.
简介:Thepresent-daytopographyoftheItalianpeninsularesultsfromtheinteractionsbetweencrustal-mantleandsurfaceprocessesoccurringsincetheLateMiocene.Analysisofexhumationandcoolingofcrustalrocks,togetherwithQuaternarydrainageevolution,helpstounravelthetectonic-morphologicevolutionoftheApenninesbydistinguishingend-membermodels,andhencedescribingtheorogenicbeltevolution.Thepatternofregionaltopography,erosionalhistoryandpresent-daydistributionofactivedeformationsuggeststhattheeastwardmigratingextensional-compressionalpaireddeformationbeltsmaystillcontrolthetopogra-phyofthenorthernApennines,albeitatslowerratesthaninthepast.Conversely,Quaternarydrainageevo-lutioninthecentralandsouthernApenninessuggeststhatthetopographyoftheseregionsunderwentaQuaternaryregionalarching,whichisonlypartlycon-sistentwiththepersistingmigrationofthecompres-sional-extensionalpair.
简介:与在中国的经济的快速的开发,很快变化的地震网络过时的人或物,因为仪器,技术系统和网络密度的能力发达国家正在接近那些,观察数据过时的人或物的大数量积累。怎么把这些资源用于经济建设和公共安全成为了值得学习的一个重要问题。以便改进地震预言和地震紧急情况反应,它在这篇论文被建议那个提取珍贵先锋信息,改善地震rapid报导能力和报导的延长快速的紧张工作是关键问题。集成网络资源,造统一了地震监视网络是的标准和多功能建立一个公共安全服务平台和地震观察资源愿望的前提显著地作出贡献到工程,海洋,气象学,和环境保护的领域。因此,地震网络的开发的未来方向正在探索监视资源,提高独立革新,构造一个技术平台并且扩大服务领域。
简介:TheencapsulationofstearicacidcoatednanometerCaCO3byapolystyrene(PS)networkviaemulsionpolymerizationisdescribed,whereγ-methacryloxypropyltrimethoxysilane(MPS)wasusedasanefficientcrosslinker.TheimportantfactorssuchasthetypeandamountofsurfactantandinitiatorandthecontentofCaCO3areinvestigatedaswellastheroleofMPS.IthasbeenshownthatlittlePSwasextractablewithonly0.6wt%ofMPS(relativetostyrene).Thecationicsurfactantcetyltrimethylammoniumbromide(CTAB)provedmoreeffectivethantheanionicsurfactantsodiumdodecylsulfonate(SDS).Theyieldrises,particlesbecomesmallerandsizedistributionbroadenswithincreasedamountofCTAB.Itisalsofoundthateither2,2′-azobis(isobutyronitrile)(AIBN)orammoniumpersulfate(APS)issuitableforattaininghighmonomerconversion.WithincreasedamountofCaCO3,theencapsulationratiocanbevariedfrom17.9to3.6,whilemonomerconversionandyielddecreaseslightly.FT-IRspectraoftheproductsafterextractionindicatetightencapsulationbetweenPSandCaCO3,andTEMphotographsofcompositeparticleswithwell-definedcore-shellstructuregivedirectevidenceofencapsulation.
简介:Thecoreofnetworksecurityistheriskassessment.Inthisletter,ariskassessmentmethodisintroducedtoestimatethewirelessnetworksecurity.Themethod,whichcombinesAnalyticHier-archyProcess(AHP)methodandfuzzylogicalmethod,isappliedtotheriskassessment.Fuzzylogicalmethodisappliedtojudgetheimportantdegreeofeachfactorintheaspectsoftheprobability,theinfluenceandtheuncontrollability,nottodirectlyjudgetheimportantdegreeitself.Theriskas-sessmentiscarvedup3layersapplyingAHPmethod,thesortweightofthethirdlayeriscalculatedbyfuzzylogicalmethod.Finally,theimportantdegreeiscalculatedbyAHPmethod.Bycomparingtheimportantdegreeofeachfactor,theriskwhichcanbecontrolledbytakingmeasuresisknown.Thestudyofthecaseshowsthatthemethodcanbeeasilyusedtotheriskassessmentofthewirelessnetworksecurityanditsresultsconformtotheactualsituation.
简介:In2004,China’sdigitalseismicobservationnetworkprojectbegantodeploy40setsYRY-4four-componentboreholestrainmetersinordertomonitorearthquakepreparationprocess.Thepaperdescribesobservedsolidtidalstraindiscretenessandtidalfactoranisotropy,analyzesthereliabilityofobservationaldataanddiscussesthecauseforthisphenomenon.Aftergettingridofinterferences,thenetwork,intwoyearspractice,hasobservedseveralpre-seismicstrainanomaliesatstationsclosetoepicentersespeciallyintheWenchuanMS8.0megaquake.Itshowsthatthisboreholestrainmeternetworkiscapableofmonitoringseismogenicprocess.
简介:Gatematrixlayoutproblemplaysanimportantroleinintegratedcircuitdesign,butitsoptimizationisNP-hard.Inthispaper,typicalgatelayoutproblemisanalysedandadaptedtoneuralnetworkrepresentation,furthermorethesimulatedresultsaregiven.
简介:Onthebasisofanalyzingsomeneuralnetworkstoragecapacityproblemsanetworkmodelcomprisinganewencodingandrecallingschemeispresented.Byusingsomelogicaloperationswhichoperateonthebinarypatternstringsduringinformationprocessingprocedurethemodelcanreachahighstoragecapacityforacertainsizeofnetworkframework.
简介:ThispaperaddressesamanagementmodelforIPnetworkbasedonNextGenerationOperationSupportSystem(NGOSS).ItmakesthenetworkmanagementonthebaseofalltheoperationactionsofISP,ItprovidesQoStouserservicethroughthewholepathbyprovidingend-to-endServiceLevelAgreements(SLA)managementthroughwholepath.Basedonwebandcoordinationtechnology,thispapergivesanimplementarchitectureofthismodel.
简介:Adistibutedoptimallocaldoubleloop(DOLDL)networkispresented.EmphasisislaidonthetopologyanddistributedroutingalgorithmsfortheDOLDL.Onthebasisofbuildinganabstractmodel,asetofdefinitionsandtheoremsaredescribedandproved.Analgorithmwhichcanoptimizethedoubleloopnetworksispresented.TheoptimalvaluesofthetopologicparametersfortheDOLDLhavebeenobtainedbythealgorithm,andthesenumer.calresultsareanalyzed,Thestudyshowsthattheboundsoftheoptimaldiameter(d)andaveragehopdistance(a)forthisclassofnetworksareand,respectively(Nisthenumberofnodesinthenetwork.「3≤N≤10~4).AclassofthedistributedroutingalgorithmsfortheDOLDLandtheimplementationprocedureofanadaptivefault-tolerantalgorithmareproposed.Thecorrectnessofthealgorithmhasbeenalsoverifiedbysimulating.
简介:Frictiondragprimarilydeterminesthetotaldragoftransportsystems.ApromisingapproachtoreducedragathighReynoldsnumbers(>104)areactivetransversalsurfacewavesincombinationwithpassivemethodslikearibletsurface.Fortheapplicationintransportationsystemswithlargesurfacessuchasairplanes,shipsortrains,alargescaledistributedreal-timeactuatorandsensornetworkisrequired.Thisnetworkisresponsibleforprovidingconnectionsbetweenaglobalflowcontrolanddistributedactuatorsandsensors.ForthedevelopmentofthisnetworkweestablishedatfirstasmallscalenetworkmodelbasedonSimulinkandTrueTime.TodeterminetimescalesfornetworkeventsondifferentpackagesizeswesetupaRaspberryPibasedtestbedasaphysicalrepresentationofourfirstmodel.Thesetimescalesarereducedtotimedifferencesbetweenthedeterministicnetworkeventstoverifythebehaviorofourmodel.Experimentalresultswereimprovedbysynchronizingthetestbedwithsufficientprecision.Withthisapproachweassurealinkbetweenthelargescalemodelandthelaterconstructedmicrocontrollerbasedreal-timeactuatorandsensornetworkfordistributedactiveturbulentflowcontrol.