简介：Unmannedaerialvehicles(UAVs)mayplayanimportantroleindatacollectionandoffloadinginvastareasdeployingwirelesssensornetworks,andtheUAV’sactionstrategyhasavitalinfluenceonachievingapplicabilityandcomputationalcomplexity.Dynamicprogramming(DP)hasagoodapplicationinthepathplanningofUAV,butthereareproblemsintheapplicabilityofspecialterrainenvironmentandthecomplexityofthealgorithm.BasedontheanalysisofDP,thispaperproposesahierarchicaldirectionalDP(DDP)algorithmbasedondirectiondeterminationandhierarchicalmodel.WecompareourmethodswithQ-learningandDPalgorithmbyexperiments,andtheresultsshowthatourmethodcanimprovetheterrainapplicability,meanwhilegreatlyreducethecomputationalcomplexity.

简介：Operationalsoftwaresystemsoftenexperiencean'aging'phenomenon,characterizedbyprogressiveperformancedegradationandasuddenhang/crashfailure.Softwarerejuvenationisaproactivefault-tolerancestrategyaimedtopreventunexpectedoutagesduetoaging.Anewrejuvenationstrategybasedonmeasurementandtime,andconstructsaSRNmodeltoevaluatethisstrategy.Numericalresultsshowthatthisstrategyoutweighseitherthepurelymeasurementbasedorthepurelytime-basedstrategyandcaneffectivelyimprovesystemperformance.

简介：Aswitchingvariabilityindex(SVI)constantfalsealarmrate(CFAR)detectorisproposedforimprovingthedetectionperformanceofVI-CFARdetectorsinmultipletargetsbackgrounds.Whenthepresenceofnon-homogeneityinCFARreferencewindowsisindicatedbyaVI-CFARdetector,aswitchingCFARdetectorisintroducedtooptimizetheperformanceoftheVI-CFARdetectorinhomogeneous,multipletargetsandclutteredgebackgrounds.ThestructureandparametersselectionmethodoftheSVI-CFARdetectorispresented.ComparisonswithclassicCFARdetectorsandrecentlyproposeddetectorsarealsogiven.TheoreticalanalysisandsimulationresultsshowthatSVICFARdetectormaintainsthegoodperformanceoftheVI-CFARdetectorinhomogeneousandclutteredgebackgrounds,whilegreatlyimprovingthecapacityofanti-multitargets.

简介：MostofImageQualityAssessment(IQA)metricsconsistoftwoprocesses.Inthefirstprocess,qualitymapofimageismeasuredlocally.Inthesecondprocess,thelastqualityscoreisconvertedfromthequalitymapbyusingthepoolingstrategy.Thefirstprocesshadbeenmadeeffectiveandsignificantprogresses,whilethesecondprocesswasalwaysdoneinsimpleways.Inthesecondprocessofthepoolingstrategy,theoptimalperceptualpoolingweightsshouldbedeterminedandcomputedaccordingtoHumanVisualSystem(HVS).Thus,areliablespatialpoolingmathematicalmodelbasedonHVSisanimportantissueworthyofstudy.Inthispaper,anewVisualPerceptualPoolingStrategy(VPPS)forIQAispresentedbasedoncontrastsensitivityandluminancesensitivityofHVS.ExperimentalresultswiththeLIVEdatabaseshowthatthevisualperceptualweights,obtainedbytheproposedpoolingstrategy,caneffectivelyandsignificantlyimprovetheperformancesoftheIQAmetricswithMeanStructuralSIMilarity(MSSIM)orPhaseQuantizationCode(PQC).ItisconfirmedthattheproposedVPPSdemonstratespromisingresultsforimprovingtheperformancesofexistingIQAmetrics.

简介：WhiletheNetworkCodingcooperativerelaying(NC-relaying)hasthemeritofhighspectralefficiency,SuperpositionCodingrelaying(SC-relaying)hasthemeritofhighthroughput.Inthispaper,anovelconcept,codedcooperativerelaying,ispresented,whichisaunifiedschemeoftheNC-relayingandSC-relaying.FortheSC-relayingstrategywhichcanbeconsideredone-waycodedrelayingschemewithmulti-accesschannel,theclose-formsolutionoftheoutageprobabilitiesofthebasicsignalandadditionalsignalareobtainedfirstly.Secondly,theDiversity-and-MultiplexingTradeoff(DMT)characteristicsofbasicsignalandadditionalsignalareinvestigatedentirelyaswellastheoptimalclose-formsolutions.Thecomparednumericalanalysisshowstheevaluationerrorofthroughputbasedontheclose-formsolutionisabout0.15nats,whichiswithintheacceptableerrorrange.Duetothemutualeffectbetweenthebothsourcesignals,theavailablemaximalvaluesofthetwomultiplexinggainsarelessthan1.

简介：Makingfulluseofwindpowerisoneofthemainpurposesofthewindturbinegeneratorcontrol.Conventionalhillclimbingsearch(HCS)methodcanrealizethemaximumpowerpointtracking(MPPT).However,thestepsizeofHCSmethodisconstantsothatitcannotconsiderbothsteady-stateresponseanddynamicresponse.Afuzzylogicalcontrol(FLC)algorithmisproposedtosolvethisprobleminthispaper,whichcantrackthemaximumpowerpoint(MPP)quicklyandsmoothly.ToevaluateMPPTalgorithms,fourperformanceindicesarealsoproposedinthispaper.Theyaretheenergycapturedbywindturbine,themaximumpower-pointtrackingtimewhenwindspeedchangesslowly,thefluctuationmagnitudeofrealpowerduringsteadystate,andtheenergycapturedbywindturbinewhenwindspeedchangesfast.ThreecasesaredesignedandsimulatedinMATLAB/Simulinkrespectively.ThecomparisonofthethreeMPPTstrategiesconcludesthattheproposedfuzzylogicalcontrolalgorithmismoresuperiortotheconventionalHCSalgorithms.

简介：Aseachtypeofsatellitenetworkhasdifferentlinkfeatures,itsdatatransmissionmustbedesignedbasedonitslinkfeaturestoimprovetheefficiencyofdatatransferring.Thetransmissionofnavigationintegratedservicesinformation(NISI)inaglobalnavigationsatellitesystem(GNSS)withinter-satellitelinks(ISLs)isstudiedbytakingtherealsituationofinter-satellitecommunicationlinksintoaccount.Anon-demandcomputingandbufferingcentralizedroutestrategyisproposedbasedondynamicgroupingandthetopologyevolutionlawoftheGNSSnetworkwithinwhichthesatellitenodesareoperatedinthemannerofdynamicgrouping.Dynamicgroupingisbasedonsatellitesspatialrelationshipsandthegrouproleofthesatellitenodechangesbyturnsduetoitsspatialrelationships.Theroutestrategyprovidessignificantadvantagesofhighefficiency,lowcomplexity,andflexibleconfiguration,bywhichtheestablishedGNSScanpossessthefeaturesandcapabilitiesoffeasibledeployment,efficienttransmission,convenientmanagement,structuralinvulnerabilityandflexibleexpansion.

简介：Withthedevelopmentofsmartgrid,residentshavetheopportunitytoscheduletheirhouseholdappliances(HA)forthepurposeofreducingelectricityexpensesandalleviatingthepressureofthesmartgrid.Inthispaper,weintroducethestructureofhomeenergymanagementsystem(EMS)andthenproposeapoweroptimizationstrategybasedonhouseholdloadmodelandelectricvehicle(EV)modelforhomepowerusage.Inthisstrategy,theelectricvehiclesarechargedwhenthepriceislow,andotherwise,aredischarged.Byadoptingthiscombinedsystemmodelunderthetime-of-useelectricityprice(TOUP),theproposedschedulingstrategywouldeffectivelyminimizetheelectricitycostandreducethepressureofthesmartgridatthesametime.Finally,simulationexperimentsarecarriedouttoshowthefeasibilityoftheproposedstrategy.Theresultsshowthatcrossovergeneticparticleswarmoptimizationalgorithmhasbetterconvergencepropertiesthantraditionalparticleswarmalgorithmandbetteradaptabilitythangeneticalgorithm.

简介：Afundamentalissueinopticalburstswitching(OBS)networksistosolvetheburstcontentionforthecorenode.Inthispaper,anovelpriority-basedcontentionsolutionstrategyforOBSnetworksisproposed.Whenthecontentionoccurs,theburstpriorityisconsideredfirstly,andthentheburstsegmentationmethodisusedforthelowpriorityburstsinthisstrategy.Ensuringtheintegrityofhighprioritybursts,partofthesegmentedburstscanbetransmittedtothedestinationnodeviacombiningwavelengthconversionandopticalbuffermethod.Simulationresultsshowthattheproposedschemenotonlyensurestheintegrityofhighprioritybursts,butalsoreducesthepacketlossrateofthelowpriorityburstsmaximally,sothatitcansupportgoodqualityofservice(QoS)forthenetwork.

简介：Inthispaper,anovelframework,namedasglobal-localfeatureattentionnetworkwithrerankingstrategy(GLAN-RS),ispresentedforimagecaptioningtask.Ratherthanonlyadoptingunitaryvisualinformationintheclassicalmodels,GLAN-RSexplorestheattentionmechanismtocapturelocalconvolutionalsalientimagemaps.Furthermore,weadoptrerankingstrategytoadjustthepriorityofthecandidatecaptionsandselectthebestone.TheproposedmodelisverifiedusingtheMicrosoftCommonObjectsinContext(MSCOCO)benchmarkdatasetacrosssevenstandardevaluationmetrics.ExperimentalresultsshowthatGLAN-RSsignificantlyoutperformsthestate-of-the-artapproaches,suchasmultimodalrecurrentneuralnetwork(MRNN)andGoogleNIC,whichgetsanimprovementof20%intermsofBLEU4scoreand13pointsintermsofCIDERscore.

简介：Inthispaper,wepresentasimplebutpowerfulensembleforrobusttextureclassification.Theproposedmethodusesasingletypeoffeaturedescriptor,i.e.scale-invariantfeaturetransform(SIFT),andinheritsthespiritofthespatialpyramidmatchingmodel(SPM).Inaflexiblewayofpartitioningtheoriginaltextureimages,ourapproachcanproducesufficientinformativelocalfeaturesandtherebyformareliablefeaturepondortrainanewclass-specificdictionary.Totakefulladvantageofthisfeaturepond,wedevelopagroup-collaborativelyrepresentation-basedstrategy(GCRS)forthefinalclassification.Itissolvedbythewell-knowngrouplasso.Butwegobeyondofthisandproposealocality-constraintmethodtospeedupthis,namedlocalconstraint-GCRS(LC-GCRS).Experimentalresultsonthreepublictexturedatasetsdemonstratetheproposedapproachachievescompetitiveoutcomesandevenoutperformsthestate-of-the-artmethods.Particularly,mostofmethodscannotworkwellwhenonlyafewsamplesofeachcategoryareavailablefortraining,butourapproachstillachievesveryhighclassificationaccuracy,e.g.anaverageaccuracyof92.1%fortheBrodatzdatasetwhenonlyoneimageisusedfortraining,significantlyhigherthananyothermethods.