简介:ThispaperpresentsanRFreceiverofzero-IntermediateFrequency(IF)architectureforCognitiveRadio(CR)communicationsystems.Zero-IFarchitecturereducetheimagerejectfilterandIFfilter,soitisexcellentinlowcost,compactvolume,andlowpowerdissipation.Thereceiveremploysthreedigitalattenuatorandahighgain,highlinearitylownoiseamplifiertoachievewidedynamicrangeof70dBandhighreceivingsensitivityof-81dBm.AfullybalancedI/QdemodulatorandadifferentialLocalOscillator(LO)chipsareusedtominimizethenegativeeffectscausedbysecond-orderdistortionandLOleakage.Inordertoselectan8MHz-channelfrom14continuousoneslocatedinUHFband(694-806MHz)accurately,approachofchannelselectivitycircuitsisproposed.TheRFreceiverhasbeendesigned,fabricated,andtest.Themeasuredresultshowsthatthenoisefigureis3.4dB,andtheerrorvectormagnitudeis7.5%whentheinputpoweris-81dBm.
简介:Indatacenters,thetransmissioncontrolprotocol(TCP)incastcausescatastrophicgoodputdegradationtoapplicationswithamany-to-onetrafficpattern.Inthispaper,weintendtotameincastatthereceiver-sideapplication.Towardsthisgoal,wefirstdevelopananalyticalmodelthatformulatestheincastprobabilityasafunctionofconnectionvariablesandnetworkenvironmentsettings.Wecombinethemodelwiththeoptimizationtheoryandderivesomeinsightsintominimizingtheincastprobabilitythroughtuningconnectionvariablesrelatedtoapplications.Then,enlightenedbytheanalyticalresults,weproposeanadaptiveapplication-layersolutiontotheTCPincast.Thesolutionequallyallocatesadvertisedwindowstoconcurrentconnections,anddynamicallyadaptsthenumberofconcurrentconnectionstothevaryingconditions.SimulationresultsshowthatoursolutionconsistentlyeludesincastandachieveshighgoodputinvariousscenariosincludingtheoneswithmultiplebottlenecklinksandbackgroundTCPtraffic.
简介:AmultisensordistributedextendedKalmanfilteringalgorithmispresentedfornonlinearsystem,inwhichthedynamicequationofthesystemandtheequationsofsensor’smeasurementsarelinearizedintheglobalestimateandglobalpredictionrespectivelyandthesuboptimalglobalestimatebasedonallavailableinformationcanbereconstructedfromtheestimatescomputedbylocalsensorsbasedsolelyontheirownlocalinformationandtransmittedtothedatafusioncenter.Ananalysisofthepropertiesofthealgorithmpresentedhereshowsthattheglobalestimatehashigherprecisionthanthelocaloneandsmallerlinearizationerrorthantheexistingmethod.Finally,anapplicationofthealgorithmtoradar/IRtrackingofamaneuveringtargetisillustrated.Simulationresultsshowtheeffectivenessofthealgorithm.