USF Libraries
USF Digital Collections

Towards the realization of cognitive radio

MISSING IMAGE

Material Information

Title:
Towards the realization of cognitive radio coexistence of ultrawideband and narrowband systems
Physical Description:
Book
Language:
English
Creator:
Sahin, Mustafa Emin
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Interference avoidance
Narrowband interference
Opportunistic spectrum usage
Spectrum sensing
Spectrum shaping
Dissertations, Academic -- Electrical Engineering -- Masters -- USF
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Ultrawideband and cognitive radio are two of the most important approaches that are shaping the future of wireless communication systems. At a first glance, the aims of UWB and cognitive radio do not seem to be overlapping significantly, however, there is a strong synergy between the capabilities of UWB and the goals of cognitive radio. One of the objectives of this thesis is to shed the first light on the marriage of these two important approaches.Ultrawideband (UWB) is a promising technology for future short-range, high-data rate wireless communication networks. Along with its exciting features including achieving high data rates, low transmission power requirement, and immunity to multipath effects, UWB is unique in its coexistence capability with narrowband systems.In this thesis, the details of practical UWB implementation are provided. Regarding the coexistence of UWB with licensed narrowband systems, narrowband interference (NBI)avoidance and cancelation techniques in the literature are investigated. It is aimed to emphasize that UWB is a strong candidate for cognitive radio, and this fact is proven by providing two different approaches in which ultrawideband is combined with cognitive radio to maximize the performance of unlicensed communications.
Thesis:
Thesis (M.A.)--University of South Florida, 2006.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
Statement of Responsibility:
by Mustafa Emin Sahin.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 86 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001787499
oclc - 124063768
usfldc doi - E14-SFE0001418
usfldc handle - e14.1418
System ID:
SFS0025738:00001


This item is only available as the following downloads:


Full Text

PAGE 1

TowardstheRealizationofCognitiveRadio: CoexistenceofUltrawidebandandNarrowbandSystemsby MustafaEminSahin Athesissubmittedinpartialfulllment oftherequirementsforthedegreeof MasterofScienceinElectricalEngineering DepartmentofElectricalEngineering CollegeofEngineering UniversityofSouthFlorida MajorProfessor:HuseyinArslan,Ph.D. VijayK.Jain,Ph.D. ThomasWeller,Ph.D. DateofApproval: March20,2006 Keywords:InterferenceAvoidance,NarrowbandInterferen ce,OpportunisticSpectrum Usage,SpectrumSensing,SpectrumShaping c r Copyright2006,MustafaEminSahin

PAGE 2

DEDICATION Tomywife Muberra

PAGE 3

ACKNOWLEDGMENTS First,Iwouldliketosincerelythankmyadvisor,Dr.Husey inArslanforhisguidance, encouragement,andsupportthroughoutthecourseofthisth esis.Theenthusiasmhegeneratesandpassesontohisstudentsleadstohardworkandsuc cess.Ithasbeenagenuine privilegetohavetheopportunitytodoresearchasamembero fDr.Arslan'sresearch group.Iamgratefultohimbecausehehastaughtmenotonlynu meroustopicsinwireless communicationsbutalsomanydierentaspectsofreallife. IwishtothankDr.VijayK.JainandDr.ThomasWellerforserv ingonmycommittee andforoeringvaluablesuggestions.Ihopetobeabletoben etfromtheirprofound knowledgeandexperienceinthecomingyears,too. IowemuchtoIsmailGuvencforhisstrongtechnicalassist ance,tolerancetocountless questions,and,ofcourse,hissincerefriendship.Without hissupport,Imighthavegiven uppursuingagraduatedegreeattheverybeginningofthislo ngjourney.Iamalsovery proudonhisbehalfthatIamgoingtobeabletocallhimDr.Gu vencfromnowon. SpecialthanksgotomyfriendsHasariCelebi,TevkYucek ,SadiaAhmed,Hisham Mahmoud,SerhanYarkan,Kemal Ozdemir,RamyTannious,TonyS.Price,NigelBrown, andAbdur-rubAbdur-rahman.TheyaretheoneswithwhomIwor ktogethersevendaysa week.Ihavelearnedalotofdierentthingsfromthem,botht echnicalandnon-technical. Theyenabledmetobecomefamiliarwithdierentcultures.W iththesespecialpeople,we havenotonlybeensharingthesameworkingenvironment,and butthesamelife,aswell. IamalsogratefultoourfriendsintheTurkishcommunityliv inginTampaandBrandon, Florida,especiallytoSalihErdem,formakinglifemuchmor eeasierandenjoyablethanit isinreality. Ialsowanttothankmyparentsinlawfortheirkindnessandco ntinuoussupport,and mybrotherinlaw,Ahmet,forteachingmehowtoplaysoccer.

PAGE 4

Mydeepestgratitudegoestomywife,Muberra,forherlove, allthesacricesshehas made,herrmsupport,hervastpatience,andhersteadyenco uragement.Ifshedidnot havesuchadeepunderstandingandtoleranceforthehardshi psofbeingthewifeofa graduatestudent,Icouldnoteventrytoobtainthisdegree. Iwanttothankherfrommy heartforeverythingshehasdone. Last,butbynomeansleast,mysincereappreciationgoestom yparentsandmyelder sisterforbringingmeup,leadingmetotherightdirection, andalwaysencouragingmefor pursuinghigherdegrees.Itisnotpossibletothankthemeno ugh,butIwantthemtoknow thatIwillbegratefultothemthroughoutmylife.

PAGE 5

TABLEOFCONTENTS LISTOFFIGURES iii LISTOFACRONYMS v ABSTRACT vii CHAPTER1INTRODUCTION 1 1.1OrganizationoftheThesis 2 CHAPTER2ULTRAWIDEBAND 5 2.1Introduction 5 2.2ModulationandReceiverOptionsforImpulseRadioUWB72.3EnergyDetectorReceivers 8 2.3.1SystemModel 9 2.3.2OptimumJointParameterSelection10 2.3.2.1Obtainingthe( u;v )HypotheseswithDierentSamplingApproaches 11 2.3.2.2ThresholdSelectionUsingExactandGaussianAnalysis 13 2.3.3BERPerformanceEvaluation14 2.3.3.1ExactBERPerformances152.3.3.2BERUsingtheGaussianApproximation17 2.3.4NumericalResults 17 2.4HighDataRateUWBUsingEnergyDetectorReceivers22 2.4.1ISIProblem 23 2.4.2SystemModel 23 2.4.3EectofISI 25 2.4.4ISICancelingReceiver 27 2.4.5PerformanceResults 30 2.5Conclusion 31 CHAPTER3UWBANDNARROWBANDSYSTEMS32 3.1Introduction 32 3.2EectofNBIinUWBSystems 36 3.3AvoidingNBI 41 3.3.1Multi-carrierApproach 41 3.3.2Multi-bandSchemes 43 3.3.3PulseShaping 44 i

PAGE 6

3.3.4OtherNBIAvoidanceMethods47 3.4CancelingNBI 49 3.4.1MMSECombining 49 3.4.2FrequencyDomainTechniques503.4.3Time-FrequencyDomainTechniques513.4.4TimeDomainTechniques 52 3.5Conclusion 54 CHAPTER4COGNITIVERADIO 55 4.1Introduction 55 4.2OpportunisticSpectrumUsage 56 4.3SensingtheSpectrumOpportunities 58 4.4SpectrumShaping 59 4.5Conclusion 63 CHAPTER5COGNITIVEUWB 64 5.1Introduction 64 5.2CognitiveUWB-OFDM 65 5.3ACognitiveSystemSupportedbyUWB67 5.3.1StepsofPracticalImplementation685.3.2RangeofCognitiveCommunicationsandCognitiveNetw orks69 5.3.3NumericalResults 71 5.4Conclusion 73 CHAPTER6SUMMARYANDCONCLUSIONS76 6.1SummaryofContributions 76 6.2Conclusions 78 REFERENCES 79 ii

PAGE 7

LISTOFFIGURES Figure1.ImpulseradiobasedtimehoppingUWB.6Figure2.OFDMcarriersinthefrequencydomain.7Figure3.Parameterestimationandsymboldetectioninbloc k-fadingchannelmodel.10 Figure4.Blockdiagramfortheproposedjointparameterest imationforenergy detectorreceivers. 11 Figure5.Threedierentapproachesforperformanceevalua tion.15 Figure6.Biterrorratevs. Eb=N0forCM1(BW=500MHzand2GHzcases)for bothGaussianapproximatedandexactthresholdestimates. 16 Figure7.BERvs.integrationintervalfordierentchannel models(at Eb=N0=10 dBand20dB). 18 Figure8.SNRvs.integrationintervalfordierentchannel models(at Eb=N0=20 dB). 19 Figure9.Optimumintegrationintervalvs. Eb=N0fordierentchannelmodels.20 Figure10.BERvs. Eb=N0forxedintegrationintervals,adaptiveintegrationinte rval,andadaptivesynchronizationpoint.21 Figure11.AnalyticalresultsregardingBERvs.numberoftr ainingsymbols(at Eb=N0=15dB,20dB,and25dB). 22 Figure12.(a)Thepulserepetitionperiod( )greaterthanthemaximumexcessdelay ( D ).(b) D> ,energyleaksfromonesymboltothesubsequentsymbol.24 Figure13.BERvs.datarateforchannelmodelCM1at Eb=N0valuesof15dBand 20dB. 25 Figure14.BERvs.datarateforchannelmodelCM2at Eb=N0valuesof15dBand 20dB. 26 Figure15.DecisionfeedbackequalizationbasedISIcancel ingenergydetector.28 Figure16.Successivetrainingsequencesusedforestimati ngthefeedbackltercoefcients. 28 iii

PAGE 8

Figure17.Biterrorratevs. Eb=N0fordierentchannelmodelsbeforeandafterthe ISIsuppression. 30 Figure18.Spectrumcrossoverofthenarrowbandinterferer sinUWBsystems.33 Figure19.(a)TH-UWBpulsesalongwithanarrowbandinterfe rer.(b)Reduced interferencepowerbymeansoftimegating.38 Figure20.AsimpleNBIscenarioformulti-carriermodulati onsystems.42 Figure21.Someproposedmulti-bandapproachesforWPAN:(a )TheXtremeSpectrumMotorolaproposalofadual-bandapproach.(b)Multi-bandO FDM.43 Figure22.NormalizedspectraforthesingleGaussianpulse andtwodierentGaussiandoublets. 45 Figure23.Theeectofnotchlteringonthetransmittedpul seshape.46 Figure24.(a)Asnap-shotofthespectrumintime.(b)Opport unisticspectrum utilizationemployingtimelimitedsinusoids.(c)Opportu nisticspectrum usageemployingspecialpulses. 60 Figure25.Dierentpulseshapesandtheirspectra(a)Recta ngularwindow.(b) Raisedcosinewindowswithroll-ofactors =0 : 3and =0 : 9.(c)Root raisedcosinewindowswithroll-ofactors =0 : 3and =0 : 9.(d)A highorderprolatespheroidalwaveletfunction.62 Figure26.(a),(b),(c)Separatepulsesobtainedviaraised cosinelteringthattinto dierentopportunities.(d)Sumoftheseparatepulses.(e) Binaryclassicationoffrequencybandsas'occupied'or'opportunity '.(f)Spectrum ofthedesignedpulsellingtheopportunities.63 Figure27.Networkofcognitivetransceivers(sensitivity rangesarenotdrawnto scale). 71 Figure28.BERvs.distancebetweenthenodesforUWBsignali ng.73 Figure29.RepetitionraterequiredforreliableUWBsignal ingvs.distance.74 Figure30.Probabilityofalicensedtransmitterbeingdete ctedbythecognitivenetwork. 75 iv

PAGE 9

LISTOFACRONYMS AWGNAdditiveWhiteGaussianNoiseBERBitErrorRateBPSKBinaryPhaseShiftKeyingCDMACodeDivisionMultipleAccessingCMChannelModelDFEDecisionFeedbackEqualizerDOFDegreeofFreedomDSSSDirectSequenceSpreadSpectrumDWTDiscreteWaveletTransformFCCFederalCommunicationsCommissionFECForwardErrorCorrectionFFTFastFourierTransformIEEEInstituteofElectricalandElectronicsEngineersIRImpulseRadioISIInter-symbolInterferenceLOSLineOfSightMEDMaximumExcessDelayMMSEMinimumMean-squareErrorNBINarrowbandInterference v

PAGE 10

OFDMOrthogonalFrequencyDivisionMultiplexingOOKOn-OKeyingPAMPulseAmplitudeModulationPPMPulsePositionModulationPDPPowerDelayProlePSWFProlateSpheroidalWaveletFunctionsRFRadioFrequencySDRSoftwareDenedRadioSNRSignal-to-noiseRatioSPTFSpectrumPolicyTaskForceTHTime-HoppingTRTransmittedReferenceUWBUltrawidebandWLANWirelessLocalAreaNetworkWPANWirelessPersonalAreaNetwork vi

PAGE 11

TOWARDSTHEREALIZATIONOFCOGNITIVERADIO: COEXISTENCEOFULTRAWIDEBANDANDNARROWBANDSYSTEMS MustafaEminSahin ABSTRACT Ultrawidebandandcognitiveradioaretwoofthemostimport antapproachesthatare shapingthefutureofwirelesscommunicationsystems.Ata rstglance,theaimsofUWB andcognitiveradiodonotseemtobeoverlappingsignicant ly,however,thereisastrong synergybetweenthecapabilitiesofUWBandthegoalsofcogn itiveradio.Oneofthe objectivesofthisthesisistoshedtherstlightonthemarr iageofthesetwoimportant approaches. Ultrawideband(UWB)isapromisingtechnologyforfuturesh ort-range,high-datarate wirelesscommunicationnetworks.Alongwithitsexcitingf eaturesincludingachievinghigh datarates,lowtransmissionpowerrequirement,andimmuni tytomultipatheects,UWB isuniqueinitscoexistencecapabilitywithnarrowbandsys tems. Inthisthesis,thedetailsofpracticalUWBimplementation areprovided.Regarding thecoexistenceofUWBwithlicensednarrowbandsystems,na rrowbandinterference(NBI) avoidanceandcancelationtechniquesintheliteratureare investigated.Itisaimedto emphasizethatUWBisastrongcandidateforcognitiveradio ,andthisfactisprovenby providingtwodierentapproachesinwhichultrawidebandi scombinedwithcognitiveradio tomaximizetheperformanceofunlicensedcommunications. vii

PAGE 12

CHAPTER1 INTRODUCTION Wirelesscommunicationsystemshavebeenevolvingsubstan tiallyoverthelasttwodecades. Theexplosivegrowthofthewirelesscommunicationmarketi sexpectedtocontinueinthe future,asthedemandforalltypesofwirelessservicesisin creasing.Newgenerationsof mobileradiosystemsaimatprovidinghigherdataratesanda widevarietyofapplications tothemobileusers,whileservingasmanyusersaspossible. However,thisgoalmustbe achievedundertheconstraintoflimitedavailableresourc essuchaspowerandfrequency spectrum.Giventhehighcostofpowerandscarcityofthespe ctrum,radiosystemsmust providehighercapacityandperformancethroughamoreeci entuseoftheavailableresources.Hence,inordernottolimittheeconomicandtechno logicalimprovementofthe wirelessworld,itisnecessarytondimmediatesolutionsr egardingtheusageofexisting resources.Arecentsolutionforthisproblemiscognitiver adio[1]. Cognitiveradioaimsataveryecientspectrumutilization employing smart wireless deviceswithawareness,sensing,learning,andadaptation capabilities[2].Asasolutionfor thespectrumscarcityproblem,cognitiveradioproposesan opportunisticspectrumusage approach[3],inwhichfrequencybandsthatarenotbeinguse dbytheirprimary(licensed) usersareutilizedbycognitiveradios. Sincecognitiveradioisaverynewconcept,thereisnoconse nsusonhowtoimplement it.However,ifthetargetsofcognitiveradioaretakeninto account,andthepropertiesof variouscandidatesystemsareconsidered,itisseenthatth ereisastrongmatchbetween whatcognitiveradioaimsatandwhatultrawideband(UWB)o ers. UWBisapromisingtechnologyforfutureshortandmediumran gewirelesscommunicationnetworkswithvariousthroughputoptionsincludingve ryhighdatarates.Ithasmany 1

PAGE 13

temptingfeaturessuchaslowpowerconsumption,signican tlylowcomplexitytransceivers, andimmunitytomultipatheects.Accordingtothemodernde nition,anywirelesscommunicationtechnologythathasabandwidthwiderthan500MH zorafractionalbandwidth1greaterthan0.2canbeconsideredaUWBsystem.Thefeatures ofUWBsystemsthat makethesesystemsverytemptingforcognitiveradioinclud ethat theydonotrequirealicensetoutilizethespectrum. theycancoexistwithlicensedcommunicationsystemsbyemp loyinginterference avoidanceandcancelationmethods. theirtransmissionparameterssuchaspower,pulseshape,a nddataratearehighly adaptive. theirpracticalimplementationdoesnothaveahighcost. Becauseoftheseproperties,UWBcanbeusedeitherasameans ofimplementingcognitive radio,oritcanbecombinedwithacognitivesystemtocomple mentitindierentways. Thisthesisinvestigatesultrawidebandsystemsindetail, includingpracticalimplementationaspects.CoexistenceofUWBwithothersystemsisdis cussedthoroughly,andthe appropriatenessofUWBforfulllingtherequirementsofco gnitiveradioisproven.The cognitiveradioconceptisintroduced,andvariouspropert iesofcognitiveradiosystemsare studied.Twodierentapproachesofsupplementingcogniti veradiowithultrawidebandare provided,rstofwhichisacognitiveUWB-OFDMsystem,andt hesecondone,acognitive networkwhosenodessharethespectrumsensinginformation witheachotherbymeansof UWB.1.1OrganizationoftheThesisThemaintopicscoveredinthisthesiscanbelistedasfollow s: Practicalimplementationofultrawideband(coveredinCha pter2) 1 Fractionalbandwidth=2 F H F L F H + F L ,where F H and F L aretheupperandloweredgefrequencies,respectively.2

PAGE 14

Coexistenceofultrawidebandandnarrowbandcommunicatio nsystems(coveredin Chapter3) Fundamentalsofthecognitiveradioconcept(coveredinCha pter4) Combiningcognitiveradioandultrawideband(coveredinCh apter5) Theoutlineofeachchapterisasfollows. InChapter2,thebasicsofUWBareprovided,anddierentmod ulationandreceiver optionsforUWBimplementationarediscussed.Alowcomplex ityenergydetectorwithoptimizedintegrationintervalandthresholdselectionproper tiesisproposed2.Theperformance oftheproposeddetectorisanalyzed.Forhighdatarateappl ications,theinter-symbolinterference(ISI)problemisinvestigated.Amodiedversio noftheenergydetectorthat employsdecisionfeedbackequalizationtocanceltheISIe ectisproposed3.Itseciency inenablinghighdatarateUWBcommunicationsisdemonstrat ed. InChapter3,thecoexistenceofUWBandnarrowbandsystemsi saddressed.Theeect ofnarrowbandinterference(NBI)onUWBisanalyzed.Thetec hniquesofsuppressing NBI,whichareclassiedasNBIavoidanceandNBIcancelatio nmethods,areinvestigated indetail4. InChapter4,thecognitiveradioconceptisintroduced,and theobjectivesaimedwith cognitiveradioareaddressed.Spectrumopportunityisde ned,anditisinvestigatedhowto sensethespectralopportunities.Dierentspectrumshapi ngapproachesintheliteratureare provided,andanalternativemethodbasedontheusageofrai sedcosineltersisproposed. InChapter5,withthepurposeofmaximizingtheeciencyofu nlicensedsystems,itis consideredtocombineUWBwiththecognitiveradio.Twodie rentmethodsareproposed; therstoneiscognitiveUWB-OFDM5,andthesecondoneisacognitivesystemthatshares thespectrumsensinginformationbetweenitsnodesusingUW Bsignaling. 2 Thisworkispublishedin[4] 3 Thisworkispublishedin[5] 4 Thisworkispublishedin[6] 5 Thisworkistopartlyappearin[7]3

PAGE 15

InChapter6,thecontributionsofthethesisaresummarized andpossiblefutureresearch topicsrelatedtothestudiesinthethesisareprovided. 4

PAGE 16

CHAPTER2 ULTRAWIDEBAND Inthischapter,thebasicsofUWBareprovidedanddierentm odulationandreceiver optionsforUWBimplementationarediscussed.Alowcomplex ityenergydetectorwithoptimizedintegrationintervalandthresholdselectionprop ertiesisproposed.Theperformance oftheproposeddetectorisanalyzed.Forhighdatarateappl ications,theinter-symbolinterference(ISI)problemisinvestigated.Amodiedversio noftheenergydetectorthat employsdecisionfeedbackequalizationtocanceltheISIe ectisproposed.Itseciencyin enablinghighdatarateUWBcommunicationsisdemonstrated 2.1IntroductionUltrawideband(UWB)isapromisingtechnologyforfuturesh ort-range,high-datarate wirelesscommunicationnetworks.Comparedtoothercommun icationssystems,UWBis uniqueinthatithastheexcitingfeatureofcombiningmanyd esiredcharacteristicslikethe increasedpotentialofachievinghighdatarates,lowtrans missionpowerrequirement,and immunitytomultipatheects.Twobasictechniquesconside redforimplementingUWB aretheimpulseradio(IR)andOrthogonalFrequencyDivisio nMultiplexing(OFDM). Impulseradioisbasedontransmittingextremelyshort(int heorderofnanoseconds)and lowpowerpulsesthathaveaverywidespectrum.InFig.1,ati mehoppingultrawideband (TH-UWB)systemisdemonstrated.Intheillustratedscenar io,eachinformationcarrying symbolistransmittedwithfourpulses.Pulsesoccupyaloca tioninthetime-framebased onthespecicpseudorandom(PN)codeassignedforeachuser .Twodierentcodesand thecorrespondingpulselocationsareshowninthegure.No tethatthesetwocodesare orthogonal(i.e.theydonotinterferewitheachother). 5

PAGE 17

Figure1.ImpulseradiobasedtimehoppingUWB. InUWB-OFDM,thedatabearingsymbolstreamissplitintosev erallowerratestreams, andthesestreamsaretransmittedondierentcarriers.The carriersaresinusoidswithdifferentfrequenciesandtheyarelimitedintime.Sincetheya retimelimited,theycorrespond to sinc functionsinthefrequencydomainasshowninFig.2.Fromthi sperspective,thissystemisnotdierentfromregularOFDM.Therequirementspeci ctoUWB-OFDMisthat theminimumfrequencybandoccupiedshouldexceed500MHz(o rfractionalbandwidth shouldbelargerthan0.2). Thischapterisorganizedasfollows.InSection,2.2,vario usmodulationandreceiver optionsforimpulseradioUWBarediscussed.InSection2.3, optimumselectionofintegrationintervalstart/stoptimes,andthethresholdisa ddressed.ExactandGaussian approximationmethodsforBERevaluationareanalyzed.InS ection2.4,theISIproblemis discussed,aperformanceevaluationoftheUWBsysteminthe presenceofISIisprovided, andanISIcancelationalgorithmisexplainedindetail. 6

PAGE 18

5020 5040 5060 5080 5100 5120 5140 5160 5180 -0.2 0 0.2 0.4 0.6 0.8 1 Frequency (MHz)Amplitude Figure2.OFDMcarriersinthefrequencydomain. 2.2ModulationandReceiverOptionsforImpulseRadioUWBImpulseradioisadvantageousinthatiteliminatestheneed forupanddown-conversion, andallowstoutilizelow-complexitytransceivers.Italso enablesemployingvarioustypes ofmodulations,includingon-okeying(OOK),pulseamplit udemodulation(PAM),pulse positionmodulation(PPM),andbinaryphaseshiftkeying(B PSK),aswellasdierent receivertypessuchasenergydetector,RAKE,andtransmitt edreferencereceivers. Coherentreceivers(suchasRAKEandcorrelatorreceivers) arecommonlyusedforimpulseradiosignalreceptionduetotheirhighpowerecienc ies.However,implementation ofsuchreceiversrequiresestimationof apriori channelinformationregardingthetiming, fadingcoecient,andthepulseshapeforeachindividualch anneltap.Coherentsignal receptionalsostipulateshighsamplingratesandaccurate synchronization.Ontheother hand,non-coherentreceivershavelessstringentapriorii nformationrequirementsandcan 7

PAGE 19

beimplementedwithlowercomplexity.Forexample,intrans mittedreference(TR)receivers,transmissionofthereferencepulse(s)(whichinc ludesthechannelinformation)to correlatetheinformationbearingpulse(s)eliminatesthe needforestimatingthechannel parameters.Therefore,throughouttherestofthischapter ,thefocusisgoingtobeon non-coherentreceivers,especiallyontheenergydetector 2.3EnergyDetectorReceiversEnergydetectorisanon-coherentapproachforultrawideba ndsignalreception,wherelow complexityreceiverscanbeachievedattheexpenseofsomep erformancedegradation[8]. AsopposedtomorecomplexRAKEreceivers,estimationofind ividualpulseshapes,path amplitudes,anddelaysateachmultipathcomponentisnotne cessaryforenergydetectors. Moreover,energydetectorsarelesssensitiveagainstsync hronizationerrors[9],andare capableofcollectingtheenergyfromallthemultipathcomp onents. On-okeyingisoneofthemostpopularnon-coherentmodulat ionoptionsthathasbeen consideredforenergydetectors.OOKbasedimplementation ofenergydetectorsisachieved bypassingthesignalthroughasquarelawdevice(suchasaSc hottkydiodeoperatingin square-region)followedbyanintegratorandadecisionmec hanism,wherethedecisions aremadebycomparingtheoutputsoftheintegratorwithathr eshold.Twochallenging issuesfortheenhancementofenergydetectorreceiversare theestimationoftheoptimal threshold,andthedeterminationofsynchronization/dump pointsoftheintegrator. Theeectofintegrationintervalonthesystemperformance hasbeenanalyzedbefore forenergydetectors[8,10].However,optimaljointselect ionoftheintegrationstartandstop times,andthethresholdisnotcoveredintheliterature.In thissection,thecontributions areasfollows: Theoptimaljointparameterselectionusingthebiterrorra te(BER)expressionswith exactanalysisandGaussianapproximationisaddressed,an ditisshownthatGaussian approximationworkswellonlyatlargebandwidths, 8

PAGE 20

Aframeworkisdenedforsynchronization/dumphypotheses withdierentsampling options, Whenanexactanalysisisconsidered,usingtheGaussianapp roximation forcalculatingthethreshold yieldsverysmallperformancelosses,andcanbeconsidered asa practicalalternativeforexactthresholdevaluation, TheparameterestimationrequiresexplicitBERminimizati on(ratherthanSNRmaximization)sincethestatisticscorrespondingtodierent bitsarenotidentical, Thenumberoftrainingsymbolsrequiredtoconvergetotheid ealparameterestimates isshowntobelessthanonehundredforpracticaloperatings cenarios. 2.3.1SystemModelLettheimpulseradiobasedUWBsignalreceivedforbit i inamultipathenvironmentbe representedas ri( t )=LXl =1rlbipl t l iTs + n ( t ) ; (1) where L isthenumberofmultipathcomponentsarrivingatthereceiv er, l istapindex, biisthe i thtransmittedbitwithOOKmodulation, pl( t )isthereceivedpulseshapeforthe l thpath, rland larethefadingcoecientandthedelayofthe l thmultipathcomponent, respectively,and Tsisthesymbolduration.TheadditivewhiteGaussiannoise(A WGN) withdouble-sidednoisespectraldensity N0= 2isdenotedby n ( t ).Thereceivedsignalis passedthroughabandpasslterofbandwidth B tocapturethesignicantportionofsignal spectrumwhileremovingout-of-bandnoiseandinterferenc e,resultingin~ r ( t ).Forthesake ofsimplicity,weconsidersinglepulsepersymbol;however ,thediscussioninthesequelalso (generally)appliestomultiplepulsespersymbol.Thefoll owingdecisionstatisticisused tomakeasymboldetectionbysensingifthereisenergyornot withinthesymbolinterval hi= ZT ij ~ r ( t ) j2dt1?0; (2) 9

PAGE 21

Symbol Detection Estimation Parameter t stationary '0' s '1' s '0' s'1' sFigure3.Parameterestimationandsymboldetectioninbloc k-fadingchannelmodel. where Tiistheintegrationwindowdenedbysynchronizationanddum ppoints( u;v ),and thesymboldecisionisperformedbycomparing hiwithathreshold .Observing(2),it isseenthatoptimal(joint)estimationof( u;v; )tupleisofcriticalimportanceforthe performanceofenergydetectors,aswillbediscussedthrou ghouttherestofthissection. 2.3.2OptimumJointParameterSelectionWirelesscommunicationsystemstypicallyrequiretheesti mationofchannel-relatedparametersforoptimaldemodulationofreceivedsymbols.Sincec hannelcharacteristicschange intime,theparameterestimationhastobetrackedand/orre peatedeveryonceinawhile; howoftentheparameterestimationhastoberepeateddepend sonthecoherencetimeofthe channel.AcommonlyusedmodelforUWBchannelsisablockfad ingchannelmodel[11], wherethechannelisassumedstationarywithinaspecicblo ck(e.g.for200microseconds[12]),anddierentchannelrealizationsareconsider edfordierentblocks.Therefore, theradiochannelcharacteristicsvaryinthelong-term,an dtheymaybeassumedstationary intheshort-term. Sincetheoptimalparametersforanenergydetectorwillvar yfordierentchannelrealizations,areceiverdesignthatoptimizestheperformance foraparticularchannelrealization isneeded.AsillustratedinFig.3,theparameterscanbeest imatedatthebeginningof eachblock,andthenbeusedfordemodulationofthesymbolsf ortherestoftheblock. Theproposedadaptivereceiver,whichtakesintoaccountth echangesinthechannel,is showninFig.4.Inthisreceiver,thereceivedsignalisrst amplied,bandpassltered, andsquared.Then,dierenthypothesesfor( u;v )areconsidered,andthecorresponding thresholdisestimatedforeachhypothesis.Throughoutthe restofthissection,rst,issues 10

PAGE 22

relatedtoobtainingtheintegratorstart/stophypotheses willbediscussed.Then,exact andGaussianapproachesforthresholdestimationwillbepr esented. 2.3.2.1Obtainingthe ( u;v ) HypotheseswithDierentSamplingApproaches Whenimplementinganenergydetector,specifyinganintegr ationintervalthatsacrices theinsignicantmultipathcomponentsinordertodecrease thecollectednoiseenergywill improvetheperformance.Forabetterperformanceitisalso requiredthatthereceiversynchronizeswiththestartingpointofthemultipathenergy.T herefore,theoptimalinterval, whichminimizestheBER,canideallybeachievedbyajointan dadaptivedetermination ofthestartingpointanddurationofintegration. Let u ( k )and v ( k )denotethestartinganddumppointsofthe k thhypothesis,respectively.Granularityofthe( u ( k ) ;v ( k ))pairdependsonthesamplingrate,andtheymay beobtainedusingdierentarchitectures.Below,threecon venientwaysofobtainingthe start/stoppointsformultiplehypothesisarepresented: Hypotheses xk(u(k), v(k), ) u v ,optoptxopt H{u(1),v(1)} H{u(2),v(2)} v(1) u(1) v(2) u(2) v(N) u(N) H{u(N),v(N)} argmin(BER)Nx x x12Integration Start & Stop Estimation Threshold z(t) BPF ( ) 2 LNA z(t)Figure4.Blockdiagramfortheproposedjointparameterest imationforenergydetector receivers. 11

PAGE 23

MultipleparallelintegratorbranchesEachbranchhasadierenttimeconstantandhenceadierent lengthofintegration interval.Integrationstartingpointsareadjustedusingd elayelements.Theintegrator outputsaresampledatasymbol-spacedrate,andtheeectiv egranularityis Ts=N ,where N isthenumberofintegrators.Thedisadvantageofthisappro achisthelargenumberof integratorsthatmayberequired. SingleintegratorwithhighsamplingrateThehigh-ratesamplingatarate Ts=N enablesdeterminingtheenergyinnerresolution. Thestartingandstoppointsareselectedbycombiningthese sampleenergiesinsuchaway toyieldtheoptimumtotalenergy.Thedrawbackcomparedwit hotheroptionsisthe requirementofahighspeedanalogtodigitalconverter(ADC ).MultipleparallelADCs mayalsobeconsideredtoincreasethesamplingrate. SingleintegratoremployingtrainingsequencesTrainingsequenceslongerthanusualenabletestingdiere ntintegrationintervalsina sequentialmanner.Symbol-ratesamplingoftheintegrator issucient.However,sincelarge numberoftrainingsymbolsarerequiredtoincreasethesamp lingrate,thecoherencetime ofthechannelshouldbesucientlylong.Ontheotherhand,s incesymbol-ratesampling willbeusedinthesymboldemodulationanyway,thisisthele astcompleximplementation ofthereceiver. Notethatincreasingtherateatwhichtheoutputoftheinteg ratorissampled,ineect, increasesthe`integrationtimeresolution'ofthereceive randenhancesthelikelihoodofobtainingalowerBER.However,thiscomesattheexpenseofadd itionalhardwarecomplexity. Ontheotherhand,highsamplingratesarerequiredonlywhen estimatingtheintegration start/stoptimes,andsymbol-spacedsamplingissucientd uringsymboldetection.Nevertheless,weassumeinthesequelthatusingoneoftheabove approaches,theintegration start/stophypothesisbecomeavailabletothereceiver. Asub-optimalsolution,wheretheinitialpointoftherecei vedsignalistakenasthe commonstartingpointforallpossibleintegrationduratio ns,yieldsverycloseperformance totheoptimalcase,whenthepowerdelayprole(PDP)ofthec hannelrealizationis 12

PAGE 24

exponentiallydecaying.Forexample,thechannelmodel1(C M1)in[11]rerectssucha minimumphasescenariowheresinglesynchronizationpoint performsaswell.Fordispersive channels(suchasCM4)however,therewillbesomeperforman cedegradation. 2.3.2.2ThresholdSelectionUsingExactandGaussianAnaly sis The exact optimalthreshold ( E ) kcanbecalculatedusingthecentralizedandnon-centralize d Chi-squaredistributions,correspondingtobits0and1,re spectively,andwhere k denotes thehypothesisnumber.However,thisrequiresasearchover possiblethresholdvalues inordertondtheonethatminimizestheBER,or,highsignal tonoiseratio(SNR) assumptioninordertouseasymptoticapproximationoftheB esselfunction(whichstill yieldsathresholdestimatebasedontabulateddata)[13].R elyingonthefactthatthe normalizedthresholdforpracticalSNRvaluesfallsinbetw een0 : 25and0 : 5[10],inorderto decreasethecomputationalcomplexity,hereaserialsearc hfor ( E ) kintherange( MN0+ 0 : 5 Eb;MN0+ Eb)isconsidered,where M isthedegreeoffreedom(DOF)denedby2 M = 2 BTi+1,and Ebistheaverageenergyofbits0and1,bit1havinganenergyof2 Eb. ByapproximatingtheChi-squaredistributionswithGaussi andistributions(whichbecomesmorevalidforlargeDOF),thethresholdestimates ( G ) kcanbeobtained(asan approximationto ( E ) k).Eventhoughtheseestimatesaresuboptimal,theycanbeob tained easily,withoutrequiringanysearchoverpossiblethresho ldvalues.LetthemeansandvariancesoftheChi-squaredistributionsforbits0and1begive nby 0 ;k;2 0 ;k;1 ;k; and 2 1 ;k, respectively,where,accordingto[14], 0 ;k= MN0(3) 2 0 ;k= MN2 0(4) 1 ;k= MN0+2 Eb(5) 2 1 ;k= MN2 0+4 EbN0: (6) 13

PAGE 25

ThethresholdestimateusingtheGaussianapproximationis locatedattheintersectionof thetwoGaussiandistributions,whichcanbeevaluatedfrom exp ( ( G ) k 0 ;k ) 2 2 2 0 ;k q 2 2 0 ;k= exp ( 1 ;k ( G ) k ) 2 2 2 1 ;k q 2 2 1 ;k: (7) Takingthenaturallogarithmofbothsidesandrearrangingt heterms,oneobtains C1( ( G ) k)2+ C2( G ) k+ C3=0 ; (8) wherethecoecientsaregivenby C1= 2 1 ;k 2 0 ;k; (9) C2= 2 0 ;k2 1 ;k 1 ;k2 0 ;k ; (10) C3= 2 1 ;k20 ;k 2 0 ;k21 ;k 2 2 0 ;k2 1 ;kln 1 ;k 0 ;k ; (11) with(8)beingasecondorderpolynomialequationthatcanbe easilysolvedfor ( G ) k(only oneoftherootsisappropriate)yielding opt= C2+ p C2 2 4 C1C3 2 C1: (12) Asanalternativetousingfrequenttrainingsymbols,theth resholdcanbeupdated(tracked) inadecision-directedmanneronceitisinitiallyestimate dinasimilarwaytoadata-aided channelestimation[15].2.3.3BERPerformanceEvaluationThreedierentapproachesareconsideredforevaluatingth eBERoftheenergydetector receiversassummarizedinFig.5.Duetothesquare-lawdevi ceusedinthereceiver, thedecisionstatisticsinanenergydetectorhaveaChi-squ aredistribution.First,the exactstatisticsareconsidered,andtheBERexpressionsas availableintheliteratureare 14

PAGE 26

Exact Threshold Exact Calculation BER Gaussian Approximation Gaussian Approx. ThresholdFigure5.Threedierentapproachesforperformanceevalua tion. evaluated.However,thethresholdisconsideredusingboth theexactapproach(usinga searchoverpossiblethresholdvalues)andtheGaussianapp roximation(usinganalytical expressionsobtainedinprevioussections).Later,theBER evaluationusingtheGaussian approximationoftheChi-squarestatisticsisconsidered.2.3.3.1ExactBERPerformancesWhentheexactChi-squarestatisticsofthereceivedsignal areconsidered,theBERobserved foreachhypothesiswhenusingaserialsearchoraGaussiana pproximationforthreshold estimationaredenotedby P( E ) b k;( E ) k and P( E ) b k;( G ) k ,respectively.Usingtheexact expressions,theBERsemployingeitherthresholdaregiven by P( E ) b k;k = P( E ) k; k(0 j 1) p (1)+ P( E ) k; k(1 j 0) p (0) ; (13) P( E ) k; k(0 j 1)=0 : 5 0 : 5 QM r 4 Eb N0; r 2 k N0! ; (14) P( E ) k; k(1 j 0)= e k N 0 2b M cXu =1( k=N0)M u ( M u +1) ; (15) where p (0)and p (1)aretheprobabilitiesforbit0andbit1,respectively, QMisthegeneralizedMarcumQ functionoforder M ,and 15

PAGE 27

( x )= Z1 0tx 1e tdt;x> 0(16) ( x )=( x 1)! ;x> 0 isaninteger (0 : 5)= p ; and (1 : 5)= p 2 : Theoptimumintegratorparametersaretheonesthatminimiz etheBER,i.e. ( uopt;vopt;opt)=argminu ( k ) ;v ( k ) ; k P( E ) b( k;k) : (17) AsanalternativetominimizingtheBER,onemayconsidertom aximizetheSNR(which haslesscomplexitysincenoBERexpressionsareevaluated) .However,thedenitionof 15 16 17 18 19 20 21 22 23 24 25 10 -8 10 -6 10 -4 10 -2 E b /N 0 (dB)BER Gaussian with Gaussian Threshold (500 MHz) Exact with Gaussian Threshold (500 MHz) Exact with Exact Threshold (500 MHz) Gaussian with Gaussian Threshold (2 GHz) Exact with Gaussian Threshold (2 GHz) Exact with Exact Threshold (2 GHz) Figure6.Biterrorratevs. Eb=N0forCM1(BW=500MHzand2GHzcases)forboth Gaussianapproximatedandexactthresholdestimates. 16

PAGE 28

SNRiscriticalinenergydetectors.OnemaydenetheSNRtob etheratioofthesquare ofthemean-shiftduetotheexistenceofsignaltotheoutput noisevariancewhensignalis present[16],whichisexpressedas SNR= ( 1 ;k 0 ;k)2 2 1 ;k; = 4 E2 b MN2 0+4 EbN0; (18) andtheparametersthatmaximize(18)canbeselected.Howev er,notethat(18)doesnot accountforthenoisestatisticswhensignalisnotpresent, andthusdoesnotcapturethe wholepicture.Thisisasopposedtoacoherentsystem,where noisestatisticscorresponding tobothbit-0andbit-1areidentical,andmaximizationofth eSNRimpliestheminimization oftheBER.2.3.3.2BERUsingtheGaussianApproximationFortheoreticalpurposes,anapproximateBERformulationt hatgivesafeasibleestimate for Pb( k;k)isgivenby P( G ) b( k;( G ) k)= 1 2 Q ( G ) k 0 ;k 2 0 ;k! + 1 2 Q 1 ;k ( G ) k 2 1 ;k! : (19) SincetheChi-squarestatisticscanbeapproximatedwithaG aussianforlargedegreeof freedoms,theaboveexpressionisexpectedtoapproximatet heBERatlargebandwidths, orlargeintegrationintervals.Itisalsovalidforsystems thatuselargenumberofpulses persymbol1. 2.3.4NumericalResultsComputersimulationsaredonetoanalyzetheperformanceso ftheproposedapproaches usingthechannelmodelsin[11].Tobemorespecic,inthese simulationstheenergies 1 Notethatifmorethenonepulseisusedpersymbol,andthepul sesarecombinednon-coherently,the numberofpulsescanbefoldedintotheintegrationinterval ,implyingthatthedecisionstatisticsapproach toaGaussiandistribution17

PAGE 29

0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 x 10 -8 10 -3 10 -2 10 -1 T i (ns)BER 10dB 20dB CM1 CM2 CM3 CM4 Figure7.BERvs.integrationintervalfordierentchannel models(at Eb=N0=10dBand 20dB).correspondingtothedierentchannelrealizationsandpar ametersetsareevaluated,and usedintheBERexpressions. InFig.6,theBERsobtainedusingthethreedierentperform anceanalysisapproaches (showninFig.5)arecomparedfor B =0 : 5GHzand B =2GHz.WhiletheGaussian approximationfailstoyieldcloseresultstotheexactexpr essionsfor B =0 : 5GHz,itisseen thattheapproximationerrordecreasesasthebandwidthinc reases.Ontheotherhand,for bothbandwidthspracticalestimationofthethresholdusin gtheGaussianapproximation yieldsverycloseresultswiththeexactthreshold(whichha stobecalculatedafteraserial search).Hence,theGaussianapproximatedthresholdcanbe employedtodecreasethe computationalcomplexity. Anotherobservationisthattheoptimumintegrationinterv alchangessubstantiallyfor dierentchannelmodels,implyingthefactthatsignicant gainscanbeobtainedfora 18

PAGE 30

mobiledevicewhentheintegrationintervalisadaptivelyd etermined.BoththeBERminimizationandSNRmaximizationapproachesareemployedton dtheoptimumintegration interval.TheresultsareshowninFig.7andinFig.8,respec tively.Althoughtheresulting curveshaveasimilarbehaviour,theoptimumintegrationin tervalsdeterminedbytheSNR maximizationapproachturnouttoyieldhigherBERsthanthe onesfoundwiththeBER minimization.Therefore,itisreasonabletoconcludethat minimizingtheBERisfavorable tomaximizingSNRdespiteitscomputationalcomplexity. InFig.9,thevariationoftheoptimalintegrationinterval withrespectto Eb=N0is plottedfordierentchannelmodels.Itisobservedthatthe line-of-sight(LOS)component ofCM1yieldsaparallelvariationwithCM2.Ontheotherhand ,CM3andCM4also exhibitaparallelbehaviorandtheyhavelargeroptimalint egrationvalues(andslopes)due tothemorespreaddistributionoftheirmultipathcomponen tsovertime. 5 10 15 20 25 30 35 40 45 50 55 5 10 15 20 25 30 35 40 45 50 55 T i (ns)SNR CM4 CM3 CM2 CM1 Figure8.SNRvs.integrationintervalfordierentchannel models(at Eb=N0=20dB). 19

PAGE 31

6 8 10 12 14 16 18 20 0 10 20 30 40 50 60 E b /N 0 (dB)Optimum Interval (ns) Gaussian Approx.Exact Calculation CM1 CM2 CM3 CM4 Figure9.Optimumintegrationintervalvs. Eb=N0fordierentchannelmodels. InFig.10,theBERperformancesofanon-adaptivereceivera ndtheproposedreceiver arecompared.Thenon-adaptivereceiverisassumedtohavea xed integrationintervalof 20ns,whichisareasonabledurationconsideringtheoptimu mvaluesfordierentchannel modelsgiveninFig.9.TheresultantBERcurvesarepresente dforCM1andCM4. Theperformanceoftheproposedreceiverisbetterthanthen on-adaptivereceiverwith anappropriatelyselectedxedintegrationintervalbyapp roximately1dB.Inthesame gure,thesynchronizationeectisalsoillustrated.Sync hronizationisachievedbyhaving thereceiversynchronizeitselfwiththestartingpointoft heoptimumintegrationinterval ratherthantheinitialmultipathcomponent.Itisseenthat theeectofsynchronizationis negligibleforCM1andveryslightforCM4. Intheprevioussimulations,perfectparameterestimatesf or(3)-(6)wereconsidered. Anotheranalysisinvestigateshowthenumberoftrainingsy mbolsaectstheparameter 20

PAGE 32

5 10 15 20 25 10 -7 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 10 0 E b /N 0 (dB)BER CM 4, 20ns, without synch. CM 4, 20ns, with synch. CM 4, adaptive interval + synch. CM 1, 20ns, without synch. CM 1, 20ns, with synch. CM 1, adaptive interval + synch. Figure10.BERvs. Eb=N0forxedintegrationintervals,adaptiveintegrationinte rval,and adaptivesynchronizationpoint.estimationand,asaresult,theBER.Thisisananalyticalex aminationratherthana simulation,andtherefore,practicalchannelrealization sarenotconsidered.InFig.11,the BERvs.numberoftrainingsymbolscurvesareplottedatdie rent Eb=N0values.These resultsareobtainedbytakingsamplesfromthecentralized andnon-centralizedChi-square distributionsofbit-0andbit-1,respectively.Eachsampl ecorrespondstoatrainingsymbol transmitted.Obviously,takingmoresamplesyieldsabette restimateforthesymbolenergy. Asignicantconclusionthatcanbedrawnfromthisgureist hatas Eb=N0increases,the numberoftrainingsymbolsrequiredtoconvergetotheoptim umBERincreasesaswell.The reasonforthisfactisthatasthesignalenergyrises,thepr obabilitydensityfunctionforbit1becomesbroader,andhence,moresamplesarerequiredfora moreaccurateestimation. ThetheoreticaloptimumBERsarealsoindicatedonthegure .NotethattheseBERsare 21

PAGE 33

10 20 30 40 50 60 70 80 90 100 10 -11 10 -10 10 -9 10 -8 10 -7 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 # of training symbolsBER 15 dB 20 dB 25 dB Figure11.AnalyticalresultsregardingBERvs.numberoftr ainingsymbols(at Eb=N0= 15dB,20dB,and25dB).dierentfromtheonesshowninFig.7.Thisisbecauseinthis analysis,theentiresymbol energyisconsideredratherthanonlytheenergyconnedtot heintegrationinterval. 2.4HighDataRateUWBUsingEnergyDetectorReceiversInthissection,theinter-symbolinterference(ISI)issue inimpulseradioUWBsystemsis addressed.Anenergydetectorbasedreceiverthathashighd ataratecapabilitythrough ISIcancelationisproposed.TheISIsuppressionalgorithm ofthereceiverdependsonthe usageofasimplieddecisionfeedbackequalizer(DFE),whi chdoesnotemployfeed-forward lters,butonlyfeedbacklters.Inspiteofthefactthatde cisionfeedbackequalizersare knowntocauseerrorpropagation,animportantfeatureofth eproposedDFEequalizeris 22

PAGE 34

thatitdoesnothaveerrorpropagation.TheDFEequalizerre quiressomeparametersto bemeasured.Theparameterestimationinenergydetectorre ceiverswillbeaddressed,and asimplealgorithmwillbedevelopedtoestimatethesignale nergyleakingfromonesymbol intothefollowingsymbol.2.4.1ISIProblemBecauseofitstemptingfeaturessuchasaimingatextremely highdatarateswithconsiderablylowcostcircuitry,UWBisconsideredacandidateforwi relesspersonalareanetworks (WPAN),whoserangeisupto10meters.Forthechannelmodels CM1andCM2in[11], whichareappropriateforthisrange,themaximumexcessdel ays(MED)arearound80 nsand115ns,respectively.Hence,athighdatarates,somep ortionofthetransmitted symbolenergyunavoidablyleaksintothefollowingsymbols ,leadingtointer-symbolinterference.ISIisoneoftheprimaryfactorsdegradingthedete ctionperformanceofUWB systemsandithastobesuppressedforsuccessfulimplement ationofhighdatarateUWB communications. Inordertomeetthehighdataraterequirement,UWBsystemha stoemployareceiver thatisalsocapableofhandlingISI.AlthoughRAKEreceiver scanprovideasatisfyingsolutiontothisproblem[17]-[19],theirhighcomplexityincre asesthesystemcostdramatically. Hence,alternativetransceiverdesignsareneededthathav elowcomputationalandhardwarecomplexity,whileprovidinghighdatarates.Possible candidatesarethenon-coherent receiverssuchastheenergydetectororthetransmittedref erencereceiver. 2.4.2SystemModelTheimpulseradiobasedUWBsignalreceivedforbit i inamultipathenvironmentcanbe representedas ri( t )=LXl =1rlpl( t )+ n ( t ) ; (20) 23

PAGE 35

0 5 10 15 20 25 30 35 40 45 -0.2 -0.1 0 0.1 0.2 Time (ns)Amplitude 0 5 10 15 20 25 30 35 -0.1 -0.05 0 0.05 0.1 0.15 Time (ns)Amplitude Figure12.(a)Thepulserepetitionperiod( )greaterthanthemaximumexcessdelay( D ). (b) D> ,energyleaksfromonesymboltothesubsequentsymbol. where pl( t )= 8><>: bivl t l Td + rivl t l ; for PAM vl t l Td bi + rivl t l ; for PPM L isthenumberofmultipathcomponentsarrivingatthereceiv er, l isthetapindex, biis the i thtransmittedbit, rl, vl( t )and larethefadingcoecient,thereceivedpulseshape, andthedelayofthe l thmultipathcomponent,respectively, isthedurationbetweenthe twopossiblepositionsinpulsepositionmodulation, rihasabinaryvaluedeterminingthe existenceofareferencepulse,and TdisthedelaybetweenthereferenceanddatainTR systems.Inthecaseofanenergydetector,both riand Tdbecomezero.Theadditivewhite Gaussiannoisewithdouble-sidednoisespectraldensity N0= 2isdenotedby n ( t ). 24

PAGE 36

2.4.3EectofISIIntheimpulseradiobasedultrawidebandcommunications,t hetransmittedUWBpulses gothroughahighlyfrequencyselectivechannel,andbecome dispersedintime.Letthe maximumexcessdelayofthechannelbedenotedby D ,andthepulserepetitionperiodby .InmanyworksdealingwithUWB, isassumedtobelongerthan D (illustratedinFig. 12-a).However,inpracticalhighdataratecommunications ,thepulserepetitionperiodis muchshorterthanthemaximumexcessdelay(Fig.12-b),i.e. D ThetemporaldispersivenessoftheUWBpulsecausesaconsid erableportionofthesymbolenergytoappearasapartofthefollowingsymbol,leadin gtointer-symbolinterference. Thisproblemisvalidforallkindoftransmissionschemes,h owever,hereon-okeyingis considered,whichisaspeciccaseofPAM. 0 20 40 60 80 100 120 140 160 180 200 10 -10 10 -9 10 -8 10 -7 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 Data Rate (Mbps)BER CM1 with ISI at 15 dB CM1 without ISI at 15 dB CM1 with ISI at 20 dB CM1 without ISI at 20 dB Figure13.BERvs.datarateforchannelmodelCM1at Eb=N0valuesof15dBand20dB. 25

PAGE 37

0 20 40 60 80 100 120 140 160 180 200 10 -11 10 -10 10 -9 10 -8 10 -7 10 -6 10 -5 10 -4 10 -3 10 -2 10 -1 BERData Rate (Mbps) CM2 with ISI at 15 dB CM2 without ISI at 15 dB CM2 with ISI at 20 dB CM2 without ISI at 20 dB Figure14.BERvs.datarateforchannelmodelCM2at Eb=N0valuesof15dBand20dB. TovisualizetheISIeect,simulationsareperformedconsi deringthedierentchannel modelsin[11].Inthesimulations,afthorderderivativeo ftheGaussianpulse,which satisestheFCClimitationsregardingthetransmissionba ndwidth,isused.Theresults obtainedrevealthatatadatarateof100Mbps,theaveragera tioofISItothesymbol energyis11.8%forCM1,andashighas25.89%forCM2.InFig.1 3andFig.14,the eectofISIinCM1andCM2fordataratesupto200Mbpsisdemon strated.The Eb=N0valuesusedare15dBand20dB,respectively.Atrelativelyl owdataratessuchas10 Mbps,theISIeectisalmostunnoticeable.However,asthed atarateincreases,ISIgrows considerably.Towards200Mbps,thecurvesfor15dBand20dB convergetothesame level.Thereasonforthisfactisthatathighdatarates,the eectofISIalmosttotally dominatesthenoiseeect. 26

PAGE 38

2.4.4ISICancelingReceiverTheproposedenergydetector(showninFig.15)isimplement edbytakingthesquareof thereceivedsignal,integratingit,andpassingittoadeci sionmechanism,wherethesymbol isdetermined.Tomakeasymboldecisionbysensingiftherei senergyornotwithinthe symbolinterval,thefollowingdecisionstatisticcouldbe usedifISIdidnotexist ( i )= ZT i si( t )+ n ( t ) 2dt: (21) IntheexistenceofISI,however,thevaliddecisionstatist icis b ( i )= ZT inXj =0 s2( i j )( t )+2j 1Xx =0s( i j )( t ) s( i x )( t ) + n2( t ) dt;x 0(22) where s( i j )( t )=Xl =1b( i j )rvl t l ; (23) l isthetapindex, bi jisthe jthprevioussymbolvalue,and isthenumberoftapsinside theintegrationintervalof ithsymbol.Thedecisionstatisticcanbedenotedasfollows b ( i )= ( i )+nXj =1bi j( i;j ) ; (24) where ( i;j )istheamountofleakingenergyfromthe jthprevioussymboltothe ithsymbol. Thesymboldecisionisperformedbycomparing b ( i )withathreshold b ( i )1?0 .Adetailed analysisaboutsettingtheoptimum isgivenin[4]. InordertoimprovethedetectionperformanceoftheUWBsyst em,theeectofISIhas tobecanceledbeforeabitdecisionismade.Incommunicatio nsystems,decisionfeedback equalizersarecommonlyusedforthispurpose[20].Thedeci sionmechanismoftheproposed receivermakesuseofasimpliedDFEalgorithm,whichemplo ysonlyfeedbackltersbut nofeed-forwardlters.From(24),itisobviousthatfordet ermining bicorrectly,theDFE 27

PAGE 39

BPF ( ) 2 u v LNA + Feedback Filter Detector Decision Feedback Equalizer Figure15.DecisionfeedbackequalizationbasedISIcancel ingenergydetector. requirestheestimationof ( i;j ),whicharetheequalizercoecientswithpositivereal values.Inthissection,onlytheenergyleakingfromtheimm ediatelyprevioussymbol ( i; 1) isconsidered,becausethishasthestrongesteectonthecu rrentsymbol.In ordertoestimate ( i; 1),transmittingsequencesoftrainingsymbolsbetweenthe packets ofdatasymbolsisproposed.Duringthetrainingsequences, thechannelisassumedto benon-varying,hencetheltercoecientsareconsideredc onstant.First,transmitting 0 20 40 60 80 100 120 140 160 180 200 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 Time (ns)Amplitude 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 0 Figure16.Successivetrainingsequencesusedforestimati ngthefeedbackltercoecients. 28

PAGE 40

asetof'0's,thenoiseenergy RT in ( t )2dt isestimated.Then,areasonableapproachis transmittingsequencesintheformof 100 ::: 0 (showninFig.16).Theenergyobtained duringtherst'0'followingthe'1',whichcanbeshownas bi 1( i; 1)+ ZT in ( t )2dt; (25) ismeasured.TheDFEcoecient ( i; 1)isfoundbysubtractingtheestimatednoiseenergy fromthiscompositeenergy.The'0'sfollowingtherst'0'a retransmittedasaguardband consideringthatathighdataratesISImaybesevere,andfur thersymbolsmaybeaected. Usingmultipletrainingsequencessuccessively,anaverag evaluefor ( i; 1)canbefound. Then,duringthedataprocessing,everytimewhena'1'isdet ected,theDFEmechanism subtractsthisDFEcoecientfromtheimmediatelyfollowin gsymbolenergy,andthus cancelstheISIeect.Ifthechannelishighlydispersive,t henumberoffeedbacklter tapscanbeincreasedsuchthattheeectonfurthersymbolsi ssuppressed.Therequired coecientscanbefoundtheway ( i; 1)isdetermined. Thedecisionfeedbackequalizersareknowntocauseerrorpr opagationwhenprocessing thereceiveddata.InthecaseofOOKmodulatedUWBsignals,h owever,thisproblemdoes notoccur.Keepinginmindthataftereach'1'theleakingene rgyhastoberemovedfrom thenextsymbol,if biis'0',anditismistakenlydetectedas'1',thentheDFEcoe cient isunnecessarilysubtractedfrom bi +1.If bi +1is'1',thissubtractionmaycauseittobe detectedas'0'.So,theerrorispropagatedbyonesymbolint hiscase.If bi +1isalready '0',theerrorin bidoesnotcauseittochange.If,ontheotherhand, biis'1',anditis mistakenlydetectedas'0',nosubtractionisdoneon bi +1,therefore,noerrorrelatedto DFEoccurs.Obviously,theproposedalgorithmhastheadvan tagethatanerrorarisendue toanyreasonisnotforwardedbymorethanonesubsequentsym bol,hencethereisnoerror propagation. 29

PAGE 41

2.4.5PerformanceResultsSimulationsaredonetotesttheperformanceoftheproposed energydetectorreceiver.A pulsebandwidthof2GHzisassumed,andthechannelmodelsin [11]areused.Foreach channelmodel,theoptimumintegrationdurationgivenin[4 ]isused(aslongasitdoesnot exceedthesymbolperiod).InFig.17,thebiterrorratesobt ainedatadatarateof100 Mbps,withandwithoutimplementingtheproposedISIcancel ationalgorithmarecompared forchannelmodelsCM1-CM4.Itisrevealedthatanimportant gaincanbeachievedusing theISIcancelingenergydetector. 5 10 15 20 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 Eb/N0 (dB)BER CM1, ISI eliminatedCM1 with ISICM2, ISI eliminatedCM2 with ISICM3, ISI eliminatedCM3 with ISICM4, ISI eliminatedCM4 with ISI Figure17.Biterrorratevs. Eb=N0fordierentchannelmodelsbeforeandaftertheISI suppression. 30

PAGE 42

2.5ConclusionInthischapter,afterintroducingUWBanddiscussingvario uspossiblemodulationand receivertypeoptions,optimizationofadaptiveenergydet ectorreceiversforUWBsystems isdiscussed.Theneedforthejointadaptationoftheintegr ationinterval,optimalthreshold, andthesynchronizationpoint(forcertainchannels)iscle arlydemonstrated,whichcanbe extendedtoothernon-coherentapproaches.Thresholdesti mationcanbenetfromthe computationaleasinessbroughtbytheGaussianapproximat ionofreceivedsignalstatistics, whichyieldsreasonableresultsforcertainbandwidths. Inthesecondpartofthechapter,theinter-symbolinterfer enceprobleminhighdata rateUWBsystemsemployingOOKbasedenergydetectorsisinv estigated.Theincreasing negativeeectofISIonthesystemperformancewithincreas ingdatarateisdemonstrated. InordertoovercometheISIproblem,amodiedenergydetect orthathasabuilt-insymbol decisionmechanismbasedondecisionfeedbackequalizatio nisproposed.Asimplebut cleverwayofusingtrainingsymbolswiththepurposeofesti matingthedecisionfeedback ltercoecientsisintroduced.Ithasbeenproventhatthee rrorpropagationproblem, whichisgenerallyobservedindecisionfeedbackequalizer s,doesnotexistintheproposed approach.Inthenalsection,thegainsachievablewiththe proposedenergydetectorare exhibited.Thesegainsturnedouttobeconsiderablyhigh,v erifyingthenecessityofISI cancelationandshowingtheeectivenessoftheproposedde tector. 31

PAGE 43

CHAPTER3 UWBANDNARROWBANDSYSTEMS Inthischapter,thecoexistenceofUWBandnarrowbandsyste msisaddressed.Theeectof narrowbandinterference(NBI)onUWBisanalyzed.Thetechn iquesofsuppressingNBI, whichareclassiedasNBIavoidanceandNBIcancelationmet hods,areinvestigatedin detail.3.1IntroductionUltrawidebandisbecominganattractivesolutionforwirel esscommunications,particularly forshortandmediumrangeapplications.UWBsystemsoperat eoverextremelywidefrequencybands,wherevariousnarrowbandtechnologiesalsoo peratewithmuchhigherpower levels(illustratedinFig.18).Theunlicensedusageofave rywidespectrumthatoverlaps withthespectraofnarrowbandtechnologiesbringsaboutso meconcerns.Therefore,significantamountofresearchhasbeencarriedoutlatelytoquant ifytheeectofUWBsignals onnarrowbandsystems[21]. ThetransmittedpowerofUWBdevicesiscontrolledbythereg ulatoryagencies(such astheFCCintheUnitedStates),sothatnarrowbandsystemsa reaectedfromUWB signalsonlyatanegligiblelevel.Thisway,UWBsystemsare enabledtoco-existwith narrowbandtechnologies.However,lookingatthefactfrom theotherside,theinruenceof narrowbandsignalsontheUWBsystemcanstillbesignicant ,andintheextremecase, thesesignalsmayjamtheUWBreceivercompletely.Eventhou ghnarrowbandsignals interferewithonlyasmallfractionoftheUWBspectrum,due totheirrelativelyhigh powerwithrespecttotheUWBsignal,theperformanceandcap acityofUWBsystems canbeaectedconsiderably[22].Therecentstudiesshowth atthebit-error-rate(BER) 32

PAGE 44

Figure18.Spectrumcrossoverofthenarrowbandinterferer sinUWBsystems. performanceoftheUWBreceiversisgreatlydegradedduetot heimpactofnarrowband interference[23]-[28].ThehighprocessinggainoftheUWB signalcancopewiththe narrowbandinterfererstosomeextent.However,inmanycas es,eventhelargeprocessing gainaloneisnotsucienttosuppresstheeectofthehighpo werinterferers.Therefore, eithertheUWBsystemdesignneedstoconsideravoidingthet ransmissionoftheUWB signaloverthefrequenciesofstrongnarrowbandinterfere rs,ortheUWBreceiversrequire toemployNBIsuppressiontechniquestoimprovetheperform ance,thecapacity,andthe rangeoftheUWBcommunications. NBIisnotanewproblem.Ithasbeenstudiedextensivelyforw idebandsystemslike directsequencespreadspectrum-codedivisionmultipleac cessing(DSSS-CDMA)based wirelesscommunications[29],andfortheoperationofbroa dbandOFDMsystemsinunlicensedfrequencybands[30].InDSSS-CDMAsystems,NBIispa rtiallyhandledwiththe processinggainaswellasbyemployinginterferencecancel ationtechniques.Approaches includingnotchltering[31],linearandnon-linearpredi ctivetechniques[32]-[37],adaptive methods[38]-[41],minimummeansquareerror(MMSE)detect ors[42,43],andtransform domaintechniques[44]-[50]havebeeninvestigatedextens ivelyforinterferencesuppression. Similarly,inOFDMsystems,interferencecancellationasw ellasinterferenceavoidance techniqueshavebeenstudied[30],[51]-[54].Comparedtot hesewidebandsystems,NBI suppressioninUWBisamorechallengingproblembecauseoft herestrictedpowertrans33

PAGE 45

missionandthehighernumberofnarrowbandinterferersdue totheextremelywideband occupied.Moresignicantly,incarriermodulatedwideban dsystems,beforedemodulatingthereceivedsignalboththedesiredwidebandandthenar rowbandinterferingsignals aredown-convertedtothebaseband,andthebasebandsignal issampledatleastwith theNyquistrate.SamplingattheNyquistrateallowstoempl oynumerousecientnarrowbandinterferencecancelationalgorithmsbasedonadva nceddigitalsignalprocessing techniques.However,inUWB,thedesiredsignalisalreadyi nthebaseband,whilethenarrowbandinterfererisinradiofrequency(RF).Samplingthe receivedsignalattheNyquist ratebeforethepulsecorrelatorrequiresanextremelyhigh samplingfrequency,whichis notrealizablewiththeexistingtechnology.Inadditionto thehighsamplingrate,the analog-to-digital-converter(ADC)mustsupportaverylar gedynamicrangetoresolvethe signalfromthestrongnarrowbandinterferers.Currently, suchADCsarefarfrombeing practical.Asanalternative,applyinganalog(notch)lte ringbeforethepulsecorrelation isconsidered.However,thismethodrequiresanumberofnar rowbandanaloglterbanks, sincethefrequencyandpowerofthenarrowbandinterferers canbevarious.Therefore, employinganaloglteringaddscomplexity,cost,andsizet otheUWBreceivers.Also, adaptiveimplementationoftheanalogltersisnotstraigh tforward.Asaresult,manyof theNBIsuppressiontechniquesappliedtootherwidebandsy stemsareeithernotapplicableforUWB,orthecomplexitiesofthesemethodsaretoohi ghfortheUWBreceiver requirements. Giventhelowcomplexityrequirementsinbothhardwareandc omputation,andconsideringtheotherlimitationssuchaslowpowerandlowcosttra nsceiverdesigninmanyUWB applications,theNBIproblemneedstobehandledmorecaref ully,andeectivetechniques thatareabletocopewithNBIneedtobedeveloped.Oneapproa chtodealwithNBIisto avoidthetransmissionoftheUWBsignaloverthefrequencie sofpossiblestrongnarrowbandinterferers.Attemptstowardthisgoalincludeapproa cheslikemultiband-UWB(both usingimpulsebasedandOFDMbasedtechniques)[55,56].Ano therapproachtohandle NBIistodesigninterferencecancelingreceivers.However ,asmentionedpreviously,the interferencecancellationapproachinUWBhasmorelimitat ionscomparedtotheconven34

PAGE 46

tionalNBIcancellationapproachesemployedforotherwide bandsystems.Veryrecently, someNBIcancellationtechniques,mostofwhicharebasedon thepreviousmethodsimplementedinCDMAsystems,havebeenconsideredforUWB.Ana logbandpassltering hasbeenappliedbeforethecorrelationreceiverin[57].As discussedabove,xedanalog lteringisnotanecientsolution,unlesstheinterfereri sxed(i.e.thefrequency,bandwidth,power,andchanneloftheinterfererisconstant),an dalwaysexists.In[46],notch ltering(orpeakclipping)isapplieddoingahigh-speedsa mplingbeforethecorrelation. Thefrequencydomainsignalisobtainedfromthesedigitals amplesthroughfront-endfastFourier-transform(FFT).Then,thenarrowbandinterferer sinthefrequencydomain,which arethecollectionoflargepeaksinthefrequency,areclipp edornotch-lteringisappliedon theselocations.However,asdiscussedabove,highsamplin gratebeforecorrelationmakes thepracticalandcost-eectiveimplementationofthistec hniquedicult.Modifyingand estimatingtheoptimalreceivertemplateforthecorrelati onofthereceivedsignalisanother solutionthatisproposedforpartialsuppressionofNBI[57 ,58].Byfarthemostpopularly consideredapproachistheuseofRake(multiplecorrelator s)receiveralongwithMMSE combining[59]-[62].MMSEcombiningisknowntoperformwel lwhenthenoiseisnotwhite (i.e.noiseondierentRakengersarecorrelated).Theper formanceofMMSEdependson thenumberofngers.NotealsothatRakereceiversaremuchm orecomplexcomparedto thecorrelationreceivers,andtheircomplexityincreases withthenumberofRakengers. Inthischapter,NBIinUWBsystemswillbestudied.InSectio n3.2,theeectofNBI ontheperformanceofUWBtransmissionwillbediscussed.Ap propriatemodelsforNBI sourceswillbeinvestigated.InSection3.3,techniquesfo ravoidingNBIinUWBsystem designwillbereviewed.Approachesincludingmulti-band/ multi-carriertransmissionand pulseshapingforavoidingNBIwillbediscussedbriery.InS ection3.4,NBIhandling approachesbasedoninterferencecancelationwillalsobei nvestigatedforrelaxingthesystem andtransmissionrequirements.Finally,section3.5willc oncludethechapterwiththe discussionofsomefutureresearchareas. 35

PAGE 47

3.2EectofNBIinUWBSystemsAccordingtothemoderndenition,UWBtransmissionisnotl imitedtotheimpulseradio. Anytechnologythathasabandwidthgreaterthan500MHzoraf ractionalbandwidth greaterthan0.2canbeconsideredaUWBsystem.Therefore,d ependingontheaccess technology,thesignalandinterferencemodelsmightvary. Ingeneral,theUWBsignal bandwidthisextremelylargeandthetransmittedsignalpow erisverylow.Onthecontrary, thenarrowbandsignaloccupiesamuchsmallerbandwidth,wh ereitspowerspectrumisvery high.Anotherdistinctionisthatthenarrowbandsignalism odulatedwithacarrier,andit isacontinuoustimesignal,whereastheUWBsignalcanbeaba sebandsignalcomposedof discreteshort-timepulsesaswellasacarriermodulatedsi gnal. ImpulseradiobasedUWBtransmissionhassomesimilarities tothewidelyusedspread spectrum(SS)systems.IntheSSsystemssuchasdirectseque ncing(DS),thebandwidth occupiedislargerthanthebandwidthrequiredfortransmit tingthedatabitsforasingle user.Eachuserisassignedapseudo-random(PN)sequenceof Nchips(wherethechipdurationsaremuchshorterthantheactualsymbolduration)to transmitasymbol.Without thespreading,thesametransmissionbandwidthcanbeusedt otransmitNinformation symbols.But,thespreadingoperationallowssimultaneous transmissionofinformation frommultipleusersonthesamebandwidthwithoutinterferi ngwitheachother,leadingto theCDMAtypeofmultiplexing.Therefore,eventhoughthepe akdatarateisreducedfor asingleuser,thecapacityofthesystemispreservedtoagre atextentbyallowingmultipleusersinthesystem.Inadditiontothese,spreadingprov idesimmunitytointerference sourceslikeNBI,reducesthepowerofthetransmittedsigna l(sothatitcauseslessinterferencetoothersystemssharingthesameband),allowspath diversityinthepresenceof multipathsignalsthatarelongerthanachipperiod,andlas tbutnotleast,providescovert communications. TheNBIjammingresistanceofDSSSsystemshasbeenstudiede xtensively[26].The jammingresistanceinthesesystemsisprovidedbytheproce ssinggain,whichisobtainedby spreading.Thelargerthespreadingratio(theratioofband widthsofthespreadsignaland 36

PAGE 48

theoriginalinformation),thehighertheprocessinggain, andhence,thebetterjamming resistanceisobtained.Atthereceiver,thetransmittedsp readspectrumsignalandthe narrowbandinterferergothroughadespreadingoperation, wherethereceivertakesthe widebandspreadspectrumsignalandcollapsesitbacktothe originaldatabandwidth, whilespreadingtheinterferertoawidespectrum.Asaresul t,withinthedatabandwidth, theeectofinterfererismitigated,afact,whichisreferr edasthejammingresistanceor thenaturalinterferenceimmunityofthespreadspectrumsi gnals. SimilartotheDSSSsystems,impulseradiobasedUWBalsohas inherentimmunity toNBI.Time-hoppingUWB(TH-UWB)systemscanbeconsidered asanexample.The processinggainofTH-UWBsignalismainlyobtainedbytrans mittingverynarrowpulses withaverylowdutycycle.Fig.19-ademonstratesasimplesc enariothatshowstheUWB pulsesandthecontinuoustimenarrowbandinterferer.Duri ngthereceptionofTH-UWB signals,usingamatchedlterthatbasicallyoperatesasat imegate(i.e.letstheUWB signalalongwithinterferencepassoverthedurationofthe expectedpulses,andblocks therestofthereceivedsignal),thepoweroftheinterferin gsignalisreducedsignicantly (showninFig.19-b).Asaresult,jammingresistanceagains tNBIisobtained.Notethat therewillbestillpartialinterferenceattheoutputofthe matchedlterdependingonthe processinggain,thepoweroftheinterferer,andotherfact ors. ItisnecessarytoinvestigatethemodelsoftheUWBsignalan dnarrowbandinterferers forathoroughunderstandingofNBIeectsonUWBsystems.Co nsideringabinarypulse positionmodulated(BPPM)time-hoppingUWBsignal,thetra nsmittedwaveformcanbe modeledas[63] str( t )=+ 1Xi = 1ptr( t iTf ciTc d )(26) where ptrdenotestheUWBpulse, Tfisthepulserepetitionduration, ciisthetime-hopping codeinthe i thframe, Tcisthechiptime, isthepulsepositionosetregardingBPPM, and d representsthedata,whichisabinarynumber. Dependingonitstype,thenarrowbandinterferencecanbemo deledinvariousways.For example,itcanbeconsideredtoconsistofasingletoneinte rferer,whichcanbemodeled 37

PAGE 49

Figure19.(a)TH-UWBpulsesalongwithanarrowbandinterfe rer.(b)Reducedinterferencepowerbymeansoftimegating.as i ( t )= r p 2 Picos (2 fct + i) ; (27) where r isthechannelgain, Piistheaveragepower, fcisthefrequencyofthesinusoid, and iisthephase. 38

PAGE 50

NBIcanalsobethoughtastheeectofabandlimitedinterfer er,thenthecorresponding modelisazero-meanGaussianrandomprocessanditspowersp ectraldensityisasfollows Si( f )= 8><>: Pint;fcB 2j f j fc+B 20 ;otherwise ; (28) where B and fcarethebandwidthandthecenterfrequencyoftheinterferer ,respectively, and Pintisthepowerspectraldensity. Sincethenarrowbandsignalhasabandwidthmuchsmallertha nthecoherencebandwidthofthechannel,thetimedomainsamplesoftheNBIarehi ghlycorrelatedwitheach other.Therefore,fortheinvestigationofthenarrowbandi nterferers,thecorrelationfunctionsareofprimaryinterest,ratherthanthetime-orfrequ ency-domainrepresentations. Thecorrelationfunctionscorrespondingtothesingletone andbandlimitedcasescanbe writtenas Ri( )= Pij r j2cos(2 fc ) ; (29) Ri( )=2 PintB cos(2 fc ) sinc ( B ) ; (30) respectively.Theresultingcorrelationmatricesforthe k thand l thinterferencesamples are[64] [ Ri]k;l=4 NsPij r j2j Wr( fc) j2h sin( fc ) i2cos 2 fc( k l) (31) forthesingletoneinterferer,and [ Ri]k;l=2 NsPintB j Wr( fc) j2 h 2cos 2 fc( k l) sinc B ( k l) cos 2 fc( k l ) sinc B ( k l ) cos 2 fc( k l+ ) sinc B ( k l+ ) i (32) 39

PAGE 51

forthecaseofbandlimitedinterference,where j Wr( fc) j2isthepowerspectraldensityof thereceivedsignalatthefrequency fc. TheotherstrongcandidateforUWBcommunicationsbesideth eimpulseradioisthe multi-carrierapproach,whichcanbeimplementedusingOFD M.OFDMhasbecomeavery populartechnologyduetoitsspecialfeaturessuchasrobus tnessagainstmultipathinterference,abilitytoallowfrequencydiversitywiththeuseo fecientforwarderrorcorrection (FEC)coding,capabilityofcapturingthemultipathenergy ,andabilitytoprovidehigh bandwidtheciencythroughtheuseofsub-bandadaptivemod ulationandcodingtechniques.OFDMcanovercomemanyproblemsthatarisewithhigh bitratecommunications, themostsignicantofwhichisthetimedispersion.InOFDM, thedatabearingsymbol streamissplitintoseverallowerratestreams,andthesest reamsaretransmittedondifferentcarriers.Sincethisincreasesthesymbolperiodbyt henumberofnon-overlapping carriers,multipathechoesaectonlyasmallportionofthe neighboringsymbols.Theremaininginter-symbolinterference(ISI)canberemovedbyc yclicallyextendingtheOFDM symbol.Intermsofadaptingthetransmissionparameters,O FDMoersmanypossibilities. Adaptingthetransmitpower,cyclicprexsize,modulation andcoding,andthenumberof sub-carriersaresomeofthesetransmissionparameters.In additiontoadaptationovereach packet(asinthecaseofsinglecarriersystems),OFDMalsoo ersadaptationofparameters foreachcarrieroroverasmallgroupofcarriers.Inotherwo rds,adaptationcanbedone independentlyovernarrowerbandsratherthantheentiretr ansmissionband. AstrongmotivationforemployingOFDMinUWBapplicationsi sitsresistanceto narrowbandinterference,anditsabilitytoturnthetransm ission on and o onseparate carriersdependingonthelevelofinterference.TheNBImod elsthatcanbeconsidered forOFDMincludeoneormoretoneinterferers,aswellasazer o-meanGaussianrandom processthatoccupiescertaincarriersalongwithwhitenoi seas Sn( k )= 8><>:N i + N w 2; if k1
PAGE 52

where k isthecarrierindex, K isthetotalnumberofcarriers,andN i 2andN w 2arethe spectraldensitiesofthenarrowbandinterfererandwhiten oise,respectively. 3.3AvoidingNBINBIcanbeavoidedatthereceiverbyproperlydesigningthet ransmittedUWBwaveform. IfthestatisticsregardingtheNBIareknown,thetransmitt ercanadjustthetransmission parametersappropriately.NBIavoidancecanbeachievedin variousways,anditdepends onthetypeofaccesstechnology.3.3.1Multi-carrierApproachMulti-carrierapproachcanbeonewayofavoidingNBI.OFDM, whichwasmentionedin theprevioussection,isawellknownexampleformulti-carr iertechniques.InOFDMbased UWB,NBIcanbeavoidedeasilybyanadaptiveOFDMsystemdesi gn.AsthesimpleinterferencescenarioillustratedinFig.20shows,NBIwillcorr uptonlysomecarriersinOFDM spectrum.Therefore,onlytheinformationthatistransmit tedoverthesefrequencieswill beaectedfromtheinterference.Iftheinterferedcarrier scanbeidentied,transmission overthesecarrierscanbeavoided.Inaddition,bysucient FECandfrequencyinterleaving,jammingresistanceagainstNBIcanbeobtainedeas ily.Avoidingoradaptingthe transmissionoverthestronglyinterferedcarrierscanpro videmorespectrumandpowerefciency,astheyincreasetheimmunityagainstNBI,andhenc erelaxtheFECcodingpower requirement. AttheOFDMreceiver,thesignalisreceivedalongwithnoise andinterference.After synchronizationandremovalofthecyclicprex,FFTisappl iedtoconvertthetime-domain receivedsamplestothefrequency-domainsignal.Therecei vedsignalatthe k thsub-carrier ofthe n thOFDMsymbolcanthenbewrittenas Yn;k= Sn;kHn;k+ In;k+ Wn;k| {z }NBI + AWGN; (34) 41

PAGE 53

Figure20.AsimpleNBIscenarioformulti-carriermodulati onsystems. where Sn;kisthetransmittedsymbolwhichisobtainedfromaniteset( e.g. QPSKor QAM), Hn;kisthevalueofthechannelfrequencyresponse, In;kistheNBI,and Wn;kdenotestheuncorrelatedGaussiannoisesamples.Theimpai rmentsduetoimperfectsynchronization,transceivernon-linearities,etc.canbefo ldedintothenoiseterm Wn;k. InOFDM,inordertoidentifytheinterferedcarriers,thetr ansmitterrequiresafeedback fromthereceiver.Thereceivershouldhavetheabilitytoid entifytheseinterferedcarriers. Oncethereceiverestimatesthesecarriers,therelevantin formationwillbesentbacktothe transmitter.Thetransmitterwillthenadjustthetransmis sionaccordingly.Notethatin suchascenario,theinterferencestatisticsneedtobecons tantforacertainperiodoftime. Iftheinterferencestatisticschangeveryfast,bythetime thetransmitterreceivesfeedback, andadjuststhetransmissionparameters,thereceivermigh tobservedierentinterference characteristics. Thefeedbackinformationcanbevarious,includingtheinte rferedcarrierindex,insome casestheamountofinterferenceonthesecarriers,thecent erfrequencyofNBI,thebandwidthofNBI,etc.Theidenticationoftheinterferedcarri erscanalsobevarious.One 42

PAGE 54

Figure21.Someproposedmulti-bandapproachesforWPAN:(a )TheXtremeSpectrumMotorolaproposalofadual-bandapproach.(b)Multi-bandO FDM. simpletechniqueistolookattheaveragesignalpowerineac hcarrier,andcompareit withathreshold.Iftheaveragereceivedsignalpowerofasu bcarrierisgreaterthanthe threshold,thatchannelcanberegardedasseverelyinterfe redbyNBI.Insteadofmaking aharddecisiononwhetheracarrierisinterferedornot,sof testimationofNBIpowercan alsobedone[65].3.3.2Multi-bandSchemesSimilartothemulti-carrierapproach,multi-bandschemes arealsoconsideredforavoiding NBI.RatherthanemployingaUWBradiothatusestheentire7. 5GHzbandtotransmit information,byexploitingtherexibilityoftheFCCdenit ionoftheminimumbandwidth of500MHz,thespectrumcanbedividedintosmallersub-band s.Thecombinationofthese sub-bandscanbeusedfreelyforoptimizingthesystemperfo rmance.Bypartitioningthe spectrumintosmallerchunks(whicharestilllargerthan50 0MHz),abettercoexistence withothercurrentandfuturewirelesstechnologiescanbea chieved.Thisapproachwillalso 43

PAGE 55

enableworldwideinter-operabilityoftheUWBdevices,ast hespectralallocationforUWB couldpossiblybedierentinvariouspartsoftheworld.Inm ultibandsystems,information oneachofthesub-bandscanbetransmittedusingeithersing le-carrier(pulse-based)or multi-carrier(OFDM)techniques.Fig.21showssomerepres entativemultibandschemes. Thepulse-basedapproach(asshowninFig.21-a)usesdual-b andwithbandwidthsineach bandexceeding1GHz[66].Thelowerbandoccupiesthespectr umfrom3.1GHzto4.85 GHz,andtheupperbandoccupiesthespectrumfrom6.2GHzto9 .7GHz.Thespectrum inbetweenupperandlowerbandsisnotusedforUWBtransmiss ion,sincepotentialinterferencesourceslikeIEEE802.11aoperateinthisunlice nsedband.TheOFDM-based multi-bandapproach(showninFig.21-b)uses528MHzchanne lsineachband,wherethe threelowerbandchannelsareforinitialdeploymentsandma ndatory,andtheupperbands areoptionalandforfutureuse[67].Astheradiofrequencyt echnologyimproves,theupper bandsareexpectedtobeincludedintothesystemgradually.3.3.3PulseShapingAnothertechniqueforavoidingnarrowbandinterferenceis pulseshaping.Ascanbeseen in(31)and(32),theeectofinterferenceisdirectlyrelat edtothespectralcharacteristics ofthereceivertemplatepulsewaveform.Thatmeans,ifthet ransmissionatthefrequencies whereNBIispresentcanbeavoided,theinruenceofinterfer enceonthereceivedsignalcan bemitigatedsignicantly.Therefore,designingthetrans mittedpulseshapeproperly,such thatthetransmissionatsomespecicfrequenciesisomitte d,NBIavoidancecanberealized. Anexcellentexamplefortheimplementationofthisapproac histheGaussiandoublet[68]. AGaussiandoublet,representingonebit,consistsofapair ofnarrowGaussianpulseswith oppositepolarities.Consideringthetimedelay Tdbetweenthepulses,thedoubletcanbe representedas sd( t )= 1 p 2 s ( t ) s ( t Td) : (35) 44

PAGE 56

Thecorrespondingspectralamplitudeofthedoubletisthen j Sd( f ) j2=2 j S ( f ) j2sin2( fTd) ; (36) where j S ( f ) j2isthepowerspectrumofasinglepulse.Noticethatduetothe sinusoidalterm in(36),thepowerspectrumwillhavenullsat f =n T d,where n canbeanyinteger(shown inFig.22).ThebasicideaforavoidingNBIisadjustingthel ocationofthesenullsinsuch awaythattheyoverlapwiththepeakscreatedbynarrowbandi nterferers.Bymodifying thetimedelay Td,anullcanbeobtainedatthespecicfrequencywhereNBIexi sts,and thiswaythestrongeectoftheinterferercanbeavoided.If Tdisadjustedto2ns,for example,theinterferenceslocatedattheintegermultiple sof500MHzcanbesuppressed. 3 4 5 6 7 8 9 10 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 10 Normalized Spectrum Magnitude (dB)Frequency (GHz) Spectrum of the single Gaussian pulseSpectrum of the Gaussian doublet with Td=0.5 ns Spectrum of the Gaussian doublet with Td=1 ns Figure22.NormalizedspectraforthesingleGaussianpulse andtwodierentGaussian doublets. 45

PAGE 57

4.5 5 5.5 -0.5 0 0.5 Time (ns)Amplitude6 th Order Gaussian Pulse 0 5 10 15 10 -15 10 -10 10 -5 10 0 Spectrum of 6 th Order Gaussian Pulse Frequency (GHz)Magnitude (dB) 0 5 10 15 10 -15 10 -10 10 -5 10 0 Spectrum of Notch Filtered Pulse Frequency (GHz)Magnitude (dB) 4.5 5 5.5 -0.5 0 0.5 Notch Filtered Pulse Time (ns)Amplitude Figure23.Theeectofnotchlteringonthetransmittedpul seshape. ThepurposeofavoidingNBIthroughabstainingtransmissio natfrequenciesofinterferencecanalsobecarriedoutbymakinguseofnotchltersinth etransmitter.Toaccomplish this,theparametersoftheltershavetobeadjustedsuchth atthenotchestheycreateoverlapwiththefrequenciesofstrongNBI.Whennotchltersare employedinthetransmitter, thetransmittedpulseisshapedinsuchaway(seeFig.23)tha tthecorrelationofNBIwith thepulsetemplateinthereceiverisminimized. PulseshapingtechniquesarenotlimitedtotheGaussiandou bletandnotchltering. AnotherfeasiblemethodistheadjustmentofthePPMmodulat ionparameter .Revisiting thecorrelationmatrixforasingletoneinterferergivenin (31),itisseenthat[ Ri]k;l=0 for = n=fc,where n =1 ; 2 ;:::;M M beingthenumberofpossiblepulsepositions. Therefore,aneectiveinterferenceavoidancecanbeattai nedbysetting to n=fc.Similarly, 46

PAGE 58

consideringthecorrelationmatrixcorrespondingtotheba ndlimitedinterference(32),it isseenthatcos 2 fc( k l ) =cos 2 fc( k l) ; when = n=fc: Also,inthe lightoftheknowledgethatthebandwidthoftheinterferenc e( B )ismuchsmallerthanits centerfrequency( fc),theassumption sinc B ( k l ) sinc ( B ( k l))canbemade for = n=fc.Thesetwofactsleadtotheconclusionthat[ Ri]k;lin(32)becomeszerofor thebandlimitedinterferencecase,too,when issetto n=fc. AlthoughtheadjustmentofthePPMmodulationparameter isastraightforwardway ofavoidingNBI,ithasanimportantdrawback.Thecorrelati onoutputisalsodependent on ,andforacertainvalueofitamaximumsignalcorrelationca nbeobtained.However,thisvalueof doesnotnecessarilyhavetobeequalto1 =fc.FortheAWGNcase (withoutconsideringtheNBI),thebit-error-ratefunctio nfromwhichtheoptimum can bedeterminedis[69] Q r NsAEp N0Ropt! ; (37) where Ropt= R (0) R ( opt), Nsisthenumberofpulsespersymbol, A isthepulseamplitude, Episthepulseenergy, N0= 2isthedoublesidedpowerspectraldensityofAWGN,and R ( t )istheautocorrelationfunctionofthereceivedpulse.The refore,thereisanobvious trade-obetweenmaximizing RoptandavoidingNBI,whendeterminingthe parameter. DependingonthelevelofNBIandAWGN,thisparametercanbea djustedtoprovidean optimalperformance.3.3.4OtherNBIAvoidanceMethodsForthetime-hoppingUWBsystems,itispossibletoavoidNBI byplacingnotchesinthe spectrumbyadjustingthetime-hoppingcode[70].In[71],a pulseamplitudemodulated (PAM)UWBsignalisconsidered.Eachsymbolhasadurationof Tsandiscomposedof Nspulses,givingriseto Nsframes,whichlastfor Tf= Ts=Nsandaredividedintochipswith adurationof Tc.Thepseudo-randomTHcodedeterminesthepositionofthepu lseinside theframebyselectingthechipwheretoplacethepulse.Insh ort,aPAMUWBsignalover 47

PAGE 59

asymboldurationcanbewrittenas u ( t )= AN s 1Xn =0p ( t cnTc nTf Ts) ; (38) where A f 1 ; 1 g denotestheamplitudeofthepulse,and cnistheTHcode.In[70],the spectrumshapeforthemulti-symbolcaseisgivenby Pu( f )= j W ( f ) j2 N b 1Xk =0j Tk( f ) j2; (39) where W ( f )istheFourierTransformofthetransmittedpulse, Nbisthetotalnumberof dierentTHcodesused, k isthesymbolindex,and Tk( f )=N s 1Xn =0exp ( j 2 f ( cn;kTc+ nTf+ kTs)) : (40) From(40),itisseenthatchangingthetime-hoppingcodecau sesthespectrumofthe transmittedsignaltovary.Thismeansthatbyemployingvar iousmethods,theTHcode canbeadjustedinsuchawaythatspectralnotchesarecreate datfrequenciesofstrong NBI,allowingthesystemtoavoidinterference. Inadditiontothemethodsmentioned,physicalsolutionsca nalsobeconsideredfor avoidingNBI.In[72],anNBIavoidancetechniquedepending onantennadesignisproposed. Themainideaiscreatingfrequencynotchesbyintentionall yaddinganarrowbandresonant structuretotheantenna,andthus,makingitinsensitiveto someparticularfrequencies. Thistechniqueismoreeconomicalthantheexplicitnotchl teringmethodsinceitdoes notrequireadditionalnotchlters.In[72],afrequencyno tchedUWBantennasuitableto avoidNBIisrealizedandexplainedindetail.Thisspecialpurposeantennaisobtainedby employingplanarellipticaldipoleantennasandincorpora tingahalfwaveresonantstructure, whichisobtainedbyimplementingtriangularandelliptica lnotches.Itisnecessarytonote thattheperformanceoftheantennaisreducedwithincreasi ngnumberofnotches.This 48

PAGE 60

factleadstotheideathatthefrequencynotchedantennamay notbesuccessfulenoughin avoidingnumeroussimultaneouslyexistingnarrowbandint erferers. 3.4CancelingNBIAlthoughmostoftheavoidancemethodsmentionedseemtohav eahighfeasibility,they maynotbeimplementedunderallcircumstances.Themainlim itationonthesemethods istheirdependencyontheexactknowledgeaboutnarrowband interferers.Withouthaving theaccurateinformationaboutthecenterfrequencyofthei nterference,suppressingNBIis notpossiblebymeansofanyoftheavoidancetechniquesexpl ained.Evenifthecomplete knowledgeabouttheNBIisavailable,ifthereisanabundant numberofinterferers,methods likeemployingnotchltersorchangingtheparametersofth etransmittedpulsemaylose theirpracticality.IfitisnotpossibletoavoidNBIatthet ransmissionstagebecauseofany reason,oneshouldmakeeortatthereceiversideforextrac tingandeliminatingitfrom thereceivedsignal. Throughouttheprevioussection,methodsofavoidingNBIha vebeendiscussedand limitationsontheirrealizationhavebeenmentioned.Inpr actice,UWBsystemsthatemploy onlyavoidancetechniquesarenottotallysuccessfulineli minatingNBI.Inthissection,an overviewofdierenttypesofNBIcancellationmethodswill beprovided. 3.4.1MMSECombiningOneofthepopularreceiversconsideredforUWBistheRakere ceiver.Rakereceiversare designedtocollecttheenergyofstrongmultipathcomponen ts,andwiththispurposethey employ ngers .IneachRake-nger,thereisacorrelationreceiversynchr onizedwithoneof themultipathcomponents.Thecorrelationreceiverisfoll owedbyalinearcombinerwhose weightisdetermineddependingonthecombinationalgorith mused.Theoutputofthe receiverforthe ithpulsecanbedenotedas[59] yi=M 1Xk =0dick k+ cknk; (41) 49

PAGE 61

where M isthenumberofRake-ngers, diisthedatabittransmittedonthe ithpulse, ck, kand nkaretheweightusedbythecombiner,thechannelgain,andthe noiseforthe kthmultipathcomponent,respectively,and = Z1 t = 1prx( t ) v ( t ) dt; (42) where prx( t )denotesthereceivedwaveform,and v ( t )isthecorrelatingfunction. InthetraditionalRakereceiver,whichemploysmaximalrat iocombining(MRC),the weightofthecombineristheconjugateofthegainofthepart icularmultipathcomponent ( c = ).SuchaselectionmaximizestheSNRintheabsenceofNBI.Ho wever,whenNBI exists,sinceinterferencesamplesarecorrelated,MRCisn olongertheoptimummethod. Minimummeansquareerror(MMSE)combining,whichisanalte rnativeapproach,depends onvaryingtheseweightsinsuchawaythatthemeansquareerr orbetweentherequired andactualoutputsisminimized.Intheexistenceofinterfe rence,theSNRismaximized whenMMSEweightvectorisused[73]: c = Rn 1 ; (43) where c =[ c1c2:::cM]T, isthescalingconstant, Rn 1istheinverseofthecorrelation matrixofnoiseplusinterference,and =[ 12:::M]Tisthechannelgainvector. TheNBIcancellationmethodsotherthanMMSEcombiningcanb egroupedinthree categoriesasfrequencydomain,time-frequencydomain,an dtimedomainapproaches. 3.4.2FrequencyDomainTechniquesCancellationtechniquesinthefrequencydomaincanbeexem pliedbynotchlteringin thereceiverside.Havinganestimationaboutthefrequenci esofpowerfulnarrowband interferers,notchlterscanbeusedtosuppressNBI.Thepl easantfactaboutthismethod isthatitcanbeutilizedinalmostallkindofreceivers,sot hattheUWBsystemisnot forcedtoemployacorrelationbasedreceiver.Themainweak nessoffrequencydomain 50

PAGE 62

methods,ontheotherhand,isthattheyareusefulonlywhent hereceivedsignal,whichisa superpositionoftheUWBsignalandNBIfromvarioussources ,exhibitsstationarybehavior. Ifthereceivedsignalhasatime-varyingnature,methodsth atanalyzethefrequencycontent takingthetemporalchangesintoaccountarerequired.Thes emethodsarecalledthetimefrequencyapproaches.3.4.3Time-FrequencyDomainTechniquesThemostcommonlyemployedtime-frequencydomainmethodfo rinterferencesuppressionis thewavelettransform.Similartothewell-knownFouriertr ansform,thewavelettransform alsoemploysbasisfunctions,andexpressesanytimedomain signalasacombinationof them.However,thesebasisfunctions,whicharecalledwave lets,aredierentfromthe complexexponentialsusedbytheFouriertransforminthese nsethattheyarenottime unlimited.Hence,thewavelettransformisabletorepresen tthetimelocalcharacteristics ofsignals,andisnotlimitedtostationarysignalslikethe Fouriertransform.Awaveletis denedas ab( t )= 1 j p a j ( t b a ) ; (44) where a and b arethescalingandshiftingparameters,respectively.Ift heseparametersare setas a =1and b =0,themotherwaveletisobtained.Bydilatingandshifting themother wavelet,afamilyofdaughterwaveletsareformed.Theconti nuouswavelettransformcan beexpressedas W ( a;b )= Z+ 1 1f ( t ) ab( t ) dt: (45) Theversionofthewavelettransformthatisappropriatefor computerimplementation isthediscretewavelettransform(DWT),whichisdenedas[ 45] dm;n= 1 p am0Z f ( t ) ( t am0 nb0) dt; (46) where m and n areintegers. 51

PAGE 63

ComputersrealizetheDWTnotbyusingwavelets,butemployi nglters.Aneective algorithmforperformingDWTbasedonusinglterswaspropo sedbyMallat[74].The Mallatalgorithmresultsinadetailedanalysis,inwhichth elowestfrequencycomponent isexpressedwiththesmallestnumberofsamples,whereasth elargestnumberofsamples expressthehighestfrequencycomponent. Onepossiblewayofsuppressingthenarrowbandinterferenc eusingthewavelettransform istohavethetransmitterpartoftheUWBsystemestimatethe electromagneticspectrum, andsetaproperthresholdforinterferencedetection[75]. Theinterferencelevelateach frequencycomponentisthendeterminedwiththewavelettra nsform,andcomparedto thisthresholdinordertodistinguishbetweentheinterfer edandnotinterferedfrequency components.Accordingtotheresultsofthiscomparisonste p,thetransmitterdoesnot transmitatfrequencieswherestrongNBIexists.Obviously ,thismethodisquitesimilarto themulti-carrierapproachinNBIavoidancetechniques. Methodsemployingthewavelettransforminthereceiversid eofthesystemalsoexist [76,77].Inthesemethods,wavelettransformisappliedtot hereceivedsignal,andfrequency componentswithaconsiderablyhighenergyareconsideredt obeaectedbynarrowband interference.Thesecomponentsarethensuppressedbyusin gconventionalmethodslike notchltering. Althoughthediscretewavelettransformisaveryusefultoo lforeliminatingNBI,the inabilityofcurrentADCstosampletheUWBsignalsattheNyq uistratesetsapractical limittothefeasibilityofthismethod.Therefore,theeec tivenessofDWTattheframe-rate andsymbol-ratesamplinghastobeinvestigatedthoroughly tobeabletodecideaboutthe usefulnessofthisapproachwiththeexistingtechnology.3.4.4TimeDomainTechniquesThethirdgroupofNBIcancellationmethodsisthetimedomai napproaches,whichcan alsobecalledpredictivemethods.Predictivemethodsareb asedontheassumptionthat thepredictabilityofnarrowbandsignalsismuchhighertha nthepredictabilityofwideband signals,becausewidebandsignalshaveanearlyratspectru m[29].Hence,inaUWB 52

PAGE 64

system,apredictionofthereceivedsignalisexpectedtopr imarilyrerectthenarrowband interferenceratherthantheUWBsignal.Thisfactleadstot heconsequencethatNBIcan becancelledbysubtractingthepredictedsignalfromthere ceivedsignal. Predictivemethodscanbeclassiedaslinearandnon-linea rtechniques.Lineartechniquesemploytransversalltersinordertogetanestimate ofthereceivedsignaldepending ontheprevioussamplesandmodelassumptions[36].Ifone-s idedtapsareused,thelter employedisalinearpredictionlter,whereasitisalinear interpolationlterifthetaps aredouble-sided.Itisworthtonotethatinterpolationlt ersprovedtobemoreeective incancellingNBI. CommonexamplesforlinearpredictivemethodsaretheKalma n-Bucyprediction,which isbasedontheKalman-Bucylterwithinniteimpulserespo nse(IIR),andLeast-MeanSquares(LMS)algorithmbasedonaniteimpulseresponse(F IR)structure. Non-linearmethodsarefoundtoprovideabettersolutionth anlinearonesfordirectsequence(DS)systemsbecausetheyareabletomakeuseofthe highlynon-GaussianstructureoftheDSsignals[29].However,forUWBsystems,thisis notthecasebecausesucha non-GaussianitydoesnotexistinUWBsignals. Adaptivepredictionltersareconsideredasapowerfultoo lagainstNBI.Whenan interfererisdetectedinthesystem,theadaptationalgori thmcreatesanotchtosuppressthe interferencecausedbythissource.However,iftheinterfe rervanishessuddenly,sincethere isnomechanismtorespondimmediatelytoremovethenotchcr eated,thereceivercontinues tosuppresstheportionofthewantedsignalaroundthenotch .Ifnarrowbandinterferers enterandexitthesysteminarandommanner,thisshortcomin greducestheperformance oftheadaptivesystemdramatically.Amoreusefulalgorith misproposedin[36],where ahidden-Markovmodel(HMM)isemployedtokeeptrackofthei nterferersenteringand exitingthesystem.Inthisalgorithm,thefrequencylocati onswhereaninterfererispresent aredetectedbyanHMMlter,andasuppressionlterisputth ere.Whenthesystem detectsthattheinterfererhasvanished,thelterisremov edautomatically. 53

PAGE 65

3.5ConclusionInthischapter,anoverviewofnarrowbandinterferenceinU WBsystemsisgiven.The signicanceoftheNBIproblemforUWBsystemshasbeendiscu ssed,dierentmodels forNBIhavebeenanalyzed,andtheeectsofNBIonUWBcommun icationshavebeen addressed.MethodsofdealingwithNBIhavebeenexaminedun dertwoseparatecategories asNBIavoidanceandNBIcancellationalgorithms.NBIavoid ancemethodsincluding multi-carrierapproachesandmulti-bandschemes,aswella salternativesolutionsbasedon pulseshaping,time-hoppingcodeadjustment,andantennad esignhavebeeninvestigated. Amongthecancellationtechniques,detailsofMMSEcombini ngalgorithmarepresented. Frequencydomaintechniquessuchasnotchltering,time-f requencymethodslikewavelet transformandtimedomainapproaches,particularlylinear techniques,havebeendiscussed inseparatesections. Asofnow,noneoftheavoidanceorcancellationmethodshasp rovedtobetheoptimum solutiontotheNBIproblem.Itseemsthatthemostinexpensi veandsuccessfulwayof suppressingNBIcanbeachievedbyemployinganadaptivemet hodcombiningtheavoidance andcancellationapproaches.TheUWBcommunicationscanbe initiallystartedbyapplying theproperavoidancemethodsinthetransmitterside,theni nthelightofthefeedback providedbythereceiver,theeectivenessofinterference excisioncanbedetermined,andif itisfoundthattheinterfererscannotbesuppressedsatisf actorily,NBIcancellationmethods canberuninthereceiversideofthesystem.Consideringtha tthecomputationalburden relatedtothecancellationmethodsisgenerallymuchhighe rthanavoidancemethods,such anadaptiveapproachcanbeveryusefulintermsofwiseusage ofresources. 54

PAGE 66

CHAPTER4 COGNITIVERADIO Inthischapter,thecognitiveradioconceptisintroduced, andtheobjectivesaimedwith cognitiveradioareaddressed.Spectrumopportunityisde ned,anditisinvestigatedhowto sensethespectralopportunities.Dierentspectrumshapi ngapproachesintheliteratureare provided,andanalternativemethodbasedontheusageofrai sedcosineltersisproposed. 4.1IntroductionTraditionalcommunicationsystemdesignisbasedonalloca tingxedamountsofresources totheuser.Adaptivedesignmethodologies,ontheotherhan d,typicallyidentifytherequirementsoftheuser,andthenallocatejustenoughresour ces,thusenablingmoreecient utilizationofsystemresourcesandconsequentlyincreasi ngcapacity.Pushingtheadaptive systemdesignfurtherbyintroducingadvancedattributess uchasmulti-dimensionalawareness,sensing,aswellaslearningfromitsexperiencestore ason,plan,anddecidefuture actionstomeetuserneedsleadstothe cognitiveradio concept.Ignitedbytheearlier workofMitola[1],cognitiveradioisanovelconceptforfut urewirelesscommunications, andithasbeengainingsignicantinterestamongtheacadem ia,industry,andregulatory bodies[78]. Eventhoughthereisnoconsensusontheformaldenitionofc ognitiveradio,theconcept hasevolvedrecentlytoincludevariousmeaningsinseveral contexts.Oneofitsmainaspects isrelatedtoautonomouslyexploitinglocallyunusedspect rumtoprovidenewpathstothe spectrumaccess.Otheraspectsinclude inter-operabilityacrossseveralnetworks, roamingacrossborders,whilebeingabletostayincomplian cewithlocalregulations, 55

PAGE 67

adaptingthesystem,transmission,andreceptionparamete rswithoutuserintervention, havingtheabilitytounderstandandfollowactionsandchoi cesoftheusers, andlearningovertimetobecomemoreresponsiveandtoantic ipatetheuserneeds. Cognitiveradioconceptproposestofurnishtheradiosyste mswiththeabilitiestomeasure andbeawareofparametersrelatedtotheradiochannelchara cteristics,availabilityof spectrumandpower,interferenceandnoisetemperature,av ailablenetworks,nodes,and infrastructures,aswellaslocalpoliciesandotheroperat ingrestrictions.Theprimary advantagetargetedwiththesefeaturesistoenablethecogn itivesystemstoutilizethe availablespectruminthemostecientway. Theorganizationofthechapterisasfollows.InSection4.2 ,opportunityisdened, andopportunisticspectrumusageisinvestigated.InSecti on5.2,acognitiveUWB-OFDM systemisproposed,andthedetailsregardingitsimplement ationareexplained.Finally,in Section4.5,conclusionsandpossiblefutureresearchtopi csaregiven. 4.2OpportunisticSpectrumUsageConventionally,frequencyspectrumallocationforradios ystemshasbeendoneintheform oflicensingdierentfrequencybandstoseparateapplicat ions.Inthisprocedure,alicensed userpossessestheabsoluteownershipofthespectrumitisa llocated,andthespectrumcan notbeoeredtotheusageofotherpotentialusers,evenifth elicenseduseristemporarily notmakinguseofit.Therefore,thestaticfrequencyalloca tionleadstoahighlypoor utilizationofthespectrum.IthasbeenshownbytheSpectru mPolicyTaskForce(SPTF) oftheFCCthatmanylicensedfrequencybandsarenotbeingus edforlongdurations[79]. Also,arecentexperimentconductedinNewYork,UnitedStat esinSeptember2004revealed thattheaveragedutycycleofthespectrumbetween30MHzand 3GHzwasonly13%[80]. Cognitiveradioinitiatesarevolutionregardingthespect rumallocationconsiderations byputtingforwardanewconceptcalledopportunisticspect rumusage,whichinvolvesthe 56

PAGE 68

softusageofthecurrentlicensedandunlicensedavailable spectrum.Thisconceptproposes thatlicensedbandscanbeutilizedbysecondaryusersattim eswhentheyarenotbeing usedbytheirowners,leadingtothemostecientexploitati onoftheentirespectrum.In thisopportunisticwayofspectrumusage,ithastobeguaran teedbytheunlicensedsystems thattheiroperationdoesnotaecttheprimaryusers.Altho ughthisapproachissimilarto theUWBfromthepointthatbothareunlicensed,therearetwo maindierencesbetween theopportunisticusageandUWB.First,UWBsystemsareforc edtooccupyabandofat least500MHzwidth,whichisnotthecasefortheopportunist icusage;andsecond,for UWBcommunicationsthereisastricttransmitpowerlimitat ion,whereasinopportunistic usagethetransmittedpowercanbecomparabletothepowerof licensedsystems. Asolidunderstandingoftheopportunisticspectrumusagec onceptrequiresthat opportunity isdenedclearly.Cognitiveradiosperiodicallyscanthes pectrumanddetectthe spectrathataretemporarilynotbeingusedbytheirlicense dusers,whichcanbecalled whitebands .Inmanyworks,whitebandsaredirectlytakenasthespectru mopportunities. However,therearespectral,temporalandspatialrequirem entsthatawhitebandhasto satisfyinordertobeusefulandtobeconsideredanopportun ity[81,82].Theserequirements canbelistedasfollows. Opportunityisnotaninstantaneouswhitespaceinspectrum .Itisnecessaryto monitorawhitespacecontinuouslyoveratimeframe(intheo rderofseconds)and ensurethatitdoesnotdisplayanerraticbehavior,i.e.for areasonablylongtime thenoisetemperatureinthatbandresidesbelowacertainth reshold,andtheband remainsasawhitespace. Itmaynotbereliabletoconsiderawhitebandanopportunity ifitisdetectedby onlyonesinglecognitiveradiodevice.Thereasonsinclude thatthedevicehasa limitedsensingrange,aswellasthatitmaybeexperiencing shadowing.Optimally, thespectrumhastobesensedbyanumberofcognitivenodesov eraregionthat goeswellbeyondtherangeofasinglecognitivedevice.Aban dcanbeconsidereda candidateforbeingaspectrumopportunityonlyifitisdete ctedaswhitebymany 57

PAGE 69

cognitivenodesatdierentlocationsthatexchangethespe ctrumsensinginformation witheachother. Ifafrequencybandtobeutilizedistoonarrow,itmaybehard forthecognitive radiotogenerateatemporallylimitedpulseshapethattsi ntothatband.Therefore,awhitebandhastobewiderthanacertainbandwidthfor beingtargetedfor opportunisticspectrumusage. Apparently,theserulesarerequiredtominimizetheriskof causinginterferencetolicensed systems.Besidethis,suchanopportunitydenitionisopti mumfromthepointofminimizingthecomputationalburden,aswell,becauseitsaves acognitiveradiofromdoing computationsandchangingitsparameterswithoutensuring thedependabilityofawhite band.4.3SensingtheSpectrumOpportunitiesAsithasbeenmadeclearintheprevioussection,anunavoida blerequirementforusing thespectruminanopportunisticwayisthatthecognitivera dioperiodicallyscansthe frequencyspectrum.Animportantconcernaboutspectrumse nsingisabouthowtosetthe boundariesofthespectrumtargetedbycognitiveradio.Con sideringtheextremeabundance offrequencyspectrathatmaycontainwhitebands,itisobvi ousthateveninthecaseof cognitiveradiocommunications,whichisconceptualizedt oremovethebordersaroundopen spectrumaccess,thetargetedspectrumshouldbelimited.A mongthenumerousreasons, theprimaryonesarethatthisway Itcanbepossibletosamplethereceivedsignal(afterthedo wn-conversion)atorabove theNyquistrateevenwiththecurrenttechnology,whichena blesdigitalprocessing ofthesignal. Thecomputationalburdenassociatedwithspectrumsensing canberestrictedtoa reasonablelevel,leadingtolimitedhardwarecomplexity. 58

PAGE 70

Theanalogfront-endrequiredforaverywidespectrumscan( includingawideband antenna,widebandampliersandmixers),whichhavearathe rhighcost,canbe avoided. Itcanbepreventedthatasingletypeofcognitiveradiooccu piesthemajorityofthe bandsthatareopentoopportunisticusage. Thespectrumallocationfordierenttypesofcognitiverad ios(likemobilephones, WLANmodems,andpalmdevices)canbedonebyregulatoryagen ciessuchastheFCCdependingontheintendedrangeandthethroughputrequiremen tofthespecicapplication. Fromthispointofview,highdataratecognitiveradiosaimi ngatwiderangeapplications suchasTVbroadcastsystemsshouldbeassignedwidertarget spectraatlowerfrequency bands,whereasrelativelylowdate,shortrangecommunicat iondeviceslikecordlessphones maybeallocatednarrowerbandsathigherfrequencies. Intheliterature,thereisalimitednumberofmethodspropo sedregardingtheimplementationofspectralsensingforcognitiveradio[2,83,84 ].Atthesystemlevel,spectral sensingcanbeimplementedinanindividualordistributedm anner[85].Intheindividual sensing,thecognitiveUWBdevicesensesthespectrumbyits ownmeans,anddepends onthisknowledgewhenmakingdecisions.However,becauseo fthedenitionofopportunity,itisnotthepreferredmethodforsensing.Inthedistr ibutedsensing,whichcanbe non-centralizedorcentralized,multipledevicesscanthe spectrum,andsharethegathered informationwitheachother.Innon-centralizedspectrums ensing,itisconsideredtohave anallocatedcontrolchanneltotransmitthisinformation[ 86].Incentralizedsensing,on theotherhand,itiscontemplatedtohaveacentralcontroll erthatgathersthisinformation,decidesforspectrumavailability,andallocatesdis tinctbandstodierentcognitive users[85,87].4.4SpectrumShapingAchallengingrequirementofopportunisticusageisthatth ecognitivetransceiverhasto beabletodynamicallyadaptitstransmissionparametersto operateoverawiderange 59

PAGE 71

5200 5220 5240 5260 5280 5300 5320 5340 5360 5380 5400 -80 -60 -40 -20 0 Power (dBm) 5200 5220 5240 5260 5280 5300 5320 5340 5360 5380 5400 -80 -60 -40 -20 0 Power (dBm) 5200 5220 5240 5260 5280 5300 5320 5340 5360 5380 5400 -80 -60 -40 -20 0 Frequency (MHz)Power (dBm) open to opportunistic usage FCC Part 15 Limit licensed users (a) (b) (c) Figure24.(a)Asnap-shotofthespectrumintime.(b)Opport unisticspectrumutilization employingtimelimitedsinusoids.(c)Opportunisticspect rumusageemployingspecial pulses.ofspectrumwithdierentbandwidths,whichcanbecalled spectrumshaping capability. Spectrumshapingisaccomplishedbymodifyingthetransmit tedpowerlevelandthepulse shapeinsuchawaythatthespectrumofthepulsellsthedete ctedspectrumopportunities asecientlyaspossible.Variousmethodstoimplementpuls eshapingforcognitiveradio aregivenin[88]-[92]. ApossibleoptionforllingthewhitespacesistoemployOFD Mcarriersasinthecase ofUWB-OFDM.OFDMbasedimplementationofspectrumshaping canbefoundin[88] and[89].However,whenshapingthespectrumofthetransmit tedpulse,ithastobestrictly ensuredthattheleakagefromtheopportunitybandstotheli censedsystemsintheadjacent bandsremainsatanegligiblelevel(illustratedinFig.24) .Ifthetransmittedpulsesare timelimitedsinusoids(asinthecaseofOFDM)andnowindowi ngisused,theresulting 60

PAGE 72

sidelobesmaybeunacceptablyhigh(seeFig.25-a).Therefo re,itisveryimportantto employspecialpulsesthat havesharpfall-osandsuppressedsidelobesinthefrequen cydomain arelimitedbothintimeandbandwidth, haveapulsewidthandbandwidththatcanbecontrolledsimul taneously. ProlateSpheroidalWaveletFunctions(PSWF)satisfythese requirementstoalargeextent (showninFig.25-d).Spectrumshapingmethodsemployingth ePSWFareprovidedin[90] and[91]. Analternativemethodforshapingthespectrumofthetransm ittedpulseisanimpulse radiotechnique.Thistechniqueisbasedontheusageofrais edcosine(orrootraisedcosine) lters,whichcanbeexempliedasinFig.25-b(andc).Inthi smethod,rst,thecenter frequencies fc iandbandwidths Biofeachopportunity Oifor i =1 ; 2 ;:::;N ,aredetermined, where N isthetotalnumberofopportunities.Inthenextstep,makin guseofitsawareness property,thecognitiveradioselectstheraisedcosinelt ers ri( t )thatarethemostsuitable foreach Oi.Fillingahigherpercentageofawhitespacerequiresahigh erroll-olter, whichcorrespondstoalongersymbolintime,leadingtointe r-symbolinterferenceora lowerthroughput.Hence,thecognitiveradiodeterminesth eltertobeusedaccording totheamountofavailablebandwidthandthedataraterequir ed.Theselectedltersare multipliedwithdigitallygeneratedcosinesignalsyieldi ng i( t )= cos (2 fc it ) ri( t ) : (47) i( t )canbeexempliedasinFig.26-a,b,andc,whicharegenerat edusing ri( t )with roll-ocoecients0 : 9,0 : 3and0 : 5,respectively.Eachofthesepulsesisllingoneofthe opportunitiesinFig.26-e.Thenalpulseshape(demonstra tedinFig.26-d)isobtained bytakingthesumofalltheseseparatepulses p ( t )=NXi =1i( t ) ; (48) 61

PAGE 73

-2 -1 0 1 2 0 0.5 1 Time (ns)Amplitude -3 -2 -1 0 1 2 3 -40 -30 -20 -10 0 Frequency (GHz)Power (dB) -2 -1 0 1 2 0 0.5 1 Time (ns)Amplitude -3 -2 -1 0 1 2 3 -80 -60 -40 -20 0 Frequency (GHz)Power (dB) -2 -1 0 1 2 0 0.5 1 Time (ns)Amplitude -3 -2 -1 0 1 2 3 -80 -60 -40 -20 0 Frequency (GHz)Power (dB) -2 -1 0 1 2 -1 -0.5 0 0.5 1 Time (ns)Amplitude 2 3 4 5 6 7 8 -80 -60 -40 -20 0 Frequency (GHz)Power (dB) a = 0.3 a = 0.9 a = 0.3 a = 0.9 a = 0.3 a = 0.9 a = 0.3 a = 0.9 Figure25.Dierentpulseshapesandtheirspectra(a)Recta ngularwindow.(b)Raised cosinewindowswithroll-ofactors =0 : 3and =0 : 9.(c)Rootraisedcosinewindows withroll-ofactors =0 : 3and =0 : 9.(d)Ahighorderprolatespheroidalwavelet function.anditllstheopportunitiesasshowninFig.26-f. Thecurrenttransceiversincludeananalogfront-end,whic hismostlyxedforaspecic functiontooperateoverasmallrangeoffrequencies.Sucha nanalogfront-endisnotrexible andnotprogrammable.Thisgivesrisetoanewconceptcalled softwaredenedradio (SDR),wherethisxedanalogcircuitryneedstobereplaced withsoftwareprogrammable hardware[78].TheidealSDRconceptdigitizesthereceived signalassoonaspossibleso thatarexibleradiofunctionalitycanbeobtained.Ascanbe seen,thisisachallengewith thecurrentanalog-to-digital-converter(ADC)capabilit iesandwiththeprocessingpower 62

PAGE 74

0 10 20 30 40 50 -1 -0.5 0 0.5 1 (a) Time (ns)Amplitude 0 10 20 30 40 50 -1 -0.5 0 0.5 1 Time (ns)Amplitude(b) 0 10 20 30 40 50 -1 -0.5 0 0.5 1 Time (ns)Amplitude(c) 0 10 20 30 40 50 -1 0 1 2 Time (ns)Amplitude(d) 0 250 500 750 1000 0 5 10 15 20 25 30 (e) Frequency (MHz)Power 0 250 500 750 1000 0 5 10 15 20 25 30 (f) Frequency (MHz)Power Figure26.(a),(b),(c)Separatepulsesobtainedviaraised cosinelteringthattinto dierentopportunities.(d)Sumoftheseparatepulses.(e) Binaryclassicationoffrequencybandsas'occupied'or'opportunity'.(f)Spectrumo fthedesignedpulsellingthe opportunities.available.Therefore,currently,thenewgenerationwirel esssystemsareslowlyintegrating aversionofthisconcept.4.5ConclusionInthischapter,thecognitiveradioconceptisintroduced. Itisemphasizedtheconcept ofopportunity,whichisofcrucialimportanceforcognitiv eradiosystems,shouldhavea soliddenition,andadetaileddenition,thatinvestigat estheopportunityfromspectral, temporalandspatialperspectivesisprovided.Theopportu nitysensingandspectrumshapingfeaturesofcognitiveradiosarediscussedanddierent approachesareaddressed.An alternativemethodforshapingthespectrumofthetransmit tedpulseisalsoprovided. 63

PAGE 75

CHAPTER5 COGNITIVEUWB Inthischapter,withthepurposeofmaximizingtheeciency ofunlicensedsystems,itis consideredtocombineUWBwiththecognitiveradio.Twodie rentmethodsareproposed; therstoneiscognitiveUWB-OFDM,andthesecondoneisacog nitivesystemthatshares thespectrumsensinginformationbetweenitsnodesusingUW Bsignaling. 5.1IntroductionSystemswithaspectralallocationsimilartoUWBareoftenr eferredas underlay systems. UnderthecurrentFCCregulation,underlaysystemsareallo wedtohaveaverylimited transmitpower.Thisseverepowerlimitationonunderlaysy stemsrestrictstheirusageto onlyveryshortrangeapplications.Hence,allcurrentunde rlaywirelesscommunication studiesbothfromindustryandacademyareinthedirectiono fmakingUWBsystemswork inanunderlayscenario,andaimatwirelesspersonalareane tworks(WPAN),only. Themaincontributionofthischapteristhattwodierentme thodsofcombiningunderlayUWBwithcognitiveradioareproposed.Therstmethodis acognitiveUWB-OFDM approach,whichsupplementstheunderlayUWBwithoverlayo pportunisticspectrumusage.Thepotentialbehavioroftheproposedsystemunderman ydierentscenariosis analyzed,inwhicheitherUWB-OFDMoropportunisticspectr umusagemaybemore preferable.Inthesecondmethod,acognitivesystemthatsh aresthespectrumsensing informationwithinitsnetworkviaultrawidebandispropos ed.Therangeofcognitivecommunicationsinthisscenarioisinvestigated.Bothofthese methodsaimatincreasingthe capacity,performance,range,andvarietyofcognitivecom municationsmakinguseofultrawideband. 64

PAGE 76

Theorganizationofthechapterisasfollows.InSection5.2 ,thecognitiveUWBOFDMisintroduced.InSection5.3,acognitivesystemthats haresthespectrumsensing informationviaultrawidebandisprovided.Thepracticali mplementationofthissystem isexplainedindetail.Thepossiblerangeofcognitivecomm unicationsisinvestigated.In Section5.4,asummaryofthechapteralongwithconclusions isprovided. 5.2CognitiveUWB-OFDMInthissection,acognitiveUWB-OFDMsystemthatiscapable ofswitchingbetweenUWB andopportunisticspectrumusage,whicheverismoreadvant ageous,isconsidered.Inthe UWB-OFDMcommunicationsthatwillbeinvestigatedinthiss ection,forUWBdevices withoutcognitivecapabilities,thepowerlimitationsspe ciedwiththepublishedspectral maskswillremainastheyare.ForthecognitiveUWBradios,h owever,itisexpected thattheregulatoryagenciesprovideadditionalfreedomfo rthetransmittedpower.A motivatingexampleisthefactthattheSPTFhasalreadybeen consideringalternative waysofallocatingthespectrum[79].Byraisingthepowerle vel,itisaimedtofreethe UWBdevicesfrombeingrestrictedtoshortrangeapplicatio ns. Beingabletoimplementboth,cognitiveUWB-OFDMsystemsde cidebetweenUWBOFDMandopportunisticusageaccordingtotheconditions.O neofthemaindecision criteriaisthatUWB-OFDMcanmakeinstantchangesinthespe ctrumitoccupiesby turningonandosomecarriersdependingonthespectrumusa geoflicensedsystems. Opportunisticusage,ontheotherhand,requiresthataband isscannedbyanumberof cognitiveradios,remainsavailableforacertaintime,and satisessomespectralquality conditions,andtherefore,itisnotsuitableforspeedycha nges. Opportunisticusagemaybeespeciallyattractiveforappli cationsthatrequireahigh quality-of-service(QoS)becauseofitshightransmitpowe randwidebandusage(relative tonarrowbandsystems).Also,iftheavailablebands,which maybetargetedforeither UWB-OFDMoropportunisticusage,areathighfrequencies,t helattercanbeabetter optionbecauseofthehigherpathlossatthesefrequencies. 65

PAGE 77

Although,ingeneral,opportunisticusageseemstobemorea dvantageousanddesirable thanUWB-OFDM,incertainscenarios,suchastheoneslisted below,cognitiveUWBradios mayhavetoselectUWB-OFDM. Ifsomeprimaryusershaveafrequencyhoppingsignal,andhe nce,thespectralconditionsarechangingveryfast,thecognitiveradiomaynota bletokeeptrackofthe spectrumopportunitiesanditcanswitchtoUWB-OFDM. Iftheprimaryuseristimehopping,cognitiveradiomightne edtomonitorthespectrumforanextralongtimeframeandmaystillnotdeterminet hetimingsequence, ormaybeunabletoadaptitselftocontinuouslychangingspe ctrum. Iftheprimaryuserismobile(orsteadilymovinginacertain area),itmayberiskyto usethespectrumopportunisticallybecausethecommunicat ionoftheprimarysystem canbeeasilydisturbed.Hence,UWB-OFDMcanbeemployed. Thebandsopentoopportunisticusagemaybetoomuchdivided bynarrowband systems(intoanumberofseparatednarrowbands),leadingt oextralongpulsesin time.Again,inthiscaseUWB-OFDMcanbepreferable. Ifnumerouslicensedusersjoinandleavethespectruminafr equentmanner(likein aGSMband),opportunisticusagemaynotbefeasible. Opportunisticusagerequiressettingathresholdinordert odeterminewhethera certainbandisoccupied.Ifthenoiseroorinabandischangi ngcontinuously,itmay notbepossibletodetermineareliablethreshold,andUWB-O FDMmaybepreferred. Graybands(thebandsinwhichthenoisetemperatureisnotas lowasinthewhite bands)canbeapotentialtargetforUWB-OFDM. Ifthespectrumsensingresultsfromdierentnodesinacogn itivenetworkdonot matchtoalargeextent,thismayindicatethateithersomeof thenodesarebeing shadowed(andcannotdetectprimaryusers),orthespectrum sensinginformation 66

PAGE 78

ofsomenodescannotbetransmitted(ordetected)correctly .Inbothcases,asa precaution,switchingtoUWB-OFDMcanbereasonable. Evenifthereisanadequateamountofspectrumopentoopport unisticusage,the numberofcognitiveusersthattargetatthisspectrummight betoohigh,forcing someoftheseuserstoswitchtoUWB-OFDM. 5.3ACognitiveSystemSupportedbyUWBInthissection,acognitivecommunicationssystemisconsi deredthatutilizesUWBsignaling forsharingthespectrumsensinginformation.Asitwillbes hown,thecognitivesystem benetsfromtheprocessinggainpropertyofUWBinordertoi ncreaseitsrange. Therstthingaboutthecognitivesystemthatwillbepropos edinthissectionisthat boththetransmitterandthereceiverhavetransmissionand receptioncapabilities.Inorder tomatchtheresultsofspectrumsensingoperationdonebybo thparties,eachofthemwill transmittheinformationregardingthewhitespacestheyha vedetected.Weproposethat thetransmissionofspectrumsensingresultsisdonevialow powerUWBsignalingthat complieswiththeFCCregulations.Sincethistransmission willbeaccomplishedinan underlaymanner,itcanbedonesimultaneouslywiththereal datacommunicationwithout aectingeachother. Consideringtherelativelylowthroughputneededtotransm itthesensinginformation aswellasthelowcosttransceiverrequirement,itturnsout tobeaproperoptiontouse anuncomplicatednon-coherentreceiversuchasanenergyde tector,andtoemployon-o keying(OOK)modulation.Theimplementationissuesregard ingtheOOKbasedenergy detectorreceiverssuchasestimatingtheoptimalthreshol danddeterminingtheoptimum integrationinterval,whichcanbefoundin[4],werediscus sedinChapter2. Oncebothpartiesofcommunicationreceivethespectrumsen singinformationobtained bytheotherparty,theylogically'AND'thewhitespaceseac hofthemhasdetectedseparately.Thereasonbehindthisoperationisthatfrequencyb andscanbeconsideredavailable foropportunisticusageonlyifbothpartiesclassifythema swhitespace.Dependingonthe 67

PAGE 79

knowledgeaboutthecommonwhitespaces,eachpartydesigns anewpulseshape.Since thenewpulsesdesignedbybothpartieswillbethesame,this methodhastheveryadvantageousfeaturethatitenableshighlyecientmatchedlte ringduringtherealdatatrac. Tobemoreclear,whatthereceivingpartyusesasthetemplat etomatchthereceivedpulse willbe(almost)thesameasthetransmittedpulsebytheothe rparty,leadingtoahigh correlationbetweenthem,andhence,toasuccessfulmatche dltering. 5.3.1StepsofPracticalImplementationStepsofpracticalimplementationofopportunisticspectr umusagecanbesummarizedas follows. Capturinganysignalinthetargetspectrum(e.g.2.1-2.3GH z)byusinganappropriate antennaandanalogbandpasslter Analyzingthespectralcontentofthereceivedsignaleithe rbyanalogmeans: { Sweepingthetargetspectrumviaamixerandpassingthedown -convertedsignal throughanIFlterwithanarrowpass-band(toobtainanacce ptablefrequency resolution) { Determiningtheexistenceofsignalsinsidethetargetspec trumbyusinganenergydetectorthatconsistsofasquarelawdevice,anintegr ator,andacomparator { Labelingthebinaryoutputoftheenergydetectoras occupied or whitespace (see Fig.26-e) orbydigitalmeans: { Down-convertingthesignaltoanIFfrequency { SamplingthesignalatarateabovetheNyquistrateanddigit izingit(applying analog-to-digitalconversion) { ComputingthepowerspectraldensityeitherbytakingtheFa stFourierTransform(FFT)andsquaring,orbyapplyingmoreadvancedpowere stimationmethodssuchasWelch'smethodorThomson'smultitapermethod 68

PAGE 80

{ Comparingtheobtainedpowerspectraldensitywithapre-de terminedpower threshold.Thisway,determiningtheoccupiedspectraandt hewhitespaces Comparingthebandwidthsofthewhitespaceswithapre-dete rminedminimumbandwidthinordertondoutiftheyarewideenoughtobeutilized Transmittingtheinformationregardingtheusablewhitesp acestotheotherpartyvia UWBsignaling Receivingthewhitespacedatafromtheotherparty Findingthecommonwhitespaces Designingapulseshapethatutilizesasmuchusablewhitesp aceaspossible Initiatingthecognitivecommunicationandrepeatingthee ntiresensingprocessata regularperiod 5.3.2RangeofCognitiveCommunicationsandCognitiveNetw orks Incognitiveradiocommunications,inordertomakesuretha ttheintendedfrequency spectrumisnotinuse,bothpartiesofcommunicationhaveto scanthespectrumandinform eachotheraboutthespectralconditions.Therefore,there shouldnotbeagapbetweenthe sensingrangesofthem.Ifthesensingrangesarenotatleast partiallyoverlapping,there isalwaysariskthatalicenseduserlocatedinsidethegapbe tweenthesensingranges isnotdetected.Therefore,thereceivingsensitivityofbo thpartieshasanintegralrole indeterminingtherangeofcommunication.Assumingarathe rhighsensitivityaround 120 dBm to 130 dBm andfreespacepropagation,inwhichthetransmittedpower( Ptx) andreceivedpower( Prx)arerelatedtoeachotherbytheFriisequation(ignoringth esystem lossandantennagains) Prx= Ptx2 (4 )2d2; (49) where isthewavelength,and d isthedistance.Withtheseassumptions,itisseenthat thescopeofcognitiveradioislimitedto50 m to150 m ,whichiscomparabletotherangeof 69

PAGE 81

WLANs.Ifthetargetedrangeofcommunicationsiswiderthan thislevel,orifthecognitive devicesareexperiencingshadowingandarehardlyabletode tecttheexistenceoflicensed users,anetworkofcollaboratingcognitivenodesmaybeane ectivesolution. Inthispartofthissection,wefocusonacognitivenetwork( seeFig.27)whosenodes communicatewitheachotherusingUWBtoexchangespectrumi nformation.Thefactthat UWBsignalingisproposedmayseemtobecontradictingwitht heaimofincreasingthe rangeofcognitiveradiobecauseofthelimitedrangeofUWB. However,lookingatthebit errorrate(BER)expressionforOOKmodulatedUWBsignals Q r NsAEp 2 N0! ; (50) where Nsisthenumberofpulsespersymbol, A isthepulseamplitude, Episthereceived pulseenergy,andtheadditivewhiteGaussiannoise(AWGN)h asadoublesidedspectrum ofN 0 2,itisseenthatincreasing Ns,whichcanbeaccomplishedbyrepeatedtransmissionof data,resultsinlowerBER.Thisfactleadstoaveryadvantag eousfeatureofUWBcalled the processinggain .Byapplyingthenecessaryamountofprocessinggain,itcan bemade possiblethatthefarthestnodesinacognitivenetworkcans harethespectrumsensing information.Althoughthiscomesattheexpenseofloweredt hroughput,itisnotalimiting factorinthiscasebecauseaquitelowdatarateisenoughtot ransmitthespectrumsensing information.Byenablingallthenodesinacognitivenetwor ktotalktoeachothervia UWB,thereisnoneed eithertoallocateaseparatechannelforsharingthesensin ginformation, ortoemployacentralizedcontrollerthatcollectssuchinf ormation,processesit,and transfersittoothercognitiveusers. Thesensinginformationreceivedfromalltheothernodesin thenetworkcanbecombined ineachnode,andpulsedesigncanbedoneaccordingtothecom monwhitespaces. Increasingthenetworksizeresultsinanincreasedprobabi lityofoverlappingwithlicensedsystems.Thisfactsetsapracticallimittothesizeo fthecognitivenetwork,because 70

PAGE 82

Figure27.Networkofcognitivetransceivers(sensitivity rangesarenotdrawntoscale). continuingtoenlargethenetworkthecommonwhitebandsbec omelessandless,andafter somepointtheiramountbecomesinsucienttoensurethemin imumquality-of-service.For thedetailsofhowthecommonwhitebandsaregoingtobeshare dbythecognitivenodes inthenetwork,thereadercanbereferredto[93]and[94].5.3.3NumericalResultsComputeranalysisandsimulationsareperformedregarding thepracticalimplementationof cognitiveradiocommunications.Thesearerelatedtothetr ansmissionofspectrumsensing resultsviaUWB,therangeofcognitivecommunications,and thecapabilityofacognitive networktodetectalicensedsystem.Inthesimulationsrega rdingtheUWBsignaling,the channelmodel CM 3in[95],whichcorrespondstoanoceenvironmentwithline -of-sight (LOS),isutilized.Thefrequencyrangeis3.1-3.6GHz,ther eferencepathloss35.4dB,the 71

PAGE 83

pathlossexponentis1.63,thereceiverantennanoisegure is17dB,theimplementation lossis3dB,thethroughputis20Mbps,andtheintegrationin tervalis30ns. Atheoreticalanalysiswasperformedtoinvestigatetheper formanceofOOKmodulated UWBdatatransmissiondependingonthedistancebetweenaco gnitivetransmitter-receiver pair.Accordingto[95],thepathlossassumedcanbeshownas L ( d )= L0+10 nlog10( d d0) ; (51) wherethereferencedistance( d0)issetas1 m L0isthepathlossat d0,and n isthepath lossexponent.Theaveragenoisepowerperbitis N = 174+10 log10( Rb) ; (52) where Rbisthethroughput.InFig.28,theeectofdistanceonthepro babilityoferroris demonstrated.TheresultsshowthattheBERsobtainedforup to40 m arestillacceptable. Forfurtherdistances,however,someprocessinggainisde nitelyneeded.Theprocessing gainisobtainedbyrepeatedtransmissionofthesameinform ation.Asimulationwasdone thatinvestigatesthenumberofrepetitionsrequiredinord ernottoexceedtheBERobtained at40 m ,whichcorrespondsto10 3 : 2.Thenumberofrepetitionsneededvs.thedistanceis showninFig.29. Asimulationisdonetoinvestigatetheeectofthenumberof nodesontheprobabilityof alicensedsystembeingdetectedbythecognitivenetwork.F ig.27demonstratesanetwork composedofcognitiveradiodevices.Thenodesinthenetwor karerandomlydistributed ina200 m x200 m areainsideabuilding.Itisassumedthatthereisalicensed transmitter, whichisa GSM 900cellphonetransmittingat 60 dBm ,whoselocationisrandom,aswell. Dependingonthelevelofthenodesensitivity,thenumberof nodesrequiredtomakea reliabledetectionmightvary.Theresultsofthissimulati onaredemonstratedinFig.30. 72

PAGE 84

0 20 40 60 80 100 120 140 160 10-10 10-8 10-6 10-4 10-2 100 Distance (m)BER Figure28.BERvs.distancebetweenthenodesforUWBsignali ng. 5.4ConclusionThecontinuouslyincreasingneedforfrequencyspectrumre quirestoincreasetheeciency ofspectrumusage.Therefore,itisnecessarytodeveloprex ibleandadaptableradioaccess technologiesthatcantakeadvantageoftheavailablespect ruminanopportunisticway. Inthischapter,twodierentmethodsforcombiningUWBwith cognitiveradioare provided.First,itisshownthatthemarriageofOFDMbasedU WBwithopportunistic spectrumusagewillopenthedoorsforfurtherimprovements inspectraleciency,and bringaboutconceptsthatwillallowthejointunderlayando verlayusageofthespectrum. Althoughthiscomesattheexpenseofincreasedhardwarecom plexityrelativetopureUWB, itismadeclearthattheadvantagesofcognitiveUWB-OFDMwo uldpayoforthisincrease. Second,acognitivesystemisproposedthatbenetsfromUWB indistributingthespectrum 73

PAGE 85

40 60 80 100 120 140 160 1 2 3 4 5 6 7 8 9 10 Distance (m)Number of Repetitions Figure29.RepetitionraterequiredforreliableUWBsignal ingvs.distance. sensinginformation.Itisshownthatsuchasystemcanmakeu seoftheprocessinggain propertyofultrawidebandtoincreasethetargetedrangeof cognitivecommunications. 74

PAGE 86

2 4 6 8 10 12 14 16 18 20 0 10 20 30 40 50 60 70 80 90 100 Number of NodesDetection Probability (%) Node Sensitivity = -130 dBm Node Sensitivity = -120 dBm Figure30.Probabilityofalicensedtransmitterbeingdete ctedbythecognitivenetwork. 75

PAGE 87

CHAPTER6 SUMMARYANDCONCLUSIONS 6.1SummaryofContributionsThethesisfocusesontheimplementationofultrawidebanda nditscoexistencewithnarrowbandsystemsinordertoleadtowardstherealizationoft hecognitiveradioconcept. Thenovelcontributionsofthethesiscanbesummarizedasfo llows. PracticalimplementationofimpulseradioUWBusingenergy detectorreceivers: Theneedforthejointadaptationoftheintegrationinterva l,optimalthreshold,and thesynchronizationpointisclearlydemonstrated. Itisshownthatthresholdestimationcanbenetfromthecom putationaleasiness broughtbytheGaussianapproximationofreceivedsignalst atistics,whichyields reasonableresultsforcertainbandwidths. Theinter-symbolinterferenceprobleminhighdatarateUWB systemsemploying OOKbasedenergydetectorsisinvestigated. TheincreasingnegativeeectofISIonthesystemperforman cewithincreasingdata rateisdemonstrated. InordertoovercometheISIproblem,amodiedenergydetect orthathasabuilt-in symboldecisionmechanismbasedondecisionfeedbackequal izationisproposed. Asimplebutcleverwayofusingtrainingsymbolswiththepur poseofestimatingthe decisionfeedbackltercoecientsisintroduced.Ithasbe enproventhattheerror propagationproblemdoesnotexistintheproposedapproach 76

PAGE 88

Coexistenceofultrawidebandandnarrowbandsystems;supp ressingnarrowbandinterference: NBIavoidancemethodsincludingmulti-carrierapproaches andmulti-bandschemes, aswellasalternativesolutionsbasedonpulseshaping,tim e-hoppingcodeadjustment, andantennadesignhavebeeninvestigated. Amongthecancellationtechniques,detailsofMMSEcombini ngalgorithmarepresented. Frequencydomaintechniquessuchasnotchltering,time-f requencymethodslike wavelettransformandtimedomainapproaches,particularl ylineartechniques,have beendiscussedinseparatesections. Opportunisticspectrumusage: Theconceptofopportunity,whichisofcrucialimportancef orcognitiveradiosystems, isdenedfromspectral,temporalandspatialperspectives Theopportunitysensingandspectrumshapingfeaturesofco gnitiveradiosarediscussedanddierentapproachesareaddressed.Analternati vemethodforshapingthe spectrumofthetransmittedpulseisprovided. CognitiveUWB: CombiningOFDMbasedUWBwithopportunisticspectrumusage isproposedto improvethespectraleciencyandtobringaboutconceptsth atwillallowthejoint underlayandoverlayusageofthespectrum. AcognitivesystemisproposedthatbenetsfromUWBindistr ibutingthespectrum sensinginformation.Itisshownthatsuchasystemcanmakeu seoftheprocessinggain propertyofultrawidebandtoincreasethetargetedrangeof cognitivecommunications. 77

PAGE 89

6.2ConclusionsUWB,OFDM,andcognitiveradioaretermsthatthewirelessco mmunityhasalreadybeen heavilyexposedtoovertherecentyears.Itisanticipatedt hatinthenearfuture,thewireless communitywillbeencounteringthatthesetermsaremention edjointlyinthecontextof thespectrumeciencyandopportunisticspectrumusage.In thelightofthisexpectation, inthisthesis,detailsofpracticalUWBimplementationare given.Thecognitiveradio conceptisintroduced,anditsrequirementsandobjectives areexplainedindetail.Itis shownthatUWBcancoexistwithnarrowbandsystemsbyutiliz ingNBIavoidanceand cancelationmethods,andhence,itisproventhatUWBisaver yappropriatecandidate bothforimplementingorsupportingcognitiveradio.Twose parateapplicationsascase studiesinwhichcognitiveradioiscombinedwithultrawide bandareprovided.Therst casestudyisanexampleofhowUWBcanbeemployedasameansof implementingcognitive radio,whereasthesecondonedemonstrateshowUWBcansuppl ementacognitiveradio system.Byprovidingtheseapplications,itisaimedtoopen thedoorsforthemarriageof UWBwithcognitiveradioandtomotivatewirelesscommunica tionsresearcherstoinvolve UWBintheircognitiveradiostudies. 78

PAGE 90

REFERENCES [1]J.Mitola,\Cognitiveradioforrexiblemobilemultimed iacommunications,"in Proc. MobileMultimediaCommun.(MoMuC'99) ,Nov.1999,pp.3{10. [2]S.Haykin,\Cognitiveradio:Brain-empoweredwireless communications," IEEEJ. Select.AreasCommun. ,vol.23,no.2,pp.201{220,Feb.2005. [3]S.Mangold,Z.Zhong,K.Challapali,andC.-T.Chou,\Spe ctrumagileradio:radioresourcemeasurementsforopportunisticspectrumusag e,"in Proc.IEEEGlobal Telecomm.Conf.(GLOBECOM) ,vol.6,Dallas,TX,Dec.2004,pp.3467{3471. [4]M.E.Sahin,I.Guvenc,andH.Arslan,\Optimizationofen ergydetectorreceiversfor UWBsystems,"in Proc.IEEEVehic.Technol.Conf. ,vol.2,Stockholm,Sweden,May 2005,pp.1386{1390. [5]M.E.SahinandH.Arslan,\Inter-symbolinterferencein highdatarateUWBcommunicationsusingenergydetectorreceivers,"in Proc.IEEEInt.Conf.onUltra-Wideband Zurich,Switzerland,Sep.2005,pp.176{179. [6]H.ArslanandM.E.Sahin, UltraWidebandWirelessCommunications .Hoboken,NJ: JohnWiley&Sons,Inc.,2006,chapterNarrowbandInterfere nceIssuesinUltrawidebandSystems,tobepublished. [7]||,\CognitiveUWB-OFDM:Pushingultrawidebandbeyond itslimitviaopportunisticspectrumusage," JournalonCommun.Networks(JCN) ,June2006,toappear. [8]M.WeisenhornandW.Hirt,\Robustnoncoherentreceiver exploitingUWBchannel properties,"in Proc.IEEEUltrawidebandSyst.Technol.(UWBST) ,Kyoto,Japan, May2004,pp.156{160. [9]A.RabbachinandI.Oppermann,\Synchronizationanalys isforUWBsystemswitha low-complexityenergycollectionreceiver,"in Proc.IEEEUltrawidebandSyst.Technol. (UWBST) ,Kyoto,Japan,May2004,pp.288{292. [10]P.A.HumbletandM.Azizoglu,\Onthebiterrorrateofli ghtwavesystemswith opticalampliers," J.ofLightwaveTechnology ,vol.9,no.11,pp.1576{1582,Nov. 1991. [11]J.Foerster,\IEEEP802.15workinggroupforwirelessp ersonalareanetworks (WPANs),channelmodelingsub-committeereport-nal,"Ma r.2003.[Online]. Available:http://www.ieee802.org/15/pub/2003/Mar03/ 79

PAGE 91

[12]B.Mielczarek,M.Wessman,andA.Svensson,\Performan ceofcoherentuwbrakereceiverswithchannelestimators,"in Proc.IEEEVehic.Technol.Conf. ,vol.3,Orlando, FL,Oct.2003,pp.1880{1884. [13]S.Paquelet,L.M.Aubert,andB.Uguen,\Animpulseradi oasynchronoustransceiver forhighdatarates,"in Proc.IEEEUltrawidebandSyst.Technol.(UWBST) ,Kyoto, Japan,May2004,pp.1{5. [14]J.Proakis, DigitalCommunications ,4thed.NewYork:McGraw-Hill,2000. [15]V.Lottici,A.D'Andrea,andU.Mengali,\Channelestim ationforultra-wideband communications," IEEEJ.Select.AreasCommun. ,vol.20,no.9,pp.1638{1645,Dec. 2002. [16]N.F.Krasner,\Ecientsearchmethodsusingenergydet ectors{maximumprobability ofdetection," IEEEJ.Select.AreasCommun. ,vol.4,no.2,pp.273{279,March1986. [17]A.Rajeswaran,V.Somayazulu,andJ.Foerster,\RAKEpe rformanceforapulsebased UWBsysteminarealisticUWBindoorchannel,"in IEEEInt.Conf.Commun.(ICC) vol.4,Anchorage,AK,May2003,pp.2879{2883. [18]A.Klein,I.Brown,D.R.,D.Goeckel,andJ.Johnson,C.R .,\RAKEreceptionfor UWBcommunicationsystemswithintersymbolinterference, "in IEEEWorkshopon SignalProc.AdvancesWirelessCommun.(SPAWC2003) ,Rome,Italy,June2003,pp. 244{248. [19]X.Peng,F.Chin,andA.Madhukumar,\Performancestudi esofanover-sampling multi-channelequalizerforamulti-bandUWBsystem,"in Proc.IEEEUltraWideband Syst.Technol.(UWBST) ,Reston,VA,Nov.2003,pp.413{417. [20]A.Duel-HallenandC.Heegard,\Delayeddecision-feed backsequenceestimation," IEEETrans.Commun. ,vol.37,no.5,pp.428{436,May1989. [21]R.Johnk,D.Novotny,C.Grosvenor,N.Canales,andJ.Ve neman,\Time-domainmeasurementsofradiatedandconductedUWBemissions," IEEEAerospaceandElectron. Syst.Mag. ,vol.19,no.8,pp.18{22,Aug2004. [22]J.Foerster,\Ultra-widebandtechnologyenablinglow -power,high-rateconnectivity (invitedpaper),"in Proc.IEEEWorkshopWirelessCommun.Networking ,Pasadena, CA,Sep.2002. [23]J.R.Foerster,\Theperformanceofadirect-sequences preadultra-widebandsystem inthepresenceofmultipath,narrowbandinterference,and multiuserinterference,"in Proc.IEEEVehic.Technol.Conf. ,vol.4,Birmingham,AL,May2002,pp.1931{1935. [24]J.ChoiandW.Stark,\Performanceofautocorrelationr eceiversforultra-wideband communicationswithPPMinmultipathchannels,"in Proc.IEEEUltrawidebandSyst. andTechnol.(UWBST) ,Baltimore,MD,May2002,pp.213{217. [25]W.Tao,W.Yong,andC.Kangsheng,\Analyzingtheinterf erencepowerofnarrowbandjammingsignalonUWBsystem,"in Proc.IEEEPersonal,Indoor,MobileRadio Commun.(PIMRC) ,Singapore,Sep.2003,pp.612{615. 80

PAGE 92

[26]L.ZhaoandA.Haimovich,\Performanceofultra-wideba ndcommunicationsinthe presenceofinterference," IEEEJ.Select.AreasCommun. ,vol.20,pp.1684{1691, Dec.2002. [27]G.DurisiandS.Benedetto,\Performanceevaluationof TH-PPMUWBsystemsinthe presenceofmultiuserinterference," IEEECommun.Lett. ,vol.7,no.5,pp.224{226, May2003. [28]R.Tesi,M.Hamalainen,J.Iinatti,andV.Hovinen,\Ont heinruenceofpulsedjammingandcolourednoiseinUWBtransmission,"in Proc.FinnishWirelessCommun. Workshop(FWCW) ,Espoo,Finland,May2002. [29]X.WangandH.V.Poor, WirelessCommunicationSystems:AdvancedTechniquesfor SignalReception .1sted.,PrenticeHall,2004. [30]D.Zhang,P.Fan,andZ.Cao,\Interferencecancellatio nforOFDMsystemsinpresence ofoverlappednarrowbandtransmissionsystem," IEEEConsum.Electron. ,2004. [31]J.ChoiandN.Cho,\Narrow-bandinterferencesuppress ionindirectsequencespread spectrumsystemsusingalatticeIIRnotchlter,"in Proc.IEEEInt.Conf.Acoustics, Speech,SignalProcessing(ICASSP) ,vol.3,Munich,Germany,April1997,pp.1881{ 1884. [32]L.RuschandH.Poor,\Narrowbandinterferencesuppres sioninCDMAspreadspectrumcommunications," IEEEPersonalCommun.Mag. ,vol.42,pp.1969{1979,Apr. 1994. [33]||,\Multiuserdetectiontechniquesfornarrowbandin terferencesuppressionin spreadspectrumcommunications," IEEETrans.Commun. ,vol.42,pp.1727{1737, Apr.1995. [34]J.Proakis,\Interferencesuppressioninspreadspect rumsystems,"in Proc.IEEEInt. Symp.onSpreadSpectrumTechniquesandApplications ,vol.1,Sep.1996,pp.259{266. [35]L.Milstein,\Interferencerejectiontechniquesinsp readspectrumcommunications," in Proc.IEEE ,vol.76,June1988,pp.657{671. [36]C.Carlemalm,H.V.Poor,andA.Logothetis,\Suppressi onofmultiplenarrowband interferersinaspread-spectrumcommunicationsystem," IEEEJ.Select.AreasCommun. ,vol.18,no.8,pp.1365{1374,August2000. [37]P.AzmiandM.Nasiri-Kenari,\Narrow-bandinterferen cesuppressioninCDMA spread-spectrumcommunicationsystemsbasedonsub-optim umunitarytransforms," IEICETrans.Commun. ,vol.E85-BNo.1,pp.239{246,Jan.2002. [38]T.J.LimandL.K.Rasmussen,\Adaptivecancellationof narrowbandsignalsin overlaidcdmasystems,"in Proc.IEEEInt.WorkshopIntel.SignalProcessingand Commun.Syst. ,Singapore,Nov.1996,pp.1648{1652. [39]H.FathallahandL.Rusch,\Enhancedblindadaptivenar rowbandinterferencesuppressionindsss,"in Proc.IEEEGlobalTelecommun.Conf.(GLOBECOM) ,vol.1, London,UK,Nov.1996,pp.545{549. 81

PAGE 93

[40]W.-S.Hou,L.-M.Chen,andB.-S.Chen,\Adaptivenarrow bandinterferencerejection inDS-CDMAsystems:aschemeofparallelinterferencecance llers," IEEEJ.Select. AreasCommun. ,vol.20,pp.1103{1114,June2001. [41]P.-R.Chang,\Narrowbandinterferencesuppressionin spreadspectrumCDMAcommunicationsusingpipelinedrecurrentneuralnetworks,"i n Proc.IEEEInt.Conf.UniversalPersonalCommun.(ICUPC) ,vol.2,Oct.1998,pp.1299{1303. [42]H.V.PoorandX.Wang,\Code-aidedinterferencesuppre ssioninDS/CDMAspread spectrumcommunications," IEEETrans.Commun. ,vol.45,no.9,pp.1101{1111, Sept.1997. [43]S.Buzzi,M.Lops,andA.Tulino,\Time-varyingmmseint erferencesuppressionin asynchronousds/cdmasystemsovermultipathfadingchanne ls,"in Proc.IEEEInt. Symp.onPersonal,IndoorandMobileRadioCommun. ,Sep.1998,pp.518{522. [44]M.Medley,\Narrow-bandinterferenceexcisioninspre adspectrumsystemsusing lappedtransforms," IEEETrans.Commun. ,vol.45,pp.1444{1455,Nov.1997. [45]A.Akansu,M.Tazebay,M.Medley,andP.Das,\Waveletan dsubbandtransforms: fundamentalsandcommunicationapplications," IEEECommun.Mag. ,vol.35,pp. 104{115,Dec.1997. [46]R.D.Weaver,\Spreadspectruminterferencesuppressi onusingadaptivetimefrequencytilings,"in Proc.IEEEAsilomarConf.SignalsSyst.Comput.PacicGrov e CA,Nov2003,pp.1881{18840. [47]B.Krongold,M.Kramer,K.Ramchandran,andD.Jones,\S preadspectruminterferencesuppressionusingadaptivetime-frequencytiling s,"in Proc.IEEEInt.Conf. Acoustics,Speech,SignalProcessing(ICASSP) ,vol.3,Munich,Germany,April1997, pp.1881{18840. [48]Y.ZhangandJ.Dill,\Ananti-jammingalgorithmusingw aveletpacketmodulated spreadspectrum,"in Proc.IEEEMilitaryCommun.Conf. ,vol.2,Nov1999,pp.846{ 850. [49]E.Pardo,J.Perez,andM.Rodriguez,\Interferenceexc isioninDSSSbasedonundecimatedwaveletpackettransform," IEEEElectron.Lett. ,vol.39,no.21,pp.1543{1544, Oct.2003. [50]T.Kasparis,\Frequencyindependentsinusoidalsuppr essionusingmedianlters," in Proc.IEEEInt.Conf.Acoustics,Speech,SignalProcessing (ICASSP) ,vol.3, Toronto,Canada,April1991,pp.612{615. [51]R.LowdermilkandF.Harris,\Interferencemitigation inorthogonalfrequencydivisionmultiplexing(ofdm),"in Proc.IEEEInt.Conf.UniversalPersonalCommun. (ICUPC) ,vol.2,Cambridge,MA,Sep1996,pp.623{627. [52]R.Nilsson,F.Sjoberg,andJ.LeBlanc,\Arank-reduced lmmsecancellerfornarrowbandinterferencesuppressioninofdm-basedsystems," IEEETrans.Commun. ,vol.51, no.12,pp.2126{2140,Dec2003. 82

PAGE 94

[53]S.Vogeler,L.Broetje,K.-D.Kammeyer,R.Rueckriem,a ndS.Fechtel,\Blindbluetoothinterferencedetectionandsuppressionforofdmtran smissionintheismband," in Proc.IEEEAsilomarConf.onSignals,Syst.,Computers ,vol.1,PacicGrove,CA, Nov2003,pp.703{707. [54]M.GhoshandV.Gadam,\Bluetoothinterferencecancell ationfor802.11gwlanreceivers,"in Proc.IEEEInt.Conf.Commun.(ICC) ,vol.2,Anchorage,AK,May2003, pp.1169{1173. [55]S.Roy,J.Foerster,V.Somayazulu,andD.Leeper,\Ultr awidebandradiodesign:the promiseofhigh-speed,short-rangewirelessconnectivity ," IEEEProceedings ,vol.92, no.2,pp.295{311,Feb2004. [56]A.Batra,J.Balakrishnan,G.Aiello,J.Foerster,andA .Dabak,\Designofamultiband OFDMsystemforrealisticUWBchannelenvironments," IEEETrans.Microw.Theory Techniq. ,vol.52,no.9,pp.2123{2138,Sep2004. [57]T.IkegamiandK.Ohno,\Interferencemitigationstudy foruwbimpulseradio,"in Proc.IEEEPersonal,Indoor,MobileRadioCommun.(PIMRC) ,vol.1,Sep2003,pp. 583{587. [58]R.WilsonandR.Scholtz,\Templateestimationinultra -widebandradio,"in Proc. IEEEAsilomarConf.onSignals,Syst.,Computers ,PacicGrove,CA,2004. [59]I.Bergel,E.Fishler,andH.Messer,\Narrowbandinter ferencesuppressioninimpulse radiosystems,"in IEEEConf.onUWBSyst.Technol. ,Baltimore,MD,May2002,pp. 303{307. [60]H.Sheng,A.Haimovich,A.Molisch,,andJ.Zhang,\Opti mumcombiningfortime hoppingimpulseradioUWBrakereceivers,"in Proc.IEEEUltrawidebandSyst.Technol.(UWBST) ,Nov2003. [61]N.BoubakerandK.Letaief,\AlowcomplexityMMSE-RAKE receiverinarealistic UWBchannelandinthepresenceofNBI,"in Proc.IEEEWirelessCommun.NetworkingConf.(WCNC) ,vol.1,NewOrleans,LU,March2003,pp.233{237. [62]D.Cassioli,M.Z.Win,F.Vatalaro,andA.F.Molisch,\P erformanceoflow-complexity RAKEreceptioninarealisticUWBchannel,"in Proc.IEEEInt.Conf.Commun. (ICC) ,vol.2,NewYork,Apr.2002,pp.763{767. [63]M.Z.WinandR.A.Scholtz,\Impulseradio:Howitworks, IEEECommun.Lett. vol.2,no.2,pp.36{38,Feb.1998. [64]X.ChuandR.Murch,\Theeectofnbionuwbtime-hopping systems," IEEETrans. onWirelessCommun. ,vol.3,no.5,pp.1431{1436,Sep2004. [65]T.YucekandH.Arslan,\Noiseplusinterferencepower estimationinadaptiveOFDM systems,"in Proc.IEEEVehic.Technol.Conf. ,Stockholm,Sweden,May2005,acceptedforpublication. 83

PAGE 95

[66]R.Fisher,R.Kohno,H.Ogawa,H.Zhang,andK.Takizawa, \IEEE P802.15workinggroupforwirelesspersonalareanetworks( WPANs),DS-UWB proposalupdate,"May2004.[Online].Available:www.uwbf orum.org/documents/ 15-04-0140-04-003a-merger2-proposal-ds-uwb-p%resent ation.ppt. [67]\Ultrawideband:High-speed,short-rangetechnology withfar-reachingeects, multibandOFDMalliance,"Sep2004.[Online].Available:w ww.multibandofdm.org/ papers/MBOA\_UWB\_White\_Paper.pdf. [68]A.TahaandK.Chugg,\Atheoreticalstudyontheeectso finterferenceuwbmultiple accessimpulseradio,"in Proc.IEEEAsilomarConf.onSignals,Syst.,Computers vol.1,PacicGrove,CA,Nov.2002,pp.728{732. [69]I.GuvencandH.Arslan,\PerformanceevaluationofUWB systemsinthepresenceof timingjitter,"in Proc.IEEEUltraWidebandSyst.Technol.Conf. ,Reston,VA,Nov 2003,pp.136{141. [70]L.PiazzoandJ.Romme,\Spectrumcontrolbymeansofthe THcodeinUWBsystems,"in Vehic.Technol.Conf. ,vol.3,Apr.2003,pp.1649{1653. [71]||,\Onthepowerspectraldensityoftime-hoppingimpu lseradio,"in IEEEConf. UltrawidebandSyst.Technol.(UWBST) ,May.2002,pp.241{244. [72]H.Schantz,G.Wolenec,andE.Myszka,\Frequencynotch edUWBantennas,"in IEEE Conf.UltrawidebandSyst.Technol.(UWBST) ,vol.3,Nov.2003,pp.214{218. [73]S.Verdu, MultiuserDetection .1sted.Cambridge,UK:CambridgeUniversityPress, 1998. [74]S.Mallat,\Atheoryformultiresolutionsignaldecomp osition:thewaveletrepresentation," IEEEPatternAnal.andMachineIntell. ,vol.11,no.7,pp.674{693,1989. [75]R.Klein,M.Temple,R.Raines,andR.Claypoole,\Inter ferenceavoidancecommunicationsusingwaveletdomaintransformationtechniques ," Electron.Lett. ,vol.37, no.15,pp.987{989,July2001. [76]M.Medley,G.Saulnier,andP.Das,\Radiometricdetect ionofdirect-sequencespread spectrumsignalswithinterferenceexcisionusingthewave lettransform,"in IEEEInt. Conf.onCommun.(ICC94) ,vol.3,May1994,pp.1648{1652. [77]J.Patti,S.Roberts,andM.Amin,\Adaptiveandblockex cisionsinspreadspectrum communicationsystemsusingthewavelettransform,"in AsilomarConf.onSignals, Syst.,Computers ,vol.1,Nov.1994,pp.293{297. [78]W.D.Horne,\Adaptivespectrumaccess:Usingthefulls pectrumspace,"in Proc. 31stAnnualTelecommunicationsPolicyResearchConferenc e(TPRC03) ,Oct.2003. [79]FederalCommunicationsCommission,\SpectrumPolicy TaskForceReportET Docketno.02-135,"Nov.2002. 84

PAGE 96

[80]\SpectrumOccupancyReportforNewYorkCityduringthe Republican ConventionAugust30-September1,2004,"Jan.2005.[Onlin e].Available: http://www.sharedspectrum.com/?section=measurements [81]M.P.Olivieri,G.Barnett,A.Lackpour,A.Davis,andP. Ngo,\Ascalabledynamic spectrumallocationsystemwithinterferencemitigationf orteamsofspectrallyagile softwaredenedradios,"in Proc.IEEEInt.Symp.DynamicSpectrumAccessNetworks (DySPAN)2005 ,Baltimore,MD,Nov.2005,pp.170{179. [82]H.ZhengandL.Cao,\Device-centricspectrummanageme nt,"in Proc.IEEEInt. Symp.DynamicSpectrumAccessNetworks(DySPAN)2005 ,Baltimore,MD,Nov. 2005,pp.56{65. [83]T.A.Weiss,J.Hillenbrand,andF.K.Jondral,\Adivers ityapproachforthedetection ofidlespectralresourcesinspectrumpoolingsystems,"in Proc.48thInt.Sci.Colloq. Illmenau,Germany,Sep.2003. [84]D.Cabric,S.M.Mishra,andR.W.Brodersen,\Implement ationissuesinspectrum sensingforcognitiveradios,"in Proc.IEEEAsilomarConf.onSignals,Syst.,Computers ,vol.1,PacicGrove,CA,Nov.2004,pp.772{776. [85]J.Hillenbrand,T.Weiss,andF.Jondral,\Calculation ofdetectionandfalsealarm probabilitiesinspectrumpoolingsystems," IEEEComm.Letters ,vol.9,no.4,pp. 349{351,April2005. [86]X.JingandD.Raychaudhuri,\Spectrumco-existenceof IEEE802.11band802.16a networksusingtheCSCCetiquetteprotocol,"in Proc.IEEEInt.Symp.Dynamic SpectrumAccessNetworks(DySPAN)2005 ,Baltimore,MD,Nov.2005,pp.243{250. [87]G.GanesanandY.Li,\Cooperativespectrumsensinginc ognitiveradionetworks," in Proc.IEEEInt.Symp.DynamicSpectrumAccessNetworks(DyS PAN)2005 ,Baltimore,MD,Nov.2005,pp.137{143. [88]T.A.WeissandF.K.Jondral,\Spectrumpooling:aninno vativestrategyforthe enhancementofspectrumeciency," IEEECommun.Mag. ,vol.42,no.3,pp.8{14, Mar.2004. [89]J.PostonandW.Horne,\DiscontiguousOFDMconsiderat ionsfordynamicspectrum accessinidleTVchannels,"in Proc.IEEEInt.Symp.DynamicSpectrumAccess Networks(DySPAN)2005 ,Baltimore,MD,Nov.2005,pp.607{610. [90]R.Dilmaghani,M.Ghavami,B.Allen,andH.Aghvami,\No velUWBpulseshaping usingprolatespheroidalwavefunctions,"in Proc.IEEEInt.Symp.onPersonal,Indoor andMobileRadioCommun.PIMRC2003 ,vol.1,Beijing,China,Sept.2003,pp.602 {606. [91]K.Wallace,B.Parr,B.Cho,andZ.Ding,\Performancean alysisofaspectrallycompliantultra-widebandpulsedesign," IEEETrans.onWirelessCommun. ,vol.4,no.5, pp.2172{2181,Sept.2005. 85

PAGE 97

[92]M.G.DiBenedettoandL.DeNardis,\TuningUWBsignalsb ypulseshaping," Special IssueonSignalProc.inUWBCommun.,EurasipJournalonSign alProc.,Elsevier Publishers ,2005,toappear. [93]H.ZhengandL.Cao,\Device-centricspectrummanageme nt,"in Proc.IEEEInt. Symp.DynamicSpectrumAccessNetworks(DySPAN)2005 ,Baltimore,MD,Nov. 2005,pp.56{65. [94]R.Thomas,L.DaSilva,andA.MacKenzie,\Cognitivenet works,"in Proc.IEEEInt. Symp.DynamicSpectrumAccessNetworks(DySPAN)2005 ,Baltimore,MD,Nov. 2005,pp.352{360. [95]A.Molisch,K.Balakrishnan,C.C.Chong,S.Emami,A.Fo rt,J.Karedal,J.Kunisch, H.Schantz,U.Schuster,andK.Siwiak,\IEEE802.15.4achan nelmodel-nal report,"Sep.2004.[Online].Available:http://www.ieee 802.org/15/pub/TG4a.html. 86


xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001787499
003 fts
005 20070508083559.0
006 m||||e|||d||||||||
007 cr mnu|||uuuuu
008 070508s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001418
040
FHM
c FHM
035
(OCoLC)124063768
049
FHMM
090
TK145 (ONLINE)
1 100
Sahin, Mustafa Emin.
0 245
Towards the realization of cognitive radio:
b coexistence of ultrawideband and narrowband systems
h [electronic resource] /
by Mustafa Emin Sahin.
260
[Tampa, Fla] :
University of South Florida,
2006.
3 520
ABSTRACT: Ultrawideband and cognitive radio are two of the most important approaches that are shaping the future of wireless communication systems. At a first glance, the aims of UWB and cognitive radio do not seem to be overlapping significantly, however, there is a strong synergy between the capabilities of UWB and the goals of cognitive radio. One of the objectives of this thesis is to shed the first light on the marriage of these two important approaches.Ultrawideband (UWB) is a promising technology for future short-range, high-data rate wireless communication networks. Along with its exciting features including achieving high data rates, low transmission power requirement, and immunity to multipath effects, UWB is unique in its coexistence capability with narrowband systems.In this thesis, the details of practical UWB implementation are provided. Regarding the coexistence of UWB with licensed narrowband systems, narrowband interference (NBI)avoidance and cancelation techniques in the literature are investigated. It is aimed to emphasize that UWB is a strong candidate for cognitive radio, and this fact is proven by providing two different approaches in which ultrawideband is combined with cognitive radio to maximize the performance of unlicensed communications.
502
Thesis (M.A.)--University of South Florida, 2006.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
500
Title from PDF of title page.
Document formatted into pages; contains 86 pages.
590
Adviser: Huseyin Arslan, Ph.D.
653
Interference avoidance.
Narrowband interference.
Opportunistic spectrum usage.
Spectrum sensing.
Spectrum shaping.
690
Dissertations, Academic
z USF
x Electrical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.1418