OFDM在衰减信道内的仿真(英文)(Matlab)

CSMSimulationReportNINGLIOFDMInFadingChannels140747151CommunicationSystemModellingSimulatingOFDMCommunicationSystemsinFadingChannelsDepartmentofElectronicandElectricalEngineeringMScinWirelessandOpticalCommunicationsAuthor:NINGLIStudentNO.:14074715Date:21/01/2015CSMSimulationReportNINGLIOFDMInFadingChannels1407471521.IntroductionMultipathFadingandSimulationofRayleighFadingChannelsInwirelesscommunications,multipathfadingisapropagationphenomenonthatsignalstransmittedfromasourcearriveatthereceiverantennaintwoormorepaths,resultinginconstructiveanddestructiveinterferencebetweensignalsarisingfrommultiplepaths[1].Rayleighfadingisoneofthemodelsthatareusedtodescribetheformoffadingthatoccurswhenmultipathpropagationhappens.Whentherearealargenumberofpaths,thecentrallimittheoremcanbeappliedtomodelthetime-variantimpulseresponseofthechannelasacomplex-valuedGaussianrandomprocess[2].AsnapshotoftheRayleighchannelresponseisshowedasfollow:Figure1.1:AsnapshotoftheRayleighFlatFading[3].TheobjectiveofthefirstpartofthisreportistobuildasignalmodelforISIchannels.Inthiswork,L-rayRayleighfadingchannelswillbesimulatedastheISIchannels.Inthechannel,powerisnormalizedtobeequalandthetimedelaybetweeneachrayislimitedto2.Meanwhile,Lcomplexfadingcoefficientsα𝐿arealsogoingtobegeneratedinwhicheachofthemhasaRayleighdistributedamplitudeandauniformlydistributedphase.Incombinationofafunctionf(assumedtoberectangular)andfadingcoefficients,aToeplitzchannelmatrix𝑔𝑚,𝑛willbebuilt.Essentially,asequenceofNindependentQPSKsymbolswillbecreatedwhichactsasaroleoftheinputsignalofthissimulation.Eventually,therelationshipbetweenthesimulationinputandoutput,alongwiththesystemsymbolerrorrate(SER),canbeobtained,withtheinfluenceofasequenceofNindependentnoisesamples.Multi-carrierSystemandtheSimulationoftheOFDMSystemThefirstpartoftheintroductiontalksaboutthesevereimpactofmultipathpropagationinwirelesscommunications.Inordertomeettherequirementoftransmittinglargeamountofdatainradiochannels,thephenomenonofmultipathpropagationmustbeovercome.Multi-carriersystemisthenintroduced.OFDMisaCSMSimulationReportNINGLIOFDMInFadingChannels140747153specialmulti-carriermodulationwhichcomprisesalargenumberofsub-carriersandthelowdataratesub-carriersarecombinedtoformahighdataratetransmission[4].Figure1.2:OFDMintimeandfrequencydomains[5].AnotherfundamentalconceptofOFDMistheorthogonalityofthesub-carriers.InOFDMthefrequencyspectrumoverlapsthespectrumofadjacentsub-carriers,howeverbymakingthesignalsorthogonaltoeachotherthereisnomutualinterference.Hence,thespectrumefficiencyisincreased.Becauseofthelargenumberofsmalldatarateorthogonalsub-carriers,OFDMhastheadvantageoversinglecarriersystemthatitisveryresiliencetointerference,whichleadstobetterperformanceinmultipathpropagation.ThesecondpartofthesimulationisaboutOFDMsystemsimulation.TheprocessofOFDMisfirstapplyInverseDiscreteFourierTransform(IDFT)tothesignalwithcyclicprefix(CP)added.Thenatthereceiverside,theCPneedstoberemovedbeforeapplyingDiscreteFourierTransform(DFT).AdiagonalmatrixHisnecessarytobeobtainedwhichmeanstoconvertISIintoparallelchannelstoeliminatetheinfluenceofISI.EventuallythesymbolerrorrateoftheOFDMsystemissimulated.ComparisonsandExtraSimulationsSomeextrasimulationsandcomparisonsarealsoofgreatimportanceandveryhelpfulinunderstandingthetheory.Theresultsinthefirsttwopartsofthereport—theSERinISIsinglecarrierchannelsandtheSERinOFDMsystems—willbecompared.InthelabweuseQPSKtogenerateasequenceofsymbols.Intheextrasimulation,acontrastsimulationontheSERofBPSKagainsttheSERofQPSKwillbefirstcarriedout.Next,differentnumbersoflambda(numberofpath)willbetestedtoseeitsaffectionontheSER,onISIchannelsandOFDMparallelchannelsinturn.CSMSimulationReportNINGLIOFDMInFadingChannels1407471542.DiscussionandResults2.1SignalModelforISIchannels2.1.1SimulationofL-RayEqual-PowerRayleighFadingChannelsInthefirstpartofthesimulation,aMATLABfunction“poissrnd(lambda)”isusedtogenerateaPoissonrandomvariableLasthenumberofraysofthefadingchannel,inwhichlambda(𝜆)isthemeanvalueofrays.AsLstandsforthenumberofmultipath,itisthendefinedtobenolessthan2.Here𝜆issetto5.Intheory,thelargertheLis,inanotherwordthemorepathsinthechannel,theserioustheISIis.Next,complexfadingcoefficients(α)aregeneratedforeachoftheLraychannels.RayleighdistributedamplitudeanduniformlydistributedphasearefulfilledbyuseofGaussianrandomvariable“randn()”inbothrealandimaginaryparts.Thecodeis:Alpha=(randn(1,L)+1j*randn(1,L))Weneedtonormalizethepowerofthechannel.Thetotalpower=E[α12]+E[α22]+…+E[α𝐿2].Ifthereisnonormalization,theresultwillbe2LwhereforeachE[α𝑛2]=2.Bydividing√2inbothrealandimaginaryparts,eachelementofαisnormalized——E[α𝑛2]=（𝑟𝑎𝑛𝑑𝑛√2）2+（j𝑟𝑎𝑛𝑑𝑛√2）2=12+12=1.However,fortheaveragetotalpowerofLrays,anadditional1√𝐿isneededinordertocompletethenormalization.Eventially,thecodeis:Alpha=(randn(1,L)+1j*randn(1,L))/sqrt(2);NormalA=Alpha/sqrt(L)AfterthetimedelayoftheLraymodelisgenerated,theresultantamplitudeandphaseofthechannelareplotted,showninFigure.2.1.Hereforconvenience,delayt(0)issetto0.(a)(b)Figure2.1:The(a):am