of 89/89
MEE09:58 CHANNEL ESTIMATION FOR LTE DOWNLINK Asad Mehmood Waqas Aslam Cheema This thesis is presented as part of Degree of Master of Science in Electrical Engineering Blekinge Institute of Technology September 2009 Blekinge Institute of Technology School of Engineering Department of Signal Processing Supervisor Prof. Abbas Mohammed Examiner Prof. Abbas Mohammed

CHANNEL ESTIMATION FOR LTE DOWNLINK

  • View
    221

  • Download
    3

Embed Size (px)

Text of CHANNEL ESTIMATION FOR LTE DOWNLINK

  • MEE09:58

    CHANNEL ESTIMATION FOR LTE

    DOWNLINK

    AsadMehmood

    WaqasAslamCheema

    ThisthesisispresentedaspartofDegreeofMasterofScienceinElectricalEngineering

    BlekingeInstituteofTechnology

    September2009

    Blekinge Institute of Technology School of Engineering Department of Signal Processing Supervisor Prof. Abbas Mohammed Examiner Prof. Abbas Mohammed

  • Abstract

    3GPPLTEistheevolutionoftheUMTSinresponsetoeverincreasingdemandsfor

    highqualitymultimediaservicesaccordingtousersexpectations.Sincedownlink

    isalwaysan important factor incoverageandcapacityaspects,specialattention

    hasbeengiveninselectingtechnologiesforLTEdownlink.Noveltechnologiessuch

    asorthogonalfrequencydivisionmultiplexing(OFDM)andmultipleinput,multiple

    output (MIMO), can enhance the performance of the current wireless

    communicationsystems.Thehighdataratesandthehighcapacitycanbeattained

    byusing theadvantagesof the two technologies.These technologieshavebeen

    selectedforLTEdownlink.

    Pilotassistedchannelestimationisamethodinwhichknownsignals,calledpilots,

    aretransmittedalongwithdatatoobtainchannelknowledgeforproperdecoding

    of received signals. This thesis aims at channel estimation for LTE downlink.

    ChannelestimationalgorithmssuchasLeastSquares(LS),MinimumMeanSquare

    Error(MMSE)havenbeenevaluatedfordifferentchannelmodelsinLTEdownlink.

    Performanceof these algorithmshasbeenmeasured in termsofBitErrorRate

    (BER)andSymbolErrorRate(SER).

  • ACKNOWLEGEMENT

    All praises and thanks to Almighty ALLAH, themost beneficent and themost

    merciful,who gave us the all abilities and helped us to complete this research

    work.

    We would like to express our sincere gratitude to our supervisor Prof. Abbas

    Mohammed for his support and guidance during the course of this work. His

    encouragementandguidancehasalwaysbeena sourceofmotivation forus to

    explore various aspects of the topic. Discussions with him have always been

    instructiveandinsightfulandhelpedustoidentifyourideas.

    Finally,weareverygratefultoourparents,brotherandsistersfortheirsacrifices,

    unremittingmotivationandeverlasting loveandtheircontinuoussupportduring

    ourstayinBTH.

  • 1

    Table of Contents

    Abstract..................................................................................................................................2

    List of Figures.......................................................................................................................4

    Chapter1Introduction..........................................................................................7

    1.1) Introduction...............................................................................................................7

    1.2) Objectives...................................................................................................................9

    1.3) OutLineoftheMasterThesis..................................................................................10

    Chapter2OverviewofLTEPhysicalLayer...........................................................11

    2.1)Introduction..................................................................................................................11

    2.2)ObjectivesofLTEPhysicalLayer...................................................................................12

    2.3)FrameStructure............................................................................................................12

    2.3.1)Type1FrameStructure.............................................................................................12

    2.3.2)Type2FrameStructure.............................................................................................13

    2.4)PhysicalResourceandSlotstructure............................................................................15

    2.5)LTEDownlinkReferenceSignalsStructure...................................................................16

    2.6)LTEDownlinkParameters............................................................................................17

    2.7)MultipleAntennaTechniques.......................................................................................18

    2.7.1)SpatialDiversity.........................................................................................................19

    2.7.1.1)ReceiveDiversity.....................................................................................................20

    2.7.1.2)TransmitDiversity...................................................................................................21

    2.7.1.3)CyclicDelayDiversity(CDD)....................................................................................23

    2.7.1.4)SpaceFrequencyBlockCoding(SFBC)....................................................................25

    2.7.2)SpatialMultiplexing...................................................................................................26

    2.7.3)Beamforming............................................................................................................28

    Chapter3LTEDownlinkSystemModel..............................................................32

  • 2

    3.1)Introduction..................................................................................................................32

    3.2)GeneralDescription......................................................................................................33

    CHAPTER4RadioPropagationModels....................................................................36

    4.2.1)LargeScalePropagationModel.................................................................................39

    4.2.2)MediumScalePropagationModel............................................................................40

    4.2.3)SmallScalePropagationModel.................................................................................40

    4.3)PropagationaspectsandParameters...........................................................................41

    4.3.1)DelaySpread..............................................................................................................42

    4.3.2)CoherenceBandwidth...............................................................................................42

    4.3.3)DopplerSpread..........................................................................................................43

    4.3.4)CoherenceTime.........................................................................................................43

    4.4)StandardChannelmodels............................................................................................44

    4.4.1)SISO,SIMOandMISOChannelModels.....................................................................44

    4.4.2)MIMOChannelModels..............................................................................................45

    4.4.3)EffectofSpatialCorrelationonMIMOPerformance................................................46

    4.4.4)ITUMultipathChannelModels..................................................................................48

    4.4.4.2)ITUVehicularA(V30,V120andV350)...............................................................50

    4.4.5)ExtendedITUmodels.................................................................................................51

    Chapter5ChannelEstimationinLTE..........................................................................54

    5.1)Introduction..................................................................................................................54

    5.2)SignalModel.................................................................................................................54

    5.3)PilotassistedChannelEstimation.................................................................................56

    5.3.1)LeastSquareEstimation............................................................................................57

    5.3.2)RegularizedLSEstimation..........................................................................................58

    5.3.3)DownSamplingMethod............................................................................................58

    5.3.4)MinimumMeanSquareEstimation...........................................................................59

    5.4)Equalization..................................................................................................................60

    5.5)PerformanceComparisonofChannelEstimation........................................................62

    Schemes...............................................................................................................................62

    Chapter6ChannelEstimationforMultipleAntennaSystems....................................69

  • 3

    6.1)Introduction..................................................................................................................69

    6.2)SFBCinLTE....................................................................................................................70

    6.3)ChannelEstimationandDecoding................................................................................71

    6.4)NumericalResultsandPerformanceanalysis...............................................................73

    Chapter7Conclusions...................................................................................................78

    References...........................................................................................................................80

  • 4

    List of Figures

    Figure2.1:Framestructureoftype1(Tsisexpressingbasictimeunitcorrespondingto

    30.72MHz)............................................................................................................................13

    Figure2.2:Framestructuretype2(for5msswitchpointperiodicity)..............................14

    Figure2.3:DownlinkResourcegrid[3]................................................................................15

    Figure2.4:AllocationofReferenceSymbolsfortwoantennatransmissions......................17

    Figure2.5:Receivediversityconfigurations........................................................................21

    Figure2.6:Transmitdiversityconfigurations......................................................................23

    Figure2.7:CDDfortwoantennaconfiguration..................................................................24

    Figure2.8:SpaceFrequencyBlockCodingSFBCassumingtwoantennas..........................26

    Figure2.9:22AntennaConfiguration(HereM=N=2).......................................................28

    Figure2.10a:Classicalbeamformingwithhighmutualantennacorrelation....................30

    Figure2.10b:Precoderbasedbeamformingincaseoflowmutualantennacorrelation.31

    Figure3.1LTEDownlinksystemmodelwith22MIMO.....................................................33

    Figure4.1:Signalpropagationthroughdifferentpathsshowingmultipathpropagation

    phenomena..........................................................................................................................37

    Figure4.2:Powerdelayprofileofamultipathchannel.......................................................37

    Figure4.3:Receivedsignalpowerlevelofatimevaryingmultipathpropagation.............39

    channel.................................................................................................................................39

    Figure4.4:Theincreaseinthecapacitywiththeincreaseinnumberofantennas.The.....47

    capacityofthesystemincreaseslinearlywithincreasingnumberofantennas................47

    Figure4.5:BERcurvesfor2x2MIMOsystemsusingflatfadingRayleighchannelwith

    differentcorrelationvalueswhichshowthatBERdecreaseswithlowcorrelationvalues..48

    Figure4.6:ChannelImpulseResponsesaccordingtoITUstandardswhicharetobeusedin

    simulationsforchannelestimationofLTE...........................................................................50

    Figure5.1:EqualizeroptionsforLTE,intimedomainandfrequencydomain....................61

    Figure5.2:BERperformanceofLTEtransceiverfordifferentchannelsusingQPSK

    modulationandLMMSEchannelestimation.......................................................................64

    Figure5.3:SERperformanceofLTEtransceiverfordifferentchannelmodelsusingQPSK

    modulationandLMMSEchannelestimation.......................................................................65

  • 5

    Figure5.4:BERperformanceofLTEtransceiverfordifferentchannelmodelsusingQPSK

    modulationandLSestimation.............................................................................................65

    Figure5.5:SERperformanceofLTEtransceiverfordifferentchannelmodelsusingQPSK

    modulationandLSchannelestimation................................................................................66

    Figure5.6:BERperformanceofLTEtransceiverfordifferentchannelmodelsusing16QAM

    modulationandLMMSEchannelestimation.......................................................................66

    Figure5.7:SERperformanceofLTEtransceiverfordifferentchannelmodelsusing16QAM

    modulationandLMMSEchannelestimation.......................................................................67

    Figure5.8:SERperformanceofLTEtransceiverfordifferentchannelmodelsusing16QAM

    modulationandLSchannelestimation................................................................................67

    Figure5.9:SERperformanceofLTEtransceiverfordifferentchannelmodelsusing16QAM

    modulationandLSchannelestimation................................................................................68

    Figure6.1:Systemmodelforsimulationof2x2MIMOOFDMusingSFBC.........................71

    Figure6.2:BERperformanceofLTEtransceiverwithmultipleantennasforITUPedestrian

    AchannelmodelusingQPSKmodulationandLMMSEchannelestimation........................74

    Figure6.3:SERperformanceofLTEtransceiverwithmultipleantennasforITUPedestrian

    AchannelmodelusingQPSKmodulationandLMMSEchannelestimation........................75

    Figure6.4:BERperformanceofLTEtransceiverwithmultipleantennasforITU................75

    PedestrianAchannelmodelusingQPSKmodulationandLMMSEchannelestimation......75

    Figure6.5:SERperformanceofLTEtransceiverwithmultipleantennasforITUPedestrian

    AchannelmodelusingQPSKmodulationandLMMSEchannelestimation........................76

    Figure6.6:BERperformanceofLTEtransceiverwithmultipleantennasforITUVehicularA

    channelmodelusing4QAMmodulationandLMMSEchannelestimation........................76

    Figure6.7:SERperformanceofLTEtransceiverwithmultipleantennasforITUVehicularA

    channelmodelusing4QAMmodulationandLMMSEchannelestimation........................77

  • 6

    List of Tables Table2.1:UplinkDownlinkConfigurationsforLTETDD..14

    Table2.2:LTEDownlinkParameters.........18

    Table4.1:AveragePowersandRelativeDelaysofITUMultipathChannelModelsfor

    PedestrianAandPedestrianBcases49

    Table4.2:AveragePowersandRelativeDelaysforITUVehicularATestEnvironment51

    Table4.4.1:PowerDelayProfileforExtendedITUPedestrianAModel..52

    Table4.4.2:PowerDelayProfileforExtendedITUVehicularAModel.52

    Table4.4.3:PowerDelayProfileforExtendedTypicalUrbanModel.53

  • 7

    Chapter1Introduction

    1.1) Introduction

    During the last decade along with continued expansion of networks and

    communications technologies and the globalization of 3rdGenerationofMobile

    Communication Systems, the support for voice and data services have

    encounteredagreaterdevelopmentcomparedto2ndGenerationSystems.Atthe

    sametimetherequirementsforhighqualitywirelesscommunicationswithhigher

    datarates increasedowingtousersdemands.Ontheotherhand,theconflictof

    limited bandwidth resources and rapidly growing numbers of users becomes

    exceptional,sothespectrumefficiencyofsystemshouldbeimprovedbyadopting

    some advanced technologies. It has been demonstrated in both theory and

    practice that some novel technologies such as orthogonal frequency division

    multiplexing (OFDM) andmultiple input,multiple output (MIMO) systems, can

    enhance the performance of the currentwireless communication systems. The

    highdataratesandthehighcapacitycanbeattainedbyusingtheadvantagesof

    the two technologies. From a standardization perspective 3G era is nowwell

    advanced.Whileenhancements continue tobemade to leverage themaximum

    performance from currentlydeployed systems, there is abound to the level to

    whichfurther improvementswillbeeffective.Iftheonlypurposeweretodeliver

    superiorperformance, then this in itselfwouldbe relativelyeasy toaccomplish.

    The added complexity is that such superior performance must be delivered

    throughsystemswhicharecheaper from installationandmaintenanceprospect.

    Users have experienced an incredible reduction in telecommunications charges

    and theynowanticipate receivinghigherquality communication servicesat low

  • 8

    cost.Therefore,indecidingthesubsequentstandardizationstep,theremustbea

    dualapproach;insearchofsubstantialperformanceenhancementbutatreduced

    cost.LongTermEvolution (LTE) is thatnext stepandwillbe thebasisonwhich

    futuremobile telecommunications systemswill be built. LTE is the first cellular

    communication system optimized from the outset to support packetswitched

    dataservices,withinwhichpacketizedvoicecommunicationsarejustonepart.

    The 3rd Generation Partnership Project (3GPP) started work on Long Term

    Evolution in 2004 with the description of targets illustrated in [1]. The

    specifications associated to LTE are formally identified as the evolved UMTS

    terrestrialradioaccessnetwork(EUTRAN)andtheevolvedUMTSterrestrialradio

    access (EUTRA). These are collectively referred to by the project name LTE. In

    December 2008, release 8 of LTE has been approved by 3GPPwhichwill allow

    network operators to appreciate their deployment plans in implementing this

    technology. A few motivating factors can be identified in advancing LTE

    development; enhancements inwire line capability, the requirement for added

    wirelesscapacity, theneed forprovisionofwirelessdataservicesat lowercosts

    and the competition to the existing wireless technologies. In addition to the

    continued advancement in wire line technologies, a similar development is

    required for technologies to work fluently with defined specifications in the

    wirelessdomain.3GPPtechnologiesmustmatchandgobeyondthecompetition

    with other wireless technologies which guarantee high data capabilities

    including IEEE802.16.To takemaximumadvantageofavailable spectrum, large

    capacity is an essential requirement. LTE is required to provide superior

    performancecomparedtoHighSpeedPacketAccess(HSPA)technologyaccording

    to 3GPP specifications. The 3GPP LTE release 8 specification defines the basic

    functionality of a new, highperformance air interface providing high user data

    rates in combination with low latency based on MIMO, OFDMA (orthogonal

    frequency division multiple access), and an optimized system architecture

  • 9

    evolution (SAE) asmain enablers. The LTE solutionprovides spectrum flexibility

    withscalabletransmissionbandwidthbetween1.4MHzand20MHzdependingon

    the available spectrum for flexible radio planning. The 20MHz bandwidth can

    provideup to150Mbpsdownlinkuserdata rateand75Mbpsuplinkpeakdata

    ratewith22MIMO,and300Mbpswith44MIMO.Asummaryofrelease8

    canbefoundin[2].

    1.2) Objectives

    Indeciding the technologies to comprise in LTE,oneof the key concerns is the

    tradeoffbetweencostof implementationandpracticaladvantage.Fundamental

    to this assessment, therefore, has been an enhanced understanding different

    scenariosoftheradiopropagationenvironmentinwhichLTEwillbedeployedand

    used.

    Theeffectofradiopropagationconditionsonthetransmittedinformationmustbe

    estimated inorder to recover the transmitted informationaccurately.Therefore

    channelestimationisavitalpartinthereceiverdesignsofLTE.Inthisthesiswork,

    a detailed study of standard channel models based on ITU and 3GPP

    recommendations for LTE has been done. The main focus of the work is to

    investigate and evaluate the channel estimation techniques such asMinimum

    Mean SquareChannelEstimation, Least SquareChannelEstimation and Down

    SampledChannel ImpulseResponse Least Square Estimation for LTE down link.

    Thereforea link levelsimulatorbasedonLTEphysical layerspecifications[3]has

    been presented. This simulator emulates channel estimation algorithms for

    standard channel models defined for LTE, using MIMOOFDM and multilevel

    modulation schemes in LTE down link between the eNodeB and the user

    equipment(UE).Theperformanceofthelinklevelsimulatorismeasuredinterms

  • 10

    of bit error rate (BER) and symbol error rate (SER) averaged over all channel

    realizationsofdifferentpropagationenvironments.

    1.3) OutLineoftheMasterThesis

    Thisthesisworkisdividedintosevenchapters:

    Chapter1abouttheintroductionofLTEdescribingthebackground,theroleofthetechnologyinthepresentmobilecommunicationsystemsandthemotivationofthismasterthesis.

    Chapters2givesdetailsaboutLTEAir Interface featuresdescribingLTEdownlinkfamestructureandthetransmissiontechniquesusedinLTE.

    In Chapter 3, block diagram of systemmodel used in the simulation of LTEdownlinkphysicallayerispresented.

    Chapter 4 gives details of radio propagation models for LTE. The chapterdescribes the basics ofmultipath channelmodeling following standard channelmodelsforUMTSandLTEincludingSISOandMIMOchannelmodelsbasedonITUrecommendations.

    Chapter 5 evaluates the channel estimation algorithms including MinimumMean SquareChannelEstimation, Least SquareChannelEstimation and DownSampledChannelImpulseResponseLeastSquareEstimationusingchannelmodelsdescribedinChapter4forSingleInputSingleOutput(SISO)systems.

    In chapter 6 the channel estimation algorithms are evaluated for MIMOsystems.

    Chapter 7 goes over the main points of this thesis work with concludingremarksandproposes futurework thatcanbedonewith the simulatorused inthisthesisinordertocontinueinvestigationwithinLTEAirInterface.

  • 11

    Chapter2OverviewofLTEPhysicalLayer

    2.1)Introduction

    As compared to previous used cellular technologies like UMTS (universalmobile

    technology systems)orhigh speeddownlinkpacketaccess (HSDPA), thePhysical

    Layer of LTE is designed to deliver high data rate, low latency, packetoptimized

    radio access technology and improved radio interface capabilities. Wireless

    broadband internet access and advanced data services will be provided by this

    technology.

    LTE physical Layer will provide peak data rate in uplink up to 50Mb/s and in

    downlinkupto100Mb/swithascalabletransmissionbandwidthrangingfrom1.25

    to20MHztoaccommodatetheuserswithdifferentcapacities.Forthefulfillments

    oftheaboverequirementschangesshouldbemadeinthephysicallayer(e.g.,new

    codingandmodulationschemesandadvancedradioaccesstechnology).Inorderto

    improvethespectralefficiencyindownlinkdirection,OrthogonalFrequencyDivision

    MultipleAccess (OFDMA), togetherwithmultipleantenna techniques isexploited.

    Inaddition,tohaveasubstantialincreaseinspectralefficiencythelinkadaptionand

    frequencydomain scheduling are exercised to exploit the channel variation in

    time/frequencydomain. LTEair interfaceexploitsboth timedivisionduplex (TDD)

    andfrequencydivisionduplex(FDD)modestosupportunpairedandpairedspectra

    [4,5]. The transmission scheme used by LTE for uplink transmission is SCFDMA

    (SignalCarriesFrequencyDivisionMultipleAccess).Formoredetaileddescriptionof

    LTEphysicallayercoveringuplinkanddownlinkinsee[6,7].

  • 12

    2.2)ObjectivesofLTEPhysicalLayerTheobjectivesofLTEphysicallayerare;thesignificantlyincreasedpeakdatarates

    up to 100Mb/s in downlink and 50Mb/s in uplinkwithin a 20MHz spectrum

    leadingtospectrumefficiencyof5Mb/s,increasedcelledgebitratesmaintainsite

    locationsasinWCDMA,reduceduserandcontrolplanelatencytolessthan10ms

    and less than 100ms, respectively [8], to provide interactive realtime services

    suchashighqualityvideo/audioconferencingandmultiplayergaming,mobilityis

    supportedforupto350km/horevenupto500km/handreducedoperationcost.

    It also provides a scalable bandwidth 1.25/2.5/5/10/20MHz in order to allow

    flexible technology to coexist with other standards, 2 to 4 times improved

    spectrum efficiency the one in Release 6 HSPA to permit operators to

    accommodate increased number of customerswithin their existing and future

    spectrumallocationwithareducedcostofdeliveryperbitandacceptablesystem

    andterminalcomplexity,costandpowerconsumptionandthesystemshouldbe

    optimizedforlowmobilespeedbutalsosupporthighmobilespeedaswell.

    2.3)FrameStructure

    TwotypesofradioframestructuresaredesignedforLTE:Type1framestructure

    is applicable to FrequencyDivisionDuplex (FDD) and type2 frame structure is

    relatedtoTimeDivisionDuplex(TDD).LTEframestructuresaregivenindetailsin

    [3].

    2.3.1)Type1FrameStructure Type1 framestructure isdesigned for frequencydivisionduplexand isvalid for

    bothhalfduplexand fullduplexFDDmodes.Type1 radio framehasaduration

    10msandconsistsofequallysized20slotseachof0.5ms.Asubframecomprises

    twoslots, thusone radio framehas10subframesas illustrated in figure2.1. In

  • 13

    FDDmode,halfofthesubframesareavailablefordownlinkandtheotherhalfare

    availableforuplinktransmissionineach10msinterval,wheredownlinkanduplink

    transmissionareseparatedinthefrequencydomain[2].

    Figure2.1: Frame structureof type1(Ts isexpressingbasic timeunitcorresponding to30.72MHz)

    2.3.2)Type2FrameStructure Type2 framestructure is relevant forTDD; the radio frame iscomposedof two

    identicalhalfframeseachonehavingdurationof5ms.Eachhalfframe isfurther

    divided into5subframeshavingdurationof1msasdemonstrated in figure2.2.

    Twoslotsof length0.5msconstituteasubframewhich isnotspecialsubframe.

    ThespecialtypeofsubframesiscomposedofthreefieldsDownlinkPilotTimeslot

    (DwPTS), GP (Guard Period) and Uplink Pilot Timeslot (UpPTS). Seven uplink

    downlink configurations are supported with both types (10ms and 5ms) of

    downlinktouplinkswitchpointperiodicity.In5mdownlinktouplinkswitchpoint

    periodicity,specialtypeofsubframesareusedinbothhalfframesbutitisnotthe

    case in10msdownlinktouplink switchpointperiodicity, special frameareused

    onlyinfirsthalfframe.Fordownlinktransmissionsubframes0,5andDwPTSare

    always reserved. UpPTS and the subframe next to the special subframe are

    always reserved for uplink communication [3]. The supporting downlinkuplink

    configurationisshownintable2.1whereUandDdonatethesubframesreserved

    Sub-frame 0 (1ms) Sub-frame 9(1ms)

    42 3 195 201

    One radio frame, Tf = 307200Ts = 10 ms

    One slot, Tslot = 15360Ts = 0.5

  • 14

    foruplinkanddownlink, respectively,andSdenotes the reservedsubframesas

    illustratedintable2.1.

    Figure2.2:Framestructuretype2(for5msswitchpointperiodicity)

    Table2.1UplinkDownlinkconfigurationsforLTETDD[3]

    UplinkDownlinkconfiguration

    DownlinktouplinkSwitchpointperiodicity

    SubframeNumbers

    0 1 2 3 4 5 6

    7

    8

    9

    0 5ms D

    S

    U U U D S

    U U U

    1 5ms D

    S

    U U D D S

    U U D

    2 5ms D

    S

    U D D D S

    U D D

    3 10ms D

    S

    U U U D D D D D

    4 10ms D

    S

    U U D D D D D D

    5 10ms D

    S

    U D D D D D D D

    6 5ms D

    S

    U U U D S

    U U D

    T=1msOne slot, 0.5ms

    Half-frame#1(5ms) Half-frame#2(5ms)

    Sub-frame#0 DwPT GP UpPTS UpPTS GP DwPT

    One radio-frame# Tf=307200Ts=10ms

    Subframe#5 Subframe#9Subframe#6 Subframe#2 Subframe#4Subframe

  • 15

    2.4) PhysicalResourceandSlotstructureIn each available slot the transmitted signal can be seen as a timefrequency

    recourse gird,where each recourse element (RE) corresponds tooneOFDM sub

    carrier during OFDM symbol interval. The number of subcarriers is being

    determinedby the transmissionbandwidth.Fornormalcyclicprefix (CP)eachslot

    contains seven OFDM symbols and in case of extended cyclic prefix, 6 OFDM

    symbolsareslottedin ineachtimeslot.Thedifferent lengthsofCParemention in

    Table2.1. InthisworkwehaveusedtwotypesofCP lengths,shortandextended,

    withType1framestructure.

    k=0 l=0

    Figure2.3:DownlinkResourcegrid[3]

    12subcarriers

    7OFDMSymbols

    ResourceBlockRBof84RE

    ResourceElementRE

    (k,l)

    Oneslot0.5ms

  • 16

    InLTEdownlinkaconstantsubcarriersspacingof15kHz isutilized.Infrequency

    domain, 12 subcarriers are grouped together to form a Resource Block (RB)

    occupyingtotal180kHzinoneslotdurationasillustratedinfigure2.3.Incaseof

    shortCP,lengtharesourceblockcontains84resourceelements(RE)andforlong

    CP the number of RE is 74. Formultiple antenna schemes, therewill be one

    resourcegirdperantenna[3,9].Forallavailablebandwidths,thesizeofresource

    blocksisthesame.

    2.5)LTEDownlinkReferenceSignalsStructure

    InordertocarryoutcoherentdemodulationinLTEdownlink,channelestimation

    is needed at the receiver end. In case ofOFDM transmission known reference

    symbolsareaddedintotimefrequencygridforchannelestimation.Thesesignals

    are called LTE Downlink Reference signals [10]. For time domain, reference

    symbols are slottedin in the first and the third last elements of resource grid,

    where as reference signals are insertedoverevery six subcarriers in frequency

    domain.Foranaccuratechannelestimationoverentiregirdandreducingnoisein

    channel estimates, a twodimensional timefrequency interpolation/averaging is

    required overmultiple reference symbols. One reference signal is transmitted

    fromeach antenna toestimate the channelquality corresponding toeachpath

    when amultiple antenna scheme is applied. In this case, reference signals are

    mappedondifferentsubcarriesofresourcegridfordifferentantennastorefrain

    from interference. Resource elements used to transmit reference signals from

    antenna1 arenot reusedon antenna2 for data transmission; theseplaces are

    filled with zeros. Allocation of these reference symbols is shown in figure 2.4

    [4,10].

  • 17

    Figure2.4:AllocationofReferenceSymbolsfortwoantennatransmissions

    2.6)LTEDownlinkParameters

    AsmentionedinabovesectionthatLTEropesscalablebandwidth,sothenumber

    of subcarries also changes while keeping subcarriers spacing up to 15 kHz.

    Dependingonthedelayspread,twoCPlengths(shortandextended)areallowed.

    Italsoaimsat supportingdifferent scenarios, indoor,urban, suburbanand rural

    forbothkindsofmobility conditions (LowandHigh)ofmobile terminal ranging

    from350Km/hto500km/h.WhileusingFDDconfigurationsameframestructure

    Antenna 1

    Reference Signal

    Not used

    Antenna 2

    Frequency

    Time

  • 18

    and parameters are used in uplink and downlink. These parameters are

    summarizedinTable2.2[11].

    Transmission BW

    1.25 MHz

    2.5 MHz

    5 MHz

    10 MHz

    15 MHz

    20MHz

    Sub-carrier Duration

    Tsub

    0.5ms

    Sub-carrier Spacing

    fspace

    15kHz

    Sampling Frequency

    fs

    1.92 MHz

    3.84MHz

    7.68 MHz

    15.36MHz

    23.04 MHz

    30.72 MHz

    FFT and NIFFT

    128

    256

    512

    1042

    1536

    2048

    Number of occupied sub--carries

    NBW

    75

    150

    300

    600

    900

    1200

    Number of OFDM symbols per

    Sub-frame Short/Long (CP)

    7/6

    CP Length (s / sample)

    short

    (4.69/9)6 (5.21/10)1

    (4.69/18)6 (5.21/20)1

    (4.69/18)6(5.21/40)1

    (4.69/72)6 (5.21/80)1

    (4.69/108)6 (5.21/120)1

    (4.69/144)6 (5.21/160)1

    Long

    (16.67/32)

    (16.67/64)

    (16.67/128)

    (16.67/256)

    (16.67/384)

    (16.67/512)

    Table2.2LTEDownlinkParameters

    2.7)MultipleAntennaTechniques

    Broadly,multipleantennatechniquesutilizemultipleantennasatthetransmitter

    or/and receiver incombinationwithadaptivesignalprocessing toprovidesmart

    arrayprocessing,diversitycombiningorspatialmultiplexingcapabilityofwireless

    system [10,12].Previously, inconventional signalantenna systems theexploited

    dimensions are only time and frequency whereas multiple antenna systems

    exploitanadditional spatialdimension.Theutilizationof spatialdimensionwith

    multipleantenna techniques fulfills the requirementsofLTE; improvedcoverage

  • 19

    (possibility for larger cells), improved system capacity (moreuser/cell),QoSand

    targeted date rates are attained by using multiple antenna techniques as

    described in [13].Multiple antenna techniques are the integrated part of LTE

    specificationsbecausesomerequirementssuchasuserpeakdataratescannotbe

    achievedwithouttheutilizationofmultipleantennaschemes.

    The radio link is influenced by the multipath fading phenomena due to

    constructive and destructive interferences at the receiver. By applyingmultiple

    antennasatthetransmitteroratthereceiver,multipleradiopathsareestablished

    betweeneachtransmittingandreceivingantenna.Inthiswaydissimilarpathswill

    experienceuncorrelated fading. To have uncorrelated fading paths, the relative

    locationofantennasinthemultipleantennaconfigurationsshouldbedistantfrom

    each other. Alternatively, for correlated fading (instantaneous fading) antenna

    arraysarecloselyseparated.Whetheruncorrelatedfadingorcorrelatedfading is

    required depends on what is to be attained with the multiple antenna

    configurations (diversity, beamforming, or spatialmultiplexing) [10].Generally,

    multiple antenna techniques can be divided into three categories (schemes)

    dependingontheirdifferentbenefits;spatialdiversity,beamformingandspatial

    multiplexingwhichwillbediscussedfurtherinthefollowingsections.

    2.7.1)SpatialDiversity

    Conventionally,themultipleantennasareexercisedtoachieveincreaseddiversity

    to encounter the effects of instantaneous fading on the signal propagating

    throughthemultipathchannel.Thebasicprinciplebehindthespatialdiversity is

    thateachtransmitterandreceiverantennapairestablishesasinglepathfromthe

    transmittertothereceivertoprovidemultiplecopiesofthetransmittedsignalto

    obtain an improvedBERperformance [14]. Inorder to achieve large gainswith

    multiple antennas there should be low fading correlation between the

  • 20

    transmittingandthereceivingantennas.Lowvalueofcorrelationcanbeachieved

    when interantennaspacing iskept large.It isdifficulttoplacemultipleantennas

    on amobile device due size restrictions depending upon the operating carrier

    frequency. An alternative solution is to use antenna arrays with cross

    polarizations, i.e., antenna arrayswith orthogonal polarizations. The number of

    uncorrelatedbranches(paths)availableatthetransmitteroratthereceiverrefers

    tothediversityorderandthe increase indiversityorderexponentiallydecreases

    with the probability of losing the signal. To achieve spatial diversity for the

    enhancementofconvergeorlinkrobustnessmultipleantennascanbeusedeither

    at the transmitter side or at the receiver side.We will discuss both transmit

    diversitywheremultiple antennas are used at the transmitter (MISOmultiple

    input signaloutput),and receivediversityusingmultiple receiveantenna (SIMO

    signalinput multipleoutput). On the other hand, MIMO channel provides

    diversityaswellasadditionaldegreeoffreedomforcommunication.

    2.7.1.1)ReceiveDiversity Thereceivediversity isthemoststraightforwardandcommonlyutilizedmultiple

    antenna configuration which relies on the use of 2 (number of receive

    antennas)antennasat the receiver toachieve spatialdiversity.Receiveantenna

    diversity will improve system performance when the signals from different

    antennas are optimally combined in such a way that the resulting signal

    demonstrates a reduced amplitude variations when compared to the signal

    amplitudefromanyoneantenna.Herediversityorder isequaltothenumberof

    receiveantennasinSIMOconfigurationanditcollectsmoreenergyatthereceiver

    to improvesignal tonoiseratioascompared toSISO (signal inputsignaloutput)

    configuration. Receive diversity configuration is depicted in figure 2.5. Two

    different combining methods are used to implement receive diversity (e.g.,

    selectioncombiningandgaincombining).Inselectioncombining,thebranchwith

  • 21

    thehighestSNR isselectedbythecombiner fordetectionand ingaincombining

    thelinearcombinationofallbranchesisusedfordetection[16].

    Figure2.5:Receivediversityconfigurations

    Receivediversity isanuncomplicatedpossibilitytoboostthe linkreliabilitybut it

    becomes limited when the size of the receiver is small. In mobile radio

    communications the sizeof cellphones isbecoming smallerand smaller so it is

    very hard task to place several antennaswith enough spacing on such a small

    device to achieve uncorrelated channels [17]. For further study on receive

    diversityanddifferentcombiningmethodsrelatedtoSIMOsystemscanbefound

    in[10,15,16,17,18].

    2.7.1.2)TransmitDiversity The transmit diversity scheme relies on the use of 2 antennas at the

    transmitter side in combination with precoding in order to achieve spatial

    diversitywhentransmittingasingledatastream[4,19].Usuallytransmitdiversity

    necessitates theabsolutechannel informationat the transmitterbut itbecomes

    feasible to implement transmit diversitywithout the knowledge of the channel

    TX

    RX

  • 22

    with space time block coding [19]. The simplest of the diversity techniques is

    Alamoutispacetimecoding(STC)scheme[20].Transmitdiversityconfigurationis

    illustratedinfigure2.6.

    The use of transmit diversity is common in the downlink of cellular systems

    becauseitischeaperandeasytoinstallmultipleantennasatbasestationthanto

    putmultipleantennasateveryhandhelddevice. In transmitdiversity tocombat

    instantaneous fadingandtoachieveconsiderablegain in instantaneousSNR,the

    receiver isbeingprovidedwithmultiple copiesof the transmitted signal.Hence

    transmit diversity is applied to have extended converge and better link quality

    whentheusersexperienceterriblechannelconditions.

    InLTE,transmitdiversity isdefinedonlyfor2and4transmitantennasandthese

    antennasusuallyneedtobeuncorrelatedtotakefulladvantageofdiversitygain.

    Asdiscussedinsection1,toachievelowmutualcorrelationbetweensignalpaths

    (channels) antennas have to be sufficiently separated relative to the carrier

    wavelengthorshouldhavedifferentpolarizations[9].

    LTEphysical layersupportsbothopen loopandclosed loopdiversityschemes. In

    open loop scheme channel state information (CSI) is not required at the

    transmitter, consequentlymultipleantennas cannotprovidebeamforming,only

    diversitygaincanbeachieved.Ontheotherhand,closed loopschemedoesnot

    entailchannel state information (CSI)at transmitterand itprovidesboth spatial

    diversityandbeamformingaswell.

    Byemployingcyclicdelaydiversityandspace frequencyblockcoding,open loop

    transmitdiversity canbe accomplished in LTE. In addition, LTE also implements

    closelooptransmitdiversityschemessuchasbeamforming.

  • 23

    Figure2.6:Transmitdiversityconfigurations

    2.7.1.3)CyclicDelayDiversity(CDD)

    In delay diversity, delayed replicas of the same signal are transmitted from

    differentantennasatthebasestationinordertoachievediversity.Ifthechannel

    is not time dispersive in itself, delay diversity is used to transform antenna

    diversity into frequency diversity. The use of delay diversity is completely

    transparent to themobile terminal,which only observes single radio linkwith

    supplementary frequency selectivity [9,21]. Cyclic delay diversity (CDD) is the

    edition of generalized delay diversity (GDD)which is used to increase channel

    frequency selectivity as seen by the receiver in OFDM system [22]. The CDD

    operatesblockswisethatswhyitisapplicableforOFDMbasedsystem1.Incyclic

    delaydiversitycircularlydelayedcopiesofthe identicalsetofOFDMsymbolson

    the identical set of OFDM subcarriers are transmitted from different transmit

    antennas. In order to attain cyclic delay (instead of linear delay) over the Fast

    FourierTransform(FTT)size,adelayisintroducedbeforetheCP.Thecyclicdelay

    is added before the CP so that any value of delay can be employed without

    1OFDMisablockbasetransmissionscheme

    RX

    TX

  • 24

    changing the delay spread of the channel. Accumulation of cyclic time delay is

    equivalent to applying the phase shift in frequency domain before OFDM

    modulation as demonstrated in figure 7bwhere Sj corresponds to the complex

    modulatedsymbolswhicharemappedon1stantennaandphaseshiftedversionof

    the samemodulated symbolsaremappedon2ndantenna.Theuseof the same

    delay toall subcarriers in timedomain results in linearly increasingphase shift

    across the subcarriers with increasing subcarriers frequency. Different sub

    carriers followdifferentspatialpathsthroughmultipathchannelwhichgivesrise

    diversityeffectandincreasedfrequencyselectivity.ForfurtherreadingonCDD[9,

    16,22]. The general principle of CDD for two antenna configuration can be

    depictedinfigure2.7a[4].

    (a)

    (b)

    Figure2.7:CDDfortwoantennaconfiguration

    RX

    Cyclic Shift

    Antenna 1

    Antenna 2

    Antenna2 S0 S1ej Sn2e

    j (n2) Sn1e

    j(n1)

    Antenna1 S0 S1 Sn2 Sn1

    Frequency (OFDM subcarriers)

  • 25

    2.7.1.4)SpaceFrequencyBlockCoding(SFBC)

    InLTE,transmitdiversity is implementedbyusingSpaceFrequencyBlockCoding

    (SFBC). SFBC is a frequency domain adaptation of renowned Spacetime Block

    Coding(STBC)whereencodingisdoneinantenna/frequencydomainsratherthan

    inantenna/timedomains.STBCisalsorecognizedasAlamouticoding[23].Sothis

    SFBC is merely appropriate to OFDM and other frequency domain based

    transmissionschemes.

    TheadvantageofSFBCoverSTBC isthat inSFBCcoding isdoneacrossthe sub

    carrierswithinthe intervalofOFDMsymbolwhileSTBCappliescodingacrossthe

    numberofOFDM symbolsequivalent tonumberof transmitantennas [23].The

    implementation of STBC is not clearcut in LTE as it operates on the pairs of

    adjacentsymbolsintimedomainwhileinLTEnumberofavailableOFDMsymbols

    in a subframe is often odd. The operation of SFBC is carried out on pair of

    complexvaluedmodulationsymbols.Hence,eachpairofmodulationsymbolsare

    mappeddirectlytoOFDMsubcarriersoffirstantennawhilemappingofeachpair

    ofsymbolstocorrespondingsubcarriersofsecondantennaarereverselyordered,

    complexconjugatedandsignedreversedasshowninfigure2.8.

    Forappropriatereception,mobileunitshouldbenotifiedaboutSFBCtransmission

    and linear operation has to be applied to the received signal. The dissimilarity

    between CDD and SFBC lies in how pairs of symbols are mapped to second

    antenna. Contrarily to CDD, SFBC grants diversity onmodulation symbol level

    while CDDmust rely on channel coding in combinationwith frequency domain

    interleavingtoprovidediversityinthecaseofOFDM[4,10].

    The symbols transmitted from two transmitted antennas on every pair of

    neighboringsubcarriersarecharacterizedin[9]asfollows:

  • 26

    1 12 2

    2.1

    wherex k denotesthesymbolstransmittedfromantennaport ponthekth

    subcarrier.Thereceivedsymbolcanbeexpressedasfollows:

    2.2

    2.3

    where channelresponseofatsymbol transmittedfromantenna ,and

    istheadditivewhiteGaussiannoise.

    Figure2.8:SpaceFrequencyBlockCodingSFBCassumingtwoantennas

    2.7.2)SpatialMultiplexingTheuseofmultipleantennasatboththetransmitterandthereceivercanbenefit

    frommultipath fading to provide additional diversity and to improve signalto

    noiseratiocomparedtoSISOsystems.Thisadvantageofmultipleantennascanbe

    used to provide higher data rates by efficient utilization of SNR over the air

    Space

    Frequency

    Space

    Frequency (OFDM subcarriers)

    S1* S0

    * Sn+1 Sn

    *

    S0 S1 Sn Sn+1

    Frequency domain OFDM symbol

    OFDM modulation

    OFDM modulation

    RX

  • 27

    interface,bythetechniquesocalledspatialmultiplexing.Spatialmultiplexingcan

    provide substantial increase indata ratesby transmittingdifferentdata streams

    over different parallel channels provided by themultiple transmit and receive

    antennas, while using the same bandwidth and with no additional power

    expenditure.ItisonlypossibleinMIMOchannels[24].InMIMOsystems,increase

    incapacityislinearlyrelatedtothenumberofthetransmit/receiveantennapair.

    Consider aMIMO system withM transmit and N receive antennas, the radio

    channel for this system will consist of MN ideally uncorrelated paths, as

    illustrated in Figure 2.9. This configuration offers , parallel

    channels that permit simultaneously transmission of L data streams and the

    receiversignaltonoiseratiocanbemadetoincreaseinproportiontotheproduct

    M . A singlebit stream is split into twohalfratebit streams,modulatedand

    transmittedsimultaneously frombothantennaswhichcancause interference to

    each other at the receiver, and hence spatially multiplexed streams are

    overlappeddue topropagation through themultipathchannel.Thereforeat the

    receiver side inference cancellation is employed to separate the different

    transmittedsignals.Forspatialmultiplexingtechniqueseveraldecodingalgorithms

    are developed for interference cancellation for the narrowband frequency flat

    fadingcase. Ina lowcomplexityreceiver,MMSE technique isemployedwhich is

    discussed indetails inchapter5.AccordingtoFigure2.9,receivedsignalscanbe

    expressedasin[10]:

    . 2.4

    whereHis22channelmatrixsrepresentsthetransmitteddatasymbolsandnis

    22noisevector.

    The spatialsignaturesof the two signalsarewell separatedunder the favorable

    channel circumstances. The receiver having the knowledge about channel

  • 28

    statistics can differentiate and recover data symbols and . Assuming no

    noise, theestimatesofdatasymbolss ands from the receivedsignaland the

    channelcoefficientscanbeobtainedasfollows:

    2.5

    Where

    All substreams aremultiplexed into original symbols stream after decoding. In

    addition,differentparallelchannelscanbemadeindependentofeachotherifsome

    knowledgeofthechannel isavailableatthetransmitter,thusbyusingclosedloop

    spatial multiplexing interfering signals at the receiver are significantly reduced.

    Furtherdetailsaboutspatialmultiplexingcanbefoundin[4,9,10,18].

    Figure2.9:22AntennaConfiguration(HereM=N=2)

    2.7.3)Beamforming

    Ingeneral,beamformingcanbedefinedastheshapingofoverallantennabeam

    inthedirectionoftargetreceivingantennato increasethesignalstrengthatthe

    S1,S2

    MIMOCoding

    MIMOCoding

    S0,S1,S2,S3,..S0,S3

    h00

    h10

    h11

    h01

    y0,y2

    y1,y3

    S0,S1,S2,S3,..

  • 29

    receiverinproportiontothenumberoftransmitantennas.TheMultipleantennas

    canprovidebeamforming in addition to spatialdiversity if some knowledgeof

    relativechannelphases isavailableatthetransmitterwhich isnotarequirement

    forspatialdiversityandspatialmultiplexingtechniques.

    Inmultipleantennatransmissionschemes,beamformingcanbeemployedonthe

    basis of low and high mutual antenna correlation. In high mutual antenna

    correlation scheme the separation between antennas is relatively small as

    illustrated in Figure 2.10a and the overall transmissionbeam canbe steered in

    different direction by applying different phase shifts to the signal to be

    transmitted.Withdifferentphase shiftsapplied toantennashavinghighmutual

    correlationissometimesreferredasclassicalbeamformingwhichcannotprovide

    any diversity against radio channel fading but just an increase in the received

    signal strength. In conventional beamforming to achieve optimal SNR over

    wireless link the samedate symbol is transmitted simultaneously fromdifferent

    antennasaftera complexweight isapplied toeach signalpathwith theaimof

    steeringtheantennaarray[10,18].

    Lowmutualantennacorrelationusuallyinvolveseitheralargeantennaseparation

    as depicted in Figure 2.10b or different antenna polarizations.However, in this

    caseeach antenna transmits simultaneously a general complex2 valueweighted

    combinationof twodata symbols.Due to lowmutual antenna correlationeach

    antennamayexperiencedifferentchannelinstantaneousgainsandphases.

    The different complex weights (precoding matrix) are applied to symbols

    S , S , S , S , to be transmitted from different antennas. After applying pre

    codingweights, two separate data streams are transmitted from two different

    antennassimultaneouslyasspatialmultiplexing.AsillustratedinFigure2.10b,the

    2GeneralComplexmeansbothamplitudeandphaseofthesignaltobetransmittedondifferentantennas.

  • 30

    data symbol is transmitted from theupperantennaduring first symbol time

    which isa linear combinationof twodata symbols, s and s .During the same

    time,datasymbol istransmittedwithdifferentcombinationsofthesamedatasymbolsfromthelowerantenna,thusefficientlydoublingupthedatarate.Sothe

    relationbetweenthetransmitteddataandthe inputsymbols isdemonstratedas

    follows:

    . 2.6

    where corresponds22precodingmatrixand corresponds21transmitted

    symbolsmatrix.

    Figure2.10a:Classicalbeamformingwithhighmutualantennacorrelation

    Signal to be transmitted

    ej1

    ej2

    RxFraction of a wave length

    Beam structure

  • 31

    , , , , Figure2.10b:Precoderbasedbeamformingincaseoflowmutualantennacorrelation

    Several wave lengths

    Pre coding matrix

    [W]

    Channel information

    x0 ,x2

    x1,x3

    Rx

  • 32

    Chapter3LTEDownlinkSystemModel

    3.1)Introduction

    TogiveadetaileddescriptionofLTEdownlink, systemmodelbasedonmultiple

    antennatransmissionschemesandOFDMisthemotivatingforceinthewritingof

    thischapter.Thecombinationofnewtechnologies,i.e.,MIMOOFDMisemployed

    tofulfilltheLTEradiointerfacerequirements[25].MIMOtechniquesdiscussedin

    Chapter 2 have been considered as the key approach for providing required

    bandwidth efficiency and high data rates in the evolution of future generation

    mobile communication systems. On the otherhand, OFDM is the special case

    multicarrier transmission scheme which is extremely attractive to implement

    multicarriermodulationschemestocombat ISI inmultipathfadingenvironments

    withhigh spectralefficiency forLTEdownlink transmission.OFDMusesmultiple

    overlappingbutorthogonalcarriersignalsinsteadofsinglecarriertoachievehigh

    spectralefficiencyandhighdata rates.The LTEdownlink systemmodelused in

    simulationisdepictedinFigure3.1withtwotransmitandtworeceiveantennas.

  • 33

    Figure3.1LTEDownlinksystemmodelwith22MIMO

    3.2)GeneralDescription

    In LTE downlink transmission, OFDM symbols are generated by performing

    differentmanipulationsonthesinglebinaryinputdatastreamasshowninFigure

    3.1.

    In first step single binary input data stream is converted into two parallel data

    streamsbytheserialtoparallelconverter.Followingeachparalleldatastream is

    modulatedwith different subcarriers frequency by using different constellation

    Bit to symbolmapping

    Physical Resource mapping

    Bit to symbolmapping

    Serial to Parallel converter

    Zero Padding

    IFFT

    CP

    Pilot insertion

    Antenna mapping

    Input data

    Bit to symbol mapping

    Physical Resource de-mapping

    Bit to symbol mapping

    P/S

    Converter

    MIMO Detection

    FFT

    CP

    Removal

    Pilot Extraction

    Antenna de-mapping

    Channel Estimation

    Zero un-padding

    BRE

    Computation

  • 34

    mapping schemes (QPSK, 16QAM, 64QAM) defined in [26]. The Constellation

    mappingisawayofassigningsinusoidshavinguniqueamplitudeand/orphaseto

    the input binary data. These higher modulation or constellation schemes are

    requiredtoachievebandwidthefficiencyandhighdataratesinLTE.

    Subsequently, according to LTE downlink subframe structure as described in

    chapter 2 the complex constellation symbols of each data stream and pilot

    symbols are mapped on OFDM resource gird. The different pilot symbols

    arrangement foreachantenna isused for thepurposeofchannelestimationas

    documented in [27].After the physical resourcemapping zero padding is done

    becausethesamplingrateismuchhigherthanthetransmissionbandwidthofthe

    system.Inzeropaddingthe lengthofthesignalspectrum is increasedbyspecific

    numberofzeros.ThezeropaddedsignalsareappliedtoIFFTblockforthesakeof

    OFDMmodulation. InOFDMmodulation the serial inputdata stream isdivided

    into anumberofparallel streams and then these lowrateparallel streams are

    transmitted over different subcarriers simultaneously. Theminimum frequency

    separation is required between these subcarriers tomaintain orthogonality of

    theircorrespondingtimedomainwaveforms.Infrequencydomainthesedifferent

    subcarriersoverlap.Thus,availablebandwidth isused inefficientway. IFFT isan

    efficient algorithm used to generate an OFDM symbols and in complexity

    reductionof the transmitter.FinallyCPsare insertedbefore the transmission to

    OFDMmodulatedsignal. Ifchannel isnottimedispersive,thetransmittedOFDM

    signal can be demodulatedwithout any interference.On the other hand if the

    channel is time dispersive then the Orthogonality will be lost between the

    subcarriers because of the overlapping of the correlation interval of the

    demodulator foronepathwith the symbol intervalof theotherpaths.Thiswill

    resultinintersymbolinterferenceaswellasintercarrierinterference.Tocombat

    theseinterferencesandtoovercometimedispersionofradiochannel,CPisused

  • 35

    inOFDM system. For further readingonOFDMworking,details canbe seen in

    [28].

    Thereceiversidebasicallyperformsthereverseoperationofthetransmitterwith

    some additional operations (e.g., channel estimation) in order to find the

    estimatesofmultipathchannel to recover informationproperly [25]. In the first

    step, cyclic prefixes are removed from the received data symbols. FFT is

    performed to recover the modulated symbol values of all subcarriers and to

    convertthereceivedsignalintofrequencydomain.Toperformchannelestimation

    (seechapter5),referencesymbolsareextractedfromeachsubframe.Torecover

    original data streams, the received complex symbols from both antennas are

    deliveredtoMIMOdetectionstage.Afterthisthecomplexconstellationsymbols

    aredemappedintobinaryvaluesandthisparalleldataisconvertedinserialdata

    tofindouttheestimatesofthetransmitteddata[25].

  • 36

    CHAPTER4RadioPropagationModels

    4.1)Introduction

    Fromthebeginningwirelesscommunicationsthere isahighdemandforrealistic

    mobile fading channels.The reason for this importance is thatefficient channel

    modelsareessential for theanalysis,design,anddeploymentofcommunication

    system forreliabletransferof informationbetweentwoparties.Correctchannel

    modelsarealso significant for testing,parameteroptimizationandperformance

    evolutionof communication systems.Theperformanceand complexityof signal

    processingalgorithms, transceiverdesignsandsmartantennasetc.,employed in

    futuremobilecommunicationsystems,arehighlydependentondesignmethods

    used tomodelmobile fading channels. Therefore, correct knowledgeofmobile

    fadingchannelsisacentralprerequisiteforthedesignofwirelesscommunication

    systems[29,30,31].

    The difficulties inmodeling awireless channel are due to complex propagation

    processes. A transmitted signal arrives at the receiver through different

    propagation mechanisms shown in figure 4.1. The propagation mechanisms

    involvethefollowingbasicmechanisms: i)freespaceor lineofsightpropagation

    ii)specularreflectionduetointeractionofelectromagneticwaveswithplaneand

    smoothsurfaceswhichhave largedimensionsascomparedtothewavelengthof

    interacting electromagnetic waves iii) Diffraction caused by bending of

    electromagneticwavesaroundcornersofbuildings iv)Diffusionorscatteringdue

    tocontactswithobjectshavingirregularsurfacesorshapeswithsizesoftheorder

    ofwavelengthv)Transmissionthroughobjectswhichcausepartialabsorptionof

    energy[16,29].Itissignificantheretonotethatthelevelofinformationaboutthe

  • 37

    environmentachannelmodelmustprovide ishighlydependentonthecategory

    of communication system under assessment. To predict the performance of

    narrowbandreceivers,classicalchannelmodelswhichprovide informationabout

    signalpowerleveldistributionsandDopplershiftsofthereceivedsignals,maybe

    sufficient.Theadvanced technologies, (e.g.,UMTSand LTE)buildon the typical

    understandingofDopplerspreadandfading;alsoincorporatenewconceptssuch

    as timedelay spread,directionofdepartures (DOD),directionof arrivals (DOA)

    andadaptivearrayantennageometry [30].Thepresenceofmultipaths (multiple

    scatteredpaths)withdifferentdelays,attenuations,DODandDOAgives rise to

    highlycomplexmultipathpropagationchannel.

    Figure 4.1: Signal propagation through different paths showingmultipath propagationphenomena

    Power

    Delay

    Figure4.2:Powerdelayprofileofamultipathchannel

  • 38

    4.2)MultipathPropagationChannel

    Multipath signals arrive at the receiverwith different propagation path lengths

    called multipath taps. The different propagation path lengths cause different

    propagation time delays. Figure 4.2 illustrates power delay profile (PDP) of a

    multipath channel with three distinct paths. Depending on the phases the

    multipath signals interact either constructively or destructively at the receiver.

    Thismakespowerofchanneltapstimevaryingresultinginfadingdipsasshownin

    figure4.3.Thepowerdistributionof channel taps isdescribedbyadistribution

    functiondependingonthepropagationenvironment.Themostseveremultipath

    channel isRayleighfadingchannel inwhichthere isno lineofsightpathandthe

    channeltapsareindependent.InthecaseofRicianfadingchannelthefadingdips

    are low due to the presence of line of sight component in addition to the

    dispersedpaths.Aradiochannelcanbecharacterizedasnarrowbandorwideband

    dependingonthechannelcharacteristicsanddurationofasymbol.Duetoabove

    propagationmechanisms,theimpulseresponseofamultipathchannelconsistsof

    aseriesofpulsesspreadintimeorfrequency.Ifthetimedifferencebetweenthe

    firstandthelastpulseissmallerthanthedurationofasymbol,thenthesystemis

    callednarrowbandsystem.Ifthetimedifferencebetweenthefirstand lastpulse

    is greater than the duration of a symbol, then the system is calledwideband

    system.Widebandcharacteristicsof impulseresponseareoftenusedtodescribe

    thebehaviorofthemultipathchannel.

  • 39

    Figure4.3:Receivedsignalpowerlevelofatimevaryingmultipathpropagationchannel

    The degradation in the received signal level due tomultipath effects can be

    classified into large scale path loss component, medium scale slow varying

    componentwith lognormal distribution and small scale fast fading component

    withRayleighorRiciandistributiondependingontheabsenceorpresenceofLOS

    component between the transmitter and receiver [29]. Thus, a three stage

    propagationmodelcanbeusedtodescribeawirelesscellularenvironment.

    4.2.1)LargeScalePropagationModel

    Largescalepropagationmodelisusedtocharacterizethereceivedsignalstrength

    by averaging the amplitude or power level of the received signal over large

    transmitterreceiverseparationdistances in the rangeofhundredsor thousands

    ofawavelength.The large scalemodelsareoftenderived frommeasureddata.

    However,semiempiricalmodelsareemployed insmallerareastoachievehigher

    accuracy.Thetheoreticalmodelsareusedwhicharethenfittedtomeasureddata

    toobtaindesiredmodelforaparticularpropagationscenario.

    0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1-35

    -30

    -25

    -20

    -15

    -10

    -5

    0Channel Fast Fading of Different Paths

    Time in seconds

    Pow

    er in

    dBs

    Path1Path2Path3Path4Path5Path6

  • 40

    4.2.2)MediumScalePropagationModel

    Mediumscalepropagationmodeldeterminesthegradualvariationsof the local

    mean amplitude or the localmean power of the received signal over a time

    variantmultipath channelwhen themobile stationmovesoverdistances larger

    than a few tens or hundreds of awavelength. Some existing componentswill

    disappearwhilenewcomponentswillappear.Fornarrowbandmeasurements,itis

    observedthatvariationsofthelocalmeanpowerofthereceivedsignalfollowlog

    normal distribution which is called slow fading or shadowing. The mean and

    standard deviations are determined from large scale propagationmodel in the

    environmentof interest.Theshadowing iscausedbyobstructions like treesand

    foliage. The wideband channel is characterized by the timevariant channel

    impulseresponse.

    , 4.1

    where istheamplitude,isthephaseandisthedelayofthe component

    arrivingatparticulartimet.Theimpulseresponseofwidebandchannelconsistsof

    series of independently fading paths received at different time delays. Each

    independentpathofthewidebandchannelcanbetreatedasnarrowbandchannel

    withfadingcharacterizedbylognormaldistribution.

    4.2.3)SmallScalePropagationModel Thismodel is used to characterize the rapid variations of the received signal

    strength due to changes in phases when amobile terminalmoves over small

    distancesontheorderofafewwavelengthsorovershorttimedurationsonthe

    order of seconds. Since the mean power remains constant over these small

    distances, small scale fading can be considered as superimposed on large scale

  • 41

    fadingforlargescalemodels.Themostcommondescriptionofsmallscalefading

    isbymeansofRayleighdistribution.

    0 4.2

    Thisdistribution represents the sumofamplitudesof the sameorderofa large

    number of uncorrelated components with phases uniformly distributed in the

    interval (0,2).Theparameter istherootmeansquarevalueofthereceived

    signal. In thecaseof lineof sightcomponent, the signal fadingcanbemodeled

    usingRiciandistribution.

    . 0 4.3

    where representsmodifiedBesselfunctionoffirstkindandzeroorder, isthe

    amplitude of the line of sight component (LOS). If there is no LOS component,

    RiciandistributioncanbesimplyexpressedasRayleighdistribution.

    Incaseofwidebandmodeling,eachpathoftheimpulseresponsecanbemodeled

    as Rayleigh distributedwith uniform phase except LOS cases. In LOS, first path

    followsRiciandistribution.Acommonmethodusedofwidebandchannelmodels

    istappeddelaylinemodelinwhichthetapsfollowabovementioneddistributions.

    4.3)PropagationaspectsandParameters

    Thebehaviorofamultipathchannelneedstobecharacterizedinordertomodel

    thechannel.TheconceptsofDopplerspread,coherence time,anddelayspread

    and coherence bandwidth are used describe various aspects of themultipath

    channel.

  • 42

    4.3.1)DelaySpread

    To measure the performance capabilities of a wireless channel, the time

    dispersionormultipathdelayspreadrelatedtosmallscalefadingofthechannel

    needstobecalculatedinaconvenientway.Onesimplemeasureofdelayspreadis

    the overall extent of path delays called the excess delay spread. This is only

    convenient way because different channels with the same excess delays can

    exhibit different power profiles which have more or less impact on the

    performance of the system under consideration. A more efficient method to

    determinechanneldelayspreadistherootmeansquare(rms)delayspread( )

    which isastatisticalmeasureandgivesthespreadofdelayedcomponentsabout

    themean valueof the channelpowerdelayprofile. Mathematically, rmsdelay

    spreadcanbedescribedas secondcentralmomentof thechannelpowerdelay

    profile[29]whichiswrittenasfollows:

    = 4.4

    where, = isthemeanexcessdelay

    4.3.2)CoherenceBandwidth

    When the channel behavior is studied in frequency domain than coherence

    bandwidthf isofconcern.The frequencyband, inwhichtheamplitudesofall

    frequency components of the transmitted signal are correlated, i.e.,with equal

    gainsand linearphases, isknownascoherencebandwidthof thatchannel [30].

    The channel behavior remains invariant over this bandwidth. The coherence

    bandwidthvaries in inverseproportiontothedelayspread.Amultipathchannel

    canbe categorized as frequency flat fadingor frequency selective fading in the

    followingway.

  • 43

    Frequencyflatfading:Achannel isreferredtoasfrequencyflat ifthecoherence

    bandwidthf B,whereBisthesignalbandwidth.Allfrequencycomponentsof

    thesignalwillexperiencethesameamountoffading.

    Frequencyselectivefading:Achannel isreferredtoasfrequencyselective ifthe

    coherencebandwidth f B. In thiscasedifferent frequencycomponentswill

    undergodifferentamountoffading.Thechannelactsasafiltersincethechannel

    coherencebandwidthislessthanthesignalbandwidth;hencefrequencyselective

    fadingtakesplace[32].

    4.3.3)DopplerSpreadThe Doppler spread arises due to themotion ofmobile terminal. Due to the

    motionofmobile terminal through thestandingwave theamplitude,phaseand

    filteringappliedtothetransmittedsignalvarywithtimeaccordingtothemobile

    speed [33].Foranunmodulatedcarrier,theoutput istimevaryingandhasnon

    zerospectralwidthwhichisDopplerspread.Forasinglepathbetweenthemobile

    terminalandthebasestation,therewillbezeroDopplerspreadwithasimpleshift

    ofthecarrierfrequency(Dopplerfrequencyshift)atthebasestation.TheDoppler

    frequencydependsontheangleofmovementofthemobileterminalrelativeto

    thebasestation.

    4.3.4)CoherenceTimeThetimeoverwhichthecharacteristicsofachanneldonotchangesignificantlyis

    termedascoherencetime.ThereciprocaloftheDopplershift isdescribedasthe

    coherence timeof thechannel.Mathematicallywecandescribecoherence time

    asfollows:

    . .

    , where is root mean square vale of Doppler spread.

    Thecoherencetimeisrelatedtothepowercontrolschemes,errorcorrectionand

  • 44

    interleaving schemesand to thedesignof channelestimation techniquesat the

    receiver.

    4.4)StandardChannelmodels

    Standardchannelmodelscanbedevelopedbysettingupframeworkforgeneric

    channelmodelsandfindingsetofparametersthatneedtobedeterminedforthe

    descriptionofthechannel.Theothermethodistosetupmeasurementcampaigns

    and extracting numerical values of parameters and their statistical distributions

    [34].WhendesigningLTE,differentrequirementsareconsidered;userequipment(UE)

    and base station (BS) performance requirementswhich are crucial part of LTE

    standards,RadioResourceManagement (RRM) requirements toensure that the

    availableresourcesareused inanefficientwaytoprovideendusersthedesired

    qualityofservice,theRFperformancerequirementstofacilitatetheexistenceof

    LTEwithothersystems (e.g.,2G/3G)systems [35].Thestandardchannelmodels

    playavitalroleintheassessmentoftheserequirements.Inthefollowingsection,

    some standard channelmodels arediscussedwhich areused in thedesign and

    evolutionoftheUMTSLTEsystem.

    4.4.1)SISO,SIMOandMISOChannelModels

    COST projects, Advanced TDMA Mobile Access, UMTS Code Division Testbed

    (CODIT)conductedextensivemeasurementcampaignstocreatedatasetsforSISO,

    SIMOandMISOchannelmodelingandtheseeffortsformthebasisforITUchannel

    models which are used in the development and implementation of the third

    generationmobilecommunicationsystems[9].COSTstandsfortheEuropeanCo

    operation in theFieldofScientificandTechnicalResearch.SeveralCostefforts

    were dedicated to the field of wireless communications, especially radio

  • 45

    propagationmodeling,COST 207 for thedevelopmentof SecondGenerationof

    Mobile Communications (GSM), COST 231 for GSM extension and Third

    Generation (UMTS) systems, COST 259 "Flexible personalized wireless

    communications (19962000) and COST 273 "Towards mobile broadband

    multimedia networks (20012005)". These projects developed channel models

    basedonextensivemeasurementcampaigns includingdirectionalcharacteristics

    ofradiopropagation(Cost259andCost273)inmacro,microandpicocellsandare

    appropriate for simulations with smart antennas and MIMO systems. These

    channelmodelsformthebasisofITUstandardsforchannelmodelsofBeyond3G

    systems.DetailedstudyofCOSTprojectscanbefoundin[36,37,38].

    The research projectsATDMA and CODITwere dedicated towideband channel

    modeling specifically channel modeling for 3rd generation systems and the

    corresponding radio environments. The wideband channel models have been

    developed within CODIT using statisticalphysical channel modeling approach

    while stored channel measurements are used in ATDMA which are complex

    impulseresponsesfordifferentradioenvironments.Thedetailsoftheseprojects

    canbefoundin[31].

    4.4.2)MIMOChannelModels

    Multipleinput multipleoutput wireless communication techniques offer the

    promise of increased spectral efficiency throughput and quality of service for

    MIMOcommunicationsystems[39,40].Asanexample,figure4.5showsthatthe

    capacity increases linearly with the increase in number of antennas. The

    performance of MIMO communication systems is highly dependent on the

    underlyingpropagation conditions.The spatial characteristicsofa radio channel

    have significant effects on the performance of MIMO systems. The MIMO

  • 46

    techniquestaketheadvantageofmultipatheffectsintheformofspatialdiversity

    to significantly improveSNRbycombining theoutputsofdecorrelatedantenna

    arrayswith lowmutual fading correlation. The other technique to increase the

    effective data rate of aMIMO system usingmultiple antenna arrays is spatial

    multiplexingwhichcreatesmultipleparallelchannelsbetweenthetransmitterand

    the receiversides.Multipleantennasat the transmitterand/or the receiverside

    can be used to shape the overall beam in the direction of a specific user to

    maximize the gain. This technique is called beam forming. Multiple antennas

    transmission techniques are described in details in chapter 2. The largeMIMO

    gainscanbeachievedbylowspatialcorrelation.Theantennaseparation,interms

    ofwavelengthof theoperating frequency,has significant impactson the spatial

    correlation.Toachieve low fadingcorrelation, theantennaseparationshouldbe

    large. The small sizes of wireless devices restrict large antenna separation

    depending upon the wavelength of the operating frequency. An alternative

    solution to achieve low correlation is to use antenna arrays with cross

    Polarizations (i.e., antenna arrays with polarizations in orthogonal or near

    orthogonalorientations)[18,41].

    4.4.3)EffectofSpatialCorrelationonMIMOPerformance

    TheMIMO transmission schemes, spatial diversity and spatialmultiplexing, can

    substantially improve theperformanceandovercome theundesirablemultipath

    effects ifthespatialdimension isproperlyconfiguredto leveragetherichnessof

    multipathenvironment [18,35,39].Thediversitygaincanbeachievedonlywhen

    thereislowcorrelationbetweenthetransmittingandthereceivingantennas.The

    spatialcorrelationhasalsoasignificant impactonthecapacity limitsofaMIMO

    channel [9]. The capacity of a channel without channel knowledge at the

    transmittercanbedeterminedas[35].

  • 47

    4.5

    Where, and arethenumberoftransmitandreceiveantennas,respectively

    and isthe x identitymatrix.Thesymbol representsthesignaltonoise

    plusinterferenceratio(SINR)andHisthechanneltransferfunction.Theoperator

    representstheHermitiantransposeoperation.

    It is obvious that for a given number of transmit antennas and signal to

    interferenceratio,thespatialcorrelationvalueofMIMOchanneldeterminesthe

    theoreticalcapacitylimits.Figure4.6illustratestheimpactofspatialcorrelationof

    the performance of radio link quality of MIMO channel. It illustrates the

    performanceof system in termsofbiterror rate.Alamouti space timecoding is

    usedwithdiversityorderof4 ( =2x2=4) in flat fadingRayleighchannelwith

    differentcorrelationvalues.TheresultsdemonstratethatSNRrequiredtosupport

    differentvaluesofbiterrorratevariesdependingondifferentcorrelationvalues.

    -5 0 5 10 15 200

    5

    10

    15

    20

    25

    SNR in dB

    Cap

    acity

    bits

    /s/H

    z

    nt = 1 , nr = 1nt = 2 , nr = 2nt = 4 , nr = 4nt = 8 , nr = 8

    Figure4.4:Theincreaseinthecapacitywiththeincreaseinnumberofantennas.Thecapacityofthesystemincreaseslinearlywithincreasingnumberofantennas

  • 48

    The diversity gain can be achieved by low correlation values. To achieve low

    correlationvalues,theseparationbetweenantennaarraysshouldbelarge.

    Figure4.5:BERcurvesfor2x2MIMOsystemsusingflatfadingRayleighchannelwithdifferentcorrelationvalueswhichshowthatBERdecreaseswithlowcorrelationvalues

    4.4.4)ITUMultipathChannelModels

    The ITU standardmultipath channelmodelsproposedby ITU [42]used for the

    developmentof3G 'IMT2000'groupofradioaccesssystemsarebasicallysimilar

    in structure to the 3GPPmultipath channelmodels. The aim of these channel

    models istodevelopstandardsthathelpsystemdesignersandnetworkplanners

    forsystemdesignsandperformanceverification. Insteadofdefiningpropagation

    modelsforallpossibleenvironments,ITUproposedasetoftestenvironments in

    [42] that adequately span the all possible operating environments and user

    mobility. Inourworkweused ITU standard channelmodels forpedestrian and

    vehicularenvironments.

    0 2 4 6 8 10 12 14 16 18 2010-4

    10-3

    10-2

    10-1

    SNR (dB)

    BE

    R

    BER vs SNR for 2x2 MIMO with different correlations

    2x2-STC-corr-02x2-STC-corr-0.52x2-STC-corr-0.72x2-STC-corr-0.99

  • 49

    4.4.4.1)ITUPedestrianA,B

    WeusedbothPedestrianA andPedestrianB channelmodels inourwork. The

    mobile speed isconsidered tobe3km/h ineachof thesecases.ForPedestrian

    models thebase stationswith low antennasheight are situatedoutdoorswhile

    thepedestrianuserarelocatedinsidebuildingsorinopenareas.Fadingcanfollow

    Rayleigh or Rician distribution depending upon the location of the user. The

    numberoftaps incaseofPedestrianAmodel is3whilePedestrianBhas6taps.

    The average powers and the relative delays for the taps ofmultipath channels

    basedonITUrecommendationsaregivenintable4.1[42].

    Tap No Pedestrian-A Pedestrian-B Doppler Spectrum

    Relative

    Delay (ns)

    Average

    Power(dB)

    Relative

    Delay (ns)

    Average

    Power(dB)

    1 0 0 0 0 Classical

    2 110 -9.7 200 -0.9 Classical

    3 190 -19.2 800 -4.9 Classical

    4 410 -22.8 1200 -8 Classical

    5 NA NA 2300 -7.8 Classical

    6 NA NA 3700 -23.9 Classical

    Table 4.1: Average Powers and Relative Delays of ITU multipath PedestrianA andPedestrianBcases

  • 50

    4.4.4.2)ITUVehicularA(V30,V120andV350)

    Thevehicularenvironmentiscategorizedbylargemacrocellswithhighercapacity,

    limitedspectrumand large transmitspower.The receivedsignal iscomposedof

    multipath reflectionswithout LOS component. The received signal power level

    0 20 40 60 80 100 1200

    0.2

    0.4

    0.6

    0.8

    1

    Tap No

    Am

    plitu

    de

    Pedestrian A

    0 20 40 60 80 100 1200

    0.2

    0.4

    0.6

    0.8

    1

    Tap No

    Am

    plitu

    e

    Pedestrian B

    0 20 40 60 80 100 1200

    0.2

    0.4

    0.6

    0.8

    1

    Tap No

    Am

    plitu

    e

    Vehicular A

    Figure4.6:Channel ImpulseResponsesaccording to ITU standardswhichare tobeused insimulationsforchannelestimationofLTE

  • 51

    decreaseswithdistance forwhichpass lossexponentvariesbetween3and5 in

    thecaseofurbanandsuburbanareas.Inruralareaspathlossmaybelowerthan

    previouswhile inmountainous areas, neglecting the path blockage, a path loss

    attenuationexponentcloserto2maybeappropriate.

    Forvehicularenvironments,weused the ITUvehicularAchannelmodels inour

    work. Themobile speed considered is 30 km/h, 120 km/h and 350 km/h. The

    propagation scenarios forLTEwith speeds from120km/h to350km/harealso

    defined in [43] tomodelhighspeedscenarios (e.g.,highspeed trainscenarioat

    speed 350km/h). The maximum carrier frequency over all frequency bands is

    f=2690MHzand theDoppler shiftat speedv=350km/h is900Hz.Theaverage

    powersand the relativedelays for the tapsofmultipath channelsbasedon ITU

    recommendationsaregivenintable4.2[42].

    Tap No

    Average

    Power(dB)

    0 -1.0 -9.0 -10.0 -15.0 -20.0

    Relative

    Delay(ns)

    0 310 710 1090 1730 2510

    Table4.2:AveragePowersandRelativeDelaysforITUVehicularATestEnvironment

    4.4.5)ExtendedITUmodelsThe analysis done by ITUR showed that evolution of 3G systems to future

    generation networkswill require technology changes on large scalewhile new

    quality of service (QoS) requirements will require increased transmission

    bandwidth.SoLTEchannelmodelsrequiremorebandwidthascomparedtoUMTS

    channelmodelstoaccountthat factthatchannel impulsesareassociatedtothe

    delayresolutionofthereceiver.TheLTEchannelmodelsdevelopedby3GPPare

  • 52

    based on the existing 3GPP channel models and ITU channel models. The

    extended ITUmodels for LTE were given the name of Extended PedestrianA

    (EPA),ExtendedVehicularA(EVA)andExtendedTU(ETU).Thesechannelmodels

    areclassifiedonthebasisoflow,mediumandhighdelayspreadwherelowdelay

    spreads are used to model indoor environments with small cell sizes while

    mediumandhighdelayspreadsareusedtomodelurbanenvironmentswithlarge

    cells.Thehighdelay spreadmodels are according toTypicalUrbanGSMmodel

    [43].Thepowerdelayprofilesforthesechannelmodelsaregiven intables4.3.1,

    4.3.2and4.3.3,respectively[43].

    TapNo

    Average

    Power(dB)

    0.0 1.0 2.0 3.0 8.0 17.2 20.8

    Excess

    Delay(ns)

    0.0 30 70 80 110 190 410

    Table4.3.1:PowerDelayProfilesforExtendedITUPedestrianAModel

    TapNo

    Average

    Power(dB)

    0.0 1.5 1.4 3.6 0.6 9.1 7.0 12 16.9

    Excess

    Delay(ns)

    0.0 30 150 310 370 710 1090 1730 2510

    Table4.3.2:PowerDelayProfilesforExtendedITUVehicularAModel

  • 53

    TapNo

    Average

    Power(dB)

    1.0 1.0 1.0 0.0 0.0 0.0 3.0 5.0 7.0

    Excess

    Delay(ns)

    0.0 50 120 200 230 500 1600 2300 5000

    Table4.3.3:PowerDelayProfilesforExtendedTypicalUrbanModel

    The above Extended ITU Models are also used in this work. The Doppler

    frequencies for these channelmodels are defined in a similar way to that of

    EUTRA.TheDoppler frequencies for LTE channelmodelswith low,mediumand

    highDopplerconditionsare5Hz,70Hzand900Hzrespectively[43].Thefollowing

    combinationsofdelayspreadandDopplerspreadareproposedin[43];extended

    pedestrianA 5 Hz, extended vehicularA 5Hz, Extended VehicularA 70 Hz and

    ExtendedTypicalUrban70Hz.

  • 54

    Chapter5ChannelEstimationinLTE

    5.1)Introduction

    In this chapter, channel estimation techniques used in LTE down link are

    described.Channelestimation isavitalpartof receiversdesignsused inmobile

    communicationsystems.Theeffectofthechannelonthetransmittedinformation

    mustbeestimatedinordertorecoverthetransmittedinformationcorrectly.The

    estimationofchanneleffectsisoftenbasedonanapproximateunderlyingmodel

    of the radio propagation channel [44]. The receiver can precisely recover the

    transmitted information as long as it can keep track of the varying radio

    propagationchannels.Channelmodelsaredescribedinthechapter4.

    5.2)SignalModel

    Inthiswork,weconsideroneOFDMsymboltoperformchannelestimationinLTE

    down link. For the purpose of channel estimation we consider the channel

    frequency response vector and a diagonal matrix containing the transmitted

    frequencydomainsamplesasfollows:

    5.1

    where is a diagonal matrix with transmitted data samples,

    reference symbols or zeros, contains unknown channel frequency

    response coefficients to be estimated and is the noise vector. The

    channel frequencyresponse (CFR)canbeexpressed intermsofchannel impulse

    response (CIR) as . So the channel impulse response can bewritten as

    follows[45]:

  • 55

    5.2

    Inaboveequation isDFTmatrixwhichcanbewrittenasfollows:

    , ,

    , ,

    5.3

    The channelestimation isdoneusing transmitted reference symbols,which are

    insertedamongdatasubcarriersasperdesignspecificationsof3GPPLTE[3].The

    referencesymbolsstructureisdiscussedinchapter2.Beforeintroducingchannel

    estimation schemeswe consider some simplifications in our signal structure in

    ordertoreducecomplexityoftheestimator[45].

    Only the channel taps with significant energy are considered. So we

    considerfirstLcolumns(whereLrepresentsthe lengthofthechannel)of

    thematrixFcorrespondingtothemaximumdelayattapL1.

    ThereferencesymbolsareslottedinaccordingtoLTEdesignspecifications

    so, only rows of the matrix F corresponding to the position of these

    referencesymbolsareconsideredforthediagonalmatrixX.Nowequation

    2canbewrittenas:

    5.4

    where, isadiagonalmatrixcontainingreferencesymbols.The

    vectorYistheoutputand isthetruncatedwhiteGaussiannoise.

    The vector represents the unknown CIR coefficients to be

    estimated. The matrix is a Fourier matrix associated with

    transmittedreferencesymbolswhichcanbewrittenasfollows:

  • 56

    5.5

    wherei thestartingisreferencesymbolindexandcanbespecifiedas2

    N FTwithN FTbeingtheDFTsize.

    Similar toequation4,we canwrite for the transmitteddata symbols as

    follows:

    5.6

    1

    5.3)PilotassistedChannelEstimation

    Thepilotassistedchannelestimationprocessconsistsoftwosteps;firststatistical

    estimation of the channel at OFDM tones consisting of reference symbols is

    determinedusingstatisticalmethods includingSquares (LS)andMinimumMean

    Squares(MMSE)estimates.

    Different pilots assisted channel estimation schemes can be employed for the

    estimationof thechanneleffectson the transmittedsignal.Theresponseof the

    channelatthedatasubcarriers issubsequentlydeterminedby interpolation.The

    interpolatorsusedforthepurposeofestimationarelinear,secondorder,cubicor

  • 57

    time domain interpolators derived from both the statistical and deterministic

    pointofview[46,47,48].Variouspublicationscanbefoundthatdealwithoneor

    all these estimation criteria for pilot assisted channel estimation of OFDM

    applicationsfromCIRorCFRprospective[49,50,51,52,53,54].

    In this chapter,weattempt togiveanoverviewof theseestimation techniques

    withareferencetoLTEdownlinkstructuredescribedinchapter2.

    5.3.1)LeastSquareEstimation

    Least Square based parameter estimation approach aims at determining the

    channel impulseresponse from theknown transmittedreferencesymbols in the

    followingway:

    , , , 5.7

    whereGLS CN is theestimated channel frequency responseon the subcarriers

    which contains reference symbols. This response can be interpolated over full

    frequency range in order to obtain the channel frequency response for the

    subcarriers carrying data symbols. The interpolation can be performed in time

    domainorfrequencydomain.

    Thesignalreceivedintimedomaincanbeexpressedasfollows:

    5.8

    ThechannelcanbeestimatedusingLeastSquares,intimedomaininthefollowing

    way[45]:

    5.9

    Solvingtheabovetwoequations,wegettheexpressionforLSestimates.

  • 58

    5.10

    The term FLHAHA FL FLH in this equation is constant and can be computed

    offline regardless of the time varying nature of the channel. This makes LS

    estimatorcomputationallysimple. Howeveran illconditionedproblemoccurs in

    straightwayapplicationofLSestimatorduetomatrix inversion.Theproblemcan

    besolvedinthefollowingtwoways[46].

    5.3.2)RegularizedLSEstimation

    In this method, small constant term is added to the diagonal entries for

    regularizingtheEigenvaluesofthematrixtobeinverted.Inthiscase,thechannel

    impulseresponsebecomesoftheform,

    5.11

    Thevalueofhastobeselectedsuchthattheinversematrixisleastperturbed.

    5.3.3)DownSamplingMethod

    The second solution to the illconditioningproblem is foundby considering the

    fact that not all portion of the spectrum is used due to LTE OFDM symbols

    structure. Consider the case of 5 MHz bandwidth, the number of used

    (modulated)subcarriersis300.Thesamplingfrequencyinthiscaseis7.