AdaptiveReceiverAlgorithmsforMIMOsystems.docx

AdaptiveReceiverAlgorithmsforMIMOsystems.docx

《AdaptiveReceiverAlgorithmsforMIMOsystems.docx》由会员分享，可在线阅读，更多相关《AdaptiveReceiverAlgorithmsforMIMOsystems.docx（18页珍藏版）》请在冰豆网上搜索。

AdaptiveReceiverAlgorithmsforMIMOsystems

ADAPTIVERECEIVERALGORITHMSFORSINGLE-USERSPACE-TIME

BLOCKCODEDMIMOSYSTEMS

I.INTRODUCTION

ThisprojectreviewssomereceiveralgorithmsforSpace-timeBlockCodedmutiple-inputmultiple-output（MIMO）systems.Asspace-timeprocessingwasshowntobeefficienttouseatwirelesscommnication,thistopicisfirstdialedwith.Usingthisstructure,itispossibletoenhancethesystemperformanceandcapacity.Asaconsequenceofspace-timeprocessing,Space-timecodingwasintroduced,whereinformationbearingsignalsarespreadedintimeandspace（byusingmultipletransmitantennas）.Wewillfocusonasubsetofthesecodes:

Space-timeBlockcodes（STBC）.ThestructureofSTBC’swillbeintroducedandexistencyofthesecodesfordiffrentscenarioswillbediscussedandsometheoremswillbepresented.Afterthesesteps,modelsforMIMOsystemswillbepresentedforbothflatandfrequencyselectivefading.

UsingthestructureofSTBC’scomplexityofreceiveralgorithmslevereges.RecentlytherehavebeenlotsofresearchesonthetopicofreceiveralgorithmsforSTBcodedsystems.Here,theapproachesofthoseworkswillbeshownandsomeofthemwillbeperformedbysimulations.

II.SPACE-TIMEPROCESSINGFORWIRELESSCOMMUNICATIONS

Spacetimeprocessingreferstothesignalprocessingperformedinthespatialandtemporaldomainonsignalsreceivedatortransmittedframamantennaarray,inordertoimproveperformanceofwirelessnetworks[1].Mainproblemsinwirelsscommunicationssuchasco-channelinterferenceduetocellularfrequencyreuse,multipathfadingdecreasesystemqualityandcapacity.ByusingSpace-timeprocessing,thatisaddingmoreantennaatreceiverortransmitter,systemperformancecanbeimproved.Briefly,space-time（ST）processingcanimprovenetworkcapacity,coverageandqualitybyreducingco-channelinterference（CCI）whileenhancingdiversityandarraygain.

Usingextraspacecomponent（morereceiveortransmitantenna）enablesinterferencecancelationinawaythatisnotpossiblewithsingleantennasystems.ThedesiredsignalandCCIalmostalwaysarriveattheantennaarraywithdistictandwellseperatedspatialsignatures,thusallowingthesystemtoexploitthisdifferencetoreduceCCI.Moreover,STtransmitsystemscanusespatialselectivitiytotransmitsignalstodesiredmobilewhileminmizinginterferenceforothermobiles,i.e,smartantennasystems.Additionally,STprocessinginthereceivercanbeusedtoincreasediversitygainandsuppressinter-symbolinterference（ISI）.

Inthenexttopic,wirelesschannelmodelwillbeintroduced,duetobefamiliarwiththedegradationeffectsofwirelesschannels.

II.A.TheWirelessChannel

Thepropagatingradiosignalsareaffectedbythephysicalchannelinvariousways[1].Asignalpropagatingthroughwirelesschannelusuallyarrivesatthereceiverfromdifferentpaths,referedasmultipath.Thesepathsarisefromscattering,reflection,refractionofthesignalsfromsurroundedobjects.Thereceivedsignalismuchweakerthanthetransmittedoneduetotheeffectofpath-loss,slowfading（long-termfading）andfastfading（short-termfading）.Multipathpropagationresultsinthespreadingofthesignalindiffrentdimentions.Thesearedelay（time）spread,Doppler（orfrequency）spreadandanglespread.Infreespacepropagationpath-losscaneasilyfound.Fadingcanbecalculatedasmultiplicationofbothshortandlongtermfadingcomponents.Slowfadingcanbecharacterizedbyadistributionwhichisaffectedbyantennaheights,operatingfrequenciesandtypeofenvironment.OneofthemostprominentstatisticalmodelisOkamuramodel.FastfadingcomponenthasRayleighdensityfunctionifthereisnodirectpathfromsignalingparts.Rayleighdistributionisasfollows,

（1）

Ifthereisadirectpath,fastfadingcomponentwillhaveRiciandensityfunction,whichisasfollows,

A≥0ver≥0

（2）

Effectsofmultipathinsmallscalefadingisdetailedin[2].

II.B.ReceivedSignalModel

IngeneralchannelmodelforMIMOsystemincludephysicalparametersuchaspathgain,delayandangleofarrival.Butinmostcaseitisconvenienttousesampledmodelofreceivedsignal.Ifwethinkthatreceivedsignalx（t）issampledatt=t0+kT,thentheoutputmaybewrittenas

x（k）=Hs（k）+n（k）（3）

whereHisthechannelresponse（mxN）matrixthatcapturesallofthemultipatheffects.MisnumberofantennasandNisthechannellength.Hisassumedbetime-invarianthere,sofadingisconstant.s（k）containingthetransmittedsymbolsisdefinedas

s（k）=

（4）

Itisoftenconvenienttohandlesignalsinblocks.ThereforewemaycollectMconsecutivesnaoshotsofx（.）correspondingtimeinsatantsk,....,k+M–1andneglectinginterference,weget

x（k）=HS（k）+N（k）（5）

Usingmodel（5）somecommonSTalgorithmswillbeshowninnexttopicduetogivesomeinferenceonwhatisgoingtobeusedforreceiveralgorithmsonthenextsections.

II.C.STAlgorithms

Heresingleusercasewillbedialedandinterferencefromotheruserswillbetreatedasadditivenoise.

II.C.1.MLandMMSE

OnecriterionforoptimalityusedinSTprocessingisMaximumLikelihood（ML）.Maximumlikelihoodsequenceestimation（MLSE）seekstoestimatethedatasequencewhichismostlikelytohavebeensentgeventhereceivedsignalvector.AnotherfrequentlyusedcriterionisMinimumMeanSquareError（MMSE）.InMMSEweobtainanestimateofthetransmittedsignalasaspace-timeweightedsumofthereceivedsignalandseektominimizethemeansquareerrorbetweentheestimateandthetruesignal.

MLSE:

Usingthemodelin（5）,wewillassumethatNisspatiallyandtemporallywhiteandGaussian,andthereisnointerference.TheMLSEproblemisreducedtofindSsoastosatisfythefollowingcriterion

（6）

wherethechannelHisassumedtobeknown.ThisisgeneralizationofthestandardMLSEproblemiasnowthechannelisdefinedinspaceandintime.

MMSE:

Inapplicationswhereinterferenceispresent,wewillnothaveknowledgeoftheinterferencestatistics.WecannolongeruseMLSEandonealternativeapproachisMMSE.InMMSE,weseektofindaspace-timefilterthatlinearlycombinesthearrayoutputsuchthatthedifferencebetweenitsscalaroutputandthetruesignalisminimized.TheMMSEcriterionis

（7）

whereDisadelaychosentocentertheSTfilter.ThesolutiontothisLeastSquare（LS）problemfollowsfromthewellknownprojectiontheorem.Definingthespace-timemNxmN

covariancematrix

andWisgivenby

（8）

Therearemanytrade-offsbetweenthisapproach,LMS,andRLSthatspansnumericalstability,convergencespeedandcomputationalcomplexity.

II.C.2.BlindandNon-Blind（TrainingBased）Methods

Intheabovemethods,weassumethatthechanelisknownatreceiver.Inpractice,Hmaynotbeknown,butanestimatebyusingtrainingsignalscanbeused.Analternativeapproachconsistsof“blind”methodsthatdonotusetrainingsymbols,andinsteadusethepropertiesofthereceiversignaltodetermineHandS.Here,someoftheseproperties（structures）aredescribed.

II.C.2.a.SpatialStructure

ArrayManifold:

Asitissaidthatchannelmatrixincludesmultipatheffects,whichcontainsdirectionofsignalarrivals.Knowingarraymanifold,A,helpsdeterminethearrayresponsevector.Aincludestheeffectsofarraygeometry,elementpatternsandobjectsnearthebasestations.

II.C.2.b.TemporalStructures

ConstantModulus（CM）:

Inmanywirelessapplication,thetransmittedwaveformhasaconstantenvelope,i.e,GMSK（GaussianMinimumShiftKaying）inGSM.

FiniteAlphabet（FA）:

Thisstructureunderliesalldigitallymodulatedschemes.Themodulatedsignalislenearornon-linearmapofanunderlyingFA,i.e,QPSKwhichhasonly4differentphasecomponent.

DistancefromGaussianity:

ThedistributionofdigitallymodulatedsignalsisnotGaussian.Thispropertycanbeexploitedtoestimatethechannelfromhigh-orderstatistics（HOS）.

Cyclostationarity:

Usingcyclostationaritycanleadtosecond-orderstatistics-basedalgorithmstoidentify/equalizechannelHthataremoreattractivethanHOStechniques.Byusingcyclostationaritystructure,whichisbyoversamplingintemporallyandspatially,resultsinHmatrixtobetallandfullrank.TallnessisoneofthekeypropertyforblindestimationofH.

Toeplitz:

Thisisapowerfulstructurethatcapturestheunderlyingconvolutionprocesses,usuallyconstrainingX,HorStobeofblockToeplitzform.

Moredetailedexplanationofthesestructuresandinplementationscanbefoundin[1]andalsomethodsforchannelestimationandequalizationcanbefoundin[3].

III.SPACE-TIMEBLOCKCODING（STBC）

Theinformationcapacityofwirelesscommunicationssystemsincreasedramaticallybyemployingmultipletransmitandreceiverantennas.Anefficientapproachtoincreasedatarateoverwirelesschannelistoemplycodingtechniquesappropriatetomultipletransmitantennasinamelyspace-timecoding.Space-timecodingisdesignedformultipletransmitantennas.STcodesintroducetemporalandspatialcorrelationintosignalstransmittedfromdifferentantennas,inordertoprovidediversityatthereceiverandcodinggainoveranuncodedsystemwithoutsacrificingthebandwidth.Thespatial-temporalstructureofthesecodescanbeexploitedtofartherincreasethecapacityofwirelesssystemwitharelativelysimplereceiverstructure.STcodingandsignalprocessingtechniqueswithmultipletransmitantennashavebeenadoptedtoCDMA-2000andW-CDMA.

Next,somesub-classesofSTBC’swillbeintroduced.Astransmittedsignalcanbemodulatedwit