cqi

CQIMeasurementandReportinginLTE:ANewFrameworkKambizHomayounfarandBijanRohani†PHYBIT,INC.AMBuilding5F,2-3-3HigashiGotanda,Shinagawa,Tokyo141-0022,JapanE-mail:[e-mail]†PHYBIT,INC.10-10CendexCenter,120LowerDeltaRoad,Singapore169208E-mail:[e-mail]hchannelqualitymeasurementinLong-TermEvolution(LTE)ishowtodefinechannelquality.Anotherishowtoreportit.WepresentanewframeworkforcalculatingtheChannelQualityIndicator(CQI)anddescribemethodsforefficientCQIreportsthatcomplywithLTEsignalingschemes.KeywordsLTE,CQI,AMC1IntroductionCommunicationoveratime-varyingradiochannelissubjecttoradiochannelimpairmentssuchasaddi-tivewhiteGaussiannoise(AWGN),flatandfrequency-selectivefading,andlog-normalshadowing.Thesein-troducelossesinthereceivedinformationanddegradethequalityofthedeliveredservice.Ifchannelbecomestoobadandthelinkbecomesuseless,andifitbecomestoogoodthenthelinkisusingupresourcesunnecessar-ily.Intime-varyingradiochannelsboththesescenarioscanariseinthelifeofaconnection.Differentappli-cationsrequiredifferentqualityofservice(QoS)levels:whatisgoodforvoicemaynotbegoodforvideo.Toen-surethattheQoSforaspecificapplicationismetundervaryingradiochannelconditions,radiolinkadaptationtechniquesbecomenecessary.Thisinvolvesradiolinkqualitymeasurementandcontrol.Measurementoftheradiolinkqualityismainlydoneatthereceiverandentailsestimationofoneoranumberofradiolinkmea-suressuchasthereceivedsignalstrength(RSS),thesignal-to-noiseratio(SNR),thebit-error-rate(BER)be-foreorafterthechanneldecoder,etc.Thecontrolpartofradiolinkadaptationinvolvesadaptingthemodu-lation,coding,and/orpowerofthetransmittedsignalwithinsystemcapabilitiesandconstraintsbasedontheradiolinkqualitymeasurements.Radiolinkadaptationatthetransmitterisdoneinresponsetolinkadaptationcommands/requestinanattempttomaintainQoSclosetoitsintendedtargetvalue.Effectivenessofradiolinkadaptationreduceswithincreasedtimevariationofthechannel.Thecommands/requestscanbecomeoutdatedinfastchannelsandresultinirrelevantadjustmentsinthetransmittedsignalparameters.Inthispaper,anewframework,namely,statisticalra-diolinkqualitycontrol(SRLQC),forthedesignofACMcontrolalgorithmsisdisclosed.Thekeydifferencebe-tweenthismethodandthestate-of-the-artisthatSR-LQCdoesnotrelyonanaccuratemappingbetweenSNRandBERtodecidethebestACMschemeforthechannelcondition.2ANewFrameworkforRadioQualityControlTheradiolinkqualityisrandominnatureandcanberepresentedbyarandomtime-series.Itshouldthenbepossibletoapplystatisticalprocesscontrol(SPC)toradiolinkadaptation.Thefollowingsectionsshowhow.2.1StatisticalProcessControlSPCisacollectionoftried-and-truemethodsfromablendofstatisticsandcontrolengineering[1].Ithasbeensuccessfullyappliedinindustriessuchasindus-trialautomationandchemicalengineeringformonitor-ingandcontrolofsophisticatedprocesses.InSPC,theoutputofaprocessisviewedasbeingrandominnature,andprovidespowerfultoolsformonitoringandcontrolofprocessesbasedonthestatisticsoftheprocess.InthecontextofSPC,aprocessisineitherofthefollowingtwostates:•Undercontrol:inthisstatetheprocessisonlyaf-fectedbycommoncauses.Commoncausescannotberemoved,andtheprocessvariationsareonlyduetothesecommoncauses.Theprocessinthisstateisstationary.•Outofcontrol:theprocessisaffectedbyspecialcauses.Theprocessvariationsareduetobothcom-1monandspecialcausesinthisstate.Theprocessisnon-stationaryinthisstate.Inordertorestoretheprocesstothestateofcontrol,specialcausesmustbeidentifiedandremoved.AnexampleofatypicalrandomprocessisshowninFigure1.Herethesamplesoftheprocesshavebeenrepresentedbythetime-seriesxkplottedagainstthetimeindexk.Forconvenience,takexktobehaveanormaldistribution(seeFigure2)withitsProbabilityDensityFunction(pdf)givenbyf(xk).Soxkcanvaryrandomlyarounditsmeanμxwhentheprocessinastateofcontrol.Inthisstateonlycommoncausesarepresentandtheprocessisstationary.Furthermore,thesamplevaluesxklieintheintervalμx±2σxwithaprobabilityof0.954,andfallintheintervalμx±3σxwithaprobabilityof0.997.Thatistosay,iftheprocessisinastateofcontrol,itssamplevaluesmustalmostcertainlyfallwithintheμx±3σxrange.Ifitdoesnot,somethingiswrongandadaptationisnecessary.The±3σxlimitsontheprocessvariationsareknownastheactionlevelsanddefinetheboundariesbeyondwhichtheprocessisdeemedtobeoutofcontrolornon-stationary.Hence,actionmustbetakensothattheprocesscanreturntoastateofcontrol.Inpractice,the±2σxareusedforprocessmonitoring.Theseareknownaswarninglevelsand,asthenamesuggests,canbeusedasalarmstoindicatethattheprocessisshowingsignsofgoingoutofcontrol(becomingnon-stationary).TheprocessmonitoringmethodillustratedinFigure1isknownasaShewhartchart.Itisasimpleandprac-ticalwayofmonitoringindustrialprocesses.However,therearemorepowerfulschemesthatcanbeadopteddependingontheneedsoftheprocesstobecontrolled.!#Figure1:Exampleofmonitoringthestateofaprocess(shownasxk)bycomparingtheobservedtime-seriesagainstcontrollevels$%&’(%)*)%+$,-./0%)*)%[phone]:Figure2:ControllevelsforaNormalProbabilityDis-tributionFunction3Statistic