SpectrumsensingfunctionIEEE1122

IEEEStd802.22-2011IEEEStandardforWirelessRegionalAreaNetworksPart22:CognitiveWirelessRANMediumAccessControl(MAC)andPhysicalLayer(PHY)specifications:PoliciesandproceduresforoperationintheTVBands576Copyright©2011IEEE.Allrightsreserved.AnnexC(informative)SensingThisannexcontainsdescriptionsofanumberofsensingtechniques.Asensingtechniqueisanimplementationofthespectrumsensingfunction.Thereareseveralclassificationsofsensingtechniques.Firstasensingtechniquecanbeclassifiedaseithersignalspecificorblind.Asignalspecificsensingtechniqueisbasedonfeaturesofspecificsignaltype.Ablindsensingtechniquedoesnotrelyonfeaturesofaspecificsignaltype.TheWRANcanbeginsensingachannelusingacoarsesensingtechnique.Ifasignalisdetectedthenthechannelisoccupied.Ifasignalisnotdetectedthenfinesensingcanbeusedtosenseforweakersignals.Asensingtechniquecanbeclassifiedaseitherafinesensingtechniqueoracoarsesensingtechnique.Afinesensingtechniqueisabletodetectthepresenceofasignalattherequiredsignalpowerlevel.Acoarsesensingtechniquemaynotbeabletodetectthepresenceofasignalattherequiredsignalpower;however,itmaystillbeusefulfordetectinghigher-powersignalsoftenisashorterperiodoftime.Eachsubclauseinthisannexdescribesaspecificsensingtechnique.Thefirstthreesensingtechniquesareblindsensingtechniquesandthenextelevenaresignalspecificsensingtechniques.Performanceforeachofthesensingtechniquesisalsoincludedineachsubclause.TheprimaryperformancemetricistherequiredSNRatwhichtheprobabilityofdetectionisgreaterorequalto0.9forallmultipathconditions.FortheATSCsignalsthemultipathwasmodeledbyusingtwelverepresentativesignalsfromIEEE802.22Contribution22-07-0359-00-0000[6]H.Foreachsensingtechniquethesensingtimeisalsoincluded.InmanycasesifthesensingtimeisincreasedtherequiredSNRwilldecrease.Somesensingtechniquesaresensitivetouncertaintyinthenoisepower.InotherwordstheestimateofthenoisepowerˆNPiswith!Δofthetruenoisepower,ˆNNEPPP#whereEP$%&&%Insomesensingtechniquesthedetectorthresholdisafunctionofthenoisepowerestimate,soanerrorinthatestimatecanaffectthedetectorperformance.IfthesensingtechniqueissensitivetonoiseuncertaintythentherequiredSNRisgiveforvariousvaluesofthenoiseuncertaintyparameterΔ.C.1BlindsensingtechniquesAblindsensingtechniquedoesnotdependonspecificsignalfeatures.Therearethreeblindsensingtechniquesdescribedinthisannex:theenergydetector,theeigenvaluesensingtechniqueandthemulti-resolutionsensingtechnique.IEEEStd802.22-2011IEEEStandardforWirelessRegionalAreaNetworksPart22:CognitiveWirelessRANMediumAccessControl(MAC)andPhysicalLayer(PHY)specifications:PoliciesandproceduresforoperationintheTVBands577Copyright©2011IEEE.Allrightsreserved.C.1.1Energy(power)detectorTheenergy,orpower,detectorisablinddetectorthatdoesnotrelyonfeaturesofaspecificsignaltype.Thissensingtechniquemaynotbeabletomeettheco-channelsensingrequirements;however,thissensingtechniqueisasimplemethodofquicklydeterminingifasignalispresentinthechannel,exceptforthecaseofweaksignals.Hencethisisacoarsesensingtechnique.TosensechannelNwiththissensingtechniquethesensingreceiveristunedtothatchannelandtheRFsignalisconvertedtoanintermediatefrequency(IF)wherea6MHzwidefilterisusedtofilteroutsignalsoutsidethechannelN.AfterIFfilteringthesignalisdown-convertedtoabase-bandsignal.Thebase-bandsignalisband-limitedto!3MHz.Theband-limitedsignaliscomplex(in-phaseandquadrature)sampledat6MHz.Thesampledsignalis,()()()ynxnwn#where()xnisthesignalcomponentand()wnisthenoisecomponent.Thepowerof()xnisSPandthepowerof()wnisNP.Theteststatisticforthissensingtechniqueis,11()*()MnTynynM$Thisisanestimateofthesignalpower.Misthenumberofsamples.IfthesummationwasnotscaledbydividingbyMthenTwouldbeanestimateofthesignalenergy.ThechoiceofwhetherornottoscalebyMdoesnotaffecttheperformanceofthesensingtechnique.Sincethisisarathersimpleteststatisticitispossibletowriteananalyticformulaforitprobabilitydensityfunction.Thisisnotalwayspossibleforasensingtechnique.Themeanandvarianceoftheteststatisticare,[]SNETPP#2()var[]SNPPTM#Foralargenumberofsamples(1M),bytheCentralLimitTheorem,wecanapproximatetheteststatisticasaGaussianrandomvariable,2()(),SNTSNPPftNPPM%&##'()*Giventhisprobabilitydensityfunctionfortheteststatisticonecanwriteaformulaforthedetectorthresholdbasedontherequiredfalsealarmprobability,1()1FANQPPM+,-.#/012where()Qxisthetailprobabilityofanormalizedzero-meanGaussianrandomvariable,andisgivenby,IEEEStd802.22-2011IEEEStandardforWirelessRegionalAreaNetworksPart22:CognitiveWirelessRANMediumAccessControl(MAC)andPhysicalLayer(PHY)specifications:PoliciesandproceduresforoperationintheTVBands578Copyright©2011IEEE.Allrightsreserved.dxxpxQxe!#$2/221)(Thesensingtechniquecanberepresentedasathresholdcomparison.IfT%&thenthechannelisclassifiedasoccupiedandifT%'thenthechannelisclassifiedasvacant.Theprobabilityofdetectiondependsonthesignalpower.Theformulafortheprobabilityofdetectionisgivenby,1()DSNSNMPQPPPP%()$#*#+,*-.Theperformanceofthissensingtechniqueindetectingweaksignalsreliesheavilyonaccurateknowledgeofthevalueofthenoisepower.Thisdependencyonaccurateknowledgeofthenoisepowerca