APdeployment.Inourfuturework,aheterogeneousAP deploymentwillbeconsideredtoreducethecost. ACKNOWLEDGEMENT ThisworkwassupportedbytheNationalNaturalScience FoundationofChinaunderGrantno.61701027,theBei- jingMunicipalNaturalScienceFoundationunderGrantno. 4182055andno.L182024,YoungEliteScientistsSponsorship ProgrambyCAST,andTalentInnovationProjectofBIT. (a)(b) Fig.6.Massiveaccessperformancecomparisonofthepro- REFERENCES posedschemesandthebaselinescheme:(a)AUDperfor- mance;(b)CEperformance.[1]M.Shirvanimoghaddam,M.Dohler,andS.J.Johnson,“Massivenon- orthogonalmultipleaccessforcellularIoT:Potentialsandlimitations,” IEEECommun.Mag.,vol.55,no.9,pp.55-61,Sep.2017. [2]C.Bockelmann,N.K.Pratas,G.Wunder,S.Saur,etal,“Towards Toverifythesuperiorityoftheproposedschemes,acon- massiveconnectivitysupportforscalablemMTCcommunicationsin5G ventionalmassiveaccesssolutionbasedonthetraditional networks,”IEEEAccess.,vol.6,pp.28969-28992,May.2018. networkarchitectureisemployedasthebaselinescheme, [3]M.Hasan,E.Hossain,andD.Niyato,“Randomaccessformachine-to- machinecommunicationinLTE-advancednetworks:Issuesandapproach- whereeachBSonlyseekstodetecttheusersfromitsowncell es,”IEEECommun.Mag.,vol.51,no.6,pp.86-93,Jun.2013. whiletreatingtheinter-cellinterferenceasnoise[18].Because [4]A.Laya,L.Alonso,andJ.Alonso-Zarate,“Istherandomaccesschannel unlessBSscooperate,itisunlikelythateachBScanobtain ofLTEandLTE-AsuitableforM2Mcommunications?Asurveyof alternatives,”IEEECommun.SurveysTuts.,vol.16,no.1,pp.4-16,1st knowledgeabouttheout-of-cellusers. Quart.,2014. Fig.6(a)comparestheAUDperformanceofmassiveaccess [5]Z.Zhang,X.Wang,Y.Zhang,andY.Chen,“Grant-freeratelessmultiple basedoncloudcomputingandfogcomputingparadigms,as access:Anovelmassiveaccessschemeforinternetofthings,”IEEE Commun.Lett.,vol.20,no.10,pp.2019-2022,Oct.,2016. wellasthebaselinescheme.Ascanbeobserved,thetwo [6]L.Liu,E.G.Larsson,W.Yu,P.Popovski,C.Stefanovic,andE.de proposedschemesoutperformthebaselinescheme.Sincethe Carvalho,“Sparsesignalprocessingforgrant-freemassiveconnectivity: inter-cellinterferencecanberecoveredviathecooperationAfutureparadigmforrandomaccessprotocolsintheinternetofthings,” IEEESignalProcess.Mag.,vol.35,no.5,pp.88-99,Sep.2018. ofAPs.Moreover,allthethreeschemescanachieveabetter [7]X.Shao,X.Chen,C.Zhong,J.Zhao,andZ.Zhang,“Auni?eddesign detectionperformancebyequippingmoreantennasateachAP ofmassiveaccessforcellularinternetofthings,”IEEEInternetofThings (RRH,F-APorBS),asalargerMcanenhancethestructuredJ.,vol.6,no.2,pp.3934-3947,Apr.2019. c [8]X.Shao,X.Chen,andR.Jia,“Adimensionreduction-basedjointactivity sparsityobservedfrommultiplereceiveantennas. detectionandchannelestimationalgorithmformassiveaccess,”IEEE InFig.6(a),fortheAUDperformancecomparisonof Trans.SignalProcess.,vol.68,pp.420-435,2020. [9]B.ShimandB.Song,“Multiuserdetectionviacompressivesensing,” twoprocessingparadigms,itisclearthat,byincreasingthe IEEECommun.Lett,vol.16,no.7,pp.972-974,July.2012. numberofF-APsforcooperation,i.e.,N,thedetection co [10]B.Wang,L.Dai,T.Mir,andZ.Wang,“Jointuseractivityanddata performanceoffogcomputingparadigmwillapproachthat detectionbasedonstructuredcompressivesensingforNOMA,”IEEE ofcloudcomputingparadigm.ItisworthnoticingthatfogCommun.Lett.,vol.20,no.7,pp.1473-1476,July.2016. [11]B.K.Jeong,B.Shim,andK.B.Lee,“MAP-basedactiveuseranddata computingevenoutperformscloudcomputingwhenN=3. co detectionformassivemachine-typecommunications,”IEEETrans.Veh. Thisisbecausethechannelstrengthsfromaspeci?cactive Technol.,vol.67,no.9,pp.8481-8494,Sep.2018. usertofarawayAPsareapproximatezero,itmaycausesome[12]L.LiuandW.Yu,“MassiveconnectivitywithmassiveMIMO-Part I:Deviceactivitydetectionandchannelestimation,”IEEETrans.Signal performancelossbyconsideringtheseAPswhenthesparsity Process.,vol.66,no.11,pp.2933-2946,Jun.2018. ratioisre?nedasline7ofAlgorithm1.Furthermore,thereisa [13]M.Ke,ZGao,YWu,XGao,andRSchober“Compressivesensing tradeoffbetweentheperformanceandthecostofpracticalAPbasedadaptiveactiveuserdetectionandchannelestimation:Massive accessmeetsmassiveMIMO,”IEEETrans.SignalProcess.,vol.68,pp. deployment.Thefogcomputingcanreapafasterresponse 764-779,2020. timeandabetterperformancewhileincreasesthecostof [14]X.Xu,X.Rao,andV.K.N.Lau,“Activeuserdetectionandchannel APdeployment.Fig.6(b)depictstheCENMSEperformanceestimationinuplinkCRANsystems,”inProc.Int.Conf.Commun(ICC), Jun,2015,pp.2727-2732. oftheconsideredschemes,whichfurtherdemonstratesthe [15]Q.He,T.Q.S.Quek,Z.Chen,Q.Zhang,andS.Li,“Compressivechan- superiorityoftheproposedfogcomputingbasedscheme. nelestimationandmulti-userdetectioninC-RANwithlow-complexity methods,”inIEEETrans.WirelessCommun.,vol.17,no.6,pp.3931- 3944,Jun.2018. V.CONCLUSION [16]Q.He,Q.Zhang,T.Q.S.Quek,Z.Chen,andS.Li,“Distributed optimizationinfogradioaccessnetworks?channelestimationandmulti- Thispaperinvestigatesthegrant-freemassiveaccessand userdetection,”inProc.Int.Symp.ModelingandOptimizationinMobile, proposestwoAUDandCEschemesbasedoncloudcomputingAdHoc,andWirelessNetwork(WiOpt),May,2018,pp.1-8. [17]S.Dustdar,C.Avasalcai,andI.Murturi,“Edgeandfogcomputing: andfogcomputingparadigms,respectively.Bycooperatingall Visionandresearchchallenges,”inProc.Int.Conf.Service-OrientedSyst APsinthenetwork,theproposedschemescanachieveamuch Engineer(SOSE),Apr,2019,pp.96-105. bettermassiveaccessperformancethanthenon-cooperative[18]Z.Chen,F.Sohrabi,andW.Yu,“Multi-cellsparseactivitydetection formassiverandomaccess:MassiveMIMOversuscooperativeMIMO,” solutionbasedonthetraditionalnetworkarchitecture.Fur- IEEETrans.WirelessCommun.,vol.18,no.8,pp.4060-4074,Aug.2019. thermore,comparedtocloudcomputingparadigms,thefog computingparadigmreapsafastresponsetimeandevena betterAUDandCEperformance,butincreasethecostof |
|