畢設(shè)外文原文及譯文

北京聯(lián)合大學(xué)畢業(yè)設(shè)計(jì)〔論文〕任務(wù)書題目：OFDM調(diào)制解調(diào)技術(shù)的設(shè)計(jì)與仿真實(shí)現(xiàn)專業(yè)：通信工程指導(dǎo)教師：張雪芬學(xué)院：信息學(xué)院學(xué)號(hào)：班級(jí)：1101B姓名：徐嘉明一、外文原文EvolutionTowards5GMulti-tierCellularWirelessNetworks:AnInterferenceManagementPerspectiveEkramHossain,MehdiRasti,HinaTabassum,andAmrAbdelnasserAbstract—Theevolvingfifthgeneration(5G)cellularwirelessnetworksareenvisionedtoovercomethefundamentalchallengesofexistingcellularnetworks,e.g.,higherdatarates,excellentend-to-endperformanceanduser-coverageinhot-spotsandcrowdedareaswithlowerlatency,energyconsumptionandcostperinformationtransfer.Toaddressthesechallenges,5Gsystemswilladoptamulti-tierarchitectureconsistingofmacrocells,differenttypesoflicensedsmallcells,relays,anddevice-to-device(D2D)networkstoserveuserswithdifferentquality-of-service(QoS)requirementsinaspectrumandenergy-efficientmanner.Startingwiththevisionsandrequirementsof5Gmulti-tiernetworks,thisarticleoutlinesthechallengesofinterferencemanagement(e.g.,powercontrol,cellassociation)inthesenetworkswithsharedspectrumaccess(i.e.,whenthedifferentnetworktierssharethesamelicensedspectrum).Itisarguedthattheexistinginterferencemanagementschemeswillnotbeabletoaddresstheinterferencemanagementprobleminprioritized5Gmultitiernetworkswhereusersindifferenttiershavedifferentprioritiesforchannelaccess.Inthiscontext,asurveyandqualitativecomparisonoftheexistingcellassociationandpowercontrolschemesisprovidedtodemonstratetheirlimitationsforinterferencemanagementin5Gnetworks.Openchallengesarehighlightedandguidelinesareprovidedtomodifytheexistingschemesinordertoovercometheselimitationsandmakethemsuitablefortheemerging5Gsystems.IndexTerms—5Gcellularwireless,multi-tiernetworks,interferencemanagement,cellassociation,powercontrol.I.INTRODUCTIONTosatisfytheever-increasingdemandformobilebroadbandcommunications,theIMT-Advanced(IMT-A)standardshavebeenratifiedbytheInternationalTelecommunicationsUnion(ITU)inNovember2010andthefourthgeneration(4G)wirelesscommunicationsystemsarecurrentlybeingdeployedworldwide.ThestandardizationforLTERel-12,alsoknownasLTE-B,isalsoongoingandexpectedtobefinalizedin2014.Nonetheless,existingwirelesssystemswillnotbeabletodealwiththethousand-foldincreaseintotalmobilebroadbanddata[1]contributedbynewapplicationsandservicessuchaspervasive3Dmultimedia,HDTV,VoIP,gaming,e-Health,andCar2xcommunication.Inthiscontext,thefifthgeneration(5G)wirelesscommunicationtechnologiesareexpectedtoattain1000timeshighermobiledatavolumeperunitarea,10-100timeshighernumberofconnectingdevicesanduserdatarate,10timeslongerbatterylifeand5timesreducedlatency[2].Whilefor4Gnetworksthesingle-useraveragedatarateisexpectedtobe1Gbps,itispostulatedthatcelldatarateoftheorderof10Gbpswillbeakeyattributeof5Gnetworks.5Gwirelessnetworksareexpectedtobeamixtureofnetworktiersofdifferentsizes,transmitpowers,backhaulconnections,differentradioaccesstechnologies(RATs)thatareaccessedbyanunprecedentednumbersofsmartandheterogeneouswirelessdevices.Thisarchitecturalenhancementalongwiththeadvancedphysicalcommunicationstechnologysuchashigh-orderspatialmultiplexingmultiple-inputmultiple-output(MIMO)communicationswillprovidehigheraggregatecapacityformoresimultaneoususers,orhigherlevelspectralefficiency,whencomparedtothe4Gnetworks.Radioresourceandinterferencemanagementwillbeakeyresearchchallengeinmulti-tierandheterogeneous5Gcellularnetworks.Thetraditionalmethodsforradioresourceandinterferencemanagement(e.g.,channelallocation,powercontrol,cellassociationorloadbalancing)insingle-tiernetworks(evensomeofthosedevelopedfortwo-tiernetworks)maynotbeefficientinthisenvironmentandanewlookintotheinterferencemanagementproblemwillberequired.First,thearticleoutlinesthevisionsandrequirementsof5Gcellularwirelesssystems.Majorresearchchallengesarethenhighlightedfromtheperspectiveofinterferencemanagementwhenthedifferentnetworktierssharethesameradiospectrum.Acomparativeanalysisoftheexistingapproachesfordistributedcellassociationandpowercontrol(CAPC)isthenprovidedfollowedbyadiscussionontheirlimitationsfor5Gmulti-tiercellularnetworks.Finally,anumberofsuggestionsareprovidedtomodifytheexistingCAPCschemestoovercometheselimitations.II.VISIONSANDREQUIREMENTSFOR5GMULTI-TIERCELLULARNETWORKS5Gmobileandwirelesscommunicationsystemswillrequireamixofnewsystemconceptstoboostthespectralandenergyefficiency.Thevisionsandrequirementsfor5Gwirelesssystemsareoutlinedbelow.·Datarateandlatency:Fordenseurbanareas,5Gnetworksareenvisionedtoenableanexperienceddatarateof300Mbpsand60Mbpsindownlinkanduplink,respectively,in95%oflocationsandtime[2].Theend-to-endlatenciesareexpectedtobeintheorderof2to5milliseconds.Thedetailedrequirementsfordifferentscenariosarelistedin[2].·Machine-typeCommunication(MTC)devices:Thenumberoftraditionalhuman-centricwirelessdeviceswithInternetconnectivity(e.g.,smartphones,super-phones,tablets)maybeoutnumberedbyMTCdeviceswhichcanbeusedinvehicles,homeappliances,surveillancedevices,andsensors.·Millimeter-wavecommunication:Tosatisfytheexponentialincreaseintrafficandtheadditionofdifferentdevicesandservices,additionalspectrumbeyondwhatwaspreviouslyallocatedto4Gstandardissoughtfor.Theuseofmillimeter-wavefrequencybands(e.g.,28GHzand38GHzbands)isapotentialcandidatetoovercometheproblemofscarcespectrumresourcessinceitallowstransmissionatwiderbandwidthsthanconventional20MHzchannelsfor4Gsystems.·MultipleRATs:5Gisnotaboutreplacingtheexistingtechnologies,butitisaboutenhancingandsupportingthemwithnewtechnologies[1].In5Gsystems,theexistingRATs,includingGSM(GlobalSystemforMobileCommunications),HSPA+(EvolvedHigh-SpeedPacketAccess),andLTE,willcontinuetoevolvetoprovideasuperiorsystemperformance.Theywillalsobeaccompaniedbysomenewtechnologies(e.g.,beyondLTE-Advanced).·Basestation(BS)densification:BSdensificationisaneffectivemethodologytomeettherequirementsof5Gwirelessnetworks.Specifically,in5Gnetworks,therewillbedeploymentsofalargenumberoflowpowernodes,relays,anddevice-to-device(D2D)communicationlinkswithmuchhigherdensitythantoday’smacrocellnetworks.Fig.1showssuchamulti-tiernetworkwithamacrocelloverlaidbyrelays,picocells,femtocells,andD2Dlinks.Theadoptionofmultipletiersinthecellularnetworkarchitecturewillresultinbetterperformanceintermsofcapacity,coverage,spectralefficiency,andtotalpowerconsumption,providedthattheinter-tierandintratierinterferencesarewellmanaged.·Prioritizedspectrumaccess:Thenotionsofbothtrafficbasedandtier-basedPrioriti-eswillexistin5Gnetworks.Traffic-basedpriorityarisesfromthedifferentrequirementsoftheusers(e.g.,reliabilityandlatencyrequirements,energyconstraints),whereasthetier-basedpriorityisforusersbelongingtodifferentnetworktiers.Forexample,withsharedspectrumaccessamongmacrocellsandfemtocellsinatwo-tiernetwork,femtocellscreate“deadzones”aroundtheminthedownlinkformacrousers.Protectionshould,thus,beguaranteedforthemacrousers.Consequently,themacroandfemtousersplaytheroleofhigh-priorityusers(HPUEs)andlowpriorityusers(LPUEs),respectively.Intheuplinkdirection,themacrocellusersatthecelledgetypicallytransmitwithhighpowerswhichgenerateshighuplinkinterferencetonearbyfemtocells.Therefore,inthiscase,theuserprioritiesshouldgetreversed.AnotherexampleisaD2Dtransmissionwheredifferentdevicesmayopportunisticallyaccessthespectrumtoestablishacommunicationlinkbetweenthemprovidedthattheinterferenceintroducedtothecellularusersremainsbelowagiventhreshold.Inthiscase,theD2DusersplaytheroleofLPUEswhereasthecellularusersplaytheroleofHPUEs.·Network-assistedD2Dcommunication:IntheLTERel-12andbeyond,focuswillbeonnetworkcontrolledD2Dcommunications,wherethemacrocellBSperformscontrolsignalingintermsofsynchronization,beaconsignalconfigurationandprovidingidentityandsecuritymanagement[3].Thisfeaturewillextendin5Gnetworkstoallowothernodes,ratherthanthemacrocellBS,tohavethecontrol.Forexample,consideraD2DlinkatthecelledgeandthedirectlinkbetweentheD2DtransmitterUEtothemacrocellisindeepfade,thentherelaynodecanberesponsibleforthecontrolsignalingoftheD2Dlink(i.e.,relay-aidedD2Dcommunication).·Energyharvestingforenergy-efficientcommunication:Oneofthemainchallengesin5Gwirelessnetworksistoimprovetheenergyefficiencyofthebattery-constrainedwirelessdevices.Toprolongthebatterylifetimeaswellastoimprovetheenergyefficiency,anappealingsolutionistoharvestenergyfromenvironmentalenergysources(e.g.,solarandwindenergy).Also,energycanbeharvestedfromambientradiosignals(i.e.,RFenergyharvesting)withreasonableefficiencyoversmalldistances.ThehavestedenergycouldbeusedforD2Dcommunicationorcommunicationwithinasmallcell.Inthiscontext,simultaneouswirelessinformationandpowertransfer(SWIPT)isapromisingtechnologyfor5Gwirelessnetworks.However,practicalcircuitsforharvestingenergyarenotyetavailablesincetheconventionalreceiverarchitectureisdesignedforinformationtransferonlyand,thus,maynotbeoptimalforSWIPT.Thisisduetothefactthatbothinformationandpowertransferoperatewithdifferentpowersensitivitiesatthereceiver(e.g.,-10dBmand-60dBmforenergyandinformationreceivers,respectively)[4].Also,duetothepotentiallylowefficiencyofenergyharvestingfromambientradiosignals,acombinationofdifferentenergyharvestingtechnologiesmayberequiredformacrocellcommunication.III.INTERFERENCEMANAGEMENTCHALLENGESIN5GMULTI-TIERNETWORKSThekeychallengesforinterferencemanagementin5Gmulti-tiernetworkswillariseduetothefollowingreasonswhichaffecttheinterferencedynamicsintheuplinkanddownlinkofthenetwork:(i)heterogeneityanddensedeploymentofwirelessdevices,(ii)coverageandtrafficloadimbalanceduetovaryingtransmitpowersofdifferentBSsinthedownlink,(iii)publicorprivateaccessrestrictionsindifferenttiersthatleadtodiverseinterferencelevels,and(iv)theprioritiesinaccessingchannelsofdifferentfrequenciesandresourceallocationstrategies.Moreover,theintroductionofcarrieraggregation,cooperationamongBSs(e.g.,byusingcoordinatedmulti-pointtransmission(CoMP))aswellasdirectcommunicationamongusers(e.g.,D2Dcommunication)mayfurthercomplicatethedynamicsoftheinterference.Theabovefactorstranslateintothefollowingkeychallenges.·Designingoptimizedcellassociationandpowercontrol(CAPC)methodsformulti-tiernetworks:OptimizingthecellassociationsandtransmitpowersofusersintheuplinkorthetransmitpowersofBSsinthedownlinkareclassicaltechniquestosimultaneouslyenhancethesystemperformanceinvariousaspectssuchasinterferencemitigation,throughputmaximization,andreductioninpowerconsumption.Typically,theformerisneededtomaximizespectralefficiency,whereasthelatterisrequiredtominimizethepower(andhenceminimizetheinterferencetootherlinks)whilekeepingtheFig.1.Amulti-tiernetworkcomposedofmacrocells,picocells,femtocells,relays,andD2Dlinks.Arrowsindicatewirelesslinks,whereasthedashedlinesdenotethebackhaulconnections.desiredlinkquality.SinceitisnotefficienttoconnecttoacongestedBSdespiteitshighachievedsignal-to-interferenceratio(SIR),cellassociationshouldalsoconsiderthestatusofeachBS(load)andthechannelstateofeachUE.TheincreaseinthenumberofavailableBSsalongwithmulti-pointtransmissionsandcarrieraggregationprovidemultipledegreesoffreedomforresourceallocationandcell-selectionstrategies.Forpowercontrol,thepriorityofdifferenttiersneedalsobemaintainedbyincorporatingthequalityconstraintsofHPUEs.Unlikedownlink,thetransmissionpowerintheuplinkdependsontheuser’sbatterypowerirrespectiveofthetypeofBSwithwhichusersareconnected.Thebatterypowerdoesnotvarysignificantlyfromusertouser;therefore,theproblemsofcoverageandtrafficloadimbalancemaynotexistintheuplink.Thisleadstoconsiderableasymmetriesbetweentheuplinkanddownlinkuserassociationpolicies.Consequently,theoptimalsolutionsfordownlinkCAPCproblemsmaynotbeoptimalfortheuplink.Itisthereforenecessarytodevelopjointoptimizationframeworksthatcanprovidenear-optimal,ifnotoptimal,solutionsforbothuplinkanddownlink.Moreover,todealwiththisissueofasymmetry,separateuplinkanddownlinkoptimalsolutionsarealsousefulasfarasmobileuserscanconnectwithtwodifferentBSsforuplinkanddownlinktransmissionswhichisexpectedtobethecasein5Gmulti-tiercellularnetworks[3].·DesigningefficientmethodstosupportsimultaneousassociationtomultipleBSs:ComparedtoexistingCAPCschemesinwhicheachusercanassociatetoasingleBS,simultaneousconnectivitytoseveralBSscouldbepossiblein5Gmulti-tiernetwork.Thiswouldenhancethesystemthroughputandreducetheoutageratiobyeffectivelyutilizingtheavailableresources,particularlyforcelledgeusers.ThustheexistingCAPCschemesshouldbeextendedtoefficientlysupportsimultaneousassociationofausertomultipleBSsanddetermineunderwhichconditionsagivenUEisassociatedtowhichBSsintheuplinkand/ordownlink.·Designingefficientmethodsforcooperationandcoordinationamongmultipletiers:Cooperationandcoordinationamongdifferenttierswillbeakeyrequirementtomitigateinterferencein5Gnetworks.CooperationbetweenthemacrocellandsmallcellswasproposedforLTERel-12inthecontextofsoftcell,wheretheUEsareallowedtohavedualconnectivitybysimultaneouslyconnectingtothemacrocellandthesmallcellforuplinkanddownlinkcommunicationsorviceversa[3].Ashasbeenmentionedbeforeinthecontextofasymmetryoftransmissionpowerinuplinkanddownlink,aUEmayexperiencethehighestdownlinkpowertransmissionfromthemacrocell,whereasthehighestuplinkpathgainmaybefromanearbysmallcell.Inthiscase,theUEcanassociatetothemacrocellinthedownlinkandtothesmallcellintheuplink.CoMPschemesbasedoncooperationamongBSsindifferenttiers(e.g.,cooperationbetweenmacrocellsandsmallcells)canbedevelopedtomitigateinterferenceinthenetwork.Suchschemesneedtobeadaptiveandconsideruserlocationsaswellaschannelconditionstomaximizethespectralandenergyefficiencyofthenetwork.Thiscooperationhowever,requirestightintegrationoflowpowernodesintothenetworkthroughtheuseofreliable,fastandlowlatencybackhaulconnectionswhichwillbeamajortechnicalissueforupcomingmulti-tier5Gnetworks.Intheremainingofthisarticle,wewillfocusonthereviewofexistingpowercontrolandcellassociationstrategiestodemonstratetheirlimitationsforinterferencemanagementin5Gmulti-tierprioritizedcellularnetworks(i.e.,whereusersindifferenttiershavedifferentprioritiesdependingonthelocation,applicationrequirementsandsoon).Designguidelineswillthenbeprovidedtoovercometheselimitations.Notethatissuessuchaschannelschedulinginfrequencydomain,timedomaininterferencecoordinationtechniques(e.g.,basedonalmostblanksubframes),coordinatedmulti-pointtransmission,andspatialdomaintechniques(e.g.,basedonsmartantennatechniques)arenotconsideredinthisarticle.IV.DISTRIBUTEDCELLASSOCIATIONANDPOWERCONTROLSCHEMES:CURRENTSTATEOFTHEARTA.DistributedCellAssociationSchemesThestate-of-the-artcellassociationschemesthatarecurrentlyunderinvestigationformulti-tiercellularnetworksarereviewedandtheirlimitationsareexplainedbelow.·ReferenceSignalReceivedPower(RSRP)-basedscheme[5]:AuserisassociatedwiththeBSwhosesignalisreceivedwiththelargestaveragestrength.AvariantofRSRP,i.e.,ReferenceSignalReceivedQuality(RSRQ)isalsousedforcellselectioninLTEsingle-tiernetworkswhichissimilartothesignal-to-interference(SIR)-basedcellselectionwhereauserselectsaBScommunicatingwithwhichgivesthehighestSIR.Insingle-tiernetworkswithuniformtraffic,suchacriterionmaymaximizethenetworkthroughput.However,duetovaryingtransmitpowersofdifferentBSsinthedownlinkofmulti-tiernetworks,suchcellassociationpoliciescancreateahugetrafficloadimbalance.Thisphenomenonleadstooverloadingofhighpowertierswhileleavinglowpowertiersunderutilized.·Bias-basedCellRangeExpansion(CRE)[6]:TheideaofCREhasbeenemergedasaremedytotheproblemofloadimbalanceinthedownlink.ItaimstoincreasethedownlinkcoveragefootprintoflowpowerBSsbyaddingapositivebiastotheirsignalstrengths(i.e.,RSRPorRSRQ).SuchBSsarereferredtoasbiasedBSs.ThisbiasingallowsmoreuserstoassociatewithlowpowerorbiasedBSsandtherebyachieveabettercellloadbalancing.Nevertheless,suchoff-loadedusersmayexperienceunfavorablechannelfromthebiasedBSsandstronginterferencefromtheunbiasedhigh-powerBSs.Thetrade-offbetweencellloadbalancingandsystemthroughputthereforestrictlydependsontheselectedbiasvalueswhichneedtobeoptimizedinordertomaximizethesystemutility.Inthiscontext,abaselineapproachinLTE-Advancedisto“orthogonalize”thetransmissionsofthebiasedandunbiasedBSsintime/frequencydomainsuchthataninterference-freezoneiscreated.·AssociationbasedonAlmostBlankSub-frame(ABS)ratio[7]:TheABStechniqueusestimedomainorthogonalizationinwhichspecificsub-framesareleftblankbytheunbiasedBSandoff-loadedusersarescheduledwithinthesesub-framestoavoidinter-tierinterference.Thisimprovestheoverallthroughputoftheoff-loadedusersbysacrificingthetimesub-framesandthroughputoftheunbiasedBS.Thelargerbiasvaluesresultinhigherdegreeofoffloadingandthusrequiremoreblanksubframestoprotecttheoffloadedusers.GivenaspecificnumberofABSsortheratioofblankovertotalnumberofsub-frames(i.e.,ABSratio)thatensurestheminimumthroughputoftheunbiasedBSs,thiscriterionallowsausertoselectacellwithmaximumABSratioandmayevenassociatewiththeunbiasedBSifABSratiodecreasessignificantly.AqualitativecomparisonamongthesecellassociationschemesisgiveninTableI.ThespecifickeytermsusedinTableIaredefinedasfollows:channel-awareschemesdependontheknowledgeofinstantaneouschannelandtransmitpoweratthereceiver.Theinterference-awareschemesdependontheknowledgeofinstantaneousinterferenceatthereceiver.Theload-awareschemesdependonthetrafficloadinformation(e.g.,numberofusers).Theresource-awareschemesrequiretheresourceallocationinformation(i.e.,thechanceofgettingachannelortheproportionofresourcesavailableinacell).Thepriority-awareschemesrequiretheinformationregardingthepriorityofdifferenttiersandallowaprotectiontoHPUEs.Alloftheabovementionedschemesareindependent,distributed,andcanbeincorporatedwithanytypeofpowercontrolscheme.Althoughsimpleandtractable,thestandardcellassociationschemes,i.e.,RSRP,RSRQ,andCREareunabletoguaranteetheoptimumperformanceinmulti-tiernetworksunlesscriticalparameters,suchasbiasvalues,transmitpoweroftheusersintheuplinkandBSsinthedownlink,resourcepartitioning,etc.areoptimized.B.DistributedPowerControlSchemesFromauser’spointofview,theobjectiveofpowercontrolistosupportauserwithitsminimumacceptablethroughput,whereasfromasystem’spointofviewitistomaximizetheaggregatethroughput.Intheformercase,itisrequiredtocompensateforthenear-fareffectbyallocatinghigherpowerlevelstouserswithpoorchannelsascomparedtoUEswithgoodchannels.Inthelattercase,highpowerlevelsareallocatedtouserswithbestchannelsandverylow(evenzero)powerlevelsareallocatedtoothers.Theaggregatetransmitpower,theoutageratio,andtheaggregatethroughput(i.e.,thesumofachievableratesbytheUEs)arethemostimportantmeasurestocomparetheperformanceofdifferentpowercontrolschemes.TheoutageratioofaparticulartiercanbeexpressedastheratioofthenumberofUEssupportedbyatierwiththeirminimumtargetSIRsandthetotalnumberofUEsinthattier.Numerouspowercontrolschemeshavebeenproposedintheliteratureforsingle-tiercellularwirelessnetworks.Accordingtothecorrespondingobjectivefunctionsandassumptions,theschemescanbeclassifiedintothefollowingfourtypes.·Target-SIR-trackingpowercontrol(TPC)[8]:IntheTPC,eachUEtracksitsownpredefinedfixedtarget-SIR.TheTPCenablestheUEstoachievetheirfixedtarget-TABLEIQUALITATIVECOMPARISONOFEXISTINGCELLASSOCIATIONSCHEMESFORMULTI-TIERNETWORKSSIRsatminimalaggregatetransmitpower,assumingthatthetarget-SIRsarefeasible.However,whenthesystemisinfeasible,allnon-supportedUEs(thosewhocannotobtaintheirtarget-SIRs)transmitattheirmaximumpower,whichcausesunnecessarypowerconsumptionandinterferencetootherusers,andtherefore,increasesthenumberofnon-supportedUEs.·TPCwithgradualremoval(TPC-GR)[9],[10],and[11]:Todecreasetheoutagera-tiooftheTPCinaninfeasiblesystem,anumberofTPC-GRalgorithmswereproposedinwhichnon-supportedusersreducetheirtransmitpower[10]oraregraduallyremoved[9],[11].·Opportunisticpowercontrol(OPC)[12]:Fromthesystem’spointofview,OPCallocateshighpowerlevelstouserswithgoodchannels(experiencinghighpath-gainsandlowinterferencelevels)andverylowpowertouserswithpoorchannels.Inthisalgorithm,asmalldifferenceinpath-gainsbetweentwousersmayleadtoalargedifferenceintheiractualthroughputs[12].OPCimprovesthesystemperformanceatthecostofreducedfairnessamongusers.·Dynamic-SIRtrackingpowercontrol(DTPC)[13]:Whenthetarget-SIRrequirementsforusersarefeasible,TPCcausesuserstoexactlyhittheirfixedtarget-SIRsevenifadditionalresourcesarestillavailablethatcanotherwisebeusedtoachievehigherSIRs(andthusbetterthroughputs).Besides,thefixed-target-SIRassignmentissuitableonlyforvoiceserviceforwhichreachingaSIRvaluehigherthanthegiventargetvaluedoesnotaffecttheservicequalitysignificantly.Incontrast,fordataservices,ahigherSIRresultsinabetterthroughput,whichisdesirable.TheDTPCalgorithmwasproposedin[13]toaddresstheproblemofsystemthroughputmaximizationsubjecttoagivenfeasiblelowerboundfortheachievedSIRsofallusersincellularnetworks.InDTPC,eachuserdynamicallysetsitstarget-SIRbyusingTPCandOPCinaselectivemanner.Itwasshownthatwhentheminimumacceptabletarget-SIRsarefeasible,theactualSIRsreceivedbysomeuserscanbedynamicallyincreased(toavaluehigherthantheirminimumacceptabletarget-SIRs)inadistributedmannersofarastherequiredresourcesareavailableandthesystemremainsfeasible(meaningthatreachingtheminimumtarget-SIRsfortheremainingusersareguaranteed).Thisenhancesthesystemthroughput(atthecostofhigherpowerconsumption)ascomparedtoTPC.Theaforementionedstate-of-the-artdistributedpowercontrolschemesforsatisfyingvariousobjectivesinsingle-tierwirelesscellularnetworksareunabletoaddresstheinterferencemanagementprobleminprioritized5Gmulti-tiernetworks.ThisisduetothefactthattheydonotguaranteethatthetotalinterferencecausedbytheLPUEstotheHPUEsremainwithintolerablelimits,whichcanleadtotheSIRoutageofsomeHPUEs.ThusthereisaneedtomodifytheexistingschemessuchthatLPUEstracktheirobjectiveswhilelimitingtheirtransmitpowertomaintainagiveninterferencethresholdatHPUEs.Aqualitativecomparisonamongvariousstate-of-the-artpowercontrolproblemswithdifferentobjectivesandconstraintsandtheircorrespondingexistingdistributedsolutionsareshowninTableII.ThistablealsoshowshowtheseschemescanbemodifiedandgeneralizedfordesigningCAPCschemesforprioritized5Gmulti-tiernetworks.C.JointCellAssociationandPowerControlSchemesAveryfewworkintheliteraturehaveconsideredtheproblemofdistributedCAPCjointly(e.g.,[14])withguaranteedconvergence.Forsingle-tiernetworks,adistributedframeworkforuplinkwasdeveloped[14],whichperformscellselectionbasedontheeffective-interference(ratioofinstantaneousinterferencetochannelgain)attheBSsandminimizestheaggregateuplinktransmitpowerwhileattainingusers’desiredSIRtargets.Followingthisapproach,aunifieddistributedalgorithmwasdesignedin[15]fortwo-tiernetworks.Thecellassociationisbasedontheeffective-interferencemetricandisintegratedwithahybridpowercontrol(HPC)schemewhichisacombinationofTPCandOPCpowercontrolalgorithms.Althoughtheaboveframeworksaredistributedandoptimal/suboptimalwithguaranteedconvergenceinconventionalnetworks,theymaynotbedirectlycompatibletothe5Gmulti-tiernetworks.Theinterferencedynamicsinmulti-tiernetworksdependssignificantlyonthechannelaccessprotocols(orscheduling),QoSrequirementsandprioritiesatdifferenttiers.Thus,theexistingCAPCoptimizationproblemsshouldbemodifiedtoincludevarioustypesofcellselectionmethods(someexamplesareprovidedinTableI)andpowercontrolmethodswithdifferentobjectivesandinterferenceconstraints(e.g.,interferenceconstraintsformacrocellUEs,picocellUEs,orD2DreceiverUEs).AqualitativecomparisonamongtheexistingCAPCschemesalongwiththeopenresearchareasarehighlightedinTableII.Adiscussiononhowtheseopenproblemscanbeaddressedisprovidedinthenextsection.V.DESIGNGUIDELINESFORDISTRIBUTEDCAPCSCHEMESIN5GMULTI-TIERNETWORKSInterferencemanagementin5GnetworksrequiresefficientdistributedCAPCschemessuchthateachusercanpossiblyconnectsimultaneouslytomultipleBSs(canbedifferentforuplinkanddownlink),whileachievingloadbalancingindifferentcellsandguaranteeinginterferenceprotectionfor