USF Libraries
USF Digital Collections

Resource allocation methodologies with fractional reuse partitioning in cellular networks

MISSING IMAGE

Material Information

Title:
Resource allocation methodologies with fractional reuse partitioning in cellular networks
Physical Description:
Book
Language:
English
Creator:
Aki, Hazar
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Adaptive Cluster Size
Capacity Maximization And Optimization
Channel Allocation
Grade Of Service
Overlaid Cellular Architectures
Dissertations, Academic -- Engineering Electrical Engineering -- Masters -- USF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: Conventional cellular systems have not taken full advantage of fractional frequency reuse and adaptive allocation due to the fixed cluster size and uniformed channel assignment procedures. This problem may cause more fatal consequences considering the cutting-edge 4G standards which have higher data rate requirements such as 3GPP-LTE and IEEE 802.16m (WiMAX). In this thesis, three different partitioning schemes for adaptive clustering with fractional frequency reuse were proposed and investigated. An overlaid cellular clustering scheme which uses adaptive fractional frequency reuse factors would provide a better end-user experience by exploiting the high level of signal to interference ratio (SIR). The proposed methods are studied via simulations and the results show that the adaptive clustering with different partitioning methods provide better capacity and grade of service (GoS) comparing to the conventional cellular architecture methodologies.
Thesis:
Thesis (M.S.E.E.)--University of South Florida, 2011.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Hazar Aki.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains 60 pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004836
usfldc handle - e14.4836
System ID:
SFS0028105:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2011 flu ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004836
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Aki, Hazar
0 245
Resource allocation methodologies with fractional reuse partitioning in cellular networks
h [electronic resource] /
by Hazar Aki.
260
[Tampa, Fla] :
b University of South Florida,
2011.
500
Title from PDF of title page.
Document formatted into pages; contains 60 pages.
502
Thesis
(M.S.E.E.)--University of South Florida, 2011.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
3 520
ABSTRACT: Conventional cellular systems have not taken full advantage of fractional frequency reuse and adaptive allocation due to the fixed cluster size and uniformed channel assignment procedures. This problem may cause more fatal consequences considering the cutting-edge 4G standards which have higher data rate requirements such as 3GPP-LTE and IEEE 802.16m (WiMAX). In this thesis, three different partitioning schemes for adaptive clustering with fractional frequency reuse were proposed and investigated. An overlaid cellular clustering scheme which uses adaptive fractional frequency reuse factors would provide a better end-user experience by exploiting the high level of signal to interference ratio (SIR). The proposed methods are studied via simulations and the results show that the adaptive clustering with different partitioning methods provide better capacity and grade of service (GoS) comparing to the conventional cellular architecture methodologies.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
590
Advisor:
Arslan, Huseyin .
653
Adaptive Cluster Size
Capacity Maximization And Optimization
Channel Allocation
Grade Of Service
Overlaid Cellular Architectures
690
Dissertations, Academic
z USF
x Engineering Electrical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4836



PAGE 1

Resource Allocation Methodologies with Fractional Reuse Partitioning in Cellular Networks by Hazar Ak A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering Department of Electrical Engineering College of Engineering University of South Florida Maj or Professor: H seyin Arslan, Ph.D. Richard Gitlin, Sc.D. Wilfrido Moreno, Ph.D. Date of Approval: March 22, 2011 Keywords: Adaptive Cluster Size, Channel Allocation, Overlaid Cellular Architectures, Capacity Maximization and Optimization, Grade of Service. Copyright

PAGE 2

DEDICATION TomyGrandmother, whowatchesmeoverallthetime,fromthesky...

PAGE 3

ACKNOWLEDGEMENTS Firstofall,IwouldliketothankmyadvisorDr.HuseyinArs lanforhissupportand guidance.Itwasapleasuretobeapartofhisresearchteam.I alsowouldliketothank bothmycommitteemembers,Dr.RichardGitlinandDr.Wilfri doMorenoforsharing theirvaluablecommentsandprofoundknowledge. IowemydeepestgratitudetomyteammatesM.CenkErturk,Sa diaAhmed,MuratKarabacak,SabihGuzelgoz,AliGorcin,AlphanSah in,M.BahadrCelebi, Ibrahim Demirdogen,Dr.SerhanYarkan,Dr.MustafaEminSahin,H asanBasriCelebi,AliRza Ekti.Thegrouphasbeenasourceoffriendshipsaswellasgoo dadviceandcollaboration. MytimeatUSFwasmadeenjoyableinlargepartduetothefrien dsandgroupsthat becameapartofmylife.IamgratefulfortimespentwithEmre Seyyal,RebekaDavidova, BojanaZivanovic;forRandallandSydneyWest'shospitalit yasInishedupmydegree, andformanyotherpeopleandmemories.Ialsowouldliketoth ankmytruebestfriends CerenSaglamandGozdeCobanoglufortheircompany.Eve nthoughweweremilesaway, theymanagedtoencouragemebyusingalmosteverytechnolog yindigitalcommunication. Lastly,Iwouldliketothankmyfamilyforalltheirlove,pat ience,andencouragement. Thisthesisanddegreewouldnothavebeenpossiblewithoutm ymom,dadandmysisters FulsenandCerenwhosupportedmeineverywayandeverystepi nallmypursuits.

PAGE 4

TABLEOFCONTENTS LISTOFTABLES iii LISTOFFIGURES iv ABSTRACT v CHAPTER1INTRODUCTION 1 1.1TheNecessityofReusingtheResources11.2ChannelAssignmentStrategies 2 1.3FrequencyReuseConcept 3 1.4FractionalFrequencyReuse 5 1.5ResourceAllocationApproachesinMacrocell-Femtocel lNetworks7 CHAPTER2OVERLAIDCELLULARARCHITECTURES9 2.1ReuseDistanceandClusterSize 9 2.2FrequencyReuseFactor 11 2.3CellRadiusRatio 12 CHAPTER3FRACTIONALREUSEPARTITIONINGSCHEMES19 3.1MaximalFractionalReusePartitioning(MFRP)193.2OptimalFractionalReusePartitioning(OFRP)22 CHAPTER4GRADEOFSERVICEPOINTOFVIEWINFRACTIONALFREQUENCYREUSE 24 4.1GradeofService 24 4.2GoS-orientedFRP 25 CHAPTER5COMPARISONSANDNUMERICALRESULTS28 5.1FundamentalDierencesBetweentheProposedMethods285.2CapacityAnalysis 31 5.3RegionalGradeofService 32 5.4EectoftheUserDistribution 33 5.5GradeofServiceLevel 35 CHAPTER6RESOURCEALLOCATIONINMACROCELL-FEMTOCELLNETWORKS 38 6.1CapacityofMacrocellwithFractionalReusePartitioni ng39 6.1.1MaximalFractionalReusePartitioning(MFRP)in Macrocell-FemtocellNetwork41 i

PAGE 5

6.1.2OptimalFractionalReusePartitioning(OFRP)inMacr ocellFemtocellNetwork 42 6.2CapacityofFemtocells 44 CHAPTER7CONCLUSIONANDFUTUREWORK47REFERENCES 49 ii

PAGE 6

LISTOFTABLES Table3.1DescriptionofparametersandnotationforFRPsch emes.20 Table5.1Parametersandtheirvaluesthatareusedinthenum ericalanalysis.31 Table6.1Descriptionofparametersandnotationformacroc ell-femtocell networkscenario. 40 iii

PAGE 7

LISTOFFIGURES Figure1.1Frequencyreuseconceptwithclustersizeof7.3Figure1.2Conventionalfractionalfrequencyreuseconcep t.6 Figure1.3Basicstructureoffemtocell-macrocellnetwork .7 Figure2.1Reusedistanceandcellradius. 9 Figure2.2Possibleclustersizevaluesfor0 i 10and i j 10.10 Figure2.3Co-channelinterferenceforworstcasecelledge assumption.12 Figure2.4CRRfor pthorderoverlaidcellularsystem.13 Figure2.5SIRvs.CRRfornandreceiversensitivity, n =4andsensitivity=18dB. 16 Figure2.6SIRvs.CRRfornandreceiversensitivity, n =6andsensitivity=18dB. 17 Figure2.7SIRvs.CRRfornandreceiversensitivity, n =2andsensitivity=8dB. 18 Figure5.1PartitioningrowchartsforvariousFRPschemes. 30 Figure5.2Eectivenumberofchannelvs.totalnumberofchan nel.32 Figure5.3GoSvs.totalnumberofchannelsoffourconcentri cregions( R1through R4)inthreedierentschemes(Conventional,MFRP,OFRP).33 Figure5.4Eectivenumberofchannelvs. 1and 2inOFRP.34 Figure5.5Eectivenumberofchannelvs.totalnumberofchan nelfor GoS-orientedFRP. 35 Figure5.6AverageGoSforvarious r vs.totalnumberofchannelsfor GoS-orientedFRP. 37 Figure6.1CRRfor pthorderoverlaidcellularsystemandbandwidthpartitioningforFRP 39 iv

PAGE 8

ABSTRACT Conventionalcellularsystemshavenottakenfulladvantag eoffractionalfrequencyreuse andadaptiveallocationduetothexedclustersizeandunif ormedchannelassignmentprocedures.Thisproblemmaycausemorefatalconsequencescon sideringthecutting-edge 4Gstandardswhichhavehigherdataraterequirementssucha s3GPP-LTEandIEEE 802.16m(WiMAX).Inthisthesis,threedierentpartitionin gschemesforadaptiveclusteringwithfractionalfrequencyreusewereproposedandin vestigated.Anoverlaidcellular clusteringschemewhichusesadaptivefractionalfrequenc yreusefactorswouldprovidea betterend-userexperiencebyexploitingthehighlevelofs ignaltointerferenceratio(SIR). Theproposedmethodsarestudiedviasimulationsandtheres ultsshowthattheadaptive clusteringwithdierentpartitioningmethodsprovidebett ercapacityandgradeofservice (GoS)comparingtotheconventionalcellulararchitecture methodologies. v

PAGE 9

CHAPTER1 INTRODUCTION Cellularsystemshavebeenawellestablishedsolutionforw irelesscommunicationover thelastthreedecades.Thearchitectureofthecellularsys temallowsresourceutilizationin thepower-spacedomain.Sincethefrequencyspectrumislim itedandexpensiveresource, ecientlymakeuseofittoachieveoptimumsystemcapacityi sextremelyimportant. 1.1TheNecessityofReusingtheResources Themaindesignobjectiveoftheearlymobileradiosystemsw astoplaceasingle, highpoweredantennaonatallbuildingorahighmountain.Ev enthoughthisapproach oeredawidecoveragearea,letalonethehugeamountofpower requirementsandhealth concerns,itwasalmostimpossibletoreusetheresourcewit houtcreatingsevereinterference totheotherusers.Forinstance,in1970s,BellMobileSyste mscouldonlysupporttwelve simultaneouscallsinNewYorkCity.Thesefactschangedthe wirelesscommunication conceptentirelyandsingle,highpoweredantennasbecamem any,lowpoweredtransmitters andwidecoverageareasbecamecells.Cellsmayhavevarious radiiintheranges(1kmto 30km),theboundariesofthemcanoverlapbetweenneighbori ngcellsandalsolargecells canbepartitionedintosmallercells. Besidestheinherentdesigncharacteristicsofthecellula rconcept,theemergingtechnologiesinwirelesscommunicationsystemshavealsobroughtcr ucialsetbackstothenetwork. Bothendusersandserviceprovidersdesirecuttingedgeper formancefromtheirdevices andtechnologies.Letalonehighqualityvoicecommunicati onormultimediamessaging; richweb-basedimplementations,audio-videostreamingan dInternetbrowsingarethemain applicationsthatdemandhigherperformancefromthesyste m.However,thelawofna1

PAGE 10

tureisviableinthewirelessdomainanditsurfacestheprob lemofscarcityinthemost importantresource:frequencyspectrum.Ontheonehand,ma jorlicensedbands,which aremostlyallocatedfortelevision-radiobroadcastingor militaryapplications.Ontheother hand,unlicensedbandsarealreadyovercrowdedbythenumer ousdevicesandtechnologies fromcordlessphones,wirelessLANstomicrowaveovensandh omeentertainmentproducts. Thus,reusingthelimitedresourcewithinthedesignatedba ndisamusttoovercomethe scarcityissues.1.2ChannelAssignmentStrategies Theconventionalmethodofallocatingchannelsiscalledx edchannelassignment(FCA), inwhichxednumberofchannelsareassignedtoeachcellacc ordingtoapresetclustersize inordertosatisfythedesiredsignalqualityatthecelledg e.EventhoughFCAschemes aresimpleandstraightforwardmethods,theydonotadaptto changingtracconditions anduserdistributions.Moreover,inFCA,theoverallavera gegradeofservice(GoS)ofthe systemisthesameastheGoSinacell.Sincetracincellular systemscanbenon-uniform withdistributedusers,axedallocationofchannelsinace llmayresulthighblockingin somecells,whileothersmighthavealargenumberofsparech annels[1],[2].Thiscould resultpoorresourceutilization. Non-uniformchannelallocationandchannelborrowingmeth odsareintroducedin[3],[4] tosolvethedrawbacksoftheFCAschemesanddierentchannel borrowingalgorithmsare oeredin[5].However,underheavytracconditions,thecha nnelborrowingmethods couldbeinadequateaswellbyincreasingtheblockingproba bilitywhichleadsareduction inchannelutilization[6]. InordertoovercomethesedecienciesofFCAschemes,dynam icchannelallocation (DCA)isoered.InDCA,allchannelsareplacedinapoolandas signedtoanewcall whenthesignaltointerferenceratio(SIR)criterionissat ised.Simulationsin[7],[8]and analysisin[9]showthatunderheavytracintensity,DCAco uldperformworse.The 2

PAGE 11

performanceanalysisofthecellularmobilesystemsusingd ynamicchannelallocationand theirdrawbacksarediscussedin[10],[11],[12]. Hybridchannelallocation(HCA)schemes[7],[13],[14]are thecombinationoftheboth FCAandDCAmethods.InHCA,channelsaredividedintoxedan ddynamicgroups.The dynamicchannelallocationisdonewhenthexedchannelsar eoccupied.Thisimproves theperformancegreatlybyproducingasignicantincrease inchannelutilization.However; hugeamountofcomputingandthereforehighcomputationalc omplexityisrequiredfor channelrearrangementsinthelargesystems[2].1.3FrequencyReuseConcept Thefundamentalideainthecellulararchitectureisthecap abilityofecientlyreusing theresourceinadesignatedregion.Withtheconceptofsmal lercoverageareaandreusing thesamefrequencies,agroupofresourcecanbeallocatedto eachbasestationandtherefore todierentcells. Figure1.1Frequencyreuseconceptwithclustersizeof7. 3

PAGE 12

Frequencyreuseisamethodofreusingthechannels/frequen ciesinordertoimprove bothcapacityandspectraleciencyasitisshowninFig.1.1 .Neighboringcellsshould havedierentfrequenciesbutthecellnexttotheneighborin gonecouldhavesameoperating frequencieswithoutsueringfrominterference.Inotherwo rds,thesamefrequencyshould beusedatleasttwocellsapartfromeachother. Frequencyreuseconceptalsoallowsserviceproviderstoha vemorecustomerswithin agivenlicenseandlimitedresource.Sinceitwasrstintro ducedtothecommunication domainbyBellLaboratoriesat1974,everygenerationandal mosteverystandardhas benetedfromthefrequencyreusemethodologies.Severale xamplesofgenerationswith standardswhichusetheconceptasfollows: 1G:AMPS,TACS, 2G:GSM,IS-95, 3G:WCDMA,CDMA2000,TD-SCDMA Ontheotherhand,itisimportanttokeepinmindthat,reusin gfrequenciesbydividing theallocatedbandandthenrepeatingthesameassignmentpr oducesatrade-obetween networkcapacityandreceptionqualityasfollows: Furtherawayseparatedcellswithsamefrequencyallocatio nsmayoeraguaranteein solvingtheinterferenceproblems;however,highernumber ofdivisionsofthespectrum willdecreasethetotalcapacityofthesystem. Ifthedivisionofthespectrumislimitedwithsmallnumbers ,capacityofthesystem willincrease;butsamefrequencieswhichareallocatedtot heneighboringcellsinclose approximationwillcausesevereinterferenceproblems. Itisunquestionablethateectivereuseoftheresourcesasa noverlaidmannercan highlyimprovetheeectivenessofthesystem.However,allo catingadedicatedchunkof spectrumineachcellmaynotbesucientforachievingthema ximumcapacityandspectral eciency.Severalothermethodologieswhicharebasedonba sicfrequencyreuseconcept 4

PAGE 13

wereintroducedinliterature.ForexampleinGSM,world'sm ostpopularstandardin2G formobilesystems,ReusePartitioning(RP)[15],[16]wasu sedasaveryusefultechnique forachievinghigherspectrumeciencybyusingregionalcl ustersizesineachindividual cell. Inreusepartitioning,acellisdividedintoseveralconcen tricSIRregionsandeachregion hasadierentclustersize.AmobileclosetoitsBSisassigne dachannelwithasmaller reusedistance,whereasamobilefarfromitsBSwithlowsign alqualityisassignedachannel withalargerreusedistance.Thisschemeallowslesscomple xsystemstoperformeciently withoutneedforadaptivemodulationschemesandpowerrear rangements.Moreavailable resourcecanbeoeredtotheenduserwithhavingasmallerclu stersizewhichleadstothe ultimategoalinoverlaidcellularsystemsisachievingthe clustersizeof1(N=1).However, withtheuniformusageofN=1,themostcelledgeusersaresue redfrominter-carrier interferenceandthiswillcauseadegradationinthetotals ystemcapacityandlowdata rateintransmission.1.4FractionalFrequencyReuse Restrictionscausedbytheuniformreusefactorcanbesolve dbyFractionalFrequency Reuse(FFR)method[17].TheFFRschemeseparatesthecellar eaintotwodierent geographicalregions:theinnercellareaclosetothebases tationandoutercellareanearto thecelledge.Theglobalreusefactorisfractionedtothein nercellareathereforedierent clustersizesorfrequenciesareusedintwodierentregions asitisshowninFig.1.2.In itssimplestform,FFRimplementsareusescheme-N(N > 1)systeminordertoprevent unacceptablelevelsofinterferencethatmightbeexperien cedbycelledgeusers. TheFFRschemewasrstlyproposedforGlobalSystemforMobi leCommunications (GSM)networks,however,itisalsoaveryeectivesolutionf orthe4Gstandardssuchas 3GPP-LTE[18]andIEEE802.16m(WiMAX)[19],[20]whereaggr essivespectrumreuse forachievinghighdatarateisnecessary. 5

PAGE 14

Figure1.2Conventionalfractionalfrequencyreuseconcep t. IntheconventionalFFRscheme,bothinnerandouterregionr eusefactorsarexedfor theeverycellandtheycannotbechangedindependentlybyan individualbasestations. Severalothermethodshavebeenproposedintheliterature: DynamicFractionalFrequency Reuse(DFFR),wherebothuserregionsareconsideredtobein thesametotalcellarea, wasoeredin[21].In[22],theauthorsoeredIncrementalFre quencyReuse(IFR)scheme. However;underoverloadscenarios,theIFRmethodcouldnot performbetterthanthe conventionalFFRscheme[23]. In[24],[25]theSoftFrequencyReuse(SFR)schemes,whichh asbeenadoptedinthe 3GPP-LTEsystem[26],[27],wereinvestigated.Thebasicid eaoftheSFRschemeisto applyN=1toinnerregionandN=3tothecell-edgeusers.Thep erformancewiththeusage oftheSFRschememightbeadvanced,comparedtotheclassica lreuseof1system,butthe resourcesarestillunderutilizedsincetheFRFisxedforb othareas[23].Aimingatthe limitationsoftheSFRscheme,theEnhancedFractionalFreq uencyReuse(EFFR)scheme wasoeredin[23],whichintendstokeeptheadvantagesofthe SFRmethodwhileavoiding itslimitationsespeciallyinoverloadsituations.Thestu dyinthisthesiscanbeconsidered asaonestep-forwardofFFRschemesintermsofpartitioning thecellulararchitecturein anoverlaidmannerfordierentSIRlevels. 6

PAGE 15

1.5ResourceAllocationApproachesinMacrocell-FemtocellNetworks Wirelessoperatorshavebeentryingtoexpandthelimitsoft hecellulararchitecturein ordertosatisfytheendusers'needsandapplications'requ irements.Besidespartitioning thecellinthecenterofsinglebasestationandallocatingt heresourcesaccordingly,several accesspointsinthesamecellandinthesamecoverageareaca nbeestablishedinorderto achievebettercapacityaswell. Thecellswithdierentpowerlevelsandthereforedierentco verageareascanbebuilt asamulti-tieredmannerinordertoenhancethecapacityoft hewirelesssystems[28],[29] asitisshowninFig.1.3.Femtocells,whicharelow-powersh ort-rangehomebasestations, haveastrongpotentialtoimprovethecapacityofnextgener ationwirelesssystemssince theyoerbetterlinkqualitiesandwiderspectrumresources forconnectedsubscribers[30]. Channelandresourceallocationtechniqueshavebecomemor echallengingwhenthe architectureofthesystemisbasedonmulti-tierednetwork s.Spectrumsharingapproaches formacrocell-microcellnetworkshavebeeninvestigatedi ntermsofcapacity,hand-orates andvelocityofthemobileusersin[31],[32].However,thea rchitectureofamacrocellFigure1.3Basicstructureoffemtocell-macrocellnetwork 7

PAGE 16

femtocellnetworkisratherdierent.Inmacrocell-femtoce llnetworks,sincefemtocellbase stationsareuserdeployed,wirelessoperatorsmaynothave fullycontrolonthenumberof femtocells,numberofusersineachfemtocellandinterfere nceconditions.Therefore,various scenariosshouldbeinvestigatedinordertooptimizethema crocell-femtocellnetwork. Inmacrocell-femtocelltwo-tierednetworks,theresource isallocatedbyusingeither sharedorsplitspectrummethods.Theformerapproachintro ducesthereuseofco-channel frequencyforsharingthespectrum.Inthismethod,cross-t ierinterferencemaycausecrucial setbackstothesystem.Inthesplitspectrumapproach,thea llocatedspectrumispartitioned betweentiers.Eachnetworkcanuseitsownsegmentofresour ceandthereforecross-tier interferenceisprevented[30]. Inordertomaximizethecapactiyofamacrocell-femtocelln etwork,adaptiveaccess operationoffemtocells[33],hybridresourceallocation[ 34]andadaptivetransmitpowers [35]haveformerlyproposedintheliterature.Capacitymax imizationinsplitandshared spectrummethodsarealsoinvestigatedindierentdomainss uchasdistributedantennas, macrocell-microcellnetworks,wiredrelaynetworks,mult i-hopwirelessnetworks,ad-hoc networks,andcognitivenetworks[36],[37]. Fractionalfrequencyreusemethodsarealsobenecialfort hetwo-tierednetworkstructures.InterferencemanagementschemesbasedonFFRformac rocell-femtocellnetworks areinvestigatedin[38]and[39]whereaLTEbasedarchitect ureisproposed.Also,afemtocellgatewayserverwhichhasthelocationandthroughput informationofthefemtocell basestationsisoeredasapartofthe3GPPtechnologyin[40] .Theservercanmanage theinterferencebyusingFFR[41].Adjustingtheclustersi zeaccordingtotheoperating environmentwasoeredin[42]. 8

PAGE 17

CHAPTER2 OVERLAIDCELLULARARCHITECTURES Limitingthesignalpowerofthetransmitterstoacertainle vel,frequencyreuseconcept isintroduced,wherethesamechunkoffrequenciesisusedov erandoverattightlyparceled cells.Onewaytoincreasethecapacityofacellulararchite ctureistoreducethechannel reusedistanceinthesystemwhileconsideringthereceiver sensitivityforSIR. 2.1ReuseDistanceandClusterSize Theco-channel(frequency)reuseratioincellulararchite cturewhichisdenedas[43]: D R = p 3 N; (2.1) Figure2.1Reusedistanceandcellradius. 9

PAGE 18

where D istheminimumrequireddistance(reusedistance)betweena nytwoco-channel cellsinacellularsystem, R istheradiusofthehexagonalcelland N isthenumberof individualcellsinacellgroup,clustersize,asitisshown inFig.2.1.Itisimportantto notethat, N canassumeonlytheintegervaluesi.e.,1,3,4,7,9,12,13,1 6,...asgenerally presentedbytheseries: N = i2+ j2+ ij; (2.2) while i 0and j i: (2.3) Applyingthisequationandtherequiredconditionsforallp ossiblevaluesof0 i 10 and i j 10givesthepossibleclustersizevaluesasitisshowninFig .2.2. Eventhoughtheoryoersinnitenumberofclustersizevalue s,inpractical,itisusually selectedas3,4,7,or12.Clustersizewhicharebelow3(only possiblesolutionisN=1)will suerfromtheunacceptablelevelofinterference(unlessot hercompansationsaremadelike Figure2.2Possibleclustersizevaluesfor0 i 10and i j 10. 10

PAGE 19

inCDMAandOFDMA-basedsystems)andaboveN=21willwasteth esystemresources bydividingthetotalfrequencybandintotoomanychunks.2.2FrequencyReuseFactor Therateofreusingtheresourcedenesthefrequencyreusef actor.Itcanbeformedas 1/NwhereNisthenumberofcellswhichcannotusethesamechu nkoffrequencies/channels forcommunication.Commonvaluesforthefrequencyreusefa ctorare1/3,1/4,1/7,1/9 and1/12. Inthecaseofkdirectionalsectoralantennaswithdierentd irectionsonthesameaccess pointorbasestation,communicationcanbeavailableforkd ierentregionsinthesame cell.kisusuallydenedas3.Reuseratioofk/Nindicatesad ivisioninresourcewithink directionalantennaspercell.Previouslyusedfrequencyr eusefactorandsectorexamples inthetechnologiesasfollows: NorthAmericanAMPS:3/7, MotorolaNAMPS:6/4, GSM:3/4. Theultimategoalofthewirelesscommunicationsystemsist omaintainclustersizeof 1ineverycell.However,duetotheinterferenceissues,iti snotpossibletoachievethegoal withoutmakingcompensationsinotherdomains.Forexample incodedivisionmultiple access(CDMA)basedsystems,frequencyreusefactorof1isa chieved.InCDMA-based architectures,neighboringcellsusetheexactsamefreque ncies;however,usersareseparated bycodesintransmissionineachbasestationratherthanfre quencies. CDMA-basedsystemsarenottheonlyoneswhichmakedierenti nterpretationsin achievingtheclustersizeof1.Orthogonalfrequency-divi sionmultipleaccess(OFDMA) basedsystems(forinstanceLTE)arealsodesignedwithafre quencyreusefactorof1.Becausesignalisnotseparatedacrossthefrequencybandinth esesystems,inter-cellresource managementiscrucialtoallocatethenecessaryresourcesa ndlimittheinterference. 11

PAGE 20

Recentlyalsoorthogonalfrequency-divisionmultipleacc essbasedsystemssuchasLTE arebeingdeployedwithafrequencyreuseof1.Sincesuchsys temsdonotspreadthesignal acrossthefrequencyband,inter-cellradioresourcemanag ementisimportanttocoordinates resourceallocationbetweendierentcellsandtolimitthei nter-cellinterference[44]. 2.3CellRadiusRatio Conventionalapproachincellularsystemsconsider N asaxednumberandtheydesign thearchitectureontheassumptionoftheworstcaseSIRfort heneighboringclusterswhich isgiveninFig.2.3isasfollows[43]: SIRmin 0 : 5 1 p 3 N 1n+ 1 p 3 N +1n+ 1 p 3 Nn; (2.4) where n istheenvironmentalpathlossexponent. Themaindesigncriteriaistooerserviceevenintheenduser sinthefurthestplacein thedesignatedcellarea.ThisisthereasonwhytheSIRlevel oftheuserswhicharecloseto Figure2.3Co-channelinterferenceforworstcasecelledge assumption. 12

PAGE 21

Figure2.4CRRfor pthorderoverlaidcellularsystem. thebasestation(BS)ishighersince(2.4)iscalculatedtot heworstcasecell-edgescenario inFig.2.3. Theratiooftheinnercellradiustotheouteroneisdenedto be cellradiusratio ( CRR )denotedby asfollows: m= Rm R (2.5) where mistheCRRof mthregion.Forinstance,innermostconcentricSIRregion's CRRiscalculatedas 1=R 1 R.Anillustrationoftheproposed pthorderoverlaidcellular systemwhichuses p dierentclusteringsizesisgiveninFig.2.4.Itisimportan ttodene thenewbordersofthecellaccordingtothecellradiusratio .Since Rp= R atthecelledge and R0= R atthebasestation,theborderlineCRRsforthecellcanbede nedasfollows: 0=0(basestation),and p=1(celledge) : (2.6) 13

PAGE 22

AnalyzingtheworstcasescenarioSIRgivenin(2.4)andFig. 2.3,thederivationof (2.14)canbeasfollows: PD= PT A Rn; (2.7) where PDreceivedsignalpowerofthedesiredBS, PTisthetransmitpower,Aisantenna gains,Ristheradiusofthehexagonalcell(celledge).Byca lculatingthereceivedpower fromtheinterferingBSsas: PI= PT A 2 ( D R )n+ 2 ( D )n+ 2 ( D + R )n : (2.8) ThereforeSIRlevelofacelledgeusercanbegivenasfollows : PD PI= 1 = 2 R D Rn+ R Dn+ R D + Rn(2.9) = 1 = 2 1 ( D=R ) 1n+ 1 D=Rn+ 1 ( D=R )+1n: (2.10) Byreplacing R with Rmin(2.10),wecanobtaintheworstcaseSIRforeachconcentri c regionas: PD PI m= 1 = 2 1 ( D=R m ) 1n+ 1 D=R mn+ 1 ( D=R m )+1n: (2.11) Byusing m=R m R,wecanrewrite(2.1)as D Rm= p 3 N m: (2.12) Thereforebyusing(2.12)in(2.11)wecanderive(2.14)as: PD PI m= 0 : 5 1 p 3 N= m 1n+ 1 p 3 N= m +1n+ 1 p 3 N= mn: (2.13) 14

PAGE 23

SIR ( m) 0 : 5 1 p 3 N= m 1n+ 1 p 3 N= m +1n+ 1 p 3 N= mn: (2.14) Fig.2.5presentsthedesiredclustersizefordierentCRRva lues.Itisimportantto notethatthepathlossexponent n isselectedas4fortheprovidedscenarioandlog-normal shadowing;smallscalefadingisnotconsideredasweintrod ucealong-termaverageSIRfor theconceptdemonstration.Assumingthatthereceivercoul dtolerate18dBofSIR,itis obviousthattheclustersizewillbechosenas7sincelowerc lustersizescouldnotsatisfy celledge( p=1)SIR.However,whenusersareclosertotheBS,SIRishighe rforthegiven receiversensitivity.Anadaptiveclusteringmechanismus ingfractionalreusepartitioning wouldtakeadvantageofitanddecreasetheclustersizeasit isshownintheenvelopein Fig.2.5. Variousotherpathlossexponentsandreceivertolerancesa reinvestigatedinFig.2.6 andFig.2.7inordertoshowthefeasibilityofthedierentfr actionalpartitioningschemes. Environmentalcircumstancesorreceiverdesignspecicat ionsdonotcauseanydrawback forfractioningthereuseratioanditisstillfavorablefor thesystemtoadapttheclustersize accordingtothedierentradiiandSIRvalues.InFig.2.6,pa thlossexponentischosen as6(i.e.non-lineofsight,veryrichscatterersanddiract ionsareintheenvironment) forconsideringthehighlypopulatedurbanareas.Whenther eceiversensitivelyiskeptthe sameas18dB,N=1canbeusedinalmosthalfofthecellarea.In Fig.2.7,ontheother hand,pathlossexponentisselectedas2forconsideringthe freespacemodelandreceiver sensitivityischosenas8dBfordemonstration.Inordertot oleratetheinterference(i.e. freespacemodelactslikeanempty,mirroredcorridorwitht hehighdegreeofrerectors), clustersizeisneededtoincreaseatthecelledge.Adapting theclustersizesstartingfrom N=12toN=1whileapproachingtothebasestationwithslight dierencesintheradius, fractionalreusepartitioningisstillbenecialforthesy stem. 15

PAGE 24

0 0.2 0.4 0.6 0.8 1 -10 0 10 20 30 40 50 60 70 80 CRR ( a )SIR (dB) N=1 N=3 N=4 N=7 Desired N Figure2.5SIRvs.CRRfornandreceiversensitivity, n =4andsensitivity=18dB. 16

PAGE 25

0 0.2 0.4 0.6 0.8 1 -20 0 20 40 60 80 100 120 CRR ( a )SIR (dB) N=1 N=3 N=4 Desired N Figure2.6SIRvs.CRRfornandreceiversensitivity, n =6andsensitivity=18dB. 17

PAGE 26

0 0.2 0.4 0.6 0.8 1 -10 -5 0 5 10 15 20 25 30 35 CRR ( a )SIR (dB) N=1 N=3 N=4 N=7 N=9 N=12 Desired N Figure2.7SIRvs.CRRfornandreceiversensitivity, n =2andsensitivity=8dB. 18

PAGE 27

CHAPTER3 FRACTIONALREUSEPARTITIONINGSCHEMES Inthischapter,twonewreusepartitioningschemeswereana lyzedandinvestigated. Thedenitionsoftheparametersusedforthecapacityandgr adeofservicederivationsare giveninTable6.1.3.1MaximalFractionalReusePartitioning(MFRP) Maximizingthetotaleectivecapacityisoneofthemostimpo rtantcriteriawhenthe systemisbeingdesigned.Especiallywiththecurrenthightechend-userdevices,forinstancesmartphonesandtabletPCs,therequireddataratefr omthebackhaulisextremely high.Sincetheconventionalsystemsweredesignedtooeras ervicefortheworstcase scenario(i.e.SIRissetinordertosatisfytheusersinthec ell-edge),theuserswithafairly bettersignalqualitycannottakeadvantageofit. Let ~ CnbetheeectivenumberofchannelineachconcentricSIRregio nintheconventionalclusteringschemewhereclustersizeisnotadaptive ,canbecalculatedasfollows: ~ Cn= CConv 2n 2n 1 ; where n =1 ; 2 ;:::p; (3.1) and CConv= CT ~ N ; (3.2) 19

PAGE 28

Table3.1DescriptionofparametersandnotationforFRPsch emes. Parameter Description CT Totalnumberofchannelsinthesystem. ~ N Conventionalsystem'sxedclustersize. CConv Conventionalsystem'stotalnum-berofchannelsineachcell. m Cellradiusratioof mthconcentric region, m 2f 0 ; 1 ; p g ~ Cn Requirednumberofchannelsin nthconcentricregion. Nn Adaptiveclustersizeof nthconcentricregion. C(1) n EectivenumberofchannelsinnthconcentricregionforOFRP scheme. C(2) n Allocatednumberofchannelac-cordingtotheareaofregionsforOFRPscheme. CR Remainingnumberofchannels. Ct n Fractionalcapacityof nthconcentricregionfort 2 f MFRP ; OFRP ; GoS g schemes. Ct Totaleectivenumberofchan-nelsfort 2f MFRP ; OFRP ; GoS g schemes. Pt n Blockingprobabilityofnthconcentricregionfor t 2f MFRP ; OFRP ; GoS g schemes. r DesiredGoSlevelforGoS-orientedFRPscheme. 20

PAGE 29

where CTisthetotalnumberofchannelsinthecluster, ~ N istheconventionalsystem's clustersizeand CConvisthetotalnumberofchannelineachindividualcellforthe conventionalscheme.Itisimportanttonotethat, C canalwaysbeconsideredasadesignated frequencyband,orsomeotherresourcewhichiswantedtomax imize.However;capacity oftheeachregioncanbeincreasedsinceSIRvaluecorrespon dingtoeachCRRisdierent. Maximizingthecapacityof pthorderoverlaidcellularsystemcouldbeachievedbyallocat ingmorechannelstothe innermost concentricSIRregionwhichisdenedbythesmallest clustersizei.e.,N=1,whilethecapacityofoutermost pthregioniskeptthesame: Cn= ~ Cn Nn; where n =2 ; 3 :::p; (3.3) while C1= CTpXn =2Cn; (3.4) where Cnisthechannelnumberofeachregionintheadaptiveclusteri ngscheme, Nnis adaptiveclustersizeand C1isthecapacityofthemaximizedinnermostregion.Therefor e thetotaleectivechannelnumberineachcellforMFRPscheme canbecalculatedas follows: CMFRP=pXn =1CMFRP n; (3.5) where CMFRP 1= C1; and CMFRP n= Cn Nnfor n =2 ; 3 ::p; (3.6) CMFRP nisthefractionalcapacityof nthconcentricregionforMFRPschemeand CMFRPisthetotaleectivecapacityofeachcelldenedbyadierent clustering Nn.Thusbyusing thisscheme,theeectivecapacityoftheoverlaidcellulars ystemisgreaterthanorinthe worstcasescenario(majorityofusersarecumulatedatthec elledgei.e.attheoutermost concentricSIRregion),equaltotheconventionalclusteri ngscheme.Thedetailedanalysis 21

PAGE 30

andsimulationsresultswiththecomparisonoftheotherpro posedmethodsareinChapter 5.3.2OptimalFractionalReusePartitioning(OFRP) Eventhoughmaximizingthelimitedresourceshasacruciali mportanceinthearchitecturedesign,sometimesitmightnotbethebestbothfortheus ersandforthesystem.For instance,theremaynotbeenoughusersintheclosevicinity oftheaccesspointforphysical reasons(i.e.basestationisplacedinarockymountain),in otherwords,adequateend usersintheinnermostconcentricregioncannotalwaysbegu aranteed.Moreover,extreme casescenariossuchasdisastersmightmaketheaggressiver esourceallocationunnecessary anduseless.Thisisthereasonwhyoptimizationoftheresou rcesshouldalwaysbeunder consideration. Optimizationin pthorderoverlaidcellularsystemcanbeachievedbydistribut ingthe channelsaccordingtotheregionswhilepreservingthenece ssarycapacityallocationineach region.Revisiting(3.3)intermsofoptimizationcouldbeg ivenasfollows: C(1) n= ~ Cn Nn; where n =1 ; 2 :::p; (3.7) where C(1) nistheeectivechannelnumberofeachregionforsatisfyingt hechannel allocationfortheconventionalclusteringschemewithada ptiveclustersize.Theremaining channelsduetotheusageofadaptiveschemecanbecalculate dbysubtractingthesumof thenecessarycapacityfromthetotalcapacityasfollows: CR= CTpXn =1C(1) n; (3.8) and C(2) n= CR 2n 2n 1 ; where n =1 ; 2 ;:::p; (3.9) 22

PAGE 31

where C(2) nistheallocatedchannelnumberaccordingtotheareaofregi ons.Itis importanttonotethat,inOFRPscheme, canbethoughtasuserdistributionineach cellwhichisexplainedinChapter5inmoredetail.Thetotal eectivenumberofchannel ineachcell, COFRP,canbecalculatedbyaddingthenecessarynumberofchannel inthe conventionalschemewithpartitioningcapacityaccording totheoptimizationmethodas follows: COFRP=pXn =1COFRP n; (3.10) where COFRP n= Cn+ C(2) n Nn; for n =1 ; 2 ::p: (3.11) Notethat COFRP nisthefractionalcapacityof nthconcentricregionforOFRPscheme. Thetotaleectivecapacityofeachcellwillsatisfythefoll owingequation: CMFRP COFRP CConv: (3.12) Thus,withthisscheme,theeectivecapacityoftheoverlaid cellularsystemisgreater thanorequaltotheconventionalclusteringscheme.Inthew orstcasescenario,bothMFRP andOFRPschemeswillconvergetotheconventionalmethodan dthereforetherewillbe nodegradationintheeectivechannelnumber,totalcapacit yandgradeofservice.The detailedanalysisandsimulationsresultswiththecompari sonoftheotherproposedmethods areinChapter5. 23

PAGE 32

CHAPTER4 GRADEOFSERVICEPOINTOFVIEWINFRACTIONALFREQUENCY REUSE Inthischapter,partitioningschemesintermsofblockingp robabilityperformancewas investigatedandaGoS-orientedFRPschemewasprovided.Th eGoSforcorresponding concentricSIRregionsareanalyzedseparatelyforoverlai darchitecturessincethenumber ofallocatedchannelsforeachregionaredierent.4.1GradeofService GradeofServiceistheblockingprobabilityofaconnection beingincompleteformore thanadenedtimeperiodwiththereferenceofthebusieshou rwherethetracintensityis maximum.Accesspointortheroutingequipmenthastheright toaccept,directordecline theincomingnewconnectionrequestfromtheusersinthecel lvicinity.Rejectionsdueto theheavytracloadresultinuncompletedrequests,howeve r,ifthesystemisdesigned withthemiscalculatedblockingprobability,alreadyesta blishedconnectionsinthecellwill alsobedropped. Itistheserviceprovider'sresponsibilitytomakesuretha tsucientnumberofresources areavailableforthespecicdemandleveloftheparticular area.Ifunnecessaryresource isallocatedtothecell,theremightbeanexcessivecapacit ywhichwillneverbeusedand therefore,limitedresourceswouldbewasted.Ontheotherh and,thelessthanrequired amountwouldcauseconnectiondropsinveryhighnumbers.Th isisthereasonwhyit isextremelyimportanttocalculatethecorrectGradeofSer viceandimplementittothe system. 24

PAGE 33

InordertocalculatetheGradeofService,thefollowingset ofassumptionscanbe made: Alltracthroughthenetworkisequalchancetrac,(i.e.al lcallarrivalsandterminationsareindependentrandomevents) Thereisstatisticalequilibrium,(i.e.theaveragenumber ofcallsdoesnotchange) Anycallthatencounterscongestionisimmediatelylost.[4 5] Fromthisassumptionset,ErlangdevelopedtheErlang-Bfor mulafortheblockingprobabilityinacellwhentheconventionalschemeisconsidered [46],itisshowninthefollowing: PConv=A C C PCk =0 A k k !; (4.1) and A = Au U; (4.2) where A isthetotaloeredtrac, Auisthecallarrivalrate, U isthetotalnumberof users, C istotalnumberofchannelsineachcellaccordingtotheconv entionalschemeand PConvisthetotalblockingprobabilityofthesystem. 4.2GoS-orientedFRP Usingthepreviouslydenedfractionalreusepartitioning schemes(MFRPandOFRP), GoSrequirementscanberedenedasfollows: Pt n=A C t n n C t n PC t n k =0 A kn k !; where n =1 ; 2 ::p; andt= f MFRP ; OFRP g ; (4.3) and An= Au Un(4.4) 25

PAGE 34

where Unisthenumberofusersinthe nthconcentricregion.Itisimportanttonote that,bothMFRPandOFRPschemeshavedierentGoSlevelsfore achregion,inother words,thesubscriberswhohavepaidthesameamountofmoney forthesameserviceare experiencingdierentservicequalityaccordingtotheirph ysicallocationsinthecell.If everyconcentricregionhaslowerGoSthanconventionalsch eme,thenitcanbeacceptable thatsomeareashaverelativelybetterblockingprobabilit ythanothers. Inordertosupervisethesystemfromtheoppositesite,ades iredblockingprobability canbedenedatrst,thentheallocationoftheresourcesis doneaccordingly.Amore controlledschemecanbegivenwithGoS-orientedFRPbyden ingaGoSlevel(i.e. r )for eachconcentricregioncanbegivenasfollows: PGoS n( CGoS n)=A C GoS n n C GoS n PC GoS n k =0 A kn k r; where n =1 ; 2 ::p: (4.5) Itisimportanttonotethat,fractionalcapacities,region alclustersizes,andtotalnumber ofchannelsshouldsatisfythisinequality:pXn =1Nn CGoS n CT(4.6) inordertoguaranteethattheresourceisallocatedtothere gionswiththelimitationof CT.ByprovidingadesiredlevelofGoSforeachconcentricregi on,thecapacitymaximizing partitioningcanbegivenasfollows: CGoS n= PGoS n 1( r ) ; where n =2 ; 3 ::p; (4.7) and CGoS 1= CTpXn =2CGoS n Nn: (4.8) Therefore,totalcapacityfortheGoS-orientedFRPcanbegi venas; 26

PAGE 35

CGoS=pXn =1CGoS n: (4.9) 27

PAGE 36

CHAPTER5 COMPARISONSANDNUMERICALRESULTS Inthischapter,threeproposedmethodsareimplementedint hebasicsimulationenvironment.Eectiveresourceswiththerespectofthetotalres ourcesarecomparedandGoS valuesofeachconcentricregionareinvestigated.Moreove r,theeectoftheuserdistributionintheoptimizationmethodareexplainedandandexampl eforaGoS-levelselectionis given.5.1FundamentalDierencesBetweentheProposedMethods Inordertoavoidsystemfailuresandend-useddissatisfact ions,choosingthecorrectfractionalfrequencyreusemethodiscrucialfortheservicepro vider.Sometimesoptimization maynotbetheperfectsolutionforthegivenarchitectureas wellasaggressivecapacity maximizationcouldendupwithunfairresourceallocation. Theessentialstepsofthethreeproposedmethodsaresummar izedintherowchartin Fig.5.1.Inordertocalculatethetotaleectivecapacityof eachindividualcell,boundaries oftheconcentricregions(i.e.CRRvalues)accordingtothe receiversensitivitiesshould bedeterminedasarststep.Afterthemandatoryresourceal locationineachareais done,desiredschemeischosenbetweenMFRP,OFRPorGoS-ori entedFRPmethodswhile consideringtheenduserrequirementsandthesystemspeci cations. InMFRPscheme,channelallocationisdoneaccordingtother eusepartitioningexcept fortheinnermostregion.Duetothesmallestclustersize,t heinnermostconcentricregion's capacityisdeterminedbysubtractingthesumoftheothersf romtotalcapacity.Byhaving this,withthesmallestclustersizeoftheinnermostconcen tricregion,designatedresources canbereusedmoreoftenthananyotherregions.Thismethoda llowstheserviceprovider 28

PAGE 37

toacquirethemaximumcapacitywhilepreservingthenecess ityresourceallocationineach region.Thisschemecanbeconsideredsuitableincaseswhic hthedeploymentareaisrat andend-usersarecumulatedtowardthecenter. InOFRPscheme,wheretheoptimizationisthekeypoint,inst eadofaggressiveresource distribution,reallocationoftheremainingchannelsared onebydistributingthemaccordingtotheconcentricregionsareas.Whileprotectingthemi nimumamountofitineach region,moreresourcescanbeavailableandaddedtothemreg ardingtothesurfacedimensions.Insteadofregionareas,userdistributionsineachr egioncanalsobeconsideredas adecisionpointfortheallocation.Inthisway,disastersi tuationsorunevenlydistributed surfaces/deploymentareascanbenetfromthesystem. InGoS-orientedFRPscheme,desiredminimumGoSlevelshoul dbeobtainedandthe capacityiscalculatedbyusinginversefunctionofblockin gprobabilityforouterregions. Afterndingthefractionalcapacities,innermostregion' sallocationisdonebysubtracting itfromthetotalcapacity.Itisimportanttonotethat,when GoS-orientedschemeisconsidered,sumofthemultiplicationoffractionalcapacitya ndregionalclustersizesshould alwaysbeequalorsmallerthanthetotalnumberofchannels. Afterdeterminingtheimportantaspectsofthesystemrequirements,userspecicat ionsandcomparingthetotal eectivecapacities,selectedschemecanbeimplementedtot hesystem. 29

PAGE 38

Figure5.1PartitioningrowchartsforvariousFRPschemes. 30

PAGE 39

Table5.1Parametersandtheirvaluesthatareusedinthenum ericalanalysis. Parameter Value p 4 ~ N 7 CT 1000 N1 1 N2 3 N3 4 N4 7 U 1000 Au 0.1 0 0 1 0.35 2 0.65 3 0.8 4 1 5.2CapacityAnalysis Inthissection,thenumericalresultsfortheproposedmeth odsofadaptiveclustering areproposedforMFRPandOFRPschemesandparametersaregiv enintheTable5.1.The pthorderoverlaidsystemwhere p ischosenas4accordingtothemodelinFig.2.5andfour concentricSIRregionsareconsideredforthesimulations. Fig.5.2illustratesthecapacity; intermsofeectivenumberofchannelineachcellaccordingt othetotalnumberofchannel intheclusterfortheconventionalmethodandMFRP,OFRPsch emes.Itisobviousthat, theeectivenumberofchannelisincreasingasthetotalnumb erofchannelisincreasing, asdescribedinFig.5.2. MFRPallocatestheredundantchannelstotheinnermostregi on(i.e.smallestcluster size)andthereforethismethodhasthelargestcapacityasi tisshowninFig.5.2.Onthe otherhand;OFRPdistributesthechannelsaccordingtothea reasofconcentricSIRregions assumingthatusersareuniformlydistributedinsidethece ll.EventhoughOFRPscheme's capacitycurvefallsbehindtheMFRPschemeinFig.5.2,itst illoersbetterresultsthan conventionalmethod.Itisimportanttonotethat;cotangen toftheanglebetweenthecurve andthex-axisyieldsthe clustersize inFig.5.2.Intheconventionalscheme,thecotangent 31

PAGE 40

0 100 200 300 400 500 600 700 800 900 1000 0 50 100 150 200 250 300 350 400 450 500 Total Number of ChannelEffective Number of Channel Per Cell C Conv C MFRP C OFRP Figure5.2Eectivenumberofchannelvs.totalnumberofchan nel. oftheangleis7(i.e. ~ N =7)asexpected.MFRPandOFRPscheme'scotangentoftheangl e yieldsthe averageclustersize Nave,ofthedesignedsystems.MFRP'saverageclustersize canbecalculatedfromFig.5.2as Nave=2.Maximalapproachoerstheclosestmethodfor reachingtheultimategoalinoverlaidcellularsystemswhi chisachievingtheclustersize, N =1. 5.3RegionalGradeofService Fig.5.3presentstheGoSofeachcell[resp.region]forconv entional[resp.MFRPand OFRP]schemes.Itisobviousthat;withaxednumberofusers ,(i.e. U =1000and Au=0 : 1),whilethetotalnumberofchannelincreases,theblockin gprobabilitydecreases exponentially.MFRPschemeimprovestheinnermostregion' sGoS( PMFRP 1)excessively, 32

PAGE 41

ontheotherhandtheGoSforouterregions( PMFRP 2, PMFRP 3and PMFRP 4),whichareusing higherclustersizes,areworsethantheconventionalschem e.Alternatively;usingOFRP scheme,theGoSforeachregionisdecreasedsincelessaggre ssivecapacitymaximization isappliedwhilepreservingtheGoSforeachregion.OFRPsch eme'sfourregioncurves arefallenundertheconventionalscheme'scurveandtheref ore,OFRPschemesoerbetter totalGoScomparingtotheconventionalscheme.5.4EectoftheUserDistribution WealsostudytheeectoftheuserdistributiontotheOFRPsch eme. 1and 2aredenedtobe;innerboundaryandouterboundaryoftheuse rdistributionsinsidethe 0 100 200 300 400 500 600 700 800 900 10 -3 10 -2 10 -1 10 0 Total Number of ChannelGrade of Service P 1 MFRP P 2 MFRP P 3 MFRP P 4 MFRP P Conv P 1 OFRP P 2 OFRP P 3 OFRP P 4 OFRP Figure5.3GoSvs.totalnumberofchannelsoffourconcentri cregions( R1through R4)in threedierentschemes(Conventional,MFRP,OFRP). 33

PAGE 42

concentricSIRregions.Theparameterset f 1, 2g i.e.0 <1<2< 1inwhichthe usersaredistributedinsidetheregiondenedbyn,wheren 2f 1R 2R g .Fig.5.4 showsthepercentageofeectivechannelallocationwithres pecttotheinnerandouter boundariesbasedonvarioususers'distributions.While 1increases,theeectivenumber ofchannelsdecreases,sincetheaverageSIRvaluesofusers distributedwithintheregion nisdecreasing.Moreover,while 2isincreasing,theoveralldistancebetweenBSandthe usersareincreasingandthereforetheeectivecapacitydec reases.Forinstance,when 1=0 and0 : 1 <2< 0 : 35,percentageofeectivechannelusageinthatareaiscalcu latedas 100%duetousers'SIRlevelsarewellenoughtoprovideaclus tersizeofonetotheentire system.ThissimulationshowsthatOFRPcanalsotakeadvant ageofthedistributionof theusersinsidetheconcentricSIRregionsandenhancethec apacityaccordingly. 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 10 20 30 40 50 60 70 80 90 100 b2 (outer boundary) Effective number of channel (%) b1=0 b1=0.3 b1=0.4 b1=0.5 b1=0.6 b1=0.7 b1=0.8 b1=0.9 Figure5.4Eectivenumberofchannelvs. 1and 2inOFRP. 34

PAGE 43

0 100 200 300 400 500 600 700 800 900 1000 0 100 200 300 400 500 600 700 800 Total Number of ChannelEffective Number of Channel Per Cell CConv CGoS for g =0.1 CGoS for g =0.05 CGoS for g =0.01 Figure5.5Eectivenumberofchannelvs.totalnumberofchan nelforGoS-orientedFRP. 5.5GradeofServiceLevel InordertoinvestigatethebehavioroftheGoS-orientedFRP scheme,aratherdierent simulationscenarioisconsideredcomparingtothesimulat ionsofMFRPandOFRP.Note thatthetotalnumberofchannelsandthenumberofusersinac ellshouldbechosen appropriatelytoanalyzetheGoS-orientedFRPschemesince theremightbecaseswhere thenumberofchannelscannotsatisfytheGoSlevelsdenedb y r values.Inotherwords, thereshouldbeatleastaminimumnumberofchannelsinwhich desiredGoSlevelsare satisedinthesystem.Thereforeinthisscenariotheusern umberisdenedasafunction totalnumberofchannels,i.e.,notxedasitisinthecasefo rthesimulationsofMFRPand OFRP. 35

PAGE 44

Fig.5.5showstheeectivenumberofchannelsversustotalnu mberofchannelsfor various r valuesinGoS-orientedFRPschemewiththereferenceofconv entionalcapacity. Notethatthecapacityincreasewiththeincreaseof r ,sincelesschannelscanbeallocated totheouterconcentricregionstosatisfytheirGoSrequire ment.Thereforemorechannels areassignedtotheinnermostconcentricregion,whereN=1, whichleadstoincreasein capacity. GoSperformanceforvarious r valuesareinvestigatedinFig.5.6.PartitioninginFig. 5.6showsthat,theouterregions'averageblockingprobabi litiessatisfythedesiredGoSlevels bybeingunderthe r limitsasitisshowninthesubplotinFig.5.6.Forthesakeof brevity, weprovideonlytheouterregions'GoSlevelswiththerefere nceoftheconventionalmethod sincetheinnermostregions'blockingprobabilitiesalrea dysatisfythedesiredGoSlevelsby reachingzerorapidlyduetotheaggressiveresourcealloca tion.Asitcanbeseeninboth Fig.5.5andFig.5.6,GoS-orientedFRPschemeoersbetterpe rformancebyincreasing thetotalcapacitywhileprotectingthedesiredblockingpr obabilityvaluescomparingtothe conventionalscheme.Itisimportanttonotethat,whileper formingthesimulationsinFig. 5.6,thenumberofusersinacellisconsideredasthehalfoft hetotalnumberofchannels inordertomakesurethatminimumnumberofchannelsareguar anteedinwhichdesired GoSlevelsaresatised.Notethatblockingprobabilitiesi nFig.5.3decreasesexponentially sincethenumberofusersarexedinacellwhereastheyareli nearinFig.5.6sincethe numberofusersareasthefunctionoftotalnumberofthechan nels. 36

PAGE 45

0 100 200 300 400 500 600 700 800 900 1000 10 -30 10 -25 10 -20 10 -15 10 -10 10 -5 10 0 Total Number of ChannelGrade of Service Conventional Average GoS of R2, R3, R4 for g =0.1 Average GoS of R2, R3, R4 for g =0.05 Average GoS of R2, R3, R4 for g =0.01 0 500 1000 10 -5 10 0 0.1 0.05 0.01 Figure5.6AverageGoSforvarious r vs.totalnumberofchannelsforGoS-orientedFRP. 37

PAGE 46

CHAPTER6 RESOURCEALLOCATIONINMACROCELL-FEMTOCELLNETWORKS Inthischapter,previouslyexplainedtworesourceassignm entschemescalledmaximal fractionalreusepartitioning(MFRP),optimalfractional reusepartitioning(OFRP)are proposedtomakeabandwidthpartitioninginmacrocell. Theproposedpartitioningschemescanbeconsideredasaone step-forwardofthepreviouslyproposedFFRschemesintheliteratureintermsofpa rtitioningthecellulararchitectureinanoverlaidmannerfordierentSIRlevelsasitisg iveninFig.6.1.Oncethe partitioningformacrocellforeachregionisdonewithoneo ftheaboveschemes,femtocells areallocatedwiththerestofthespectrum(i.e.,spectrumw hichmacrocellusersarenot using)withineachconcentricregion.Inbothschemes,clus teringsizeischangedaccording totheconcentricSIRregions.InMFRPscheme,abundantband widthisassignedtothe innermostregioninordertoacquiremaximumeectivecapaci tyfromtheinterestedcell. OFRPscheme,ontheotherhand,allocatesunusedbandwidtha ccordingtotheareasand thedistributionofusersintheconcentricSIRregions.The rexiblereuseofresourcemakes theseschemesveryattractivebothintermsofcapacityandt heGoS[47]. Anillustrationoftheproposed pthorderoverlaidcellularsystemwhichuses p dierent clusteringsizesisgiveninFig.6.1.Itisimportanttonote thattheassumptionswhichwere madeinChapter3arestillvalid.Thepathlossexponent n isselectedas4fortheprovided scenarioandlog-normalshadowing;smallscalefadingisno tconsideredasweintroducea long-termaverageSIRfortheconceptdemonstration.Assum ingthatthereceivercould tolerate18dBofSIR,itisobviousthattheclustersizewill bechosenas7sincelower clustersizescouldnotsatisfycelledge( p=1)SIR.However,whenusersareclosertothe BS,SIRishigherforthegivenreceiversensitivity. 38

PAGE 47

Figure6.1CRRfor pthorderoverlaidcellularsystemandbandwidthpartitioning forFRP 6.1CapacityofMacrocellwithFractionalReusePartitioning Inthissection,weanalyzetheMFRP,andOFRPschemesfromca pacitymaximization andoptimizationpointofviewsrespectively.Thedenitio nsoftheparametersusedforthe capacityaregiveninTable6.1. 39

PAGE 48

Table6.1Descriptionofparametersandnotationformacroc ell-femtocellnetworkscenario. Parameter Description i;j;k Indicesforthetiers,networksineachtierandusersineachnetwork,respectively. Ci;j;k Capacityofthe kthuserinthe ithtierand jthnetwork.Notethat, thereisonlyonemacrocell( j =1, for i =1)andseveralfemtocells ( j =1 ;::;NN ; 2,for i =2). Bi;j;k Bandwidthofthe kthuserinthe ithtierand jthnetwork. SINRi;j;k Signaltointerferenceplusnoisera-tio(SINR)ofthe kthuserinthe ithtierand jthnetwork. Pi;j;k Receivedpowerofthe kthuserin the ithtierand jthnetwork. B Totalbandwidthofthesystem. ~ N Conventionalsystem'sxedclustersize. ConvBM Conventionalsystem'stotalband-widthineachcell. n Cellradiusratioofconcentricre-gion, m 2f 0 ; 1 ; p g ~ Bn M Compulsoryamountofbandwidthin nthconcentricSIRregion. Nn Adaptiveclustersizeof nthconcentricregion. (1)Bn M Eectivebandwidthin nthconcentricregionforOFRPscheme. (2)Bn M AllocatedbandwidthaccordingtotheareaofregionsforOFRPscheme. BR Remainingportionofbandwidth. tBn l Fractionalbandwidthofnthconcentricregionfor t 2f MFRP ; OFRP g schemes, where l 2f M,F g ;MandF standsformacrocellandfemtocell,respectively. tBM Totalbandwidthofmacrocellfort 2f MFRP ; OFRP g schemes. 40

PAGE 49

6.1.1MaximalFractionalReusePartitioning(MFRP)inMacrocell-Femtocell Network Let ~ Bn MbetheeectivebandwidthineachconcentricSIRregioninthe conventional clusteringschemewhereclustersizeisnotadaptive,canbe calculatedasfollows: ~ Bn M=ConvBM 2n 2n 1 ; where n =1 ; 2 ;:::p; (6.1) andConvBM= B ~ N ; (6.2) where B isthetotalbandwidthinthecluster, ~ N istheconventionalsystem'scluster sizeandConvBMisthetotalbandwidthineachindividualcellfortheconven tionalscheme. However,capacityoftheeachregioncanbeincreasedsinceS IRvaluecorrespondingto eachCRRisdierent.Maximizingthecapacityof pthorderoverlaidcellularsystemcould beachievedbyallocatingmorebandwidthtothe innermost concentricSIRregionwhich isdenedbythesmallestclustersizei.e.,N=1,whiletheca pacityofoutermost pthregion iskeptthesame: Bn= ~ Bn M Nn; where n =2 ; 3 :::p; (6.3) while B1= B pXn =2Bn; (6.4) where Bnisthebandwidthofeachregionintheadaptiveclusteringsc heme, Nnis adaptiveclustersizeand B1isthecapacityofthemaximizedinnermostregion.Therefor e thetotaleectivebandwidthineachcellforMFRPschemecanb ecalculatedasfollows:MFRPBM=pXn =1 MFRPBn M; (6.5) 41

PAGE 50

whereMFRPB1 M= B1; andMFRPBn M= Bn Nnfor n =2 ; 3 ::p; (6.6) whereMFRPBnisthefractionalbandwidthof nthconcentricregionforMFRPscheme andMFRPBMisthetotaleectivebandwidthofeachcelldenedbyadieren tclustering Nn.Thusbyusingthisscheme,theeectivebandwidthoftheover laidcellularsystemis greaterthanorintheworstcasescenario(majorityofusers arecumulatedatthecelledge i.e.attheoutermostconcentricSIRregion),equaltotheco nventionalclusteringscheme. 6.1.2OptimalFractionalReusePartitioning(OFRP)inMacrocell-Femtocell Network Optimizationin pthorderoverlaidcellularsystemcanbeachievedbydistribut ingthe bandwidthaccordingtotheregionswhilepreservingthenec essarybandwidthallocationin eachregion.Revisiting(6.3)intermsofoptimizationcoul dbegivenasfollows:(1)Bn= ~ Bn M Nn; where n =1 ; 2 :::p; (6.7) where(1)Bnistheeectivebandwidthofeachregionforsatisfyingtheba ndwidthallocationfortheconventionalclusteringschemewithadapt iveclustersize.Theremaining bandwidthduetotheusageofadaptiveschemecanbecalculat edbysubtractingthesum ofthenecessarybandwidthfromthetotalbandwidthasfollo ws: BR= B pXn =1 (1)Bn; (6.8) and(2)Bn= BR 2n 2n 1 ; where n =1 ; 2 ;:::p; (6.9) where(2)Bnistheallocatedbandwidthaccordingtotheareaofregions. Itisimportant tonotethat,inOFRPscheme, canbethoughtasuserdistributionineachcell.The 42

PAGE 51

totaleectivebandwidthineachcell,OFRPBM,canbecalculatedbyaddingthenecessarybandwidthintheconventionalschemewithpartitionin gbandwidthaccordingtothe optimizationmethodasfollows:OFRPBM=pXn =1 OFRPBn M; (6.10) whereOFRPBn M= Bn+(2)Bn Nn; for n =1 ; 2 ::p: (6.11) NotethatOFRPBn Misthefractionalbandwidthof nthconcentricregionforOFRP scheme.Thetotaleectivebandwidthofeachcellwillsatisf ythefollowingequation:MFRPBMOFRPBMConvBM: (6.12) Thus,withthisscheme,theeectivebandwidthoftheoverlai dcellularsystemisgreater thanorequaltotheconventionalclusteringscheme.Inthew orstcasescenario,bothMFRP andOFRPschemeswillconvergetotheconventionalmethodan dthereforetherewillbe nodegradationintheeectivebandwidth,andtotalcapacity UsingthenotationgiveninTable6.1thecapacityofamacroc elluserwithineachregion canbewrittenasfollows:tCn 1 ; 1 ;k=tBn 1 ; 1 ;klog2 1+ Pn 1 ; 1 ;k In 1 ; 1 ;k+ Bn 1 ; 1 ;kN0! : (6.13) Thereforetotalcapacityofmacrocell-tierbecomestCMac=PXn =1 tCn Mac: (6.14) wheretCn Mac=N n U,1,1Xk =1 tBn 1 ; 1 ;klog2 1+ Pn 1 ; 1 ;k In 1 ; 1 ;k+ Bn 1 ; 1 ;kN0! : (6.15) 43

PAGE 52

where In 1 ; 1 ;kdenotesinterferencelevelforthe kthmacrocelluserinthe nthregion.The followingassumptionsetisconsideredforthemacrocellne twork(ASM-1): Inter-macrocellinterferenceisassumedtobenegligiblei .e., In 1 ; 1 ;k=0 ; 8 k;n Receivedpowerofeachuserisassumedtobeconstanti.e., Pn 1 ; 1 ;k= Pn M; 8 k n = 1 ; 2 :::p Bandwidthineachmacrocellisdistributedinaroundrobinf ashioni.e., Bn 1 ; 1 ;k=t B n M N n U,M; 8 k where Nn U ; 1 ; 1= Nn U,M, t = f MFRP ; OFRP g ,and Nn U ; M= NU ; M 2n 2n 1 Then,(6.15)canbeexpressedasfollows:tCMac=PXn =1 tBn Mlog2(1+n1) | {z }t C n Mac: (6.16) wheren1=P n M N n U,M t B n M N 0. 6.2CapacityofFemtocells Bandwidthoftheeachfemtocellcanbecalculatedasfollows :tBn F= B tBn M(6.17) = B pXi =1 ;i6 = n tBi M; (6.18) for n =1 ; 2 ::p: andt= f MFRP ; OFRP g : UsingthenotationgiveninTable6.1,thecapacityoffemtoc elluserk withthe jthfemtocellinthe nthregioncanbewrittenastCn 2 ;j;k= Bn 2 ;j;klog2 1+ SINRn2 ;j;k : (6.19) 44

PAGE 53

Signaltointerferenceplusnoiseratio(SINR)ofthe kthuserwiththe jthfemtocellin the nthregioncanbegivenas SINRn2 ;j;k= Pn 2 ;j;k In 2 ;j;k+tBn 2 ;j;kN0; (6.20) where In 2 ;j;kdenotesinterferencepowerofthe kthuserwiththe jthfemtocellinthe nthregionand N0isthespectraldensityofnoise.Thereforethetotalcapaci tyforthe femtocell-tier(tier-2)canbeexpressedasfollows:tCn Fem=N n N ; 2Xj =1 N n U ; 2 ;jXk =1 tCn 2 ;j;k=N n N ; 2Xj =1 N n U ; 2 ;jXk =1 tBn 2 ;j;klog2 1+ Pn 2 ;j;k In 2 ;j;k+tBn 2 ;j;kN0! : (6.21) Forthesakeofanalyticaltractability,weconsiderthefol lowingsimplifyingassumptions inassumptionset(ASM-2): Thenumberoftheusersineachfemtocellisassumedtobexed ,i.e., Nn U ; 2 ;j= NU,F; 8 j .Assumingthatfemtocellsareuniformlydistributedinama crocell, Nn N ; 2= NN ; 2( 2n 2n 1). Inter-femtocellinterferenceisassumedtobenegligible, andinter-tierinterferenceis assumedtobeconstantineachregioni.e., In 2 ;j;k= In; 8 j;k Receivedpowerisassumedtobeconstant,i.e., Pn 2 ;j;k= PF; 8 j;k Bandwidthineachfemtocellisdistributedinaroundrobinf ashion,i.e.,tB2 ;j;k=t B n F N n U ; 2 ;j=t B n F N U,F; 8 j;k 45

PAGE 54

Then,(6.21)canbeexpressedasfollows:tCFem=PXn =1Nn N ; 2 tBn Flog2(1+n2) | {z }t C n Fem: (6.22) wheren2=P F N U ; F N U ; F I n + t B n F N 0. Totalcapacityforeachpartitioningschemecanbegivenas:tCTot=PXn =1 tBn Mlog2(1+n1)+PXn =1 tBn FNn N; 2log2(1+n2) : (6.23) 46

PAGE 55

CHAPTER7 CONCLUSIONANDFUTUREWORK Inthisthesis,wepresenttheideaofpartitioningtheresou rcewithadaptiveclustersize andfractionalfrequencyreusebyexploitingthehighSIRle velandapplydynamicresource allocationmethodsaccordingtotheconcentricSIRregions .Conventionalmethodwith xedclustersizeisusedforcomparingtheproposedmethods whilerelationshipsbetween dierentchannelallocationmethodsandvariousfractional reuseschemesintheliterature aregiven.Theproposedmethodsshowthatthetotaleectivec apacityandGoSofthe conventionalschemecouldbeincreasedbyusingfractional frequencyreusewithadaptive overlayclustering. ThreenewresourceassignmentschemescalledMFRP,OFRP,an dGoS-orientedFRP areproposed.Theproposedschemesareinvestigatedinterm soffractionalfrequencyreuse withadaptiveclustersizes,whiletheGoSrequirementsare considered.Inallthreeschemes, clustersizeischangedaccordingtotheconcentricregions 'receiversensitivities.InMFRP scheme,abundantchannelsareassignedtotheinnermostreg ioninordertoacquiremaximumeectivecapacityfromtheinterestedcell.OFRPscheme, ontheotherhand,allocates unusedchannelsaccordingtotheareasurfacesanduserdist ributionstomaintainfairand optimumresourcedistributionforthesystem.GoS-oriente dFRPschemeoerscapacity maximizationsimilartoMFRPwhileresourceallocationisd oneinordertoprotectthe desiredGoS-levelsineveryconcentricregion.Therexible allocationoftheresource,adaptiveclustersizesandfractionalfrequencyreusemakethes eschemesveryattractivebothin termsofcapacityandtheGoS. Thisworkcanbeconsideredasapartofnextgenerationwirel essnetworkarchitecture suchas3GPP-LTEandIEEE802.16m(WiMAX)intermsofintelli gentFRPwhichwill 47

PAGE 56

allowincreasingcapacityinatwo-tierednetworkstructur e.Theproposedschemesinthis studyaregoingtobeusedfordesigningtierednetworks(e.g .femtocells,picocells,orwired relays)tosolvetheproblemofdead-zones. 48

PAGE 57

REFERENCES [1]M.ZhangandT.-S.Yum,\Thenon-uniformcompactpattern allocationalgorithm forcellularmobilesystems," IEEETrans.VehicularTechnology ,vol.40,no.2,pp. 387{391,March2002. [2]I.KatzelaandM.Naghshineh,\Channelassignmentschem esforcellularmobile telecommunicationsystems:Acomprehensivesurvey," IEEEPersonalCommunications ,pp.10{31,1999. [3]S.-H.Oh,\Prioritizedchannelassignmentinacellular radionetwork," IEEETrans. onCommun. ,vol.40,pp.1259{69,1992. [4]J.S.EngelandM.Peritsky,\Statisticallyoptimumdyna micserverassignmentin systemswithinterferingservers," IEEETrans.VehicularTechnology ,vol.22,pp.203{ 209,1973. [5]L.Anderson,\Asimulationstudyofsomedynamicchannel assignmentalgorithmsin highcapacitymobiletelecommunicationssystem," IEEETrans.VehicularTechnology vol.22,p.210.,1973. [6]S.TekinayandB.Jabbari,\Handoverandchannelassignm entinmobilecellularnetworks," IEEECommun.Mag. ,vol.29,1991. [7]T.J.KahwaandN.Georganas,\Ahybridchannelassignmen tschemeinlargescale cellular-structuredmobilecommunicationsystems," IEEETrans.onCommun. ,vol.26, pp.432{438,1978. [8]D.CoxandD.Reudink,\Increasingchanneloccupancyinl argescalemobileradio systems:Dynamicchannelreassignment," IEEETrans.VehicularTechnology ,vol. VT-22,pp.218{222,1973. [9]S.JordanandA.Khan,\Optimaldynamicallocationincel lularsystems," IEEETrans. onCommun. ,1993. [10]J.Zander,\Asymptoticboundsontheperformanceofacl assofdynamicchannel assignmentalgorithms," IEEEJSAC ,vol.11,pp.926{933,1993. [11]D.EverittandD.Manseld,\Performanceanalysisofce llularmobilecommunication systemswithdynamicchannelassignment," IEEEJSAC ,vol.7,1989. [12]J.ZanderandJ.Frodigh,\Capacityallocationandchan nelassignmentincellularradio systemsusingreusepartitioning," Elect.Lett ,vol.28,pp.438{440,1992. 49

PAGE 58

[13]J.SinandN.Georganas,\Asimulationstudyofahybridc hannelassignmentscheme forcellularland-mobileradiosystemswitherlang-cservi ce," IEEETrans.onCommun. ,vol.Com-9,pp.143{147,1981. [14]W.Yue,\Analyticalmethodstocalculatetheperforman ceofacellularmobileradiocommunicationsystemwithhybridchannelassignment," IEEETrans.Vehicular Technology ,vol.VT-40,pp.453{459,1991. [15]S.L.ChenandP.H.J.Chong,\Dynamicchannelassignmen twithrexiblereuse partitioningincellularsystems," IEEECommunicationsSociety ,pp.4275{79,2004. [16]P.H.J.ChongandC.Leung,\Capacityimprovementincel lularsystemswithreuse partitionings," J.ofCommun.andNetworks ,vol.3,pp.280{287,2001. [17]H.Lei,L.Zhang,X.Zhang,andD.Yang,\Anovelmulti-ce llofdmasystemstructure usingfractionalfrequencyreuse,"Sep.2007,pp.1{5. [18]\3rdgenerationpartnershipproject;technicalspeci cationgroupradioaccessnetwork; evolveduniversalterrestrialradioaccess(E-UTRA);phys icalchannelsandmodulation (release8),"3GPP,3GPPTS36.211,April2008. [19]H.Jia,Z.Zhang,G.Yu,P.Cheng,andS.Li,\Ontheperfor manceofieee802.16 ofdmasystemunderdierentfrequencyreuseandsubcarrierp ermutationpatterns," Jun.2007,pp.5720{5725. [20]\IEEEstandardforlocalandmetropolitanareanetwork s.part16:Airinterfacefor xedandmobilebroadbandwirelessaccesssystems;amendme nt2:Physicaland mediumaccesscontrollayersforcombinedxedandmobileop erationinlicensed bands,"IEEEStd802.16e-2005,Dec.2005. [21]S.AliandV.Leung,\Dynamicfrequencyallocationinfr actionalfrequencyreused ofdmanetworks," WirelessCommunications,IEEETransactionson ,vol.8,no.8,pp. 4286{4295,Aug.2009. [22]K.T.KimandS.K.Oh,\Anincrementalfrequencyreusesc hemeforanofdmacellular systemanditsperformance,"May.2008,pp.1504{1508. [23]Z.XieandB.Walke,\Enhancedfractionalfrequencyreu setoincreasecapacityof ofdmasystems,"Dec.2009,pp.1{5. [24]M.Bohge,J.Gross,andA.Wolisz,\Optimalpowermaskin ginsoftfrequencyreuse basedofdmanetworks,"May.2009,pp.162{166. [25]Y.XiangandJ.Luo,\Inter-cellinterferencemitigati onthroughrexibleresourcereuse inofdmabasedcommunicationnetworks,"April2007,pp.1{7 [26]G.T.S.G.R.A.Network,\Physicalchannelsandmodulat ion,release8,"TS-36.211, June2007. [27]Huawei,\Softfrequencyreuseschemeforutranlte,"3G PPR1-050507,May2005. 50

PAGE 59

[28]R.CoombsandR.Steele,\Introducingmicrocellsintom acrocellularnetworks:acase study," IEEETrans.Commun. ,vol.47,no.4,pp.568{576,Apr.1999. [29]L.-C.Wang,G.Stuber,andC.-T.Lea,\Architecturedes ign,frequencyplanning,and performanceanalysisforamicrocell/macrocelloverlayin gsystem," IEEETrans.Veh. Technol. ,vol.46,no.4,pp.836{848,Nov.1997. [30]V.Chandrasekhar,J.G.Andrews,andA.Gatherer,\Femt ocellnetworks:asurvey," IEEECommun.Mag. ,vol.46,no.9,pp.59{67,Sep.2008. [31]S.Kishore,L.Greenstein,H.Poor,andS.Schwartz,\So fthandoanduplinkcapacity inatwo-tiercdmasystem," IEEETrans.onWirelessCommun. ,vol.4,no.4,pp. 1297{1301,July2005. [32]T.KleinandS.-J.Han,\Assignmentstrategiesformobi ledatausersinhierarchical overlaynetworks:performanceofoptimalandadaptivestra tegies," IEEEJ.Sel.Areas Commun. ,vol.22,no.5,pp.849{861,June2004. [33]D.Choi,P.Monajemi,S.Kang,andJ.Villasenor,\Deali ngwithloudneighbors:The benetsandtradeosofadaptivefemtocellaccess,"in Proc.IEEEGlobalTelecommun. Conf.(GLOBECOM) ,Dec.2008,pp.1{5. [34]I.Guvenc,M.-R.Jeong,F.Watanabe,andH.Inamura,\Ah ybridfrequencyassignmentforfemtocellsandcoverageareaanalysisforco-chann eloperation," IEEECommun.Lett. ,vol.12,no.12,pp.880{882,Dec.2008. [35]V.Chandrasekhar,J.Andrews,T.Muharemovic,Z.Shen, andA.Gatherer,\Power controlintwo-tierfemtocellnetworks," IEEETrans.WirelessCommun. ,vol.8,no.8, pp.4316{4328,august2009. [36]V.ChandrasekharandJ.Andrews,\Spectrumallocation intieredcellularnetworks," IEEETrans.Commun. ,vol.57,no.10,pp.3059{3068,Oct.2009. [37]K.Huang,V.Lau,andY.Chen,\Spectrumsharingbetween cellularandmobilead hocnetworks:transmission-capacitytrade-o," IEEEJ.Sel.AreasCommun. ,vol.27, no.7,pp.1256{1267,Sept.2009. [38]R.-T.Juang,P.Ting,H.-P.Lin,andD.-B.Lin,\Interfe rencemanagementoffemtocell inmacro-cellularnetworks,"in Proc.IEEEWirelessTelecommun.Symp.(WTS) ,Apr. 2010,pp.1{4. [39]P.Lee,T.Lee,J.Jeong,andJ.Shin,\Interferencemana gementinLTEfemtocell systemsusingfractionalfrequencyreuse,"in Proc.Int.Conf.onAdv.Commun.Tech. (ICACT) ,vol.2,Feb.2010,pp.1047{1051. [40]3GPP,\Ts22.200v9.1.1-servicerequirementsforhome nodebsandhomeenodebs," Jun.2009. [41]I.C802.16m-09/1426,\Interferenceavoidanceandint erferencemitigationbetweenfemtocellbss,"Jul.2009. 51

PAGE 60

[42]H.-C.Lee,D.-C.Oh,andY.-H.Lee,\Mitigationofinter -femtocellinterferencewith adaptivefractionalfrequencyreuse,"in Proc.IEEEInt.Conf.onCommun.(ICC) May.2010,pp.1{5. [43]W.C.Y.Lee, MobileCellularTelecommunicationsSystems .McGraw-Hill,1989. [44]V.Pauli,J.Naranjo,andE.Seide,\Heterogeneouslten etworksandinter-cellinterferencecoordination," WhitePaper:NomorResearch ,Dec.2010. [45]J.Flood, TelecommunicationsSwitching,TracandNetworks ,2nded.NY:Prentice Hall,1998. [46]T.S.Rappaport, WirelessCommunications:PrinciplesandPractice ,2nded.NJ: PrenticeHall,2003. [47]H.Aki,M.Erturk,andH.Arslan,\Dynamicchannelalloc ationschemesforoverlay cellulararchitectures,"in Proc.IEEEWirelessTelecommun.Symp.(WTS) ,Apr.2010, pp.1{5. 52