TALLAUFER-ARRIS JEROENPUTZEYS-ARRIS UFFECALLESEN-STOFA IT SALIVE!GETTINGTO SUCCESSFULR-PHY DEPLOYMENT: DO SANDDON TS

TALLAUFER-ARRIS JEROENPUTZEYS-ARRIS UFFECALLESEN-STOFA IT SALIVE!GETTINGTO SUCCESSFULR-PHY DEPLOYMENT: DO SANDDON TS

TABLEOFCONTENTS INTRODUCTION.4 Market.5 CableMarketDriver The"NeedforSpeed".5 DistributedAccessArchitectures(DAA).7 DOCSIS3.1Introduction.10 COMPANY(STOFA)OVERVIEW.11 CompanyOverview.11 NetworkArchitectureattheStartingPoint.12 MotivationforNetworkUpgrade.12 OPERATORDRIVERSFORR-PHY.13 DriversandBenefitsofRemotePHY.13 STOFA SDRIVERSFORR-PHY.15 PLANNING.16 NetworkArchitecture.16 PlanningtheFiberandInterconnectNetwork.17 RFPlantUpgradeandUpdatingNetworkInventorySystem.19 CIN Planning.20 IEEE1588Timing.21 ProductQualificationPhases.23 InitialTestingandSolutionEvaluation.24 IntegrationintotheNetwork.25 Automation.26 DEPLOYMENTANDOPERATIONALRESULTS.29 FieldDeployment.29 2 WEB BLOG www.arris.com www.arriseverywhere.com

TABLEOFCONTENTS PERFORMANCEIMPROVEMENT.30 RFPerformance.30 Space&PowerintheHeadend.33 OperationalSimplification.37 CONCLUSION.38 KeyBenefits.38 KeyChalenges.40 MajorTakeaways.41 ABBREVIATIONS.43 RELATEDREADINGS.46 Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 3

INTRODUCTION Thispaperisanoperationaloverview ofoneofthefirstremotephydeploymentsinthe world,atadanishoperator Stofa.Beinganearlyadopter,theentireprocessofselecting thetechnology,planningthenew networkdesign,anddeployingtheproductswasan unchartedterritory.stofaandarris,theselectedvendorfortheproject,havelearned manylessonsfrom theexperience,whicharealdetailedinthispaper. Inaddition,wehaveanalyzedthebenefitsthattheRemotePHYnetworkupgradehas providedtotheoperators,includingmeasuredimprovementsinsignaltonoiseratio(snr) inthefieldandspaceandpowersavingintheheadend. TheresultswepresentdemonstratethegreatbenefitsoftransitioningtoRemotePHY. First,thereisaclearimprovementinSNRvaluesintheupstream (US)direction,whichwil alow StofatogotohighermodulationordersusingDOCSIS3.1US.Inthedownstream (DS) direction,resultsarelessconclusive.thetransitiontor-phywasaccompaniedwithamove ofdocsis3.0channelstohigherpartofthespectrum,whichmayhavecounteredthe improvement.inaddition,someelementsinthenetworklikethedropcablewerenot replaced,andmayaffectthemeasuredsnratthecablemodem (CM)side. Second,weanalyzedthespaceandpowersavingintheheadendthatresultedfrom the migrationfrom thepreviousnetworkarchitecture(modularcmts)toremotephy.the analysisshowedsignificantspacesavingsfrom about70ruto18ruintheheadendfor themaccoreandnetworksupport,andlessthanhalfthepowerrequiredintheheadend forsupportingthesefunctions.inaddition,theoperationalcomplexityisdramaticaly reduced,duetothemajorreductionindevicesthatneedtobemanaged,andthe automatedmanagementthatisintroduced.aspartoftheproject,stofa'sabilitytodo proactivenetworkmanagementactionsisgreatlyimproved,aswelasitsabilityto effectivelygrow thenetworkandcustomerofferingwithoutanytruckrolsorfieldwork. Althesebenefitswilbereviewedinthepaper,aswelasthestepsStofatookinorderto prepareandsuccessfulydeploythenew technologyinthefield,servingthousandsof subscribers. 4 WEB BLOG www.arris.com www.arriseverywhere.com

MARKET Cablemarketdriver the"needforspeed" Itcannotbedeniedthatthe"needforspeed"isdominatingourCableOperatormarketata growingrate.operatorsareverymindfulofthehighlycompetitivelandscapetheyoperate within,andtheneedtoincreasesubscriberbandwidthby50% annualy.wecanseethe projectedbandwidthdemandgrowthaccordingtonielsen'slaw infigure1below: Figure1-Nielsen slawofinternetbandwidth(growthrate=50%/yearforhighendsubs) Figure1-Nielsen slawofinternetbandwidth(growthrate=50%/yearforhighendsubs) Inordertodealwiththerapidgrowthinthedemandforbandwidth,Operatorsare searchingfornew technologiesandarchitecturesthatcanhelpthem supplythosespeeds andnew services,usinganefficientscalabledesign,withpredictableandcontrolablecost ofownership. Operatorsmayconsidermakingchangestotheirheadendnetwork,theirnodesand amplifies,theirservicegroupssizes,themodulationprofileusedondifferentareasofthe 300bpsDS 1982 6KbpsUS 1995 56KbpsDS 1995 100KbpsUS 2002 1MbpsDS 2002 1MbpsUS 2008 10MbpsDS 2008 10MbpsUS 2013 100MbpsDS 2013 35MbpsUS 2016 1GbpsDS 2016 DOWNSTREAM UPSTREAM EraofSymmetrical Services ExtendedSpectrum DOCSISw/FDX FDX DOCSIS3.1 2Gbps 2019 20Gbps 2014 100Gbps 2026 200Gbps 2030 10Gbps 2021 100Gbps 10Gbps 1Gbps 100Mbps 10Mbps 1Mbps 100Kbps 100Kbps 10Kbps 1Kbps 100bps 1982 1983 1984 1985 1986 1987 1988 1988 1989 1990 1991 1992 1993 1994 1995 1995 1996 1996 1997 1998 1999 2000 2001 2001 2002 2003 2004 2005 2006 2007 2008 2008 2009 2010 2011 2012 2013 2014 2015 2015 2016 2017 2018 2019 2020 2021 2022 2022 2023 2024 2025 2026 2027 2028 2029 2029 2030 2031 2032 5 Copyright2018 ARRISEnterprises,Inc.AlrightsReserved.

plantandmore.thechalengeforoperatorsistochoosetherightmixofadjustmentsthat canhelpthem optimizetheirnetworkandsupplytheirsubscribers'demand. Inthispaperwewilreview thecaseofstofa,adanishoperatorthathaschosento introducedocsis3.1andremotephyarchitectureintoitsnetwork,toovercomecommon Operatorchalenges. Manyoperatorsareseeingsimilarchalenges,andthereforeweareseeingasignificant changeinthecableaccessmarket,whileoperatorsareconsideringtheirnexttechnological upgradeandtheirfuturenetworkevolution. InFigure2wecanseetheS&PGlobal(Kagan)forecastformarkettransition: Milions $2,000 $1,800 $1,600 $1,400 $1,200 $1,000 $800 $600 $400 $200 $- CentralizedCAAP/CCAPCore,RemotePHY/MACPHY,andVirtualCMTS WorldwideRevenue 1.32 20.53 197.29 4.11 60.85 385.17 447.49 536.33 vcmts 544.66 RemotePHY/MACPHY CentralizedCAAP/CAAPCore 757.28 851.06 $1,516 $1,476 969.15 $1,371 $1,024 $656 $515 $302 2016 2017 2018 2019 2020 2021 2022 Figure2-CCAP/RemoteArchitecturesMarketRevenueForecast ThischartclearlyshowstheexpectedtransitiontoDistributedAccessArchitectures(DAA), whichincluderemotephyandremotemacphy.startinggradualyfrom thisyearwith earlytrialsandlimiteddeployments,wecanseeaprojectedgrowthinthepenetrationof thenew technologiesstartingsignificantlyfrom 2019andonwards. 6 WEB BLOG www.arris.com www.arriseverywhere.com

Wecanatributethistimingtomultiplefactors: ProductReadiness EquipmentprovidershavestartedtoofferRemotePHYproducts alreadythisyear,andtheywilgetsomeexperiencewithinitialdeployments[silbey]. However,someOperatorsarechoosingtowaitforthetechnologytomature,orfora specificflavoroftheproductstobecomeavailablebeforetheybegindeployments. NetworkPlanning ArchitecturalupgradessuchasthetransitiontoDAArequiremany planningphasesandalotofthought.aswilbediscussedlaterinthepaper,operators havetogothroughtheconsiderationsofwhichaccessarchitectureisthebestfitforthem, andtheymustalsogothroughdetailedplanningactivitiesinordertobepreparedforthe transitionfrom anetworkreadinessperspective,aswelasfrom anoperationalperspective. ExistingUnrealizedCapacity ManyOperatorsarecurrentlyusingdenseintegrated CCAPs,deployedinthelastfew years.manyofthesesystemsstilhaveunused capacitythatcanalow forservicesexpansionwiththeadditionoflicensesonly.so,for thoseoperators,theneedtoupgradetheirnetworkmaybelesspressing.iftheydo upgrade,thentheymayopttoutilizearchitecturesthatpermitthem tore-usetheir recentlydeployedequipment. DistributedAccessArchitectures(DAA) DistributedAccessArchitecturesrepresentanevolutioninthecableaccessnetwork structureandoperations.thedriversfordaa(andspecificalyremotephy)aredetailedin section1.c.theyincludethedesiretobringfiberclosertothehome(inordertobebeter preparedforfttxarchitecture),theneedtoreducespaceandpowerintheheadend,and thedrivetoprovidehigherbandwidthtothesubscriber. Inordertodothat,thecableindustrydesignedavarietyofDistributedAccess Architectures,wherepartsofthetraditionalintegratedCMTSorCCAParemovedtothe nodestructure,closertothesubscriber.thedifferentarchitecturesvaryintheamountof functionalitythatisbeingmovedtothenode,andthechangestheoperatorwilbemaking intheheadenddesign,andspecificalythevideoportion. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 7

IfwestartbylookingattheCentralizedAccessArchitectureinFigure3,wecanseethatal CMTSandCCAPfunctionalityiscentralizedintheheadendorhub,supportinghigh-speed data,voiceandvideoservices.theprocessedradiofrequency(rf)signalsaretransmited totheopticalnodeviaanalogcarriersoverfiber,wheretheyareconvertedtoanalog signalsovercoax. M LSR I-CCAP MSO Facility MAC PHY (DS) (DS) DAC MAC PHY (US) (US) ADC OpticalXmtr RF OpticalRcvr Signals Node/Gateway Actives Passives CentralizedAccessArchitecture-I-CCAP Fiber AmplitudeModulation AmplitudeModulation HybridFiberCoax HFCNode TrunkAmps RF Signals LineExtenders BridgerAmp Taps Taps Figure3 CentralizedAccessArchitectureDiagram WithRemotePHYinFigure4,thePHYpartoftheprocessingismovedtothenode, meaningtheqam modulation,fecanddac/adc.theheadendequipmentisresponsible forthemacprocessing,andistransmitingthemac-processedsignalstothenodevia digitalopticsintheconvergedinterconnectnetwork(cin). OOB CCAPMACCore M LSR MAC (DS) MAC (US) DEPI/UEPI/GCP 10GEPON 10GEthernet DistributedAccessArchitecture(DAA)-RemotePHY DigitalOptics(OpticalEthernet/G.709/PON) DigitalOptics RemotePHYNode PHY DAC Digital (DS) Optics DEPI/UEPI/ GCP/OOB PHY (US) ADC TrunkAmps RF Signals LineExtenders BridgerAmp Taps Taps Figure4 DistributedAccessArchitecture RemotePHYDiagram TheinternalcomponentsofaRemotePHYsystem include(per[r-phy-spec]): EthernetInterface ClockCircuitry RemotePHYPseudo-WireInterface CommonLayer1PHYCircuitry 8 WEB www.arris.com BLOG www.arriseverywhere.com

CCAP-Core RemotePHYDevice DEPI UEPI R- DTI DOCSIS L2 MAC Remote PHY PW Remote PHY PW COMMON L1 PHY Ethernet Video L2 MAC Clock Clock RF Ethernet 9 RemoteCCAPNode DigitalOptics(OpticalEthernet/G.709/PON) RF Signals LineExtenders LineExtenders TrunkAmps BridgerAmp Taps Taps DigitalOptics MAC (DS) MAC (US) M LSR OIN-G.709 NG-EPON NG-PON2 10GEPON AWGP2P OOB 10GEthernet OOB PHY (DS) Digital Optics DAC PHY (US) ADC DistributedAccessArchitecture(DAA)-RemoteCCAP MLSR Copyright2018 ARRISEnterprises,Inc.AlrightsReserved.

DOCSIS3.1introduction DOCSIS3.1istherelativelynew broadbanddataspecification,designedtoincreasethe DOCSISservicescapacityontheexistingHFCnetworks.DOCSIS3.1relieson: OFDM LDPC EnergyManagement HierarchicalQoS ActiveQueueManagement AdvancedTimingSupport ThecapacityDOCSIS3.1isexpectedtoofferissummarizedintheTable1. Upstream Downstream DOCSIS3.0 0.1Gbps 1Gbps DOCSIS3.1 1-2Gbps 10Gbps Table1-DOCSISEvolutionincreasesHFCnetworkcapacity(Source:CableLabs) SincetheDOCSIS3.1speccompletionin2013,weareseeinggradualdeploymentof DOCSIS3.1services.Somereasonsfortheslow adoptionare:headendequipment readiness,cpedeviceavailabilityandcompatibilityissuesbetweenthetwogivenitisanew spec.someoperatorsarealsodelayingthedocsis3.1introductionduetothelarge investmentrequiredforcpedeviceupgrades. 10 WEB BLOG www.arris.com www.arriseverywhere.com

Figure7showstheforecastforDOCSIS3.1servicesdeployment. 40% WesternEurope EasternEurope Asia-Pacific NorthAmerica LatinAmerica MiddleEast Africa 30% 20% 10% 0% 2017 2018 2019 2020 Figure7-DOCSIS3.1DeploymentForecast(Source:ABIResearch) COMPANY(STOFA) Companyoverview StofaislocatedinDenmarkandisdeliveringbroadbandservicestoabout500Kcable households.stofaisdenmark'ssecondlargestprovideroftelevision,broadbandand telephony. Stofa'scustomersaremostlyantennaassociations organizationsresponsiblefor providinginfrastructureservicestoacommunityofafew thousandsubscribers.the servicestheyprovideincludehigh-qualityhigh-speeddata,voiceandvideo(legacyqam video)totheirsubscribers. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 11

Networkarchitectureatthestartingpoint About60% ofstofa sfootprintisrural,butthereisastrongconcentrationofmdusin Denmark smajorcities.asistypicalinmostofeurope,cablerunsareburiedratherthan aerial(asislargelythecaseintheusoutsideurbanareas). Priortothenetworkupgradeproject,StofahadbeendeployingaModularCMTS(M-CMTS) architecture.thecmtswasdeployedintheheadend,andtheassociatedeqamswere usualydeployedintheantennafacilitieslocation. StofahasbeendeployingservicesusinganN+5plantconfiguration-meaning5levelsof amplifiersafterthefibernode.beforetheupgradeproject,forthehigh-speeddataservice groups,theoperatorhaddeployeda1downstream:2upstream nodecombiningratio. Stofahasfiberlinksdeployedbetweenitsheadendsandregionalhubs.Thetopologyis suchthatthetypicalmaximum distancebetweentheheadendandthenodelocationis10km. Motivationfornetworkupgrade In2016,Stofalaunchedaprojecttoplananetworkupgrade,andstartedsearchingforthe bestfitintermsofnetworkarchitectureandproducts.thedriversforthestofanetwork upgradeincluded: Marketcompetition growingcompetitioninthedanishmarketwasdrivingstofato lookforadvancedsolutionstoincreasethebandwidthofferingtoitssubscribers.stofa chosetoupgradeitsnetworktodocsis3.1,andgofrom an860mhzplanttoa1.2 GHzplantforDS,andupto204MHzintheUS.Theseupgradeswereintendedto increasethecapacitystofadeliveredtoitscustomersinboththeusandtheds directions:first,bytransitioningtodocsis3.1andreachingahighermodulation order;andsecond,byincreasingtheamountofspectrum availableforthedifferent services. Transitiontodigitalfiber Theyear-on-yeargrowthinbothphysicalfibercountand opticalwavelengthdivisionmultiplexing(wdm)filterstosupportthegrowingnumber ofopticalnodeswasbecominganissuewithregardtophysicalspaceatthemajor headends.eithermuchbeterutilizationofexistingfibersornew largerheadendswas 12 WEB BLOG www.arris.com www.arriseverywhere.com

needed.digitalfibersolvesthisissuebyprovidingthebenefitsofnotonlymultiplexing manynodesontothesamefiber(notcompletelyunlikewdm)butalsobyproviding thebenefitsofstatisticalmultiplexing(customeranotusingbandwidthatthe exact sametimeascustomerb).theuseoftheipprotocolontopofdigitalfiberalso enablesdynamicsharingoftrafficacrossdifferentfibers-somethingthatwas unthinkableinthepast Consolidatinghubsites Stofawaschalengedbythespaceandpowerinthehuband headendlocations.theoperationalcostofthosefacilitieswasastrainonthebudget, andtheoperatorwaslookingtoreducethespacerequiredfornetworkinfrastructures.a longerterm plannedwastoconsolidatemanyoftheheadendsintoacentralizeddata center OPERATORDRIVERSFORR-PHY DriversandbenefitsofRemotePHY PotentialdriversforDAAarchitectureinclude(alsoaccordingto[Cloonan]): Transitiontodigitaloptics WithDAA,thetransmissionfrom theheadendtothe nodesisdoneusingdigitalsignalsonthefibernetwork,asopposedtothecentralized architecture,whereitisdoneoveranalogsignalsonthefibernetwork.thedigital transmissionalowsformorelambdastobepopulatedoveronefiber.withdigital fiber,anoperatorcanuseupto80lambdasononefiber,whereaswithanalogoptics themaximum isabout32.thisalowstheoperatortobeterutilizeitsexistingfiber networkanddrivemorebandwidth.digitalfibersalsoalow forgreaterdistance betweentheheadendorhubandthenode,alowingtheoperatortocentralizeits MACcorefunctionalityindatacentersfeedingremotenodes.Inaddition,operationof digitalfibersisperceivedaseasierthananalogfibersthatmayrequirefrequenttuning andadjustments.thetransitiontoipbroughtinmanyoperationalbenefits,and significantlyincreasedthefibercapacity. Headendspaceandpowerreduction DAA'smainprincipleisaboutmoving functionalityoutoftheheadendandhub,anddowntotheplant.thereductionof functionalityintheheadendresultsinadecreaseinspaceandpowerrequiredforthe Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 13

processingfunctions,alowingforsavingsonthefacilitymaintenance,andpotentialy hubconsolidation.seepreviousstudyconductedonspaceandpowerrequirements forthedifferentaccessarchitectures[hfc-green-ulm]. BeterSNRattheend-of-line DAAmovesRFsignalprocessingfrom theheadend (whereitistransmitedtothenodewitham fiber)tothenode,eliminatingthe potentialsignaldegradationsfrom AM fiber.itcanimprovesnrwiththeuseofhigh ordersofmodulationwithmoredatatransportedoverthesamebandwidth. Facilityconsolidation Asdiscussed,RemotePHYcanhelpwithfacilitiesconsolidation, duetospaceandpowersaving.inaddition,r-phyalowsfortheheadend/hubtobe moreremotefrom thenodeandsubscriber,sincethefibersusedtocarrythedataare carryingdigitalsignals,whichhavefewerdistancelimitations.sowithr-phy,the facilitiescanpotentialybeconsolidatedtoreducefacilitiesandmaintenancecostsfor thenetwork. AdditionalbenefitsmayapplyspecificalytoRemotePHY: AbilitytoreassignMACprocessing WhenusingR-PHY,theMACprocessing capabilitiesarecentralizedinthemaccore.theremotephydevices(rpds)can dynamicalybemovedfrom onecoretoanotherasneeded,potentialyforload balancingpurposes.thisgivestheoperatorbeterresourceefficiencyandflexibility. Abilitytoselectbest-in-breedfrom CCAPcoresandnodes RemotePHYspecby CableLabs[R-PHYSpec]isdesignedsuchthatcoresandnodesfrom differentvendors caninteroperate,whentheyarebothcomplyingwiththespec. Litlechangeintheprovisioning,configurationandmanagementsystems Remote PHYarchitecturedefinesthatprovisioningandconfigurationofthenodesbedone from themaccore.havingonecentralizedpointofconfigurationissimilartotheway DOCSISnetworksoperatetoday,sominimalchangeshavetobedoneinbackoffice systemsandoperationalmodels. Beterpathtowardsvirtualization CentralizingtheMACprocessinginonelocation createsabeterpathtowardsvirtualization,sincethemacprocessingfunctionis easiertovirtualize.inthefuture,physicalappliancessuchasthemaccorecanbe virtualizedandmovedtooff-the-shelfservers,whichwilfurtherreducecost,space andpowerconsumption,andenhancetheflexibilityofresourcesassignment. 14 WEB BLOG www.arris.com www.arriseverywhere.com

STOFA'SDRIVERSFORR-PHY OfthepotentialbenefitsofRemotePHY,afew weremoreimpactfulthanothersinstofa's decisiontomigrateitsnetwork. Theheadendspaceandpowersaving,alongwithpotentialfacilitiesconsolidationwere veryimportanttostofa.thecostassociatedwithmaintainingitsheadendsandhubsisa considerablepartofitsopex,andstofawaslookingtobothdownsizeintheheadends,and alsoremoveprocessingequipmentfrom thelocalantennaassociations/communities. BetersignalqualityorSNRatthesubscriberlocationwasanotherdriverfortheproject. Stofa'smarketisverycompetitive,andtheOperatorhadtoimproveexistingcapabilitiesto supportthemarket'sdemand.betersnrwouldalow StofatomigratetoDOCSIS3.1and providebeterservicewithhighermodulationorders,whichwoulddelivermorecapacityto thesubscriberwithoutchangingthe"lastmile"wiring. Astrongmotivatortomakethesignificantinvestmentintheproject,wasthefuture expansionofthenetworkasforecastedfrom Stofa strafficengineeringnumbers.inthe comingyears,stofawilhavetocontinuetoexpanditsservices,increasingthecapacityit providestoeachsubscriber,inordertoremaincompetitive.thecurrentinvestmentinthe new technology(withalotofroom togrow)wilrequirestofatoonlyaddlicensesinthe nextnetworkupgradecycles.theoperatorwilnotneedtosendtechnicianstothefieldfor instalations,butwilonlyhavetoremotelychangetheconfigurationonthenodesalready inthefield. Inaddition,StofalikestheR-PHYarchitecturebecauseitenablestheOperatortokeepthe expensiveandmorecomplicatedequipment(hencemaccore)intheheadends,whichare ownedbystofa.thisalowsbetercontrolovertheinstalationenvironment,andoffersa futurepathtowardvirtualizingthesefunctionsandcentralizingthem indatacenters. ThecentralizationoftheMACcoreprocessingwasanotherdriverforStofatomigrateits network.first,thesingleprovisioninginterface(foramaccore)alowsstofatoperform minimalchangesintheprovisioningsystem andbackofficetools.someautomationwilbe requiredtosupportther-phynodes,butothersimplechangesreducethecostand complexityoftheupgradeproject.second,itsimplifiesthetrainingrequiredofthestofa personnelinordertomanagealconfigurationitemson"onebox".thatappliesforboth theengineeringteamsandtheoperationalteamsdebuggingissuesonthesystem. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 15

Lastly,theRemotePHYsolutionStofahaschosenalowsittoconvergealcustomer servicesononeplatform:docsis,vodandbroadcastservices.alservicesareprocessed bythemaccoreandtransmitedoverdigitalfibertother-phynode.theconvergence alowsforoperationalsimplicity(sincedatapathisunifiedforalservices),andstofaisalso benefitingfrom beingabletosendthevideosignalsefficientlyoverdigitalfiber.other architecturesforvideotransmissionexistinthemarket,butstofaoptedfortheconverged approachbecauseofitsoperationalsimplicity. PLANNING Networkarchitecture AfteritwasdecidedtodeployaconvergedvideoanddataRemotePHYintheStofa network,thefocusshiftedtowardscraftingtherightdesigntoprovidethebenefitsof R-PHY,whilematchingtheStofanetworkanditsrequirements. R-PHYcomeswithnew requirementstotheipnetwork.astheremote-phyarchitectureis literalyseparatingtwophysicalelementsthatwerepreviouslyonthesamecircuitboard, thenetworkhastoprovidethesamestabilityandfixedlow latencyforthesolutiontowork overtheexistingfibernetwork. Replacingtheclassicanalogopticalequipmentmeansthatalotoffocuswasputonthe transceivers,andnotjustforthefibersgoingtotherpds.asclassicheadendsturnintoip hubsinstead,theneedfor100/200gbackhaullinksneedstobeaddressed.stofawasalso mindfuloftemperaturelimitsforthetransceiversgoingintotherpds. Fortheactualdesignapproach thereisno"onesizefitsal,"asopticalnetworkdesignwil varyfrom OperatortoOperator.ThefolowingbrieflydescribesthedesignvariantsStofa evaluated. Aspreviouslydescribed,Stofahasmanyheadendsinruralareas,typicalywithavery limitednumberoffibersconnectingthem tothecentralnetwork.inordertoselectthe optimalnetworkarchitecture,stofachoseonetopologyrepresentingageographicarea thatincludestypicalheadendsizesanddistancesbetweenthem,asilustratedinfigure8. 16 WEB BLOG www.arris.com www.arriseverywhere.com

CoreNetwork 13km 12km Ejstrupholm (1) Brande(2) Tyregod(1) CoreNetwork 10km 7km Fare(1) Give(12) Bredbale(22) 37km Pedersholm (1) 24km 7km 4km Bilund(8) 35km Nyboe(33) 6km Molholm (12) Whatswhat: SiteName(RPDs) 17km Skaerup(1) CCAPCoreSite RPDAgg.site Remoteplant CoreNetwork Figure8 StofaSelectedReferenceArea Planningthefiberandinterconnectnetwork Stofareviewedvariousoptionsforinterconnectingtheelementsofitsnetworkdesign,as detailedinthesectionbelow. Stofaconsideredupgradingtheexistingpoint-to-pointfibernetworktoaDWDM networkin ordertoconnecttherpdstoaggregationsiteswherethemaccoresreside,asdepictedin Figure9. CoreNetwork 13km 12km Ejstrupholm (1) Brande(2) Tyregod(1) CoreNetwork 10km 7km Fare(1) Give(12) Bredbale(22) 37km Pedersholm (1) 24km Opticalring 7km 4km Bilund(8) 35km Nyboe(33) 6km Molholm (12) Whatswhat: SiteName(RPDs) DWDM 17km Skaerup(1) CCAPCoreSite RPDAgg.site Remoteplant CoreNetwork Figure9 ReferenceAreawithDWDM Solution Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 17

Aclearbenefitofthisapproachisthatactiveequipment(includingeverythingbelongingto theipnetwork)canbeconsolidatedtofewerheadends.thisconsolidationalsomeansless investmentincostly100g/200gbackhaultransceiversaseach10grpdlinkusesa dedicated10gwavelengthoverdwdm,using10gdwdm transceiversinbothends. AnotherbenefitisthatusingDWDM won tdrasticalychangethedaytodayoperationsof fiberplantandnodes. However,manychalengeswereidentifiedaswel.AsStofahaslimitedfiberbetweenthe existingheadends,a50ghz/80channeldwdm system wouldbeneededinmostareas, andinsomecaseseven80channelswouldnotsuffice(wheremorethan80rpdseach usingonedwdm channel,neededtransportacrossasinglesetoffibers).therangeofthe DWDM transceiverswouldalsobeproblematicinsomeruralareas,requiringoptical amplificationatsomesitesthatwereinitialydesignedaspassivesites.ontopofalthat, thecomplexitiesofensuringredundancyinthedwdm networkandthemuchhighercost ofdwdm transceiverscomparedto10glongreach(lr)opticswereadditionalconcernsfor Stofa. Hence,theDWDM solutionwasnotanidealfitforstofa.thenextlogicalapproachwould havebeentodesignasingle-servicenetworktosupportjusttherpds.inalayer2(l2) (switched)design,guaranteeinglow latencyandsymmetricaltrafficflowsinthenetworkis relativelysimple.operatorshavetodealwiththeknownlimitationsoflayer2networks, likethenumberofsupportedvlans,spanning-treeforredundancyetc.nevertheless,this approachisfeasible. Despiteitsbenefits,alayer2networkwasnotconsideredastherearesomesignificant disadvantages.stofa,likemanyotheroperators,isgoingtobeusingtheconverged interconnectnetwork(cin)forotherservicesaswel,makingitimpossibleor overly-complicatedtousel2inpractice.withanl2approach,anoperatormayevenneed tobuilditasacompletelyisolatednetwork notprovidinganybenefittotherestofthe business. TherewasanotherstrongargumentagainsttheuseofanL2networkintheStofa environment.withdaa,anoperatorwilbeaggregatingiptrafficwhereitsfiber infrastructureisaggregated.thisleadstophysicalsiteswithmassivebackhaulcapacity requirements,creatingwhatmightbethebiggestdownsideofdeployingdaaoverl2 networks.anoperatorwouldlikelyneedalotofexpensive100g+transceiversthatwilbe poorlyutilizedinanl2networksinceeffectiveloadbalancingofiptrafficrequiresip routing,whichinturn,requiresanl3-basednetwork. 18 WEB BLOG www.arris.com www.arriseverywhere.com

ForStofathepreferreddesignoptionwasaconvergedIP(L3)networkthatsupportsal existingservicesandfulfilstherequirementsforr-phy.beinganearlyadopter,thisdidnot leavestofawithmanyoptionsotherthantostudyroadmapsandchoosethebestfuture solutionthatcomplieswiththenetworkrequirements.(furtherdetailsofcin designare addressedinalatersection). RFplantupgradeandupdatingnetworkinventorysystem Stofadecided,aspartoftheDAAtransformation,toalsomigratetheHFCnetworkto1.2 GHzdownstream/204MHzupstream operation anecessarymovetobeabletooffer symmetricalservicesandtoprolongthelifetimeoftheinitialr-phyrolout.onthe pre-existinghfcnetwork,stofahadlimitedtopologydatalinkingcablemodems,taps, amplifiersandnodes.akeygoalforthehfcrebuildwastoimprovetopologydataforthe RFnetwork. Thiswastoalow beterintegrationwithmonitoringandpnm solutionsinordertoletstofa perform targeted,proactivenetworkmaintenance,andtobeabletoautomaticalyidentify affectedcustomersduringoutages. ItwasdecidedtoconducttheHFCrebuildinthreephasesonaperservicegroup(SG)basis: Duringthefirstphase,aSGisredesigned,resultinginupdatedandaccuratenetwork documentation.nositesurveysaredoneyet(unlessknownissuesarepresent).some, butnotal,inconsistenciesinthenetworkdocumentationarefoundandfixedatthis point.theupdateddesigndataisautomaticalyimportedintothepnm solutionto enableuseofthepnm toolstotroubleshootissuesafterphase3. Duringthesecondphase,aSGhasitspassivesandamplifiersreplaced,stilkeeping theoriginalus/dssplit.qualityassuranceduringthisphasemeansthatservicegroups aredeployedwithidenticalcomponents,guidelinesformounting,cablingetc.any inconsistenciesbetweenthenew networkdesignandtherealworldarecaptured duringthisphaseandscheduledforfieldfix.inaddition,cabinetsarecheckedfor defectsandimprovementsneededforbeterairflow.asthenew node/rpdgenerates moreheatthantheoldopticalnode,thecabinetisgivenanew lidanddoor both withventilationholes.it sasmalimprovementthatyieldsa10-15ctemperature decreaseinthenode/rpd.afterphase2,thehfcnetworkiscompletelyrebuilt(using theoldnodesandfrequencyplans). Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 19

ThethirdandfinalphaseinvolvesdeployingtheR-PHYdevice/nodeandswapping diplexfiltersandupstream modulesintheamplifiers,aswelaschangingthesplitto 1DS-1US.Anew frequencyplaniseffectivelyinoperationassoonastherpdisonline. RFfiltersoneachtapareswappedtoreflectthenew frequencyplan(broadcasttv subscriptionsaremanagedusingbandpassfilters). CIN planning DAAbringsseveralspecificrequirementstotheCIN,theconvergedIPnetwork.The solutionchosen bystofa(depictedinfigure10)isentirelyipv6basedandrequiresal devicesinthecin tobeipv6capable(don tassumethisisthecase).also,theabilityto eithersupportoreffectivelytransportieee1588timinginformationneededforrpd operationiscrucial.beinganearlyadopterofdaa,stofahadaverylimitedchoiceof possibleswitchesandroutersthatcomplytotheserequirements.thiswasmadeeven morechalengingbecauseoftherequirementsforhighlyscalablebackhaullinks(nx100g perrouter).itwouldhavebeenpossibletodeployacheaperplatform basedon40g backhaullinks,butthelifetimeofsuchasolutionwouldnotbeinlinewiththewishtorol outaplatform withaverylonglifetimeinthefieldandnetwork. Stofadecidedtobuildamulti-serviceIP/MPLSnetworktoserveastheCIN.Akeydriverfor thiswasthewishforfulflexibilityontheassociationofrpdstomaccores,andremoving anygeographicalconstraints.idealy,theabilitytosimplymoverpds(through provisioning)toanew corewouldalsomeanthattransitioningtoavirtualcoreatalater stagewouldbeseamless. Thewishtoimplement(andprovidedevicesupportfor)802.11xandMACSECshouldbe carefulyconsidered.for802.11xspecificaly,therearechalengesifrpdsaredeployedin DaisyChainorRingtopologies;wehavenoticed802.11xvendorimplementationsmaybe pre-mature.itisrecommendedthatthecin edgedevicesusedtosupportthe implementationbecompatiblewiththoserecommendedandtestedbytherpdvendor. Anotherobservation:forthenearfuture,weexpectthebackhaullinkstobeamajorcost driver-especialyifdeploying100g/200g.however,thecablelabscoherentoptics Specification[P2PCO-SP]couldhelpdrivecostsdown. 20 WEB BLOG www.arris.com www.arriseverywhere.com

CoreNetwork 13km 12km Ejstrupholm (1) Brande(2) Tyregod(1) CoreNetwork 1xAgg.router 10km 2x100Glinks(DWDM) 13x10GLR 4x10GDWDM 7km Fare(1) Give(12) Bredbale(22) 37km 1xCCAPCore 1xAgg.router 2x10Glinks(DWDM) 22x10GLR Pedersholm (1) 24km 7km 4km Bilund(8) 35km Nyboe(33) 6km Molholm (12) Whatswhat: SiteName(RPDs) DWDM 1xAgg.router 2x100Glinks(DWDM) 8x10GLR 2xAgg.routers 2x100Glinks(DWDM) 33x10GLR 14x10GDWDM CoreNetwork 17km Skaerup(1) CCAPCoreSite RPDAgg.site Remoteplant Figure10 CINNetworkDesign IEEE1588timing Theimplementationofprecisiontimeprotocol(PTP)Grandmaster(GM)andsupportingthe flow ofptptrafficinthenetworkwasoriginalyamajorconcernforstofaastherewasno in-houseknowledgeofieee1588.stofalimiteditslabteststogm devicesfrom two vendors,withtheexperiencefrom usingbothwasthatthebasicptpfunctionalityneeded fordaaisalmosttrivial.ifselectedcin devicescanprovideconsistentlow latency forwarding,idealyusingqosbasedondscpclassification,theoperatoronlyneedsto considerpossibleasymmetricroutingfortheptptraffic ilustratedinfigure11. PTPMaster t1 t4 Sync(t1) Delay_req Delay_resp(t4) meanpathdelay=[(t4-t1)-(t3-t2)]/2 offsetfrom master=[(t2-t1)-meanpathdelay] Slave (RPD) t2 t3 Master Latencyfrom MastertoSlaveis2ms A B 2ms 2ms 2ms C Latencyfrom SlavetoMasteris4ms Slave meanpathdelaycalculationassumesdelayisidenticalbothways! meanpathdelaycalculationwilbeincorectforthisdesign! Figure11 PTPLatencyCalculation Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 21

PTPusedwithRemote-PHYalowsforapacketdelayvariation(PDV)of2miliseconds whichmeansthatr-physhouldactualyworkwithupto2milisecondsofdifferencein latencyinanasymmetricroutingsituation.butwithmorefocusbeingputonlatencyin DOCSISnetworks,Stofadecidedtoaim forsymmetricalroutingforptp tobeableto implementanylatencyimprovementsdonetodocsiswithouthavingtoreworkthecin. FailoverofPTPinthecaseofGM failureornetworkissuesisstilverylooselydefinedinthe R-PHYspecs,andrecoveryimplementationsvaryfrom vendortovendor.currentlywiththe Stofaimplementation(depictedinFigure12),theworkaroundistohavetwoactiveGMsin thenetworksourcingtheptptrafficfrom thesameipv6addressbutusingdifferentprefix lengths.ineffectthismeansthattheiproutingprotocolmakestheprimarygm reachable onthenetworkuntilthelinktoitfails,inwhichcasetheroutingprotocolswitchestothe secondarygm.thisdoesnotaccountforcaseswhereonlyptpcommunicationswiththe GM fails.inthesecases,manualinterventionisneededbutwiththeholdovertimeforthe MACcoreandRPDsofatleast2hours,thisistolerable. TheStofaCIN networkisactualylicense-basedupgradeabletoptpboundaryclock support.itremainstobeseenifthisstepwilbeneededinthefuture.sofarstofahasnot neededptptestequipmenttovalidateaccuracyinthefield,becauseofthelow IPhop countinthenetwork,andsimplifiedroutestotherpds.howeveritisimportantto understandthattheoperatorwilneedtestequipmentsynchronizedtothesamesourceas itsgms(typicalygnss)tomeasureandtroubleshoottimingaccuracy,formore complicatednetworks,andforbeteranalysisoftheimpactonptpaccuracyasittraverses thenetwork. Asmoreexperienceisgained,Stofaexpectstoincreasebothdistanceandhopcount betweenccapcoresandrpds.ahandheldptpreferencemightproveveryusefulatthat stage. CINRouter Mngtnetwork sitea-ptp01a 2001:db8:1/96 CIN CINRouter 2001:db8:1/64 sitea-ptp01b Figure12 TimingDistributionNetwork CCAPCore RPD RPD 22 WEB BLOG www.arris.com www.arriseverywhere.com

Productqualificationphases Initialtestingandsolutionevaluation Asmentioned,Stofaisanearlyadopter,andwasoneofthefirstOperatorsintheworldto qualifyanddeployaremotephyarchitecturewithlivecustomers.correspondingly,the firstmaccoreandnodesthatwereusedintheinitialtestinginthestofalabwereofthe firstfew betaunitsoftheproductworldwide.thosecomponentshadtobeintegratedwith anew DHCPv6andtimingservers,andnew routerschosenforthecin ofther-phy network.giventhat,stofawantedtobebestpreparedforthetesting,andstartedasearly astheequipmentvendorcouldalow. Inordertogetahead-startonthenew R-PHYtechnology,Stofareceivedaveryearly pre-productionunitintoitslabmanymonthsbeforeitwasreadyforproduction. Asapreparationphase,StofadecidedtotesttheintegratedCCAPsolutioninthelab, althoughthesolutionitwasplanningtodeploywasremotephyandnotani-ccap solution.therationaleforthatwasthatthehardwareusedforthemaccoreoperationwas thesameasthei-ccaphardware,andtheclicommandsandgeneralflow ofoperationare thesamebetweenthei-ccapandthemaccoresolutions.stofatestedthei-ccapinthe lab,anditalsolaunchedalimitedfieldtrialwiththei-ccap,inordertolearnmoreabout thefieldbehavioroftheplatform andabilitytoconvergedocsisandvodvideoservices. Thelabandfieldtrialsgavetheengineeringandoperationteamssomeexperiencewith servicesconfiguration,monitoring,anddebugoftheplatform.afterthistrial,stofagained valuableexperienceandconfidenceontheremotephyarchitecture. TheRPDandnodewerestagedintheStofalabandconnectedinaverybasicnetwork schemetomakeitoperational,asdepictedinfigure13. Timing Server DHCP CM STB RPD Aggregation Switch MACCore Figure13 StofaInitialLabSetup Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 23

Thetestplanthatwasrunonthesetupincludingthefolowingtestsoffunctionality: NoderegistrationintheMACcore GCPtunnelestablished,timingserverconnectivity operational DOCSIS cablemodem registered,highthroughputachievedtocm VoiceoverDOCSIS VideoonDemand(VOD)services Broadcastvideo DOCSIS3.1downstream OFDM MultipleR-PHYnodesonsinglecore Routing IPv6 Routing OSPFandOSPFv3operation TerminalAccessControlerAccess-ControlSystem (TACACS) LawfulinterceptusingSNMPv3 Loadtesting,makingsureQoSismaintained,USandDS DOCSISloadbalancing makingsureloadbalancingrulesaretriggersarerespected Integratedupstream agility(dynamicmodulationprofileswitching) Sparing routingswitchmodule,downstream card,upstream card Linkredundancy totheremotephymodule Timingserverfailure connectivitylosttothetimingserver,testingtheholdover operation 24 WEB BLOG www.arris.com www.arriseverywhere.com

Integrationintothenetwork Thenew R-PHYproductmustfitintotheexistingoperator'snetworkandbeableto integratewithexistingcomponentsthatarepartofthenormalnetworkoperationand management.theintegrationwiththesecomponentshadtobetestedinadvance,and guidingdocumentsgeneratedinordertoinstructtheoperationalteam duringdeployment. Themainsystemsthatrequiredintegrationare: DHCPservers Folowinginternalanalysis,andalsoaligningwiththevendor recommendation,stofadecidedtouseseparatedhcpserversfortherpdnodesin thenetwork.itmadesenseseparatingthedhcpinstancefrom theoneusedfor subscriberdevicesandmaintainseparationofservicesforthesakeofsecurityand easeofoperation.thedhcpserversselectedhadtobeipv6supporting,tomatchthe coreandnoderoutingcapabilities.thenew DHCPserversweresetupinthelabwith ther-phydevicesandtestedwiththerestofthearchitectureforthefirsttime. TACACSserver StofamanagestheirusersandpermissionlevelsusingTACACS servers,thecorehadtobetestedalongwiththeexistingservers. Proactivenetworkmanagementsolution Inordertoimprovenetworkvisibilityand alow fastactiontohandlepotentialcustomeraffectingissues,stofadecidedtoadopt anew networkmanagementsystem,whichmadeintegrationtestingmucheasier, giventher-phyandpnm productscamefrom thesamevendor.thepnm solution chosencolectsinformationfrom theplant,alertstheoperatorofaproactivealarm notificationswhichalowsissuestobefixedbeforetheycauseanysignificantservice degradationandoroutagetothecustomer.moredetailsandcasestudiescanbe foundin[cunhapnm]. Figure14containsavisualizationofthetopologicalinformationprovidedbythetool, alowingforgeographicallocationofpotentialoutagecausingissues. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 25

Figure14 PNM TopologyView USRFnetworkalignmentandRFQualityMonitoringSolution Withtheanaloglinks becomingdigitallinksbetweenthenodeandhub,areplacementsolutionwas requiredforthelegacyanalogusrfalignmentandrfmonitoringsystemsincluding someofthefieldtools.stofadecidedtouseatoolthatconnectsremotelytotherpd andprovidesrealtimeusrfspectrum measurementsthroughwhichtheuspathcan bealigned,andwhichcanbeusedforperiodicorreactiverfnetworkmonitoringand maintenance.thistoolismeasuringtheupstream RFspectrum: UtilizingfastFouriertransform (FFT)datafrom therfburstreceiver,spectrum from aspecificdocsisupstream isdisplayed(powerdensityvs.frequency) UsestheFFTdatatomeasurethenoiselevelintheupstream channel ExamplescreensareshowninFigure15. Figure15 SoftwareToolforUSRFNetworkAlignmentandMonitoring 26 WEB BLOG www.arris.com www.arriseverywhere.com

ImplementationofanOptimizedR-PHYnodeRFspecificationenablingaseamless integrationofther-phynodewithan+5rfcascade Withsupportfrom Stofa,the vendordefinedthedocsis3.0anddocsis3.1rfperformanceandleveldesignfor ther-phynodewithrfcascadetoachieveamostoptimizedeol(endofline)mer supporting4kqam downstream and2kqam upstream operation Automation TheintroductionofRemotePHYarchitecturerequiresasignificantupgradeofthenetwork operationalmanagement.distributedaccessarchitecturesdrivemorefunctionalitycloser tothesubscriberandintothefield.thiscausesanexplosionofmanageddevicesoutinthe fieldinlocationswheretherewereonlyanalogcomponents.now withdaa,thereareafew hundredmoredigitalandmanageddevices.theremotephydevicesareipdevicesthat arepartoftheipnetwork,andrequiresoftwareupgrades,aswelassubjecttologicaland physicaloutages.althisstronglydrivestheneedformoreautomationinthenetwork,for onboarding,provisioningandmonitoringthoserpds. Ifwesummarizethedifferentfunctionsinthebackofficethatneedtobemanagedand integratedtosupporttheoperationofther-phydevices,itmaylooklikefigure16. Inventory Workflow Assurance DHCP AAA SDN CCAPDAABackOffice MACCores Timing Switching RemotePHYNodes Figure16 BackofficeManagementArchitecture Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 27

Stofaconsidereddifferentoptionsformanagingthenetwork,includingbuildingitsown automationtools.however,theoperatordecideditwilbemorebeneficialtoinvestina long-term managementsystem forthenetworkthatcanexpandtoadditionalfunctions andsupportfuturemigrationtovirtualization. Specificaly,StofadecidedtodeployanRPDmanager,whichsupportsautomationofthe onboardingofanew node.itperformsthisfunctionfrom theboxintothenetwork,with configurationrequiredfrom theoperator'sside.thisisverycriticaltooperatorslikestofa thatareplanningtodeploydozensofrpdsaweek(ormore)aspartofafastroloutplan, orneedtomanagealargerpddeployment. TheRPDmanageralowsfieldtechnicianstoscanthebarcodeontheRPDusingtheir mobiledeviceatthepointofdeployment,andloadingthatinformationintothecentralized databasetocreatethenetworkinventory.thisinformationwilincludetherpdserial number,thenodegeo-location,timeofdayandadditionalparametersrelevantforthe nodeinstalation. Afterthenodeisidentified,itismatchedtoaMACcore(accordingtotheOperator'spolicy) anddownloadedsoftwareversionforinitialconfiguration. RPD Manager MACCores DHCP Figure17 RPDManagerOnboarding 28 WEB www.arris.com BLOG www.arriseverywhere.com

InordertomanagethematchingofRPDstocores,theRPDmanagerinterfacestothe DHCPserverfortheRemotePHYdevices.Itisalsoabletodownloadconfiguration templatestothemaccores,toeliminatetime-consumingmanualclicommands. AutomatingtheRPDmanagementalowstheOperatortodobulksoftwareupgradeor configurationchanges,thusreducingthecyclesandmanualworkrequiredtomanagethe fielddevices.itsoperationdiagram isincludedinfigure17. Inthefuture,thesamenetworkmanagementsystem isplannedtoalsosupportthe provisioningofthecin,alowingforendtoendprovisioningandmonitoringoftheremote PHYnetwork,from onesinglesystem. DEPLOYMENTAND OPERATIONALRESULTS Fielddeployment Withtheobjectivetostartthefirstfieldtrial,thelabsetupwasreplicatedintoafield locationwithfourrpdsconnectedtothemaccore.thefirsttrialphaseincludedlaunching onlyonerpdintothelivenetwork,supportingdocsis,vodandbroadcastservicestolive customers.thefirstnodewas,naturaly,theonethattookthemosttimeandeffort.the dominantchalengesencounteredweremostlyfoundinthedeploymentofthe"supporting network,"andweremostlyrelatedtoroutingconfiguration.theroutestodhcpand TACACSservershadtobeestablishedandverified.Thenew CIN networkhadtobe configuredtomatchthefutureservicesexpansion,whichalsotooksometimeandeffort. Eventualy,thecutoverwasdeclaredasuccess,withnoissuesonthenetwork. Thesinglenodewasleftrunningforafew weeks,tomakesurealserviceshadgood stability.afterthattimeperiod,threemorenodeswerecutover,withnoissuesatthetime ofthemigrationorafter. Thefieldtrialwasconsideredahugesuccesswithpracticalynoissuesencounteredfor morethanamonthdespiteusingpre-productionsoftware.eventualythefulproduction deploymentwasstartedasthevendorreleasedfieldmaturesoftware. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 29

PERFORMANCEIMPROVEMENT RFperformance RFsignaldatawascapturedbeforeandafterthethreedeploymentphasesdescribed above.inthistable,wesummarizethedifferenceinaverageusanddssignaltonoise levels. DSnoisewasmeasuredandreportedbythecablemodem.USnoisewascolectedfrom ther-phynode,asshownintable2. Upstream averagedb Downstream averagedb Before After Change Before After Change Min 28,0 31,1-0,8 37,6 35,3-3,5 Avg 33,6 36,2 2,5 40,4 40,3-0,2 Max 37,4 37,7 8,2 42,0 41,8 2,1 Table2 Upstream anddownstream SNRMeasurements TheUSSNRshowsanaverage2.5dBimprovement,withupto8.2dBimprovementinsome areas.thiswilalow StofatogotohigherDOCSIS3.1modulationsinthefuture,and providehigheruscapacitytoitssubscribers. Wenotethatmildimprovementtodownstream signalqualityisseen,ontheaverage. Therearefew possiblecausesforthis: A.AdecisionwastakentomovetheDOCSISspectrum to722 906MHz,from 218 338MHz.ThiswasdonetokeepDVB-Cvideomultiplexesclearofpossibly impactinglteingressnoise.inthelongrun,thisspectrum wilbeusedforofdm whichismoreresilienttolteinterference.thefactthatsimilarsnrwasmaintained onamuchhigherfrequencyrangecanbeconsideredasanimprovementinthe networkoperation,comparedtoitspreviousstate. 30 WEB BLOG www.arris.com www.arriseverywhere.com

B.ItshouldbenotedthatnochangestotheRFdropcableweremadeexceptfor F-connectorreplacements(ifdeemednecessaryfolowingvisualscreening).Inmany casestherfdropcableshavebeenoutinthenetworkforover20to30yearswhilst exposedtovaryingenvironmentalconditions(low andhightemperatures,rain,wind andsun)andpossiblymechanicalstress(instalationconditions).alarefactors impactingtherfatenuationlossandrfshieldingqualitybetweenthetapandhome, especialyinthehigherdownstream frequencyranges(750to1.2ghz). C.Deploymentofnew 1.2GHzR-PHYnodes,RFamplifiersandRFpassivesfolow a strictandhighlyoptimizedrfnetworkdesignandimplementationprocess.rf network(r-phynodetotaprfoutlet)designcharacteristicsassumelow dropcable atenuationandcablequalityconditions.incaseswherenon-typicaldropcablerf losseshadtobecompensated,stofahadhistoricalyimplementedone-offwork aroundsolutions(suchashigherrfampoutput,instalationofalowertapvalue,etc.). Someofthesead-hoccompensationfixesmaybenegatedthroughtheRFnetwork upgradeprocess. D.ImpactofpoorinhomeRFnetworkqualityonend-of-linedownstream performance.anynetworkchanges/rfperformanceimprovementswithinthehome weredeemedoutofscopeforther-phyanddocsis3.1networkupgradeproject. ThegraphsinFigure18show thechannelsnrdistribution,beforeandafterthenetwork upgrade. Upstream SNRbeforemigration 1019 1011 925 865 760 Upstream channels 660 651 510 574 553 204 175 171 133 SNR(dB) 17 82 12 14 81 39 13 0 0 0 1 <26 (26,26,5] (26,5,27] (27,27,5] 265 (27,5,28] (28,28,5] (28,5,29] 341 417 278 384 (29,29,5] (29,5,30] (30,30,5] (30,5,31] (31,31,5] (31,5,32] (32,32,5] (32,5,33] (33,33,5] (33,5,34] (34,34,5] (34,5,35] (35,35,5] (35,5,36] (36,36,5] 332 (36,5,37] (37,37,5] (37,5,38] (38,38,5] (38,5,39] (39,39,5] (39,5,40] (40,40,5] (40,5,41] Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 31

Upstream SNRaftermigration 2663 2590 2560 Upstream channels 2002 1520 1335 425 SNR(dB) 87 92 249 343 91 2 0 0 0 1 1 4 5 0 18 22 179 27 25 18 1 <26 (26,26,5] (26,5,27] (27,27,5] (27,5,28] (28,28,5] (28,5,29] (29,29,5] (29,5,30] (30,30,5] (30,5,31] (31,31,5] (31,5,32] (32,32,5] (32,5,33] (33,33,5] 710 (33,5,34] (34,34,5] (34,5,35] (35,35,5] (35,5,36] (36,36,5] (36,5,37] (37,37,5] (37,5,38] (38,38,5] (38,5,39] (39,39,5] (39,5,40] Downstream SNRbeforemigration 3564 3077 2924 3133 2646 2578 Downstream channels 1461 1020 1057 1029 150 13 37 33 38 72 77 209 189 99 101 <32 (32,32,5] (32,5,33] (33,33,5] (33,5,34] (34,34,5] (34,5,35] (35,35,5] (35,5,36] (36,36,5] (36,5,37] (37,37,5] 431 377 321 (37,5,38] (38,38,5] (38,5,39] (39,39,5] (39,5,40] (40,40,5] (40,5,41] (41,41,5] (41,5,42] (42,42,5] (42,5,43] (43,43,5] 240 60 17 (43,5,44] (44,44,5] (44,5,45] SNR(dB) 32 WEB BLOG www.arris.com www.arriseverywhere.com

Downstream SNRaftermigration 4561 3833 3702 4105 4254 Downstream channels 3341 3327 3190 2879 2414 2432 1459 1256 1265 1213 1047 SNR(dB) 718 776 842 365 514 59 216 17 (35,35,5] (36,36,5] (37,37,5] (38,38,5] (39,39,5] (40,40,5] (41,41,5] (42,42,5] (43,43,5] (44,44,5] (45,45,5] >46 <35 (35,5,36] (36,5,37] (37,5,38] (38,5,39] (39,5,40] (40,5,41] (41,5,42] (42,5,43] (43,5,44] (44,5,45] (45,5,46] Figure18-USandDSchannelSNRdistribution Inthesediagrams,wecanseemoredetailabouttheSNRlevelsreportedforeverychannel (SC-QAM)thatisinusebycablemodemsintheexaminedpartofthenetwork. HerewecanseetheclearimprovementinUSSNR.TheDSSNRremainsdistributed,andis assumedtobeassuchfrom thereasonsmentionedinthesectionabove. Space&powerintheheadend Oneofthemaindriversforthenetworkupgradeprojecthasbeentoreducethespaceand powerneededintheheadendsfornetworkoperation.beforethenetworkupgrade,the folowinglistofdeviceswererequiredtoprovidethefulsetofservices,docsis,vodand broadcast,intheheadend: A.CMTS supportingaldocsisservices B.EdgeQAM (sinceusinganm-cmtsarchitecture) supportingdocsisphyandvod C.BroadcastEdgeQAM D.DTIserver Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 33

AdditionalheadendequipmentrequiredistheRFcombiningnetwork includingrf combinersandspliterstocombinethedifferentrfoutputsfrom theeqamsforunicast andbroadcasttraffic.theoperatorwilalsoneedopticaltransceiverstotransmitthe analogrfsignalsoverthefibertothenodes.insomeoperators'networks,thereisalsoan RFswitchatachedtotheCMTS,howeveritwasnotusedinthiscase. Insomelimitedcases,StofaplacedtheEQAM devicesinthehub,closertothesubscriber. Wewilfocusonthemoretypicaldeploymentscenariowherealequipmentisdeployedin thesameheadend. ThedevicesthatwilresideintheheadendinthenewlyestablishedR-PHYarchitectureare: A.MACcore B.Timingservers C.Aggregationrouter(s) Anoteabouttheaggregationrouters:theswitchesorroutersthatarerequiredtotransmit ther-physignalsfrom themaccoretothenodescanresideintheheadend,datacenter orhub.inmanycases,someoftheroutingdeviceswerealreadypre-existinginthe headendorareusedforothernetworkservices.inouranalysis,wehavechosentoinclude oneadditional"dedicated"aggregationrouterforthesakeofcomparisontotheprevious stateofnetworkdevicesintheheadend. ForRemotePHY,giventhedecentralizednatureofthearchitecture,PHYfunctionsare distributedinthefield,insidecabinets(inthestofacase),orfieldnodes(onapoleorin storagelocations).thosedevicestakespace,naturaly,butthatspaceisassumedto already"bethere"becausethefibernodethatisenclosingtheremotephydevicewas thereforthelegacyservices,toprovideanalogfibertransmission.insomecases,there maybeaneedtoupgradethefibernodesortheirpowersuppliestomeetthenew R-PHY requirements.theseupgradeswilincurcostandresourcesthatarenotcalculatedinthis section,giventheyareoutsidetheheadend,andmaynotbeapplicabletoaloperators. Table3summarizesthespaceandpowerconsumptionforalthedevicesthatcombinethe twoconfigurations,assumingpowerconsumptionat250candcalculatingtheamountof spaceandpowerrequiredtomake96servicegroups(sgs)operational. 34 WEB BLOG www.arris.com www.arriseverywhere.com

Device Count SpaceperUnit[RU] EstimatedTotalPower[W] CMTS-UBR10K EQAM -DOCSISandVoD EQAM -Broadcast DTIserver 2 2 1 1 18 13 1 1 4470 3000 400 80 E6000MACcore 1 16 3125 Timingserver 2 1 40 Aggregationrouter 1 1 320 Table3 SpaceandPowerrequiredforM-CMTSandR-PHYDevicesintheHeadend ItisimportanttomentionthattheR-PHYsolutioncandoubleitscapacitywithoutany additionalhardwarerequired.wehavechosentorunthecomparisonon96sgs,sincewe feelthisisagoodrepresentation,andinaddition,thelowerdensityonthelegacydevices makesthenumbersforhighercapacityverydifficulttosupportwithinonehub. Ifwecombinethosenumbersforthecaseof96SGssupport,wegetthetotalsshownin Table4. Legacyarchitecture RemotePHY Space[RU] Power[W] 64 7950 19 3485 Improvement Over3x over2x Table4 SummaryofSpaceandPowerRequiredforLegacyandR-PHYArchitectures Thedifferenceinspaceandpowerconsumptionbetweenthe"before"and"after"statesis quiteoutstanding. However,weshouldconsiderafew pointsthatmaynarrow thegapabit,butwouldstil keeptheobviousbenefitofthetransitiontor-phy: Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 35

Stofa slegacyarchitecturewasinserviceformanyyears,andincludedoldertypesof devicesthatmaynothaveleveragedcurrenttechnologyadvancementssuchasnewer chipsetsandroutingtechnologies.itissafetosaythatthereareneweri-ccap solutionsinthemarketthatwilbebeterthanthem-cmtsarchitectureintermsof spaceandpower.however,eventhosealternativescomeshortwhencomparedto RemotePHY,simplybecauseR-PHYmovesthePHYprocessingoutoftheheadend andthusremovesspaceandpowerrequiredforit. StofahadbeenusingM-CMTSarchitecturethatrequiredafew moredevicessuchas thetimingserver,andaseparateeqam.ifthestartingpointhadbeeni-cmtsor I-CCAP,itissafetoassumethattheinitialspaceandpowerconsumptionwouldhave beenlower.inaddition,inordertosupportredundancy,externalrfswitcheshave beeninuse.again,therewerealternativesinthemarketthatsupportedthatsparing protectionwithouttheneedforexternaldevices,andwithlowerspaceandpower consumption. ThelegacyarchitectureincludedexternalbroadcastQAM processing thereare solutionsinthemarketthatalreadyalow forthatprocessingtobedoneinsidethe CCAPchassis,thusreducingtheneedforthatEQAM. Asmentioned,evenconsideringalthepointsthatmaynarrow thegapbetweenthelegacy architectureandtheremotephyarchitecture,thenew designprovidessignificantspace andpowersavingsintheheadend. Thebenefitbecomesevenmoreevidentwhenweconsiderthattheexistingdevicesused intheheadendforthemaccoreprocessingcandoubletheircapacity,supportingtwoor threetimesthenumberofservicegroupsestimatedinthisanalysis,withoutanyadditional hardwarecomponentneeded. Inthefuture,theMACcorefunctioncanbevirtualized,meaningtransferredtoresideona virtualmachine,usinganofftheshelfserver.dependingontheperformanceoffuture off-the-shelfserversversusfutureappliance-basedmaccores,itispossiblethatthismay provideevengreaterspaceandpowersavings,aswelasincreasethenetworkresource assignmentflexibility. 36 WEB BLOG www.arris.com www.arriseverywhere.com

Operationalsimplification Thenew networkarchitectureintroducesbeteroperationalsimplification,comparedto thepreviousnetworkstofahaddeployed. Ifwelookatthenumberofheadenddevicesthatneedtobemanagedtosupport200 servicegroups,wecanseethesignificantdifferenceshownintable5. BeforeNetworkUpgrade AfterR-PHYIntroduction Space[RU] Pow 4CMTS 3UniversalEdgeQAM 1DTIserver 1MACcore 2Timingservers 1Aggregationrouter* 8devices 4devices Table5 NumberofManagedDevicesinDifferentArchitectures Note-thenumberofadditionalaggregationroutersdependsontheOperator's architecture. Thefewerdevicesthatneedtobemanagedmeansoperationalsimplicitybecausefewer physicaldevicesmeanfewernetworkcablesandinterfacestomanagethem.italsomeans thatthereisadecreaseinmanagementtools(softwareuiorcommandline)neededtobe usedandtrainedon(whichwilalsolowerthesupportcost).italsomeansless interoperabilityissuesbetweenthedevices,andaquickerdebuggingprocess(sincethere arefewerdevicestocheckforthesourceofissue).italsomeansfeweripaddressesneed tobeassignedandasmalerinventorythattheoperatorwilhavetokeepinordertodeal withoutages. IntheRemotePHYarchitecture,althenodesaremanagedandprovisionedthroughthe MACcore,pertheR-PHYSpec[R-PHYSpec](thereisnodirectconfigurationinterfaceinto theremotephydevices).soalthoughr-phyintroducesmanynew devicesintothefield, theyarealprovisionedandmanagedfrom onecentralpoint,whichmakesiteasierfrom anoperationalperspective,andalsoforproblem diagnosis. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 37

Accordingtofeedbackfrom Stofa,understandingandoperatingthenew architecturewas significantlyeasiercomparedtothepreviousone,duetothecentralizedmanagement.it requiredlesstraining,andshortenedthedebugcyclesforissuesinthelabandinthefield, comparedtothepreviousnetworkarchitecture. TheadditionallevelofautomationprovidedbytheRPDmanagertoolhelpedtofacilitate thequickandpainlessinstalationprocess.therpdmanagerenabledashorternew node deploymentperiodfrom roughly15minutesonthelegacyplatform tojust5minuteson thedaaplatform (includingscriptedcin,dnsandipam provisioningcurrentlydone outsideofrpdmanager).whilethismightnotsoundlikeahugeimprovement,itisa criticaltimesavingfactorwhendeployingdozensofrpdseachday. CONCLUSION Keybenefits Stofarecognizedmanybenefitsfrom thenetworkupgradeprojectandthetransitionto R-PHY. Thebenefitshighlightedbelow werethemostimportanttostofa: Betersignaltonoiseratio(SNR) Asmeasured,thetransitiontoRemotePHY providedsignificantincreaseinussnr(averageof2.5db).theimprovementofsnr onthedsdirectionisnotmanifestedinthemeasurements.wedobelievethereis someslightimprovementthatiscounteredbythetransitiontothehigherfrequency range.inaddition,thereisapotentialdssnrimprovementthatislimitedbyother elementsintheexistingnetworkthathavenotbeenupgraded,suchasthedropcable andin-homeend-of-linenetwork. PathtoefficientuseofDOCSIS3.1 Thenetworkprojectincludedaldevicesbeing upgradedto1.2ghzspectrum supportinthedownstream.thisenlargestheavailable downstream spectrum fortransmission,usingdocsis3.1abovethe1ghzprevious topbar.italsoalowsupstream highsplitof204mhz.inaddition,thesnrincreasein bothusanddsdirectionsalow forhighermodulationorderstobeusedforsignal transmission,hencealowingmorebitsperhertzontheexistinginfrastructure.these 38 WEB BLOG www.arris.com www.arriseverywhere.com

changessignificantlyincreasetheplantcapacity,andalow Stofatoincreasetheir bandwidthservicetiersofferedtoitscustomers. Spaceandpowersavingintheheadend Thespaceandpowerneededtosupport ther-phyarchitectureintheheadendwasshowntobesignificantlylowerthanthe previouslyusedarchitecture.itresultedinthreetimesthereductioninthespace required,andmorethantwicethesavingsinpower,forthesameamountofservice groups.thesavingsareveryimpactfulintermsofoperationalcost,andwilalow Stofatoconsolidateheadendsandhubs. Futuregrowthmadeeasier TheRemotePHYsolutiondeployed,aswelasthe supportingcin network,werechosenoutofconsiderationofthefuturestofaneeds. Thedesignfocusedonfuturegrowth,suchthatadditionalcapacityandcapability addedtothenetworkwilnotrequirefieldtechniciansandtruckrols.futureupgrades tothenetworkwilbedonebyremotesoftwareconfiguration,licensesenablement, andcapacityincreasesinsoftware.stofafulyfocusedonanetworkplanthatinvested inresourcesandtoolstoday,forthenetworkneedsoftomorrow. Pathtovirtualization WiththeselectedRemotePHY,DistributedAccess Architecture,theMACcorefunctionalityremainsinthecentralizedheadend,alowing forbetercontrolandaccess.thisdesignwilenablemigrationofthatfunctionalityin thefuturetovirtualizedservers,performingthesamefunctions,usingavirtualized MACcoreonoff-the-shelfservers.ThesevirtualizedMACcores,orvCores,wilbe designedaccordingtonetworkfunctionvirtualization(nfv)standards,andmanaged accordingtosoftware-definednetwork(sdn)principles,pavingthepathtoafuly virtualizedheadenddesign. Improvedoperationalsimplicity Asshown,thenetworkupgradeprovideda greateroperationalsimplicity,forengineeringteams,supporttechniciansand customerservices.thesimplicityisgainedfrom thedecreaseinthenumberofdevices tobemanagedinthenetwork,thecentralizedmanagementofaltheremotephy devicesinthefield,andtheadditionallevelofautomationobtainedusingadvanced PNM andorchestrationtools. R-PHYprojectdroveanupgradetothe DOCSISsupportingnetwork,more investmentmadeincin networkupgrade Alongwiththeaccessnetworkupgrade, StofadecidedtoinvestinupgradingitsCIN networktobeanl3supportingnetwork. Inthisdesign,giveneverythingisrouted,Stofacanloadsharetrafficdynamicaly Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 39

acrossaltheexpensivebackhaullinks,andnothavelinksdedicatedtooneserviceor another.secondly,withrpdsconnecteddirectlytotheipnetwork(usinglow-cost10g LRtransceivers),theyarenotlockedtoaspecificMACcoresite,whichgivesnetwork configurationflexibilityaswel. KeyChalenges Suchacomplexmigrationprojectcannotcomewithoutchalengesalongtheway.Some stemmedfrom thestilyoungtechnologybeingused,andsomefrom thesignificant changesthatneededtotakeplacetothenetworkandtheoperations: RemotePHYspecstability OneofthesignificantchalengesintheR-PHYprojectwas duetostofabeingaveryearlyadopterofthenew architecture.duringtheproduct developmentphases,ther-physpecificationswerestilchanging,whichcauseddelay inproductdelivery,andsomerecurringchangesinthenetworkdesignfor deployment.stofa'sstrategyhasbeentokeepeverythingstandards-basedsoquitea lotoftimehasbeenspentdiscussinghow greyareasofthespecifications could/shouldbeimplemented.closecooperationwiththeequipmentvendorwaskey forstofatobeabletostarttheroloutatsuchanearlystage. Networkplanning Inordertocreatethebestdesignforthenetwork,Stofahadto embarkonnew territorysuchasthecin routingupgrade,andthetimingdistribution architecture.stofaupgradeditsrouterstosupportfuturegrowthoftheservices,and thuspurchasedandintegratednew spineandleafrouters,creatingaresilientnew backbone.thisnetworkalsohadtosupportthetimingdistributionrequirements neededforremotephyoperation,guaranteeinglow delayandsymmetricforward andreturnpathsforthetimingsynchronizationtraffic.thiswasanew domainfor Stofa,inwhichitneededtorampupfast,posingsomechalengeforthenetwork architectsduringtheproject. Operationalprocessesupdate Aswitheverymajornetworkupgradeproject,the transitiontoanew architecturerequiredmassiveupdatestotheoperational processesdefinedforthestofaengineersandtechnicians.deepleveltrainingsand educationalsessionswereheldtobringtheteam uptospeedonthenew architecture,andtocreatenew processesthatoptimizetheinstalationprocessand networkmanagementforaltypesofservices. 40 WEB BLOG www.arris.com www.arriseverywhere.com

Changeinfieldmeasurementtools Thechangeofarchitecturewasalso accompaniedbytheneedtochangesomeofthemeasurementanddebugtoolsthat werepreviouslyusedwiththem-cmtsnetworkdesign.specificaly,therfalignment proceduresusedinthepastweredifficulttousewithr-phy,astheyrequired proprietarysupportinthenodesandheadenddevices.toresolvethat,stofa transitionedtostandard-basedmethodsofdatacolectionfrom theplant,using DOCSISstandardfeatures(likeFFTMIBs)toobtaintheSNRandnoisevaluesfrom the plant. MajorTakeaways ForOperatorsconsideringanupgradetoR-PHY,orgetingreadyforthedeployment,we believethestofastoryastoldinthispaperwilprovidealotofvalue,bypreparingthe Operatorsforthemaindecisionpointsandactivitiesyouwilundertakeaspartofthe project.herearethemainpointsforconsiderationthatwehavedrawnfrom our experience: 1.Starttestingearly Becomingfamiliarwiththeproductandarchitectureisgoing topayoffmanytimesoverdowntheroad.starttestingasearlyasyoucan,evenwith similarflavorsofthearchitecture.getingyourlabreadyforthenew architecture testingisimperative,andyouwilsaveyourselftimeifyoustartthatearly,even beforetheproductisavailable. 2.Planningiskey Onecannotunderestimatetheimportanceofpre-planningina complexprojectlikeanarchitectureupgrade.planningthelabandfieldqualification inadvancewilgivethetechniciansandnetworkengineerstimetofigureouttheir needstosupporttheeffort.inaddition,planningthedeploymentphasesinadvance wiluncovermigrationchalengesandoperationalissuesthatcangetresolvedin advance,astonotslow downtherolout. 3.CIN networkandtimingarchitectureareofgreatimportance DeployingDAAis massivelydifferentthani-cmtsorm-cmts,andthesupportingnetworksaremore complex,specificalythecin andtimingarchitectures.operatorsmustdedicate resourcestoplantheseaspectsofthenetwork,withtherightexperienceandskil sets.specificaly,thecin networkhastobedesigned,astheoperatorsmustmake suretheyunderstandtherequirementsspecifictodaa IEEE1588(PTP), non-asymmetricrouting,etc.andthendedicatetherightresourcestosupportit. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 41

4.Gradualmigration Donotunderestimatetheimportanceofstartingoutslowly itisimportanttohaveenoughtimetocapturebothprocessglitchesandtechnical issuesandtoimplementtherequiredimprovements.theroloutwilbeunlike anythingyourhfctechnicianshaveexperienced.consideraltoolsand measurementsthatwilbeimpactedbythearchitecturechangeandplanahead.itis recommendedtostartfielddeploymentwithonenode,andgradualyincreasescale forfuldeployment. 5.Plantdataquality PoorHFCnetworkdocumentationwilseriouslyslow downthe deploymentrateandresultinmiscommunicationstocustomersaboutplanned outagesthatmaygetdelayedduetooperationalreasons.itisworthconsideringasite survey,incaseswhereitmaynotbeabundantlyclear. 6.Importanceofautomation Thesignificantchangeinthenetworkisan opportunitytoautomatemoreprocesses,makingiteasierforthedeploymentteam andforthemaintenanceandsupportteamsinthelong-term.specificaly,automating therpdsmanagementalowsforunifiedinterfacetothedevices,bulkconfiguration andmanagement,inventorytrackinganddebuggingoftheunitsinthefield. 7.Partnership Alongtheentireproject,theStofaandequipmentvendor'steams sharedfultransparencyonthestatusofthedevelopmentandreadinessforthe project.theopenlineofcommunicationenabledustoquicklyreactwhenroadblocks wereencountered,andform workaroundstogetherinawaythatwouldnotdelaythe project. Werecommendweeklymeetingsbetweenthedifferentintegrationteamsandadditional ad-hocmeetings(quicklyscheduled)todiscussspecificissuesandchalenges. Inconclusion,theauthorsconsidertheRemotePHYdeploymentandnetworkupgrade projecttobehighlysuccessful.itprovidedthevaluestofawaslookingforbytransitioning tothenew technology,andguaranteesthestofanetworkcangrow andexpandinthe futuretocontinuetoprovideexcelentservicetoalitssubscribers. 42 WEB BLOG www.arris.com www.arriseverywhere.com

ABBREVIATIONS ADC CCAP CIN CM CMTS DAA DAA DAC DHCP DOCSIS DOCSIS3.0 DOCSIS3.1 DS DWDM DWDM FEC FFT Gbps GHz GM HFC HSD HSD MAC MACPHY MPEG MIB MPLS MSO OFDM OFDM NFV PHY PNM PTP QAM AnalogtoDigitalConverter ConvergedCableAccessPlatform ConvergedInterconnectNetwork CableModem CableModem TerminationSystem DecentralizedAccessArchitectures DecentralizedAccessArchitectures DigitaltoAnalogConverter DynamicHostConfigurationProtocol DataOverCableServiceInterfaceSpecification DataOverCableServiceInterfaceSpecificationversion3.0 DataOverCableServiceInterfaceSpecificationversion3.1 Downstream Densewavelengthdivisionmultiplexing Densewavelengthdivisionmultiplexing ForwardErrorCorrection FastFourierTransform GigabitsPerSecond Gigahertz GrandMaster HybridFiberCoax HighSpeedData HighSpeedData MediaAccessControlinterface DAAinstantiationthatplacesbothMAC&PHYintheNode MovingPictureExpertsGroup ManagementInformationBase MultiprotocolLabelSwitching MultipleSystem Operator OrthogonalFrequencyDivisionMultiplexing OrthogonalFrequencyDivisionMultiplexing NetworkFunctionVirtualization Physicalinterface ProactiveNetworkMaintenance PrecisionTimeProtocol QuadratureAmplitudeModulation 43 Copyright2018 ARRISEnterprises,Inc.AlrightsReserved.

ABBREVIATIONS R-MACPHY RPD R-PHY SC-QAM SDN SG SNR TACACS VLAN vcore VOD RemoteMAC-PHY RemotePHYDevice RemotePHY SingleCarrierQuadratureAmplitudeModulation SoftwareDefinedNetwork ServiceGroup SignaltoNoiseRatio TerminalAccessControlerAccess-ControlSystem VirtualLocalAreaNetwork VirtualCore VideoonDemand 44 WEB BLOG www.arris.com www.arriseverywhere.com

MEETOUREXPERT:TalLaufer TalLauferisDirector,ProductLineManagement,forCMTSandCCAPproductlinesat ARRIS.Inhercurrentrole,Talisresponsibleforvideofeaturesimplementationonthe E6000CER,andfortheE6000productdeploymentasafulCCAPplatform.Talhelps customersdesigntheirfutureccaparchitectures,themigrationproceduresand evolutionpathforvideoservicesontheplatform.inaddition,talsupportsemea customersforalcmtsproducts,andmanagescustomerengagements,deployments andstrategicplanning.talisapartofthecmts/ccapproductmanagementteam, designingthecurrentandfuturegenerationsofsoftwareandhardwarefeaturesofthe product,accordingtocustomerandmarketevolutionandrequirements.shewasa memberofthecablelabs workinggroupandhascontributedtotheccaptechnical Requirementspublication. TaljoinedARRISwiththeBigbandNetworksacquisitionin2011.Duringhertimein BigbandNetworks,TalmanagedagroupofVLSIengineers,designingand implementingdatapathfpgasaspartofthecompany'smajorproduct,mspedge QAM.TalholdsanMBA(withhonors),aswelasaB.ScinElectricalEngineeringand ComputerSciencefrom TelAvivUniversity. Copyright2018 ARRISEnterprises,Inc.AlrightsReserved. 45