Business and ownership model casestudies for next generation FTTHdeployment
DISCUSwhitepaper
FP7–ICT–GA318137 2DISCUS
Abstract:ThispaperisareviewoftwopreviouswhitepaperspublishedbytheDISCUSproject consortium in November 2013. After discussions with manystakeholderson topicsof regulationandbusinessmodels fornextgenerationfibreaccessnetworkarchitectures,wehaveanalysedanumberofcasestudiesaround the world on both open access and vertically integrated ownershipmodels,andreporthereourupdatedview.
FP7–ICT–GA318137 3DISCUS
COPYRIGHT© Copyright by The University of Dublin, Trinity College, 2016.
This document may not be copied, reproduced, or modified in whole or in part for any purpose without written permission from The University of Dublin, TrinityCollege. In addition to such written permission to copy, reproduce, or modify this document in whole or part, an acknowledgement of the authors of the document and all applicable portions of the copyright notice must be clearly referenced. All rights reserved.
DISCLAIMERAnyviewsoropinionspresentedinthisdocumentarethoseoftheauthorsanddonotnecessarilyrepresentthoseoftheindividualDISCUSconsortiumpartners.
FP7–ICT–GA318137 4DISCUS
Authors:
Name Affiliation
Marco Ruffini TCD
David B. Payne TCD
Submission date: January, 27th 2016
FP7–ICT–GA318137 5DISCUS
TABLEOFCONTENTS1 INTRODUCTIONANDBACKGROUNDTOTHEDOCUMENT.......................................61.1 STAKEHOLDERFEEDBACK................................................................................................................................72 CASESTUDIES..........................................................................................................................73 REVISITINGTHEBUSINESSMODELS............................................................................103.1 REVISITINGTHEWAVELENGTHUSAGEMODELS........................................................................................103.2 REVISITINGTHEOWNERSHIPMODELS........................................................................................................134 CONCLUSIONS.......................................................................................................................15ACKNOWLEDGMENTS...............................................................................................................15REFERENCES................................................................................................................................16ABBREVIATIONS.........................................................................................................................19
FP7–ICT–GA318137 6DISCUS
1 IntroductionandbackgroundtothedocumentThis white paper is a follow up of two technical documents [1],[2] proposed bymembersoftheDISCUS[3]consortiumin2013.Afterevaluatingthefeedbackcollectedformanumberofstakeholdersincludingmajoroperators,vendors,nationalregulatorsandgovernmentinstitutions,thisnewwhitepaperreportsanupdatedviewonbusinessandwavelengthmodelsfornextgenerationbroadbandaccessnetworks,supportedbyanumber of case studies around the world. We recommend the readers to read theassociatedprojectdeliverable[4]formoredetailedinformation.TheDISCUSarchitecture[5],[6],focusesontheideaofnodeconsolidationbyadoptingaLong-ReachPassiveOpticalNetwork(LR-PON)approachwhichallowsbypassingthemajority of local exchanges and the associated metro transmission networks, thusdirectlyconnecting(i.e.,withoutintermediatepacketprocessing)theaccessfibretotheMetro-Core (MC) nodes. These MC nodes become the only aggregation and packetprocessingnodesinthenetwork.ItalsoenvisagestheideaofaflatopticalcorewhereMetro-Core nodes are connected on a full mesh of wavelength links. The concept isvisualizedinFigure1.
Figure 1 DISCUS Long-Reach access network architecture
Themajor objective of theDISCUS project is to enable a future network thatwouldaddress three major problems arising due to the huge growth in network capacitydemand.Theseproblemsare:
• thecostofnetworkprovisionandfinancialviabilityofthetelecomssector
• theneedtoavoida“digitaldivide”beingcreatedbetweenthosecustomersindense urban areas and those in the sparser rural communities, without theneedformassivegovernmentsubsidies.
FP7–ICT–GA318137 7DISCUS
• The huge growth in electrical power required if today’s telecommunicationnetworkarchitecturesaresimplyscaledtomeetthefuturegrowth.
Solving these problems has major implications for the regulatory policy, for thedistribution and assignment of network resources at all layers to all users of thenetwork, including the service providers, and for the nature of the ownership andbusinessmodelstructuresthatneedtobesupported.
1.1 Stakeholderfeedback
Our initial white papers [1],[2] produced quite a heterogeneous feedback; we havesingledoutherethemainrecurringobservations.
• Therewas a disagreement on the fact that a partial vertical integration at thepassive and active layers would be beneficial, even if regulated, as it wasbelievedthat itwouldnotpromotecompetitionat that layerandwouldhinderinnovation. Also it was mentioned that regulation has often limited power inimposingdrasticchangesincurrentbusinessmodels.
• TherewasdoubtwhetheraPONinfrastructurecouldtrulyfacilitateopenaccessduetothesharedmediumandthatpoint-to-pointfibre(oratleastwavelength)would be a better solution to grant independent access to customers. (Webelieve this is because the access flexibility of PON architectures is notsufficientlywellunderstoodandthatbothpointtopointandPONcanofferequalflexibility for access to customers with the PON having the advantage of theoptionofprovidingthataccessflexiblyandondemand).
• Itwasarguedthatcompetition in fibreaccessnetworkcouldreduce incentivesfor operators to install fibre, especially in rural areas, where the return oninvestmentisverypoorandslow.
This white paper revisits the wavelength uses and business models taking intoconsideration the feedback received and investigation of a number of case studiesaround theworldboth inopenaccess andvertical integratednetworks.We refer thereaderstotheassociatedprojectdeliverable[4]formoredetailedinformation.
2 CasestudiesWe have considered a number of case studies on both open access and verticallyintegrated architectures, to help us understand how these different options havedeveloped over the past 10 years in different geographic areas. The cases examinedincludedcountries suchasSweden, theNetherlands, Switzerland,France, Italy, Spain,Singapore,Australia,NewZealandandNorthAmerica.Therearesomegeneralaspectsregarding the success of open access networkswhich can be learned from the actualdeployments. Some of them do not only apply for an open access regime but areapplicable for the general acceptance of FTTH networks. Since multiple recurringreferenceswereusedforcasestudies,,theyarereferencedhereasagroup[7]-[41].Firstofall,FTTH(whetheritbeopenaccessornot)hasmainlybeensuccessfulwherethere has been no or very limited broadband offering available before. The problem
FP7–ICT–GA318137 8DISCUS
arises if an alternative broadband technology exists, e.g. xDSL (FTTC/N) or cablenetwork. Also, if the price for fibre is significantly higher than the competingtechnologiesthenthetakerateisgenerallylow.Furthermore,iftheFTTHdataratesarenotcompetitive,e.g.,only10/1MbpsDS/USwhich isstillverycommoninmanydataplans,thesuccessisalsoratherlimited.Sotheanswertothequestion:“Willtheuserpayapremiumforfibre?”isusually“No”(orverylittle).Tomaximizethesuccessoperatorsshould chargeonlyprices in theorderofusual xDSLor cableoffersor at a relativelysmallpremiumforasignificantlyhigherspeedoffering. Insomecasesoperatorsevenofferedabonustotriggerpotentialcustomerstosignacontract.The acceptance of a fibre connection or take rate is one of the fundamental successcriteriaforFTTHespeciallyforopenaccessnetworks.Heretheavailablemarketsharefor any one operator is a priori smaller than for a vertically integrated operatorscenario.The two approaches of supply-driven deployment and demand-driven deploymentstrategiesneed tobedistinguished.Thesupply-drivenstrategystarts thedeploymentindependently of any guaranteed demand. The large national broadbandnetworks inSingapore, Australia andNewZealand follow this strategy. This strategy is inevitablyhigher riskas there isnoguaranteedrevenueand inat leastonecase (NewZealand)peoplehavebeenveryreluctanttoconnecttotheFTTHnetwork.Themainreasonforsuchreluctanceseemedtobethehigherpremiumbeingchargedfortheservice[41].Incaseofademand-drivenstrategythedeploymentonlystartsifacertainnumberofpre-contracts are already signed. This is the casewith Reggefiber in theNetherlandswhereabout30–40%ofhouseholdsmustsignsuchacontractbeforeReggefiberstartslayingfibreatall.ThesamestrategyholdsforDeutscheTelekominGermanyinthosecities where they planned to establish an FTTH network. This percentage is theminimum which is believed necessary to achieve a positive return on investmentaccording to numerous techno-economic investigations. Stokab in Stockholm had adifferentbutsimilarstrategyastheyfirstconnectedtopubliccustomers(governmentbuildings, schools, hospitals, etc.) and established a secure revenue basis before theextension to private customers. The first private customers were the large housingcompanieswhichfurthersecuredtherevenues.Becausethisstrategylinksinvestmentdirectlytorevenuestheyaremuchlowerfinancialriskbutsufferfromtheproblemofpotentially long delays before deployment can commence and maybe losing out toneighbouringareasthatdeployearlier.Related to this question of deployment strategy is the question: howmanynetworkoperatorscanacountryorregionaffordandberunprofitably?Someinvestigationshavestudied this aspect and showed that thenumberofpotentialoperatorsisverylimited–notmorethantwoorthree.Other studies investigated the question: where are the limits of profitable privatedeployments?According to them, infrastructurebasedcompetitionwithseveraldifferentnetworksispossibleonlyinthemostdenselypopulatedareas. In rural areaswith a lowpopulationdensitynoprivateroll-outwilleverbecash-flowpositive,evenmoreso ifthemarketshareissplitinanopenaccessregime.Thesecalculationsarebasedonsomeassumptions:atimeframeforreturn-on-investmentofabout5–7years(whichisquitea long time for typical business cases), and some assumptions about the achievableARPU(averagerevenueperuser).ThemaximumARPUwillbelimited(foracceptance
FP7–ICT–GA318137 9DISCUS
reasonsandavailabilityofdisposableincome)butthetimehorizoncanbeextendedifthe infrastructure investor is not a private company but amunicipality/utilitywhichhasdifferentcommercialconstraintscomparedtoprivateenterprises.HoweverunfaircompetitionargumentsbasedontaxpayersubsidisationcanbeleviedatsuchoperatorsandinEuropecanfallfoulofstateaidrules.Howevermunicipalities,especially inruralareas/smaller towns,areoften taking theinitiative to trigger thedeploymentofFTTHnetworks.Thesemunicipalitiessee thesenetworksasakindofnaturalmonopolyorpublicinfrastructurewhichshouldbeownedbypublicauthorities.Thenetworksthenwillbemostlyopenaccessbasedandsincethemunicipalities often don’t have enough funds to invest, a lot of the deployments arefinanced under a private-public partnership (PPP) agreement. Under EU lawmunicipalities are only allowed to invest in telecommunication infrastructure if it isdoneunderidenticalconditionsaprivateinvestorwouldinvestintheproject.OftenEUpermissionhastobegranted(e.g., inthecaseofAmsterdam’sCitynetortheAsturconnetwork).IntheUSveryoftenmunicipalitiesarenotallowed(bystatelaw)toinvestinor run telecommunication networks and there have been many legal proceedingsagainst towns thatwanted to establisha communitynetworkbasedon complaintsofthetelecomincumbentsorthecablecompanies.Utilities which are often owned by municipalities are already familiar with publicnetworks (gas, water, electricity). They have rights of way, own ducts and want toexpandtheirbusinesscase.Soalotofthemhaveinvestedintelecomnetworks,i.e.fibrenetworks.ThisisthecaseespeciallyintheNordics,Germany,theUSandSwitzerland.One interesting aspect in open access FTTH networks is the choice of networktopology, either point-to-point or point-to-multipoint (e.g., GPON). Most of theincumbentsworldwidehavechosenaPONarchitecturefortheirdeployments,whichisthecasealsofortheopenaccessnationalbroadbandnetworksinSingapore,Australiaand New Zealand. But most of the non-incumbent driven networks, especially inEurope, are established using a point-to-point topology. Even the incumbentsKPN inthe Netherlands and Swisscom are relying on this topology. The reasons for it arenamed as most future proof and most secure architecture and best-suited for openaccess, althoughwebelievesuchargumentsare lessvalid todayasNG-PON2systemsalreadygivetheabilitytousedifferentwavelengthsinthesamefibre,makingthemuchhighercostofpoint-to-pointfibresystemsdifficulttojustify.InthecaseofGPON,competitiveaccesstothenetworkispossibletodayalmostonlyviaabitstreamaccessonlayer2or3.Thistypeofaccessdoesnotallowforsignificantdifferentiation in the competitive offer. An alternative would be a kind of sub-loopunbundling (SLU) at the last splitter location. The rest of the way to the customerpremise then ispracticallyapoint-to-point connection. If thesplitter is located in thefieldnearthecustomerpremiseacompetitorwouldthenhavetoownhisownfibrelinktothatlocation.Adifferentscenario,whichisalsodescribedintheliterature,istopushthe splitters back into the central office in order to make access to the fibre moreconvenient. This however this is effectively a point to point fibre infrastructure andincreases the cost of FTTH deployment, which needs to be accounted for whenconsideringtheeconomicsofdeploymentstrategies.In addition, recent advances on the use of Software Defined Network (SDN) andNetworkFunctionVirtualisation(NFV)approachesforaccess/metronetworks(e.g.,the
FP7–ICT–GA318137 10DISCUS
CORDprojectbyAT&T[42])couldchangethescenarioswithinthenextcoupleofyears.Separatingcontrolplaneanddataplaneintheaccessinfrastructure,movingthecontrolsoftware in commodity servers throughNFVcouldproduceanewvirtualOLT that isfullyprogrammableandwillallowdifferentoperatorstotakecontrolofvirtualslicesofthePON.IndeedtheuseofSDNintheaccessnetworkhasrecentlybeeninvestigatedbyanumberofotherprojects([43],[44]).Generally,theadoptionofmultiplewavelengths(WDM)intothenetworkisseenasafinalsolutiontoallaccessproblems(especiallyinthecaseofPONs).InPONsaseparatewavelength could for example be assigned to a provider or a customer and create adirectvirtualpoint-to-point link.Howeveronlyvery fewdiscussionsor investigationsontheuseofdifferentwavelengthsinsuchafuturenetworkhaveyettakenplace,andthenextsectionprovidessomeupdateddiscussionsfromaDISCUSprojectperspective.
3 Revisitingthebusinessmodels
3.1 Revisitingthewavelengthusagemodels
Inconsiderationofthecasestudiesabove,wehavereviewedtheDISCUSdeliverablestitled: “Wavelength usage options in access networks” and “Business and ownershipmodelsforfuturebroadbandnetworks”.It is expected that the future optical access networks will be based on multi-wavelength transmission over optical fibre to provide the scalable future capacityrequired.Howeverhowwavelengthsareusedandwhat theyareused for canhaveasignificantimpactnotonlyonthecostsandefficiencyofthefuturenetworkbutalsoontheopportunitiesforcompetitionandtheservicecreationenvironment.Inourpreviouswhitepapersfourwavelengthusageoptionswereconsidered.InreferencetoFigure 2.a these differ mostly on the part of infrastructure in the central office,highlightedwiththeredcircleinthefigure.Theproposedwavelengthusagemodels,are:1) Wavelengthsassignedtoserviceproviders,visibleinFigure2.b,wheredifferent
wavelength channels are assigned to different SPs. The users tune to differentwavelengthtoaccessservicesfromdifferentSPs.
2) Wavelengths assigned to service types, visible in Figure 2.c, where differentwavelength channels are assigned to different service types, and each SP canprovidereachserviceonadifferentwavelength.SimilarservicesfromdifferentSPscanbemultiplexedintothesamewavelength.
3) Wavelength used flexibly for bandwidth management, visible in Figure 2.d,wherewavelengthsareusedtomultiplexanyproviderandanyservicetype.Thusan ONU can receive different service types from different SPs on the samewavelengthchannel.
4) Wavelengthassignedtousers,visibleinFigure2.e,whereeachuserisassignedadedicated wavelength channel. Thus each ONU is serviced by a dedicated OLT.Noticethatthisserviceistypicallyassociatedtotheuseofwavelengthmultiplexingdevices instead of optical power splitters, in which case it requires a different
FP7–ICT–GA318137 11DISCUS
architectureinthedistributionsideofthenetwork(left-handpartoffigureFigure2.a).
Fig2.a
Fig2.b Fig2.c
Fig2.d Fig2.e
Figure 2 Different wavelength assignment models for FTTP access architectures. Fig 2.a is the overall reference, while fig 2.b refers to the wavelengths assigned to service providers model; fig 2.c to the wavelengths assigned to services model; fig 2.d to the wavelength used flexibly for bandwidth management model; fig 2.e to the wavelengths assigned to user’s model.
Weconsideredthatoption2)waslikelytobeunrealisticasitwouldrequiremultipletransceiversandprotocolimplementationsperONUandmultipleOLTsatthehead-end(themetro-corenodefortheDISCUSarchitecture)toterminatethemultipleinstancesofthePONprotocol.Thisisbelievednottobecosteffectiveandpowerhungryasmanyof the transceivers protocol circuitry andOLTswould have to be operating for largeproportionsofthetimethatservicesareused.Althoughadvancesinopticsintegrationmeans that building integrated laser and photodiode arrays in a common opticalmodule with only a single fibre interface is possible and could enable ultra-fastswitching among wavelengths, it remains to be seen if the additional complexity,compared to the limitedbenefits that suchamodelwouldbring,will reacheconomicviability. But it isapossibilitythat futureONUscouldoperateovermultiplechannelssimultaneously.
Metro/Core(MC)node
Business
Remotenode
OLT
1st stagesplitting
2nd stagesplitting
…
3rd stagesplitting
…
Accessswitch
SP-1ONU
ONU
ONU
…
Residence
ONU
Mobile BS
Port-n
SP-n
PON-1
Remotenode
…
…
PON-n
RE
RE RE :Reach Extender
:Powersplitter
:Wavelength de/multiplexer
Port-1
OLTPort-n
Port-1
…
…
…
- WpSP:thewavelengthidentifiestheServiceProvider- WpST:thewavelengthidentifiestheservicetype- SW:wavelengthstransportmultipleservicesfrommultipleproviders
- WpU:thewavelengthidentifiestheenduser
FP7–ICT–GA318137 12DISCUS
Option 4) wavelengths assigned to users is the well-knownway of implementing avirtualpoint-to-pointtopologymodeloveraPONphysicallayerarchitectureandcouldbe a way of implementing the point to point architecture that has so far been thepreferredoptionsforopenaccessnetworks.Themaindrawbackisthatfromanactivenetwork perspective it presents higher cost, energy consumption and footprintcompared to TWDMPON, and limits the number of users per PON to the number ofwavelengthsavailable(i.e.,around40inpostNG-PON2systems,unlessearlyadoptionofcoherenttechnologyisevoked).WhilethisisOKconsideringcurrentGPONsplitratioof32or64, it ismuch less than splitsof256orhigherenvisagedbynextgenerationsystems and the DISCUS architecture. In addition, this method does not operatestatistical multiplexing on the PON, thus wasting capacity that could instead bedistributedtootheruserstodeliveradditionalservices.Finally,wherethewavelengthseparationisachievedthroughpassivewavelengthfilters(whichisusuallythecaseinWDM PONS), it will lock wavelength channels and thus capacity to the individualcustomeraccessfibres,withariskofossifyingdevelopmentoffurtherbusinessmodels.However,wedoenvisagethatinaPONanumberofwavelengthwillbeusedaslogicalpoint-to-pointconnectionsfordedicatedservices,suchasformobilebasestationsandultra-highcapacityservicesforbusinessesandenterprisescustomers.Option1)wherewavelengthsareassignedtoserviceprovidersmight inprinciplebean option until the control andmanagement issueswith option 3) are resolved (seebelow). By separating access, giving providers different wavelengths, it can providereasonable statistical multiplexing of capacity, without limiting the number of userswhere PONs with high split ratios are deployed. From a physical implementationperspectivehoweveritcannotbeassumedthatdifferentserviceproviderswillbeableto deploy their own terminal equipment into the same fibre infrastructure throughphysicalwavelengthunbundlingduetothepotentialinterferencebetweenwavelengthswhen operated independently. Although it is possible to deploy more stabletransmitters at the ONU to avoid interference due towavelength tuning inaccuraciesthis could increase the cost at the ONU, which is typically a low-cost device forresidential usage. However, if the physical layer is controlled by one entity, thendifferentSPscanconnectelectricallytotheOLTandbeassignedanentirewavelengthchannel using today’s technology. Themain drawback of this approach remains that,besidesnotbeingasbandwidthefficientasoption3),itdoesnotallowformultiplexingdifferentprovidersintooneONU(unlesstheONUhasmorethanonetransceiver)andthereforerestrictssimultaneousaccesstomultipleproviders.Option3) remains themost efficient in termsof capacity utilization and assignmentflexibility by dynamically matching network resource to user demand and freelyassigningcapacitybetweenusersandproviders.Ithoweverrequiresthepresenceofanentityowningandcontrollingtheactiveinfrastructuresuchasanincumbentoperator.WhilethiscouldbeoperatedinconjunctionwithmodelssuchasbitstreamandVULA,fullopenaccessoperationwillrequiremorepowerfulvirtualizationmechanismswheremultipleproviderscouldoperateoverthesamePONandbeabletocontroleveryaspectoftheirPONvirtualsliceasiftheyweremanagingaseparatephysicalsystem.Thiswillrequire enhancements on the control and management of PONs, which could beprovidedbydevelopmentsintheconceptofSoftwareDefinedAccessNetworks.Whencombinedwith such a virtualization framework, option 3) could in principle give theexactsameabilityasthepoint-to-pointmodelintermsofcustomermanagementfromaproviderperspective,withtheaddedvaluefortheendusertomultiplexservicesfrom
FP7–ICT–GA318137 13DISCUS
different providers at the same time, while being more cost effective and energyefficient.Therearehowevertwointermediatestepsthatcouldbeadoptedinthemeantimetoenablemulti-tenancyintheaccess[45].Thefirstistoreuseexistingnetworkequipmentcontrolled by the infrastructure provider through their management system: virtualnetworkoperatorscouldgetaccesstothenetworkthroughastandardsharinginterfacewhichcanproviderawaccess to themanagement layerwithoptionalmonitoringanddiagnostic functionalities (there are however serious network security issues to beaddressed for this tobecomeanacceptablesolution fornetworkoperators).Thenextstepinvolvesthedeploymentofnewhardwareintheaccessnodecapableofresourcevirtualisation, so that the virtual network operators/service providers could beassigned a virtual network slice and get full control of the equipment. For this stephowever the interface might still be into the same network operator managementsystem.ThethirdandfinalstepisthefullSDNintegration,wherethevirtualoperatorsgetaccess to theirnetworkslice througha flexibleSDNframeworkwithstandardisedAPIs.
3.2 Revisitingtheownershipmodels
This section reviews a number of principles concerning the ownership modeloriginallydevelopedin[1].Theproposeddesignprincipleswereasfollow.1) There isno duplication of passive network infrastructureused toprovidebasic
networkservices:thatisthereisonlyonefibrenetworkforeachcustomerpremisesforthemassofcustomers.
This principle is still of paramount importance and should be taken intoconsideration in access network deployment. Indeed open access networks arebased on this principle. There are however cases where this principle is notrespected, i.e. where an incumbent is not required to open up its network. In anumberof suchcases thishas ledcompetitors toorganiseanddeploya separateinfrastructure from the incumbent. So although not optimal from a networkefficiency perspective, market forces can lead to the deployment of overlappingopticalaccessnetworks.
2) Asmuchaspossibleofthefibre infrastructure(andnetworkequipment)should
besharedbyasmanycustomersaspossible.Whatthisprinciplereallystatesisthatthenumberofprovidersandtheirmodeofoperationsshouldn'tgetinthewayofmaximisingsharingofinfrastructureamongusers.Whilethisisachievablewithwavelengthusagemodel3),asstatedaboveitwould require full access virtualisation to enable a fully open accessmodel thatgoesbeyondbitstreamandVULAoverlaymodels.ThisrequiresupdatingtheOLTsaswellasthedevelopmentofanagreedandstandardisedcontrolandmanagementframework.Untilthiscanbeachievedtherearepossibleintermediatestepsusing
FP7–ICT–GA318137 14DISCUS
virtualisationof currentnetworkmanagement systems,anddifferentwavelengthassignmentmechanismssuchaswavelengthtoserviceproviders.
3) Customersshouldhavetheoptiontoaccessmultipleproviderssimultaneously
andbeabletochangeproviders“onthefly”.
4) Customers should have the option of bundled and fixed term contracts or have allservices provided from any provider at any time on a pay as you go basis ifdesired
Option3)and4)arebothvariantsofa fullyaccessibleor “openaccess” scenarioservice provisioning market. While this is the preferred option to achieve fullcompetitionattheservicelevel,itworkswithwavelengthusagemodel3),butnotwithusagemodel1),whichmightbeusedforexpediencyinthenearterm.Afewstakeholdersbelievethatthebenefitsthatcouldoccurfromsuchahighdegreeofflexibility in the service provisioning is over-rated, as they believe that servicebundlingprovidesabetteroptionforbothusersandproviders.Webelievethisisstilldebatableandcannotbetesteduntilsuchflexibleservicelevelarchitectureismade available in the market. There are certainly significant limitations in theservicebundlingmodelsusedtodaysuchas:mostbundlesarepre-structuredandcustomersmayneedtobuyintobundlesthathaveservicestheydonotuse,ortheydo not get the all the services theywould prefer because they are in a differentbundle package, or possible all the services they want are not available from asinglesupplier.Thecustomercannotgetaccesstobestofbreedforalltheserviceshe/she uses but will have to compromise on some of the services within thebundled package. The customer gets locked into a contract for a fixed term andthere is often a penalty for leaving the contract early (at best not all of theremainingtermwillberefunded).Ifwavelengthoption1)istobeconsideredhowever,thereisstilltheoptionforanew category of brokers to emerge in the market, offering bundles that are aselected mix from the services offered by different SPs. However, whether SPswould enable brokers access to such service bundleswithout regulation to forcetheminto“openaccess”remainstobeseen.
5) Thereshouldbenolock-intosingleproviders.
Fromatechnicalperspectivethereisnoreasonwhychangeofproviderscouldnotbedoneondemand,althoughinmanycases,wherepartoftheinstallationcostissubsidised by the provider (e.g., for the fibre connection or for the ONU),temporarylock-inmightberequired.
6) Thereshouldbenophysicalhardwarereconfigurationofnetworkequipment,or
infrastructure,requiredtochangeserviceprovider-allreconfigurationwouldbeviasoftwarecontrolofnetworkequipment.
From a technological perspective there are no major issues. With wavelengthoption1) thiscouldbedonebytuningtheONUtoadifferentchannel,whilewithoption 3) only network PseudoWire rearrangement might be required. The
FP7–ICT–GA318137 15DISCUS
requirementisthatofastandardizednetworkcontrolandmanagementframework(e.g., a Software Defined Access Network) allowing infrastructure, network andserviceproviderstocoordinatetheiractionsfollowingauserrequest.Thistypeofinteraction iswhatDISCUShas implemented in its controlplanedeliverablesanddemonstrationsandisakeyfeatureoftheDISCUSarchitectureproposal.
4 ConclusionsThispaperhaspresentedadiscussiononbroadbandaccessnetworkownershipandbusinessmodels,consideringanumberofcasestudiesonbothverticalintegratedandopenaccessnetworks.Indeed,manyoptionsareavailablebothforwavelengthuseandnetwork ownership in next generation fibre access network. However, the favouredsolution presented in this paper is that of using wavelengths flexibly for bandwidthmanagement without artificially restricting their use to a specific service provider,servicetypeoruser,exceptforfewspecifichigh-capacityservicesthatjustifytheuseofdedicatedwavelength channels. From an ownershipmodel perspective, the favouredsolutionremainsthatofopenaccess,as itallowssharingofcostly fibre infrastructureamong multiple market players, which could otherwise lead to a monopolisedbroadband market. Finally, it is believed that access network virtualisation andsoftware defined network, by adding open programmability and flexibility to thenetwork management infrastructure, will play an increasing role in facilitating suchopenmarketsolutions.
AcknowledgmentsThis material is based upon work supported by the European Union seventhFramework Programme (FP7/2007-2013) under grant agreement N. 318137(Collaborative project “DISCUS”). The authorswould like to thankWerner Graudszusfor is invaluablecontributiontothe investigationof thecasestudiespresented inthispaper.
FP7–ICT–GA318137 16DISCUS
References[1] M.Ruffini, etal.,Businessandownershipmodels for futurebroadbandnetworks.DISCUSFP7
projectwhitepaper,September2013.[2] M.Ruffini,etal.,Wavelengthusageoptionsinaccessnetworks.DISCUSFP7projectwhitepaper,
September2013.[3] http://www.discus-fp7.eu/[4] DISCUSdeliverableD3.7.Finalreportonregulation,policyandmulti-businessmodelusage.
January2016.[5] M.Ruffini,etal.,DISCUS:Anend-to-endsolutionforubiquitousbroadbandopticalaccess.IEEE
Com.Mag.,vol.52,no.2,February2014[6] M. Ruffini, et al., DISCUS: End-to-end network design for ubiquitous high speed broadband
services.ICTONconference,2013[7] OECD(2011):NextGenerationAccessNetworksandMarketStructure,OECDDigitalEconomy
Papers,No.183,OECDPublishinghttp://dx.doi.org/10.1787/5kg9qgnr866g-en
[8] Ooteghem,J.V.,etal.:“Canasynergeticcooperationbetweentelecomandutilitynetworkprovidersleadtoafasterrolloutoffibretothehomenetworks?”;50thFITCECongress,2011http://dx.doi.org/10.1109/FITCE.2011.6133447
[9] Tomonaga,etal.,CompetitionPromotingUnbundlingofPON,2ndInternationalConferenceonEvolvingInternet,2010http://dx.doi.org/10.1109/INTERNET.2010.36
[10] M.Lebourges,CompetitionviaInvestment,anEfficientModelforFTTHRollout,Communications&Strategies,78,2ndQ.2010,pp.45-66
[11] Domingo,etal.,DeploymentstrategiesforFTTHnetworksandtheirimpactonthebusinesscase:Acomparisonofcasestudies,20thITSBiennialConference,2014,RiodeJaneirohttp://www.econstor.eu/handle/10419/106863
[12] OECD(2011),FibreAccessNetworkDevelopmentsintheOECDArea,OECDDigitalEconomyPapers,No.182,OECDPublishinghttp://dx.doi.org/10.1787/5kg9sqzz9mlx-en
[13] D.Carmona,etal.,GPONUnbundlingforMultioperatorAccess,XVthInternationalTelecommunicationsNetworkStrategyandPlanningSyposium(NETWORKS),2012,Romehttp://dx.doi.org/10.1109/NETWKS.2012.6381701
[14] J.P.Pereira,etal.,InfrastructureSharingasanOpportunitytoPromoteCompetitioninLocalAccessNetworks,J.ofComputerNetworksandCommunications,vol.2012,ArticleID409815http://dx.doi.org/10.1155/2012/409817
[15] J.R.Schneir,etal.,EconomicImplicationsofaCo-investmentSchemeforFTTH/PONArchitectures,TelecommunicationsPolicy,Vol.37,Issue10,Nov.2013,pp.849-860
[16] A.Domingo,etal.,ModelingtheeffectofductsharinginaNGANcompetitionmarket,July2011[17] http://telecoms.com/opinion/how-viable-is-wavelength-unbundling-on-ftth-networks(retrieved
2015-09-11)[18] S.Hoernig,etal.,Architecturesandcompetitivemodelsinfibrenetworks,WIKconsult,
December2010
FP7–ICT–GA318137 17DISCUS
http://www.wik-consult.de/uploads/media/Vodafone_Report_Final_WIKConsult_2011-01-10.pdf(retrieved2015-09-11)
[19] B.Sadowski,etal.,Newchallengesinmunicipalbroadbandnetworkmanagement:fromverticalintegrationtowholesale-retailmodel,http://www.imaginar.org/taller/its2008/71.pdf(retrieved2015-09-11)
[20] S.Verbrugge,etal.,ResearchApproachtowardstheProfitabilityofFutureFTTHBusinessModels,FutureNetwork&MobileSummit(FutureNetw),June2011,Warsaw
[21] Karl-HeinzNeumann,StructuralmodelsforNBNdeployment,WIKconsult,DiskussionsbeitragNr.342,September2010https://www.accc.gov.au/system/files/Dr.%20Karl-Heinz%20Neumann%20paper.pdf(retrieved2015-09-11)
[22] A.Banerjee,etal.,TowardsTechnologicallyandCompetitivelyNeutralFibretotheHome(FTTH)Infrastructure,in[Chlamtac,etal.,“BroadbandServices:BusinessModelsandTechnologiesforCommunityNetworks(JohnWiley,2005)]http://repository.cmu.edu/cgi/viewcontent.cgi?article=1451&context=tepper(retrieved2015-09-11)
[23] A.González,ProspectsonFTTH/EP2POpenAccessModels,49thFITCECongress,2010,SantiagodeCompostela
[24] A.Alexiou,DesignAspectsofOpenMunicipalBroadbandNetworks,Proc.ofthe1stInternationalConferenceonAccessNetworks,2006http://dx.doi.org/10.1145/1189355.1189375
[25] OECD(2013),BroadbandNetworksandOpenAccess,OECDDigitalEconomyPapers,No.218,OECDPublishinghttp://dx.doi.org/10.1787/5k49qgz7crmr-en
[26] W.Lehr,etal.,BroadbandOpenAccess:LessonsfromMunicipalNetworkCaseStudies,http://people.csail.mit.edu/wlehr/Lehr-Papers_files/Lehr%20Sirbu%20Gillett%20Broadband%20Open%20Access.pdf(retrieved2015.09.11)
[27] M.Matson,etal.,StudyonLocalOpenAccessNetworksforCommunitiesandMunicipalities,infoDev2006http://www.infodev.org/infodev-files/resource/InfodevDocuments_130.pdf(retrieved2015-09-11)
[28] HeavyReading,TheBusinessCaseforMunicipalFibreNetworks,January2006http://www.ftthcouncil.eu/documents/Reports/2006/BusinessCase_for_Municipal_Fibre_Networks_2006.pdf(retrieved2015-09-11)
[29] A.Dixit,etal.,FibreandWavelengthOpenAccessinWDMandTWDMPassiveOpticalNetworks,IEEENetwork,Nov/Dec2014,pp.74-82
[30] M.Tahon,etal.,ImprovingtheFTTHbusinesscase–Ajointtelco-utilitynetworkrolloutmodel,TelecommunicationsPolicy38(2014),pp.426-437
[31] R.E.Timmerman,OpenAccessFibretotheHomeNetworking,Dissertation2009,BrighamYoungUniversity,Provohttp://scholarsarchive.byu.edu/etd/2019(retrieved2015-09-11)
[32] D.Chaffee,etal.,TheMunicipalandUtilityGuidebooktoBringingBroadbandFibreOpticstoYourCommunity,2008http://community-wealth.org/content/municipal-and-utility-guidebook-bringing-broadband-fibre-optics-your-community(retrieved2015-09-15)
[33] C.Troulos,etal.,FactorsdeterminingmunicipalbroadbandstrategiesacrossEurope,TelecommunicationsPolicy(2011),doi:10.1016/j.telpol.2011.07.008
[34] M.VanderWee,etal.,Techno-economicEvaluationofOpenAccessonFTTHNetworks,J.Opt.Commun.Netw.,Vol.7,No.5,May2015,pp.433-444
FP7–ICT–GA318137 18DISCUS
[35] DevelopmentsofNextGenerationNetworks(NGN):countrycasestudies,ITU2009[36] IntegratedOASEresultsoverview,D8.5,EUprojectOASE,2013
http://www.ict-oase.eu/public/files/OASE_D8.5_WP8_DTAG_15032013_V1.0.pdf(retrieved2015-09-15)
[37] M.Forzati,etal.,Next-GenerationOpticalAccessSeamlessEvolution:ConcludingResultsoftheEuropeanFP7ProjectOASE,J.Opt.Comm.Netw.,Vol.7,No.2,February2015,pp.109-123
[38] G.Vall-Llosera,WDM-PONTechnologyforOpenAccess,BROADNETS2010http://www.ict-oase.eu/public/files/22_Vall-llosera_WDM-PON%20technology%20for%20open%20access_BROADNETS_2010.pdf(retrieved2015-09-15)
[39] S.Verbrugge,etal.,BusinessModelsandTheirCostsforNextGenerationAccessOpticalNetworks,ICTON2012,paperTh.A2.4
[40] M.VanderWee,etal.,EvaluationoftheTechno-EconomicViabilityofPoint-to-PointDarkFibreAccessInfrastructureinEurope,J.Opt.Commun.Netw.,Vol.6,No.3,March2014,pp.238-249.
[41] F. Mirza et al., Drivers and Barriers to the Uptake of a FTTH Ultra-Fast Broadband in NewZealand. The 41st Research Conference on Communication, Information and Internet Policy,March2013.
[42] CentralOfficeRe-architectedasDatacenter(CORD),whitepaperfromwww.onosproject.org.[43] K. Kerpez, et al., Software-Defined Access Network (SDAN), IEEE Communications magazine,
Vol.52,No.9,September2014.[44] O’SHARE, An open-access SDN-driven architecture enabling multi-operator and multi-service
convergenceinsharedopticalaccessnetworks.www.oshare.ie.[45] B. Cornaglia, Fixed Access Network Sharing, Workshop on “Fibre access and core network
evolution:whatarethenextstepstowardsanintegratedend-to-endnetwork?”,ECOC2015.
FP7–ICT–GA318137 19DISCUS
Abbreviations
API Applicationprogramminginterface
ARPU Averagerevenueperuser
CORD Centralofficereimaginedasadatacentre
DSL Digitalsubscriberline
FTTC Fibretothecabinet
FTTH Fibretothehome
FTTN Fibretothenode
FTTP Fibretothepremises
GPON GigabitPON
LR-PON Long-reachpassiveopticalnetwork
NFV Networkfunctionvirtualization
NG-PON2 Next-generationPON2
OLT Opticallineterminal
ONU Opticalnetworkunit
PON Passiveopticalnetwork
PPP Private-publicpartnership
SDN Softwaredefinednetworks
SLU Sub-loopunbundling
SP Serviceprovider
VULA Virtualunbundledlocalaccess
WDM Wavelengthdivisionmultiplexing