local loop unbundling regulation: is it a barrier to ftth

LocalLoopUnbundlingregulation:isitabarriertoFTTHdeployment?

DISCUSwhitepaper

FP7–ICT–GA318137 2DISCUS

Abstract:This white paper describes how the Local Loop Unbundling regulation ofcopperpairsinaccessnetworksmightpresentabarriertoFTTHdeployment,as it inhibits infrastructure providers from developing more efficient opticalaccess architectures that benefit from the closure of many currently activecentralofficesandtheearlyretirementofthecoppercableinfrastructure.Afterintroducingtheproblem,thepaperpresentsthreepotentialsolutionstothisissue,togetherwithabriefdescriptionoftheirpossibleimplementationbyinfrastructureprovidersintheshortandmediumterm.


COPYRIGHT© Copyright by The University of Dublin, Trinity College., 2015.

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.


Authors:

Name Affiliation

David B. Payne TCD

Marco Ruffini TCD

Submission date: January, 27th 2016


TABLEOFCONTENTS1 PROBLEMDESCRIPTION......................................................................................................61.1 SUMMARYOFLLUPROBLEMS.........................................................................................................................92 POSSIBLESOLUTIONS...........................................................................................................93 IMPLEMENTATION.............................................................................................................104 CONCLUSIONS.......................................................................................................................12ACKNOWLEDGMENTS...............................................................................................................13REFERENCES................................................................................................................................14ABBREVIATIONS.........................................................................................................................15


1 ProblemdescriptionCopper pair local loop unbundling was a very successful strategy for introducingcompetition into incumbent networks enabling other competing network providers(NPs)andserviceproviders(SPs)toreuseexistingcopperpairaccessinfrastructureonanannualrentalbasisratherthanhavingtofindthehugecapitalinvestmentthatwouldberequiredtobuildnewaccess infrastructure.However, thequestioniswhetherthatregulation has now passed its usefulness and has instead become a barrier to thedevelopment of FTTH and the novel architectural changes necessary for aneconomicallyviablefuturesuperfastfibretothehome/premises(FTTH/P)broadbandnetwork.

FTTHrequiresconsiderableinvestmentintonewfibreinfrastructureformassrolloutand that large investmentrequirementhasalwaysbeenabarrier to largescaleFTTHdeploymentinEuropeandtheUSwherecompetitivemarketsexistandcompaniesneedrelativelyshorttermreturnsoncapitalinvestmentforaviablebusinesscase.

The capital investment required for network build can be split into two broadcategories:“upfront” investmentand“just intime”(JIT) investmentseeFigure1 .Thenatureandproportionsof these two typesof investmenthaveasignificant impactonthe business case and the time to positive cash flow for business case evaluation.Upfront investment is capital that is needed to build network to “pass” the potential

customersthataretargetedfor the new network andassociated services. This isrisk capital as there are nocertaintiesofrevenuewhenthis investment is made.The other JIT investmenthowever is that investmentrequired to “connect”customers to the network,it usually includes the lastdrop, the customerpremises termination andthe terminating networkservices unit (i.e., theopticalnetworkunit(ONU)

for fibre systems e.g. FTTH Passive Optical Network (PON) or point to point fibresystems(PtP). This JIT capital is only incurred when an order for service has beenplaced and therefore there is a revenue stream associated directly with thisexpenditure. This capital expenditure is therefore much lower risk with predictablereturnsontheinvestment.ThusJITexpenditurehasrelatively little impactontimetopositivecashflowcomparedtotheupfrontinvestmentcomponent.

We believe future network designs need to change the network architecture to notonly minimise total capital investment per unit of network resource but also toredistributethebalanceofthatcapitalinvestmentfromupfrontinvestmenttowardsJITinvestment. This is the basis of the network architecture proposed by the DISCUS

Figure 1 Upfront and JIT cost division for PON developments


project[1-3].Itdoesthisbymaximallysharinginfrastructurethatisprovidedupfrontwhile also being able to gracefully grow that infrastructure as demand for serviceincreasessothatgrowthinnetworkexpenditureis,asfaraspossible,associatedwithrevenuegrowth.

Figure 2 DISCUS Long-Reach access network architecture

ThearchitecturalfeaturesofDISCUSthatenablethischangeininvestmentexpenditureistheclosureofasmanynetworknodes(localexchangesorcentraloffices)aspossibleandtheuseofLong-Reachaccesstechnologythatconnectdirectlytoarelativelysmallnumberofcoreexchanges,seeFigure2.ThesecoreexchangesareinterconnectedviaaflatopticalcorenetworkwhichhasbeenshownbytheDISCUSprojecttobethelowestcost and lowest power consumption network beyond a certain user bandwidthrequirement. This results in a much simpler network, which for European-sizedcountries leads to a two-hop network topology, thus also enabling simpler protocolsandnetworkoperationsinthefuture(althoughthisaspectwasbeyondthescopeoftheDISCUSproject).

Thequestionwewanttoaddressinthiswhitepaperis:howdoesLLUregulationaffectFTTHdeploymentingeneralandthisarchitecturalsolutioninparticular?

The basic problem arises from the physical characteristics of the copper networkinfrastructure that LLU regulation was applied to. Copper cables have very limitedreachduetothepoortransmissionproperties(comparedtofibre).Thisreachproblemis evenworse as higher speed broadband technologies are operated over the copperinfrastructure forcing electronic nodes ever closer to the customers, which alsoincreasescostandtheupfrontinvestmentrequired.Thereasonisthatmulti-customerelectronicequipmenthastobeinstalledbeforeservicecanbeoffered,thusincreasingtheupfrontinvestmentcomponent,andhencetheriskandthetimetoapositivereturnon investment. FTTH on the other hand can have very long reach: by using simpleoptical amplifiers DISCUS is proposing over 100km reach, with actual distance


depending on the geotype under consideration (i.e., the longest reachmight only beusedforlessdenselypopulatedareas).

TherearetwoproblemsthatLLUcreatesthatproduceabarriertothetransitionfroma copper local loop to full FTTH deployment and to the implementation of thearchitecturalchangesrequiredtoensureitseconomicviabilityintothefuture.

1. The first problem is the increasing cost perworking copper pair as customersmove from thecoppernetworktothefibrenetwork.Coppercablesareinstalledinaphysicaltreeandbranchnetworkwiththinnercables(fewercopperpairs)neartothecustomerandmuchlargercableswith thousands of copper pairs) entering Local Exchanges/Central Offices. The cable owner(usually the incumbent operator) must maintain these cables while there are any remainingcustomers connected. Also, in general, the cost of maintaining those cables is largelyindependentofthenumberofworkingpairsanddoesnotdecreaseascustomerstakingupFTTHmoveoffthecoppercabletofibrecables,thusthelargerandmostexpensivecablestomaintainwillbethe last fewstandingwithonlyahandfulofcustomersstillconnected.Thismeansthattheoperationalcostpercopperpairpercustomerconnectedisincreasingascustomerstransferacrosstofibrecables.Theregulatoryquestionis:howcanthatcostbeabsorbedbytheremainingoperatorsthatstilluse those copper cables to access their customers? The current regulatory process usuallyinvolvesnegotiationsbetweentheregulatorandthelargestownerofthecables(usuallytheoldincumbentoperator)aboutthecoststomaintainandexpandthecoppernetworkandtoagreeafair annual rental which then becomes the upper price limit imposed on the infrastructureowner that can be charged to the LLU operator. If those annual charges cannot be flexiblyincreasedand sharedacross all the remaining LLUusers, then the cableownerhas increasingoperational costs per customer as they deploy FTTH. If this cost cannot be passed on to thecopperpairusersandhastobeabsorbedbytheFTTHdeploymentthenthepaybackperiodforthat FTTH deployment will be extended, possibly to untenable time scales. However, thedilemma for the regulator is that if those copper pair costs are passed on fairly to the LLUoperatorsthenthoseLLUoperator’scostswillincreasetopossiblyunsustainablelevels,makingthem less competitive. In addition, from the LLU operator perspective, this is not due to anyinternal inefficiency,buttoanexternalactiontakenbytheregulator/incumbentoperatoroverwhichtheywouldhavenocontrol.TheendresultofanFTTHmigrationprocesscouldeasilybethattheonlycustomersremainingonthecopperpairnetworkarethosebelongingtotheLLUoperators.

2. ThesecondproblemistheclosureoflocalexchangesthathaveLLUoperator’sequipmentwithinthem.TheDISCUSprojectisshowingthatthelowestcostandlowestpowerconsumptionfuturebroadband network is onewheremost of today’s Local Exchanges/Central Offices are closeddown and customer traffic is transported deep into the network to a small number of largeMetro-Corenodes.However, ifLocalExchangescannotbebypassedandcloseddownbecausethere are LLU operators within them then the lowest cost network cannot be realised andcustomerswilleitherhavetopaymorethannecessaryforFTTHortheFTTHnetworkwillnotbedeployed,oratbestbedeployedmuchmoreslowly.There is realevidence that this is indeedoccurringasoperators suchasBTarecontemplatingnetworkswithover1000nodes.Althoughthisisasignificantreductioncomparedwiththe5600nodes in theUKnetwork today, it is still significantly greater than theminimumproposedbyDISCUS(i.e.,ontheorderof70to100nodes)andwill result inamorecostlysolution.Also,acore architecture with 1000 nodes requires a fairly conventional hierarchical core networkwhere the core is structured into at least two and possibly three hierarchical layers. This isnecessary for networks with a large number of nodes because the number of light pathsrequired to interconnect a flat network, as proposed by DISCUS, scales as the square of thenumberofnodesandbecomesuntenablewith largenodecount.However,DISCUShasshown


theflatcoretobethelowestcostandlowestenergyconsumptiontransmissionarchitectureforfuturenetworks, if thenumberofnodescanbereducedtomuch lowervalues,as theDISCUSarchitectureenables.However,thiswillrequireclosureofLocalExchangeswhereLLUoperatorshaveequipmentinstalled.Theregulatoryproblemishowcantheseexchangesbeclosedinafairand equitablemanner whilemaintaining competition that benefits the customer base ratherthanjustforservicingapoliticalagenda?

Again the dilemma from a regulatory perspective is the following: if the full costs ofmaintainingthosebuildings,whichwouldbeclosedbytheincumbentowner,arepassedontothe LLU operators they would have significantly increased costs, which would make themuncompetitive. However, if the incumbent absorbs these costs then they are effectivelysubsidising the LLU operators andwould need to pass those cost onto the FTTH deploymentpotentially making that uneconomic and therefore at best delaying FTTH roll out or possiblymakingitnon-viableindefinitely.

1.1 SummaryofLLUproblems

ThetwoproblemswithcopperLLUunbundlingforwidescaletransitiontofullFTTHdeploymentcanthereforebesummarisedas:

1. The increasing cost per working copper line per customer as customers take up FTTHdeliveredservices.

2. The problem of LLU operators in existing Local Exchanges inhibiting closure of thoseexchangesandpreventingnecessaryarchitecturalchange.

Ifeffectiveregulatorysolutionsarenotfoundtheseproblemscouldinhibitwidescaleroll out of FTTH for a significant time if not indefinitely. This could put Europe at asevere disadvantage compared to regions of the world where FTTH does havesignificantdeployment.

2 PossiblesolutionsThese are not trivial problems and there seems to be no easy solutions but somepossibleapproachesare:

1. Forcethephysicallayerinfrastructureowners(usuallytheincumbentoperator)toabsorbtheincreasingcosts.Thisisthe“donothing”approachandiseffectivelythecurrentdefaultapproach thatEurope is taking.Theproblem is that it inhibits those infrastructureownersfrominvestinginFTTH,andthereforedoesnotsolvetheproblemsfacingfuturebroadbandnetworks that are: the digital divide which requires an FTTH solution; the economicproblemsof futurebandwidthgrowth,whichwill eventually impact theabilityofnetworkoperators to invest in network build; and the power consumption problems facing futurecommunications networks. All these problems require an architectural change thatsimplifies the network by reducing nodes and packet processing electronics and alsoprovidesthelowestcostFTTHsolution.

2. Letthemarkettakeitscourseandallowcoststoriseastheyoccurandbepassedontotheorganisations in proportion to their use. This will inevitably lead to many of the LLUoperators going out of business if they insist on staying with copper pair technology;


however, itmaybeverydifficult forsuchoperatorstoadoptfibreduetothehighupfrontcosts. Even a dark fibre leasing model would put significant financial strain on suchoperators.CoperpairLLUwasintroducedtoencouragecompetitioninthetelecomsservicebusiness: in order tomaintain competition for the benefit of customers there could be amove from thephysical layer competitionof current LLU tomuchbroader competitionattheserviceproviderlayerenabledforexampleviaaccessnetworkvirtualisation.However,astructurefortheinfrastructureandnetworkoperationslayerswouldneedtobesetupandthatmightmeangreaterregulationofthoselayerswhileleavingtheserviceprovisionlayerlargely unregulated. Such access network virtualisation could provide the competitiveserviceprovisionmarketforthebenefitofcustomersonFTTHnetworkswithouttheserviceprovidershaving toprovide thehuge investment for accessnetwork infrastructure, in thesamewayLLUprovidedthatcapabilityforthecopperaccessnetwork.

3. Restructurethetelecomsnetworkbusinesssothatretailandwholesaleserviceprovisionisseparatedfromthenetworkprovisionandoperationsbusiness.Thesimplestformofthisandpotentiallyoneofthemostefficientoptionsistohaveaninfrastructure(physicalcablesand transmission systems) andnetworkoperations (traffic/packetprocessingandnetworkOA&Mandcontroletc.)businessesasanintegratedbutregulatedentitywithseparatedandlargely unregulated retail andwholesale service industries. This enables the economies ofscale from a large infrastructure and network operations business while also enablingrestructuring of the new network architecture required for low cost and energy efficientbroadband networks. The wholesale, retail, service provision and content industries canthenbelargelyleftasseparatedandunregulatedfreemarketcompanieswherecompetitionforcustomersisthenorm.

Thesesuggestionsarejustsomeofmanyotherpossibleoptionsandcombinationsbutthebasicregulatoryproblemsthatneedtobeurgentlyaddressedarethosedescribedintheearliersectionsandsummarisedinthe“SummaryofLLUproblems”section.

3 ImplementationWebelieve a favoured solution to be the combination of the option3 in the sectionabovewithsomestrategysuchasthatsuggestedinoption2sothatthephysicallayercompetition gets replaced by a more regulated network business framework. Thenetwork businesswould sell connectivity services to provide access to the customerbasetotheservicesbusinessesonanequalaccessbasisandindeedshouldsetupthosenetwork access services so that customer contractual lock-in cannot be enforced andwould enable customers to access multiple service providers at all times. Or atminimum let different users of a network terminal, e.g. anONU, set up different andindependent service contracts. This would enable a free market approach for LLUevolution to occurwith costs falling on those using the copper network butwith anopportunity for LLU operators to migrate to the FTTH platform with differentiatedserviceofferings.A key to this is the requirement for some formof accessnetworkvirtualisation andsharingofthecapacityprovidedbyfutureFTTHnetworks.IntheDISCUSwhitepaper


titled“BusinessmodelcasestudiesfornextgenerationFTTHdeployment”[4]weprovideadditionaldetailsofdifferentpossiblemethods thatcouldbe implemented todivide the capacity between providers and users in next generation FTTH networks,whicharebrieflysummarisedhere:

1. assignwavelengthstoserviceproviders,sothatausercanmovebetweendifferentSPsbychangingtheoperatingwavelengthoftheONU.InthisschemeeachSPisassigned(atleast)onewavelengthchanneloverasharedfibre,physicallayer,infrastructure.Wefindthatthisoptionrequires implementationofcostlyONUtransmitters iftheunbundlingiscarriedoutat the wavelength level, with SPs providing their own independent network equipmentoperatingovertheassignedwavelengths.Thisissueissolvediftheopticallayeriscontrolledby one active network equipment provider (who could also be the physical layerinfrastructureprovider),whichoffersdedicatedelectricalinterfaceconnectionstotheotherSPs.

2. assign wavelengths to service type requiresmultiple transceivers at the ONU in order toreceive simultaneously different service types (e.g., IPTV, VoD, Internet,…). While in themedium to long term future integrated transmitters and receivers array could reducetechnologycoststoaffordable levels, itneedstobeseenwhethertheadvantagesbroughtabout by this solution will justify the additional complexity and energy consumptionrequired.

3. assign wavelengths to individual users, whichmeans each ONUs receives connectivity byoperating on a different wavelength.While potentially feasible today, this option is veryinefficient for delivering mass market residential broadband due to poor statisticalmultiplexing, and overall high energy consumption and footprint. It also requires someadditional form of bit stream access in situations where multiple users share an ONU.However,itisenvisagedsuchlogicalpoint-to-pointconnectionswillbeusedforlowvolume,targetedservicessuchasmobilebasestationsandbusinessandenterpriseservices.

4. assignwavelengths flexibly for capacitymanagement: thisoption implies that thephysicalinfrastructure layer and traffic processing equipment layers are operated by a networkoperator thatassigns capacity toSPsusing some formofaccessnetworkvirtualisation forservice providers to differentiate their competing products and services. This remains ourpreferredoption,andisdiscussedfurtherbelow.

Again, we invite the readers to read white paper [4] for a more comprehensivediscussiononcapacityassignmentoptionsinfuturePONs.The reason we favour the last option is that we believe it is the most efficient fordeliveringresidentialbroadbandservices(aswellasotherservicesthatdonotrequiretheuseofa fullwavelength, suchassmallcellbackhaul)as itallowsbetterstatisticalmultiplexing among services, users and providers and reduces the total number ofwavelengths required and therefore optical line terminations (OLTs) and networkswitch/routerports for terminating thoseOLTs. It shouldbestressed though that theflexiblecapacityonofferinourproposalgoesfarbeyondthatcurrentlyprovidedbybit-streamservices.Followingupon latestdevelopmentonaccessnetworkvirtualisationthrough Software Defined Networking and Network Function Virtualisation, weenvisageafullyprogrammableplatform,whereserviceproviderscanrequestcapacitydynamicallyandon-demandtotheunderlyingnetworkattherequiredgranularity,andselecttheappropriatequalityofserviceandnetworkprotectiontype.Anetworkaccesshypervisor is in chargeofpooling all available capacity, bothat thewavelength level,andwithineachPONwavelengthusingthetimedivisionmultiplexinglevel.


Indeed, recently operators have started considering the adoption of SDN in theirnetworks as a strategic upgrade to lower capital and operational spending, whilereducingserviceprovisioningtime.SDNwillallow“drivingupto95%improvementinour provisioning cycle times…” stated AT&T senior executive VP of technology andsenioroperations,andadded“…whatisournetworkofthefuture?It’samulti-service,multi-tenant platform where equipment is more flexible and able to perform manyfunctionsand,itwillbeopen,simpleandscalable”[5].Additional efforts have focusedmore specifically in the access part of the network,withprojectssuchasCORD(CentralOfficeReimaginedasaDatacentre) [6],basedontheONOSSDNcontroller,whichaimsatvirtualisingmostofthefunctionalitiesofOLTand CPE through a NFV framework, running them on commodity servers.White boxswitchesareenvisagedforbasicCPEoperations,whererequired,andfortheMAClayeroftheOLT.From here, the step towards a sliceable virtual access network is small, as thisframeworkcaneasilyembedmulti-tenancyfunctionalities,givingtheabilitytoexternalproviders to seamlessly control virtual OLTs across multiple PONs. Such ideas havealreadybeenpickedupby standardisationbodies suchas theBroadbandForum,andothercompaniesandresearchprojectsaroundtheglobe[7],[8].Itishoweverrecognisedthatthefully-virtualisedSDNsolutionmighttaketimetobedevelopedanddeployed,andintheshorttermtwointermediatestepscouldbeadoptedto enable multi-tenancy in the access [9]. The first is to reuse existing networkequipmentcontrolledbytheinfrastructureproviderthroughtheirmanagementsystem:virtualnetworkoperatorscouldgetaccesstothenetworkthroughastandardsharinginterface which can provide raw access to the management layer with optionalmonitoring and diagnostic functionalities although its network security implicationsneed to be further investigated. The next step involves the deployment of newhardware/software in the access node capable of resource virtualisation, so that thevirtual operators could be assigned a virtual network slice and get full control of theequipment. For this stephowever the interfacemight still be the same Infrastructureprovidermanagementsystem.ThethirdandfinalstepisthefullSDNintegration,wherethevirtualoperatorsgetaccesstotheirnetworkslicethroughaflexibleSDNframeworkwithstandardisedAPIs.

4 ConclusionsThe LLU regulation for the copper pair access network could be a barrier to massmarket FTTH deployment either causing delay and/or increased costs for FTTHcustomers.Themainreasonsforthisaretheincreasingoperationalcostsperworkingcopper pair as customers leave the copper access network to get services from theFTTHnetworkand thedifficultyof closingdown local exchangebuildingswhereLLUoperatorsconnecttotheircustomersandpotentiallyhaveequipmentinstalled.Thesearedifficultproblems for theregulatingbodies tosolvewithoutdamaging thecompetitive environment that LLU has successfully produced. However, the favouredsolution proposed in this white paper is to let free market forces evolve the LLUlandscapeallowingcoststofallontheusersofthecoppernetworkbutatthesametime


allowing those LLU operators to transition to a shared access FTTH network usingaccessnetworkvirtualisationapproaches toofferdifferentiatedcompetingservices totheirendusercustomers.

AcknowledgmentsThis material is based upon work supported by the European Union seventhFramework Programme (FP7/2007-2013) under grant agreement N. 318137(Collaborativeproject “DISCUS”).Theauthorswould like to thankThomasPfeiffer forhisinvaluablecontributiontothediscussionsleadingtothiswhitepaper.


References[1] http://www.discus-fp7.eu/[2] M.Ruffini,etal.,DISCUS:Anend-to-endsolutionforubiquitousbroadbandopticalaccess.IEEE

Com.Mag.,vol.52,no.2,February2014[3] M. Ruffini, et al., DISCUS: End-to-end network design for ubiquitous high speed broadband

services.ICTONconference,2013[4] D.B. Payne, M. Ruffini, Business model case studies for next generation FTTH deployment,

DISCUSwhitepaper,January2016.[5] Lightreading article: “AT&T: SDN Is Slashing Provisioning Cycle Times by up to 95%”,

http://www.lightreading.com/carrier-sdn/sdn-architectures/atandt-sdn-is-slashing-provisioning-cycle-times-by-up-to-95-/d/d-id/717582

[6] CentralOfficeRe-architectedasDatacenter(CORD),whitepaperfromwww.onosproject.org.[7] K. Kerpez, J. Cioffi, G. Ginis, M. Goldburg, S. Galli, P. Silverman. Software-Defined Access

Network(SDAN),IEEECommunicationsmagazine,Vol.52,No.9,September2014.[8] O’SHARE, An open-access SDN-driven architecture enabling multi-operator and multi-service

convergenceinsharedopticalaccessnetworks.www.oshare.ie[9] B. Cornaglia, Fixed Access Network Sharing, Workshop on “Fibre access and core network

evolution:whatarethenextstepstowardsanintegratedend-to-endnetwork?”,ECOC2015.


Abbreviations

API Applicationprogramminginterface

CPE Customerpremisesequipment

CORD Centralofficereimaginedasadatacentre

FTTH Fibretothehome

FTTP Fibretothepremises

JIT Justintime

LLU Localloopunbundling

MAC Mediumaccesscontrol

NP Networkprovideroroperator

NFV Networkfunctionvirtualization

OLT Opticallineterminal

ONU Opticalnetworkunit

PON Passiveopticalnetwork

PtP Pointtopoint

SDN Softwaredefinednetworks

SP Serviceprovider

VoD Videoondemand

local loop unbundling regulation: is it a barrier to ftth

Documents