a practical vision for optical transport networkinggrover/mesh_networking/wp004[1].pdf · a...

A Practical Vision for Optical Transport Networking

Paul Bonenfant, Curt Newton, Kevin Sparks, EveVarma, and Rod Alferness

W H I T E P A P E ROPTICAL NETWORKING GROUP

ABSTRACT

This white paper presents a practical vision forOptical Transport Networks (OTNs). In thisvision, OTNs will evolve from point-to-pointDWDM remedies to scalable, robust optical-networking applications that cater to a widevariety of client signals having equally-variedservice requirements. This paper places theevolution of OTNs in context with othernetworking trends, and addresses the fivefactors most central to real-world OpticalTransport Networking – span engineering,maintenance, survivability, interoperability,and service transparency. NOTE: A condensedversion of this white paper appeared as theintroduction to the Bell Labs Technical JournalSpecial Issue on Optical Networking, Jan.-Feb.1999.

INTRODUCTION

Commercial lightwave communications overfiber-optic cable began some twenty years ago.

Since then, researchers have held a vision ofaccessing a larger fraction of the theoretical 50 THzof available information bandwidth that single-mode fiber offers. In more recent years, acceleratedresearch has explored sophisticated photonicdevices and techniques that would allow thetransport and routing of signals in the opticaldomain. At the same time, network operators,spurred by persistent projections of vastly increaseddemands for bandwidth for services of widelyvarying characteristics, have been driven by avision of a network that could gracefully evolve tomeet unknown future demands for increasedflexibility, capacity, and reliability. Such networks,in addition to providing high-speed transmissioncapabilities, would need to incorporatemultiplexing and routing techniques appropriatefor ultra-broadband services. It was recognized thatthe efficient implementation of suchtransmission/switching functions coulddramatically reduce overall network cost.

As evidenced by the Special Issue of the Bell LabsTechnical Journal on Packet Networking [1], thepersistent projections of earlier years have finallybecome a reality. The current unprecedenteddemand for network capacity is mostly driven bythe rapidly growing demand for packet-basedservices – in particular, by Internet/Intranet-basedapplications. A conservative estimate of Internettraffic growth is that it doubles every six months. Ifthis growth rate is accurate, and continues, theaggregate bandwidth required for the Internet bythe year 2005 in the US will be in excess of 280 Tbps [2]. Transport networks currentlyoptimized for a mix of narrowband and widebandservices (64 kbps to 2 Mbps) must becomeoptimized for much larger channels carryingbroadband data, voice, and video. The combinationof an unprecedented demand for new capacity andthe emergence of very-high bandwidthapplications, at a time when many network serviceproviders are experiencing high utilization rates oftheir existing fibers, has led network planners tolook for the most expedient and cost-effectivemeans of increasing capacity.

Hence, the research vision of virtually infiniteinformation bandwidth and transport in the opticaldomain, together with network operator needs fordramatically increased network capacity and an

LUCENT TECHNOLOGIES OPTICAL NETWORKING2

associated broadband infrastructure, have becomemutually reinforcing thrusts. With individualchannels in the transport network growing togigabits/second, and with optical technologybecoming increasingly viable and cost-effective, weare now witnessing rapid large-scale deployment ofWavelength Division Multiplexed (WDM) systemsworldwide. Network operators are relying onWDM extensively to expand transmission capacityon long-haul and fiber-limited routes, currentlyachieving throughput in excess of 400 Gbps perfiber. Because the primary short-term need beingaddressed has been capacity gain, these deployedsystems have been point-to-point line systems.Beyond these point-to-point WDM applications,the next evolutionary step in reliable, scalabletransport networks will be optical networking.

But what exactly is optical networking? Like manyother networking techniques, optical networkingimplements functionality for transmission,multiplexing, routing, supervision, andsurvivability for a wide range of client signals. Asdescribed in a previous issue of the Bell LabsTechnical Journal [3], optical networking operatespredominantly in the optical domain, where theunit of bandwidth granularity – the optical channel– is much larger than in Time Division Multiplexed(TDM) networks. In an optical network, the opticalchannel (“frequency slot”) may be considered asanalogous to a time slot in a TDM system, withoptical-network elements manipulating opticalchannels (“frequency slots”) similar to the wayTDM elements manipulate time slots.

In an idealized optical network, there are noanalog engineering constraints on optical-channelrouting; the network is completely service-transparent, that is, free from service-specificfunctions that can limit flexibility. In addition,ubiquitous wavelength conversion (interchange) isavailable to minimize stranded capacity, and thereis support for a multi-vendor environment. Ofcourse, the reality is more challenging. Publictelecommunications relies on an extremely diversenetwork of networks, with widely varyingtopologies, deployed technologies, services, andunderlying business models. A practical vision foroptical networking is one that goes beyond point-to-point WDM transmission to address the practical

needs of this diverse networking environment. Itencompasses the progression of applications fromlong-haul capacity expansions toward a unifiedend-to-end optical network, spanning access,metro, and long-haul domains. Such a vision alsodescribes how optical networking supports anexpanding variety of client signals that haveequally varied service requirements – flexibility,scalability, and survivability, coupled with bit-rateand protocol independence.

This paper focuses specifically on optical-networking solutions for transport applications1, orthe Optical Transport Network (OTN). Opticaltechniques have been evolving for a wide range ofnetworking applications. However, owing to theunique combination of technology and businessforces, it is on transport networks specifically thatoptical networking offers the most immediate andextensive payoffs. While much of this paper isrelevant to the full range of optical-networkingapplications, including enterprise and residentialaccess, it specifically addresses transportapplications.

In summary, optical transport networking hascome of age. A practical vision for optical transportnetworks – one that is widely deployable, highlyreliable and maintainable, readily evolvable, anddemonstrably cost – effective - is now at hand. Thispaper lays out this practical vision, placing it incontext with other networking trends. In addition,this paper addresses the most critical factors toconsider in architecting optical transport networks.

OPTICAL TRANSPORT NETWORKING

It is essential to clearly define what is meant bythe term Optical Transport Networking. After all,

the ability to support optical transmission is notnew – SONET/SDH equipment has been usedsuccessfully in the construction of single-channeloptical transmission systems for some time.However, SONET/SDH networks and opticalnetworks differ in several respects; in particular, inhow they support capacity expansion and channelrouting. In SONET/SDH networks, once thetransmission rate of a network’s single opticalchannel has been maximized, capacity expansion

LUCENT TECHNOLOGIES OPTICAL NETWORKING 3

1 Transport networks provide the underlying high-capacity infrastructure for core interoffice, metropolitan interoffice, andbroadband business-access networks. We distinguish these transport-networking applications from other applications, suchas residential access and enterprise networking, because each may have quite different architecture requirements forcapacity, traffic management, physical topologies, operations, and reliability. At the same time, there may also be sometechnologies and other aspects that are common across all these applications.

involves adding new transmission systems inparallel over separate fibers. In optical networks,capacity expansion involves simply addingwavelengths within the same fiber (up to somepredefined maximum) and transmission system.The routing functions for SONET/SDH networksare performed by means of time slots, whereas therouting functions for optical networks areperformed by means of optical channels (frequencyslots) between wavelengths of various frequencies.

The use of WDM – more specifically, the rapiddeployment of point-to-point WDM line systems intelecommunications networks worldwide – isviewed as the first step toward optical transportnetworking. While WDM line systems alonesupport little in terms of networking functionality,the elements for WDM optical transportnetworking are on the horizon. WDM line systemswith fixed wavelength add/drop capability arebeing deployed, and optical-network elements withnodal features, such as optical add/dropmultiplexers (OADMs) and optical cross-connects(OXCs) – employing either electrical or opticalswitching matrices – have been reported to be inlaboratory and field trials. The ability of theseWDM nodal elements to add, drop, and in effectconstruct optical channel-routed networks allowsfor the manipulation of optical channels in WDMnetworks, just as time slots are manipulated inTDM networks today (see Figure 1).

Optical transport networking is defined as theability to construct WDM networks havingadvanced features, such as optical-channel routingand switching, that support flexible, scalable, andreliable transport of a wide variety of client signalsat unprecedented bandwidth

granularities – upwards of tens of Gbps per opticalchannel.

NEXT GENERATION NETWORKS

This section introduces how our practical visionof optical transport networking relates to

Lucent’s vision of the next generation network, aspresented in the Bell Labs Technical Journal SpecialIssue on Packet Networking [1]. We firstsummarize the role of optical transport networkingin the next generation network, and then examinethe evolution of transport networks from single-wavelength TDM networks to WDM linesystems, and ultimately to the vision of opticaltransport networking at the optical-channel level.

OPTICAL TRANSPORT NETWORKING IN THE NEXT

GENERATION NETWORK

Today’s TDM-based transport networks have beendesigned to provide an assured level ofperformance and reliability for predominantlyvoice and leased-line services. Proven technologies,such as SONET/SDH, have been widely deployed inthe current transport infrastructure, providinghigh-capacity transport, scalable to gigabit-per-second rates, with excellent jitter, wander, anderror performance for 64 kbps voice connectionsand leased-line applications. SONET/SDH self-healing rings enable service-level recovery withintens of milliseconds following network failures. Allof these features are supported by well-establishedglobal standards, enabling a high degree of multi-vendor interoperability. To summarize,SONET/SDH has been the transport-networkingtechnology of choice for a world dominated bycircuit-voice and leased-line services.


Figure 1. Optical Transport Network with Optical Channel Routing

Optical Transport Network (OTN)

OTN Client Connection (Optical Channel)

Metro Core Metro

Optical SubnetworksOptical Subnetworks

OTN Client(e.g. SONET/SDH,IP, ATM)

Optical layer Cross Connect (OLXC) Optical Add/Drop Multiplexer (OADM)

OTN Client

Optical Subnetworks

In contrast with today’s TDM-based transportnetworks – and, to some extent, with ATMnetworks – legacy “best-effort” IP networksgenerally lack the means to guarantee highreliability and predictable performance. The best-effort service provided by most legacy datanetworks, with unpredictable delay, jitter, andpacket loss, is the price paid to achieve maximumlink utilization through statistical multiplexing.Link utilization (for example, the number of usersper unit of bandwidth) has been an importantfigure of merit for data networks, because the linksare usually carried on leased circuits through theTDM transport network. While today’s datanetworks provide excellent connectivity, they donot enable controllable distribution of networkresources among traffic from competing users.Given the inherently bursty statistics of packet datatraffic, the fixed bandwidth pipes of TDM transportmay not be an ideally efficient solution. However,this inefficiency has traditionally been consideredof less importance than the network reliability andcongestion-isolation features that a TDM-basedtransport network provides.

The surging demand for high-bandwidth anddifferentiated data services is now challenging thisdual-architecture model of TDM-based transportand best-effort packet data networks. It is not cost-effective to extend the usefulness of best-effortnetworking by over-provisioning networkbandwidth and keeping the network lightly loaded.Further, this approach cannot always be achievedor guaranteed due to spotty demand growth, and isa particular issue for the network access domainthat is most sensitive to economic constraints ofunder-used facilities. For a commercial network, anadditional drawback of the best-effort model is thatthe network relies only on user (customer)cooperation for congestion control, by assumingthat end users will slow down their transmissionrates when significant congestion is detected. As aresult, in general, data-service providers today donot have the network infrastructure support toprovide customer-specific, differentiated-serviceguarantees and corresponding service-levelagreements.

With regard to satisfying these new networkrequirements, the Bell Labs Technical Journal SpecialIssue on Packet Networking [1] described a newdata-centric optical transport network architecture.This next generation network will dramaticallyincrease, and maximally share, backbone network-infrastructure capacity, and provide

sophisticated service differentiation for emergingdata applications. Complementing the manyservice-layer enhancements, optical transportnetworking will provide a unified, optimized layerof high-capacity, high-reliability bandwidthmanagement, and create “Optical DataNetworking” solutions for higher-capacity dataservices, with guaranteed quality.

THE VALUE OF TRANSPORT NETWORKING IN A

DATA-CENTRIC WORLD

Network architectures for cost-effective, reliable,and scalable evolution employ both transportnetworking and enhanced service layers, workingtogether in a complementary and interoperablefashion. Transport networking, whether based onSONET/SDH or optical technology, enables theservice layers to operate more effectively, freeingthem from constraints of physical topology to focuson the sufficiently large challenge of meetingservice requirements. This point is fundamental toan understanding of our vision of opticalnetworking.

Transport networking offers four primary benefits:

• It is the only solution for fast survivability, andit supports service-layer restoration withefficient shared-protection architectures.

• It enables service-layer scalability, controllingthe pace of service-layer upgrades by freeing theservice-layer logical topology from thenetwork’s physical-link topology.

• It enables you to achieves a lower-cost network,especially when upgrade-fueled lifecycle costsare included.

• It establishes a future-ready unifyinginfrastructure for efficient multi-servicenetworking in an unpredictable serviceenvironment.

These benefits are apparent when considering thecost and functionality of intermediate nodes alonghigh-capacity, long-haul service routes. As shownin Figure 2, transport networking permits aflexible, cost-effective end-to-end connectionwithout incurring the cost of service-layerprocessing at the intermediate nodes. As trafficdemand grows, the transport network evolves frommanaging broadband circuits toward managingoptical channels.


However, what if optical transport never progressesbeyond point-to-point WDM interconnectsbetween service-layer elements? Point-to-pointtransport, rather than full-functionality transportnetworking, places new demands on the servicelayer for through-traffic networking andsurvivability, on top of the rapidly growing servicelayer requirements. Already, growth and churn areforcing many data-networking systems to bereplaced or upgraded every 12 to 18 months.Analysis of real networks suggests that without thesupport of transport networking, the aggregateservice-layer capacity more than doubles in orderto compensate, and likely shortens useful service-layer lifecycles even further.

Thus, even in emerging broadband data-networking applications, a full-functionalitytransport layer provides considerable value. Froman optical transport networking perspective, theessential elements are efficient and cost-effectivecapacity expansion, flexible optical-channelbandwidth management, and survivabilitymechanisms to support improved reliability of datanetworks.

TRANSPORT CAPACITY EXPANSION

As discussed earlier, the combination of anunprecedented demand for new capacity andmaximal usage of existing cable systems has lednetwork planners to look for the most expedientand cost-effective means to increase capacity. Thus,embedded carriers must evolve their currenttransport architectures, currently optimized for amix of narrowband and wideband services (64 kbps to 2 Mbps), to become optimized for

larger-capacity channels carrying broadband data,voice, and video. WDM’s increasingly high channelcounts and large information capacity are anexcellent solution to these needs.

OPTICAL CHANNEL BANDWIDTH MANAGEMENT

Bandwidth management functions – routingchannels through a network using add/drop, cross-connect, and interchange techniques – areessential for real networks, because they providean efficient means to accommodate growth andchurn. Several levels of bandwidth-managementgranularity are necessary, reflecting a range ofservice requirements and the increasing degree oftraffic aggregation toward the core of thenetwork2. While the explosion in data traffic isdriving enhanced requirements for fine-granularitybandwidth management in the IP and ATM servicelayers, it is simultaneously creating a rapidproliferation of ultra-broadband optical channels.Optical transport networking will provide flexiblemanagement for these optical channels,complementing SONET/SDH’s bandwidthmanagement for lower-bandwidth transportchannels.

NETWORK RELIABILITY

How will next generation networks deliver dataservices with the reliability of today’s voicenetwork? While service-layer enhancements willimprove packet loss and delay, transport-networksurvivability provides the best first line of defenseagainst major network faults. Survivable transportnetworks are essential for two reasons – fastestrecovery from faults such as fiber cuts, and efficientcapacity utilization through shared-protectionmechanisms. For broadband data networks,recovery time becomes even more critical as eachfiber span carries an ever-increasing amount ofinformation. Transport protection-switch times, onthe order of tens to hundreds of milliseconds, areorders of magnitude faster than for typicalservice-layer scheme, giving multi-service providersimportant options to meet premium Service LevelAgreements. In addition to delivering the fastestrecovery from faults, the next generation networkwill include advances in survivable networkefficiency and flexibility. The survivable opticalnetwork will be able to optimize the use of


Figure 2. Complementary Optical TransportNetworking and Service Layer Networking

Service LayerNetworking(IP, ATM)

OpticalNetworking

Optical Networking Element (OADM, OXC)

..........................

.............

........

......

......

........

..

2 For a more complete treatment of bandwidthmanagement, see [4].

protection bandwidth, as well service routing, onan optical-channel basis.

LEGACY TRANSPORT NETWORK EVOLUTION

WDM is a revolutionary technology in manyrespects. However, its impact on transportnetworking is largely evolutionary, leading to thenext phase of capacity expansion and serviceindependence. From a high-level architecturalperspective, optical transport networks will followmany of the same architectural principles as theirSONET/SDH predecessors. Therefore, a basicunderstanding of SONET/SDH transportnetworking (Figure 3) is essential to understandingthe nature of practical optical transport networks.

A decade of SONET/SDH deployment hasestablished several types of network elements andnetwork topologies as the basis for transportnetworking. There are three broad classes ofnetwork elements: multiplexers, includingTerminal Multiplexers (TMs) and Add/DropMultiplexers (ADMs); regenerators; and DigitalCross-connect Systems (DCSs). Among numerousnetwork topologies, survivable rings have provento be particularly useful.

A ring comprises a set of ADMs that allow traffic toenter and exit the ring, interconnected in a loopconfiguration. The main advantage of the ringtopology is protection; the ring offers traffic two

different ways of passing from ingress to egress. Tosatisfy varied requirements, there are several ring-based protection schemes, including SONET 2-fiber and 4-fiber Bi-directional Line SwitchedRings (equivalent to SDH Multiplex Section SharedProtection Rings) and SONET Unidirectional PathSwitched Rings (akin to SDH 1+1 Sub-NetworkConnection Protection in a physical ring). For long-span applications, regenerators can be placedat approximately 40 - 80 km spacing betweenADM nodes. Rings may also exist in openconfigurations (linear add/drop, open ring). Or, inthe simplest form of ring, the ADM can beconfigured in a point-to-point configuration (forexample, 1+1 SDH Multiplex SectionProtection/SONET Line Protection).

Digital Cross-connect Systems groom traffic atvarious rates to ensure that network facilities arewell used, and are key elements in mesh-basedrestoration architectures. In a typical application,low-speed signals dropped from an ADM arerouted through a digital cross-connect system,where they may be groomed (that is, separatedand recombined according to service type ordestination), and handed off to another ADM.Traditionally, cross-connect systems have notperformed the ring facility termination functionsassociated with ADMs (for example, MultiplexSection protection switching), although LucentTechnologies’ next-generation transport system,


Figure 3. SONET Transport Network Architectures

...... Access TDM Channels : 1.5 - 155 Mbit/s

(e.g. PSTN accesstrunks, ISP accesslines) SONET/SDH

ADMDCSTM

: Add/Drop Multiplexer: Digital Cross-Connect System: Terminal Multiplexer

Central Offices

MS-SPRing2.4 - 10 Gb/s

SNCP ring155 - 622 Mb/s

MS-SPRing2.4 - 10 Gb/s

Central Offices

RT sites

(PSTN trunks. Internet backbone, FR/ATM backbone)

- - - - Core TDM channels : 45 - 622 Mbit/s

ADM

TM

TM TM TM

TM

TM

ADM

ADM

ADM

ADM

ADM

ADM

ADM ADM

ADM

ADM

DCS

DCS

DCSDCS

Linear/MeshCore/

Long Haul

MetroInteroffice

Access

the WaveStar™ BandWidth Manager, integratesthis functionality and eliminates standalone ADMs.

ADMs and regenerators are most often deployed insingle-vendor configurations for management andinterworking simplification. On the other hand,DCS systems must support more extensive multi-vendor inter-working, because they serve a“hub” role in a central office, terminating electricaland optical signals from the multi-vendor network.

It should be noted that the variety of survivabilityarchitectures deployed within inter-office and long-haul backbone networks, as well as withinaccess transport networks, reflects the critical needfor high networking reliability and performance.Along with the dramatic increase in networkcapacity that has taken place over the last decade,customers have been concerned with such issuesas: surviving catastrophic failures in cable-facilityand equipment sites; the restoration times ofcritical circuits when a wire center failure occurs;and new survivability levels for applications suchas automatic teller machines, local area networkinterconnection, and mainframe-to-mainframecomputer links.

From the perspective of evolving to opticalnetworking, many SONET/SDH transport functionsand architecture principles will transfer to newoptical transport networks as the basic unit oftransport bandwidth shifts from time slot to opticalchannel (frequency slot).

THE NEXT PHASE: POINT-TO-POINT WDMPoint-to-point WDM represents the next phase inthe evolution of transport networking, and the firstrealization of optical networks based on multiplewavelengths. WDM deployment is growing rapidlybecause it is often a more cost-effective way toexpand capacity than any alternatives, such asadding fiber or replacing current-capacity (forexample, 2.5 Gbps) Time Division Multiplexing(TDM) transport systems with new, higher-rateTDM systems. New fiber is particularly expensivefor long routes, and its installation can take toolong to satisfy some customers. Upgrading TDMsystems often turns into wholesale replacement,and TDM technology is no longer keeping pacewith the emerging ultra-broadband data backbonedemands.

Compared to new fiber or upgraded TDM systems,point-to-point WDM systems (Figure 4) offer asuperior value proposition, in particular for thelong-distance environment, where electronicregenerators are required for long, “repeatered”transmission spans.

This value proposition is based on the following:

• High-capacity path routing: the high TDMcapacity per optical channel (frequency slot)allows “direct feed” from IP routers and ATMswitches with ultra-broadband interfaces.

• Electronic regenerator cost savings: for an N-channel system, a single Optical Amplifier(OA) replaces N electronic regenerators at each


Figure 4: Point-to-Point WDM for Core/Long Haul Capacity Expansions

Core/Long HaulTransport

SONET/SDH ADM DCS TM

: Add/Drop Multiplexer: Digital Cross-Connect System: Terminal Multiplexer

DWDM OLS OA

: Optical Line System: Optical Amplifier

TM

TM

TM

ADM

ADM

ADM

DCS

DCS DCS

Fiber exhaust and ultra-broadband channels (>2.4 Gbit/s)drive deployment of pt-pt DWDM

for capacity expansionInternetIP VPNFR/ATM

Pt-Pt DWDM

Pt-Pt DWDMOA

2.4 Gbit/s FiberMS-SPRing

OLS

OLS OLS OLS

ATM

ATMIP IP

repeater site.

• Greater transport capacity per fiber: theaddition of WDM Terminal Multiplexers (WDM-TMs) exploits the inherent capacity ofoptical fiber, which is vastly underused bysingle-channel optical transmission systems.

• Preservation of investment in existingequipment: existing TDM equipment need notbe replaced, but continues to operate in parallelwith other, new TDM equipment on the samefiber.

• Unification of transport for all services:different types of services, both existing andnew, can use the same fiber, virtuallyindependent of bit rate or protocol.

In short, WDM allows service providers to tap intothe full capacity of their existing fiber plant, thusmaximizing the return on existing facilities. Inaddition, the service provider has the flexibility ofadding new services onto the existing fiber toaccommodate new capacity demands.

As illustrated in Figure 4, point-to-point WDM canbe deployed on individual spans of a TDM ringtopology, as well as on linear networks. In eithercase, the existing TDM network continues tooperate as if it were still the only application onthe fiber (for example, TDM protection is stillactive), and new applications can be added to newwavelengths without impacting the existing TDMtraffic.

Several fundamental technology advances,particularly dispersion-managed fibers coupledwith Erbium Doped Fiber Amplifiers (EDFAs), havehelped make WDM a cost-effective, deployabletechnology. Commercially available WDM systemscarry 16 or more wavelengths, each capable ofcarrying up to a 2.5-Gbps or 10-Gbps signal.Further, the rapid pace of WDM-related technologyadvances has resulted in the announcement ofhigher and higher channel-count systems. Forexample, Lucent Technologies recently unveiled an80-channel, global optical networking system, theWaveStar™ OLS 400G, delivering up to 400 Gbpson a single fiber. Other advances, such as LucentTechnologies AllWave™ fiber, are opening uppreviously unusable parts of the optical spectrumfor WDM transport, permitting even morewavelengths on each fiber.

WDM represents the first step toward opticalnetworking, because it employs wavelength-basedtransport. However, in these initial point-to-pointapplications, most of the transport-networkingfunctionality is provided by the underlying TDMsystems that use the WDM span. As network trafficgrows and WDM deployment continues, opticalchannels will increasingly become the fundamentalmedium for exchange in networks. Transport-networking functions will migrate into the opticallayer, and carriers will begin to manage capacity atthe optical-channel level. Consequently, theapplication of WDM in the transport network willquickly evolve from point-to-point capacityexpansion to scalable and robust optical transportnetworking applications that cater to an expandingvariety of client signals with equally varied servicerequirements.

THE FUTURE: OPTICAL TRANSPORT NETWORK

EVOLUTION

As previously indicated, Optical TransportNetworking (OTN) represents a natural next stepin the evolution of transport networking. As anevolutionary result, optical transport networks willfollow many of the same high-level architecturalprinciples as followed by SONET/SDH transportnetworks. For instance, both SONET/SDH and OTNare connection-oriented, multiplexed networks.Thus, optical-network topologies and survivabilityschemes will closely resemble – if not mirror –those of SONET/SDH TDM networks.

At the same time, there are some important, ifsubtle, distinctions between optical andSONET/SDH networks. The major differencesderive from the particular form of “multiplexing”technology used: digital Time Division Multiplexingfor SONET/SDH versus analog Wavelength DivisionMultiplexing for OTN. The digital versus analogdistinction turns out to have a profound effect onthe fundamental cost/performance trade-offs inmany aspects of OTN network and system design.In particular, the complexities associated withanalog network engineering and maintenanceimplications, as outlined in Section 4, account forthe majority of the challenges associated withoptical transport networking.

To satisfy the short-term need for capacity gain,large-scale deployment of WDM point-to-point linesystems will continue. As the number ofwavelengths grows, and as the distance betweenterminals grows, there will be an increasing need


to add and/or drop wavelengths at intermediatesites. Hence, flexible, reconfigurable opticaladd/drop multiplexers (OADMs) will becomeintegral elements of WDM networks. As morewavelengths become deployed in carrier networks,there will be an increased need to manage capacity,as well as to hand off signals between networks, atthe optical-channel level. In much the same waythat digital cross-connects emerged to managecapacity at the electrical layer, optical cross-connects (OXCs) will emerge to managecapacity at the optical layer. Similar to theirelectrical counterparts, OXCs will be required tosupport a multi-vendor environment, whileindividual linear and ring subnetworks based onoptical line system (OLS) terminals and OADMswill typically be single-vendor.

Figure 5 depicts an optical transport network forcore, metro-interoffice, and high-capacity business-access applications. Initially, the need foroptical-layer bandwidth management will be mostacute in the core (long distance) environment. Thelogical mesh-based connectivity found in the corewill be supported by way of physical topologies,including OADM-based shared protection rings,and OXC-based mesh restoration architectures,depending on the service provider’s desired degreeof bandwidth “overbuild” and survivability time-scale requirements. As similar bandwidth-management requirements emerge forthe metropolitan interoffice (IOF) and Accessenvironments, so too will OADM ring-based

solutions optimized for these applications: opticalshared protection rings for mesh demands, andoptical dedicated protection rings for hubbeddemands.

Hence, just as the optical amplifier (OA) was thetechnology enabler for the emergence of WDMpoint-to-point line systems, in turn, OADMs andOXCs will be the enablers for the emergence of theoptical transport network. The optical layer willcome to serve as the unifying transportinfrastructure for both legacy and converged packetnetworks. Of course, service-provider movement tooptical transport networking will be predicated onthe transfer of the required transport-networkingfunctionality to the optical layer, concurrent with thedevelopment of an OTN maintenance philosophyand associated network-maintenance features. Thisleads us to the key factors associated with realizingthe vision of optical transport networking.

BUILDING A NEW TRANSPORT LAYER:A PRACTICAL VISION FOR OPTICALTRANSPORT NETWORKING

Realizing optical transport networkingarchitectures involves carefully balancing a

web of tradeoffs in context with a set of criticalfactors, and successfully rising to meet associatedtechnical and networking challenges. Implicit inthe balancing exercise is the ubiquitous


Figure 5. Optical Transport Network Architecture

Core/Long Haul

...... Access Optical Channels 155 Mbit/s to Gbit/s

Optical Networking OXC OADM OLS OA

: Optical Cross Connect: Optical Add/drop Multiplexer: Optical Line System: Optical Amplifier

MetroInteroffice

Access

Central Offices

8 - 16 λ

Ring 40 - 80 λ

Mesh40 - 80 λ

(next generation packet core ; legacy networks)Core Optical Channels : 2.4 Gbit/s and up

OLS

OLSOLS OLS

OLS

OADMOADM

OADM

OADM

OADM

OXC

OXCOXC

--

OADM OADM

OADM

OADM

OADMOADM

OADM

OXC

cost/functionality tradeoff – in other words, atwhat price functionality? Rather than a singlesolution for all applications, a range of optical-networking architectures will arise as eachmarket segment applies its unique priorities tomake fundamental architecture tradeoffs. Inconsidering the key issues and cost/functionalitytrade-offs for next-generation optical transportnetworks, five critical factors emerge:

• Analog network engineering: In order toassure a high-performance network, it isnecessary to limit the accumulation of analogimpairments, while balancing cost with othertradeoffs. A network-engineering strategy thatenforces network integrity at subnetworkdomain boundaries, resulting in a segmentedapproach to span engineering and regeneration,will help realize this goal

• Service Transparency: In order to future-proof the network, and to simplify theengineering of the shared multi-service core, astrategy is needed to minimize the amount ofclient-dependent processing within the core ofthe optical transport network

• Survivability: While continuing to offer thefastest possible recovery from faults, a strategyis needed to ensure that the next-generationtransport network continues to offer a highdegree of survivability coupled with networkefficiency and service flexibility

• Maintenance: An essential element to realizingcost and service transparency goals is amaintenance philosophy tailored to the uniqueneeds and constraints of optical transportnetworking. This OTN maintenance philosophyshould build on prior experience with TDM andFDM systems, incorporating useful features,while avoiding the mistakes of the past

• Interoperability: As standards and underlyingtechnologies mature, near- and mid-termevolutionary strategies are required to helpensure a smooth transition from single-vendorsolutions to multi-vendor, interoperable opticaltransport networks

ANALOG NETWORK ENGINEERING

Despite the apparent similarity betweenSONET/SDH and optical-networking architectures,the difference in design and implementation of

these transport networks is fundamental. While theformer is an exercise in digital networkengineering, the latter is an exercise in analognetwork engineering, somewhat similar to thedesign of frequency division multiplexed (FDM)networks of old. The chief advantage of digitaltransmission has been the ruggedness of the digitalsignal. The limiting factor in a properly designeddigital system is not impairments introduced by thetransmission facility, but rather the accuracy of theconversion of the original analog waveform intodigital form. A major advantage has been thatsystem noise is controlled by the design of thedigital network element and is essentiallyindependent of the total length of the system.Because no appreciable degradation is incurred intime division multiplexing or demultiplexing,facility arrangements do not typically need to takethe number of previous multiplex/demultiplexoperations into account. In contrast to TDMsystems, the phase distortion and attenuation (andresulting noise degradation) suffered by channelsat the band edge of FDM systems can affect thesystem performance and so must be explicitlyconsidered in engineering the total system [5]. Inthe excitement associated with advances in optical-networking technology, the implications ofthe differences cited above have not always beensufficiently considered.

In what is termed “all-optical” networking, signalstraverse the network entirely in the opticaldomain, with no form of opto-electronic processingwithin optical-network elements. This implies thatall signal processing – including signal regeneration,routing, and frequency slot interchange (FSI), orwavelength interchange –takes place entirely in theoptical domain. Over recent years, there has beenconsiderable debate over the nature of the term“optical” in optical networking; in retrospect, it isclear that function and implementation were oftenconfused. Given analog engineering constraints,and considering the current state-of-the-art in all-optical processing technology, the global, oreven national, “all-optical” network is notattainable. In particular, opto-electronic conversionmay be required in optical network elements toprevent the accumulation of transmissionimpairments – impairments that result from suchfactors as fiber chromatic dispersion and non-linearity’s, cascading of non-ideal flat-gainamplifiers, optical signal crosstalk, andtransmission-spectrum-narrowing from cascadednon-flat filters [6]. Opto-electronic conversion canalso support wavelength interchange, which is


currently a challenging feature to realize in the all-optical domain.

In short, in the absence of commercially availabledevices that perform signal regeneration to mitigateimpairment accumulation and support wavelengthconversion in the all-optical domain, somemeasure of opto-electronic conversion should beexpected in practical optical-networkingarchitectures. Because the use of opto-electronics isoften associated with higher-cost solutions, theensuing exercise in cost reduction involvesestablishing what constitutes the minimumrequired amount of opto-electronics necessary toassure the presence of desired optical transportnetworking attributes enabled by their presence. Inthe near term, the result of this exercise is reflectedin optical-networking architectures characterizedby transparent (or “all-optical”) segments, boundedby opto-electronics, as shown in Figure 6.

Until standards for the OTN – which are in theearly stages of development [7] – are fullydeveloped, the near-term transparent subnetworksshown in Figure 6 will likely remain single-vendorsolutions. Hence OXCs, as points of multi-vendorinteroperability, are shown distinct from near-term,single-vendor OADM and OLS-based subnetworks.The emergence of OTN standards will change thisview, as described below.

SERVICE TRANSPARENCY

As we have seen in the preceding discussion,practical considerations will govern the ultimaterealization of the Optical Transport Network(OTN). Paramount among these considerations isthe network operator’s desire for a high degree ofservice transparency within the future transportinfrastructure. As the term “transparency” hastended to engender much discussion, we willattempt here to clearly define the intent of theterm “service transparency”.

SONET/SDH offers a good example of servicetransparency. Specifically, for a desired set of clientsignals targeted for transport on a SONET/SDHnetwork, individual mappings are defined forcarrying these signals as payloads of SONET/SDHserver signals (along with associated SONET/SDHoverhead to assure proper networkingfunctionality). Once a client signal has beenmapped into its SONET/SDH server signal at theingress of the SONET/SDH network, an operatordeploying such a network need not possessdetailed knowledge of the client signal until it isde-mapped at the network egress. This definition ofservice transparency is equally applicable to theOTN. Indeed, once a client signal is mapped into anoptical channel, an OTN network operator shouldnot require detailed knowledge of (nor access to)the client signal until it is de-mapped at the OTNegress. Hence, the optical-network ingress andegress points should delimit the domain of OTNservice transparency.


Figure 6. Practical Optical Transport Networking Architectures

Optical EngineeringSubnetwork

Opto-electronicProcessing Unit

Transparant Subnetworks, Bounded by opto-electronics

OTNClientConnection

OLS OXC OADM

Service transparency is an essential factor, becauseit helps control the complexity of the opticalnetwork’s role as a unifying infrastructure for alltypes of clients. In the context of optical transportnetworking, the notion of service transparency iscurrently limited to digital client signals, but thescope will ultimately expand to include analogclient signals as well. Without service transparency,one must be aware of the type of client signal onevery channel throughout the network, resulting inincreased network engineering and maintenancecomplexity. The most important factor in realizingservice transparency is to eliminate all client-specific equipment and processing betweenOTN ingress and egress points. Fortunately, at theingress/egress point, it is easier to accept client-dependent equipment, because it is generallydedicated on a per-service basis anyway. Forexample, we might expect different opto-electronicport units to be required to support a mix of 622 Mbps, 2.5 Gbps, and 10 Gbps SDH/SONET;Gigabit Ethernet; emerging IP/WDM mappings;and leased “clear-channel” optical channels. Thus,service transparency within the OTN minimallyrequires flexibility at the optical-channel layer tocarry a wide variety of digital client signals.

We will now examine the impact, and consequentimplications, on service transparency of deployingoptical-network architectures characterized byoptically transparent segments bounded by opto-electronic processing. One immediateengineering consequence of such an architecture isthat cascading opto-electronic processing units willintroduce digital jitter and wander accumulationwithin digital client signals carried on opticalchannels, unless they employ so-called 3Rregeneration to regenerate, reshape, and retimethese client signals. By virtue of this re-timingfunctionality, such devices have traditionallydepended on the client-signal bit rate. However,bit-rate dependent opto-electronics result inconstraining the service transparency of opticalnetworks. The use of “broad-band” or bit-rateindependent opto-electronic regenerators with re-timing functionality [8] will alleviate thisconstraint, thereby “opening up” the opticalnetwork to a wider variety of client signal bit rates.Such opto-electronic regenerators will enablecascaded OTN architectures composed of abalanced combination of optical transparency and“feature-enhanced” opto-electronics supporting bit-rate and protocol independence at the optical-channel layer. Bit-rate independent opto-electronic regenerators with retiming

functionality are therefore a key enabler for OTNservice transparency.

NETWORK MAINTENANCE

The evolution toward optical transport networkingalso brings further challenges to integratedtelecommunications network management. A setof requirements must be established related tofault, configuration, and performancemanagement, including the establishment of speed,latency, and robustness requirements associatedwith OTN management functions andcommunications.

Indeed, when we examine OTN managementneeds, we can recognize many that are applicableto both SONET/SDH and OTN networks. However,it is essential to avoid the trap of assuming thatwhatever has been done for SONET/SDH is directlyapplicable to the optical transport network. Someof the forces that led to SONET/SDH standardswere particular to that time in network evolution.For example, one of the main maintenanceadvantages for moving from PDH to SONET/SDHwas an opportunity to significantly expand, andmore importantly to truly standardize, capabilitiesfor monitoring the validity, integrity, and quality oftransport. SONET/SDH, with its digital signalformat, offered a relatively painless means for theseenhancements, by means of layered provision ofassociated maintenance overhead. Networkoperators saw opportunities to attempt a proactivemaintenance philosophy, fueled by acomputationally intensive collection of masses ofdata, and assign responsibilities for poorperformance where a signal crosses network-operator boundaries.

Just as SONET/SDH offered an opportunity tocorrect some of the operating problems of the PDHnetwork, so too will the OTN provide anopportunity to correct some of the operatingproblems associated with the SONET/SDHnetwork. For example:

• A reassessment of the types and uses ofperformance information is in order. It hasbecome clear that network availability has beenlargely unaffected by attempts at proactivemaintenance strategies, and by the resultantabundance of SONET/SDH performance-monitoring information. Rather, networkavailability is primarily determined bycomponents, design, and survivability decisions,


so it seems reasonable to first focus energy onquick fault localization, to speed network uprepairs. Volumes of detailed performance data atevery node within a subnetwork will notcontribute to fault localization. On the otherhand, performance data has become absolutelyessential at subnetwork ingress/egress points forverifying tariffs and troubleshooting betweenoperators. It is fortunate that performance datawill be readily available at these ingress/egresspoints, because they are very likely to haveopto-electronic processing, and thus will offerrelatively easy access to both digital and analogmeasurements

• Two maintenance needs were not fully realizedwhen the first SDH/SONET standards were putin place, and have proven difficult to realize anddeploy in a widespread, practical fashion. OTNmaintenance offers some hope that these needsmay be better satisfied. The current difficulty ofknowing the physical topology of the networkcan be resolved by providing a trace function bymeans of the Optical Supervisory Channel(OSC) on each fiber segment. Additionally,requirements for tandem connectionmonitoring on signals that traverse multipleoperator domains can be developed early in thestandards process

Again, owing to the analog nature of WDM,establishing the performance of a client signal in an

OTN is a very different proposition from that forSONET/SDH systems. The difficulties related toperformance monitoring within optical transportnetworks are inherent in the parameters that needto be measured. Specifically measured opticalparameters must be accurately correlated withdegradations in the client-signal layers in order toindicate that the optical layers are not providingthe required level of service. It is becoming clearthat establishing such a relationship using practicalmeasurement techniques remains a significantchallenge [9].

Aside from details of performance monitoring andmeasurements, a more general and fundamentalchallenge remains: what operations,administration, and maintenance (OAM)information is needed at each point in thenetwork, and how should it be carried? Thisinformation falls into two categories: informationthat must be associated with, and must follow, aparticular optical channel connection; andinformation that need not necessarily follow thisconnection. It is the former that will provide themain challenge, because the latter can be carried inthe optical supervisory channel, or through someother external means.

Our practical optical-network vision calls for amaintenance strategy aligned with an engineeringapproach of transparent optical subnetworksbounded by opto-electronics. As illustrated in


Figure 7. OTN Maintenance



Thorough Performance Info at Subnet Boundaries (especially for tariff/SLA verification)

Intra-subnet Maintenance Targets Fault Location

OTNClientConnection

OLS OXC OADM

Figure 7, the maintenance strategy, out ofnecessity, becomes a balanced combination of“opto-electronic enabled maintenance” (whereopto-electronics are present) and “off-line,” shared,non-intrusive measurement and monitoring of theoptically-transparent segments.

NETWORK SURVIVABILITY

Survivability is central to optical networking’s roleas the unifying transport infrastructure. As withmany other architectural aspects, optical-networksurvivability will bear a high-level resemblance toSONET/SDH survivability, because the networktopologies and types of network elements are verysimilar. Within the optical layer itself, survivabilitymechanisms will continue to offer the fastestpossible recovery from fiber cuts and otherphysical-media faults, as well as more efficient andflexible management of protection capacity.Recovery times will range from below 50 milliseconds to a few hundred milliseconds,depending on the particular network architectureand the type of fault. There are a variety ofsurvivability schemes that may be considered:

For point-to-point WDM line systems involvingWDM-TMs and WDM-OAs, 1+1 optical protectionswitching schemes similar to those for SDH/SONETare often considered. The addition of the OADM tothe WDM line system will allow more advanced,optical-layer protection switching schemes, againanalogous to those for SDH/SONET. Simplearchitectures for optical add/drop, such as 1+1Optical SubNetwork Connection Protection (O-SNCP), are single-ended and thus do notrequire coordinated Automatic ProtectionSwitching (APS) algorithms, protocols, andsignaling channels. They can be applied to bothlinear add/drop and ring topologies.

More complex, optical-layer protection switchingarchitectures will be capable of greater flexibilityand bandwidth efficiency, especially for meshedtraffic demand patterns. For ring topologies in corenetworks, an optical-layer version of theSDH/SONET Shared Protection Ring is a likelycandidate.

The introduction of OXCs will enable optical-layermesh-based architectures. For some applications, amesh architecture will be lowest cost, largely dueto intelligent, automated mesh restoration schemesthat can be highly optimized for efficient allocationof protection capacity. Mesh-based restoration

algorithms have been developed for SONET/SDHnetworks; however, the introduction of new opticaltechnology and associated network elements offersopportunities for enhancements in restorationspeed to support survivable optical transportnetwork architectures. For example, distributedmesh-based restoration schemes can be employedthat not only increase restoration speed, but alsoaccommodate churn due to frequent provisioningand rearrangements, with associated algorithmsthat can interact constructively with a centralizedcapacity planning system.

While the general classes of survivable optical-network architectures are similar to thosein SDH/SONET, there are some fundamentaldifferences in functionality and applications.Survivable optical networks must provide flexiblesupport for the requirements of a wide range ofclient signals, from embedded SONET/SDH ringsthat simply require unprotected access to WDMwavelengths to highly optimized optical-layerrestoration schemes capable of differentiatingamong the needs of different service classes in aconverged packet core. Additionally, survivableoptical networks must be even more bandwidthefficient in order to support the tremendousbandwidth carried on WDM links. These gains inclient networking flexibility and bandwidthefficiency also rely on a coherent multi-layersurvivability strategy in which optical layersurvivability mechanisms coexist gracefully withthose of its client signals (for example,SONET/SDH, ATM, and IP).

Issues associated with multi-layer survivabilityhave been cited within recent conferences (see[10]) and research consortia (see [11]). Forinstance, in the SONET world, consider an ATMOC-3/OC-12 based backbone network over anSONET OC-48 ring, with some of the SONET spanscarried as optical channels over an OTN system.Each of these three layers – ATM, SONET, andWDM – may have survivability mechanisms;however, the absence of a cohesive multi-layersurvivability approach may cause contentionamong the layers and result in inefficientoperations and unnecessary survivability actions.More generically, there are two pitfalls to avoid inmulti-layer survivability, as illustrated in Figure 8.The first involves the potential for considerablewasted bandwidth, because each layer reserves itsown backup resources to use during failures. Thesecond pitfall relates to the fact that a singlephysical-layer failure (such as a fiber cut) can


result in many unnecessary protection actions at theclient layers. This undesirable result can greatlyextend the end user’s measured outage from asingle fault, potentially activating the “minimumavailability” penalties of service-level agreements,as well as complicating the carrier’s maintenancetasks.

Considerable discussion is taking place within theindustry regarding possible multi-layer survivabilitystrategies. However, given the range of operatorpolicies, services, and deployed technologies, it isclear that any proposed approaches must enableflexibility that allows them to be efficiently andcost-effectively tailored to individual operatorneeds. For example, an ideal approach couldinvolve deploying only one survivability scheme

per subnetwork domain per service, coupled withavailability of provisionable hold-off timers foroptional premium service and restoration support.As a simple case, we can consider that for mostservices, an optical-layer protection or restorationscheme will be desirable, because it can offer aunifying solution. Sometimes, though, it may bebest to use the client layer’s survivability scheme,and disable optical-layer protection on that client’soptical channel.

In summary, optical network survivability willcontinue to offer the fastest possible recovery fromfaults, while catering to a wide diversity of clientsignals having varying survivability requirements.


Figure 9. Interconnecting OTN Islands



Interconnection Among Idependent "OTN Islands" via Single-ChannelClient or WDM Physical Layer Connection

OTNClientConnection

OTNClientConnection

OLS OXC OADM

Figure 8. Common Multi-Layer Survivability Pitfalls

...............

...............

........................................... ........................

Wasted BandwidthPitfall : Pitfall : Unnecessary Protection Actions

MultipleClient LayerActions ...

Up to 75% is Spare Capacity

Client (IP, ATM, STM) Clients (IP, ATM, STM)

ON

ON

ON

... From a SingleON LayerFault

X

XXX

INTEROPERABILITY

The application of any new technology, such asWDM, in the telecommunications environmentrequires that standards be developed to facilitatemulti-vendor interworking and networkinterconnection. A key aspect is to fully define theinformation that is associated with the OpticalNetwork Node Interface (NNI) – for example, theformat of the optical supervisory channels thatcarry OAM information between networkelements. Another key aspect involves the physicallayer and requires a very precise description of thephysical properties of the multi-wavelength opticalsignal. Not only is this an extremely difficult task toachieve for what is essentially an analog network-design problem, it is hampered by the richdiversity of options for optical transmission [7].For example, the optimal choice of operatingwavelengths for a WDM line system is dependenton a number of factors, including: type of fiber, OAtechnology, filter technology, span distance, andoverall target reach for the system. In general, thechoice of operating wavelengths is dictated by acombination of underlying technology choicescoupled with the envisioned application(s).

Given the range of specifications that need to bestabilized, and the pace of technology advances,interoperability is expected to follow anevolutionary path. As a first step, multi-vendorinterworking will likely be achieved by way ofsingle-channel OTN, client-level interconnection(for example, SONET/SDH interfaces conforming toexisting standards, such as ITU-T Rec. G.957). This

is shown in Figure 9; again, note that the OXCs, aspoints of multi-vendor interoperability, are distinctfrom the single-vendor OADM and OLS-basedsubnetworks. In the near future, multi-channelOTN, client-level interconnection will occur byusing standard WDM wavelengths and physical-layer parameter specifications. The abovesolutions imply interconnection among a networkof “OTN islands” [12].

In the future, single-channel or multi-channel,OTN-level interworking will occur when full OTN-NNI specifications are available, including thespecification of optical-layer overheads and theOptical Supervisory Channel. As OTN standardsmature, the level of interworking will allow foroptical channel-level continuity and opticalnetworking beyond constrained transparentsegments. As shown in Figure 10, OTN-levelinterworking will permit the independent “OTNislands” to grow, and in some cases grow together,creating larger administrative and span engineeringsubnetworks.

REALIZING THE OPTICAL NETWORKINGVISION

The transfer of transport-layer functionality tothe Optical Transport Network, concurrent with

the development of a maintenance philosophy andassociated OAM features, will enable evolutionfrom single-channel optical-transmission systems toOTNs with advanced features such as optical


Figure 10. OTN-Level Interworking

TransparentSubnetwork


OTN-Level Interworking : No longer Interconnecting OTN "Islands" via OTN-Client Level Signals OCH-Level Continuity is Maintained Across the Optical Transport Network

OCH Endpoint (OCH Continuity Maintaind Across the Network)

OTNClientConnection

OTNClientConnection

OLS OXC OADM

channel routing and wavelengthconversion/interchange. It will also facilitaterealizing the network operator goal of a flexible,scalable, and robust transport network, catering toan expanding variety of client signals havingequally varied service requirements (flexibility,scalability, and survivability coupled with bit-rateand protocol independence). The evolutionary pathto the ultimate goal of a unified transport layer willinvolve a number of key considerations, includingthe optimization of optical-layer transportfunctionality while maintaining client-independence. The optical layer, to remain future-proof, insofar as future-proofing is possible, shouldnot be optimized for any particular client signal(this includes the optical-layer control plane). Justas the transport network has evolved to new signalformats, so too will today’s “networks du jour.”We need to avoid “locking-in” the optical layer to amyopic view of the future, based solely on legacyTDM and legacy (“best-effort”) data signals.

It should be emphasized that SONET/SDH willcontinue to be an integral part of the data/optical“convergence” networking evolution. Whiletransport-networking responsibility will shift(dramatically, perhaps) over time, from theSONET/SDH layer to the optical layer, SONET/SDHwill continue to serve the networking role formodest aggregate point-to-point bandwidthdemands, such as voice trunking and many traffictypes at the edges of the network. Some newtelecommunications carriers will aggressively buildpacket-based networks from the ground up, andtherefore will not have to deal with integration oflegacy voice/TDM networks. On the other hand,the vast majority of carriers will find a way ofintegrating their current networks with emergentnetwork infrastructures based on the optical layer.

Building the high-capacity optical transportnetwork of the future depends on innovativesolutions incorporating cutting-edge technologiescoupled with sound networking principles. A Special Issue of the Bell Labs Technical Journal onOptical Networking outlines the Lucent NetworkVision for leading the transition to a network ofnetworks unified by Optical Networking,highlighting Lucent’s innovative solutions for:

• The evolution of flexible bandwidthmanagement [13][14] and survivable networkdesign [15] in the era of convergeddata/transport networking

• Architecture and protocol considerations for a

converged optical data network [16][17]

• Optical Networking – the core of the nextgeneration network infrastructure [18], fromresearch to realization [19]

• Migrating Optical Networking toward viablemetropolitan interoffice and broadbandbusiness-access transport solutions [20]

• Critical components for optical networking – optical amplifiers [21], opticaladd/drop technologies [22], and the future ofhigh-capacity transport [23]

• Building blocks of the Optical Networkinginfrastructure, from fiber [24] to next-generation optical cross-connects [25]

• Optical Networking solutions for enterprise andresidential access applications [26][27]

SUMMARY

Arevolution in data networking fueled by theexplosive growth of the Internet is in

significant measure responsible for the evolution oftransport networking toward an OpticalNetworking infrastructure. Optical TransportNetworking will leverage the transportinfrastructure in the era of data/transportconvergence by offering carriers unprecedentedarchitectural flexibility – client-protocol (and bit-rate) independence, and service differentiationby separation of optical channels for different typesof services (TDM, ATM, IP, optical leased lines, andso forth). With a practical vision for OpticalTransport Networking, a balanced consideration ofanalog network engineering, service transparency,survivability, maintenance, and interoperability willrender a cost-effective, survivable, and flexiblebroadband optical-transport infrastructure. Inshort, we do not have long to wait for an optical-transport network that can rise to the challenge ofproviding an optimized layer of high-capacity,high-reliability bandwidth management thatincludes multi-service support is close at hand.


ACKNOWLEDGMENTS

The authors would like to thank the manyfriends and colleagues in the Lucent

community who have (in one way or another)contributed to the material presented in this paper.Special thanks are extended to Jon Anderson, JohnEaves, and George Newsome for their insightfulcomments during the preparation of thismanuscript.

REFERENCES

1. Special Issue on Packet Networking, Bell Labs Technical Journal, Vol.3, No.4, Oct.-Dec. 1998

2. R. Castelli, T. Kraus, “Market Trends and Evolution for Optical Transmission Systems”, Alcatel Telecommunications Review, 3rd Quarter 1998

3. D. Al-Salameh, M. Fatehi, W. Gartner, S.Lumish, B. Nelson, and K. Raychaudhuri,Optical Networking, Bell Labs Technical Journal,Vol.3, No.1, Jan.-Mar. 1998, pp. 39-61

4. Curt Newton, “Selective Layered BandwidthManagement for the Network of Networks”,Proceedings of NFOEC’98 , Orlando, FL,September 1998, Vol.2, pp. 109 – 120

5. Members of the Technical Staff, Bell TelephoneLaboratories, Transmission Systems forCommunications, fifth edition, 1982

6. K. Bala, R. R. Cordell, E. L. Goldstein, “Thecase for opaque multiwavelength lightwavenetworks,” Proceedings of the IEEE/LEOS SummerTopical Meeting on Global InformationInfrastructure, Keystone, Colorado, August 1995

7. A. McGuire and P. Bonenfant, "Standards: TheBlueprints for Optical Networking", IEEECommunications Magazine Special Issue on "OpticalNetworking Has Arrived , February 1998, pp. 68-78

8. M. Ushirozawa, K. Asahi, A. Noda, S. Fujita,"Bit-rate-Independent SDH/SONETRegenerator for Optical Networks," Proceedingsof the IEE European Conference on Communications,ECOC'97 , Eidenburg, Scotland, September 1997

9. ITU-T COM15-121, "Signal Quality Monitoringin Optical Networks," Ericsson, August 1998

10. J. Manchester and P. Bonenfant, "Fiber OpticNetwork Survivability: SONET/OpticalProtection Layer Interworking", Proceedings ofNFOEC'96, Denver, CO, 1996, Vol. 3, pp. 907-918

11.A. Autenrieth, K. Van Doorselaere, A. Iselt, P.Demeester, K. Struyve, L. Vandendriesseche,"Simulation and Evaluation of Multi-layerBroadband Networks," Proceedings of the FirstInternational Workshop on the Design of ReliableCommunication Networks (DRCN'98 Workshop),Brugge, Belgium, 17-20 May 1998, Paper O40,pp. 1-6


12.A. Saleh, "Islands of Transparency - AnEmerging Reality in Multiwavelength OpticalNetworking," Proceedings of the IEEE/LEOSSummer Topical Meeting on Broadband OpticalNetworks and Technologies: An Emerging Reality,Monterey, California, July 1998

13.B. Doshi, P Harshavardhana, and C. Sunada-Wong, "Role of the BandwidthManager (BWM) in Broadband NetworkInfrastructure: A Quantitative Study," Bell LabsTechnical Journal, Vol.4, No.1, Jan.-Mar. 1999

14.C. Alegria, S. Bergstein, K. Dixson, C. Hunt,and M. Wilson, "The WaveStar™ BandWidthManager: The Key Building Block in the NextGeneration Transport Network," Bell LabsTechnical Journal, Vol.4, No.1, Jan.-Mar. 1999

15.B. Doshi, S. Dravida, P. Harshavardhana, O.Hauser, and Y. Wang, "Optical Network Designand Restoration," Bell Labs Technical Journal,Vol.4, No.1, Jan.-Mar. 1999

16.B. Doshi, S. Dravida, E. Hernandez-Valencia,W. Matragi, A. Qureshi, J. Manchester, and J.Anderson, "A Simplified Data Link (SDL)Protocol for High-Speed Packet Networks," BellLabs Technical Journal, Vol.4, No.1, Jan.-Mar. 1999

17.J. Anderson, J. Manchester, A. Rodriquez-Moral, and M. Veeraraghavan, "Architecturesand Protocols for IP Optical Networking," BellLabs Technical Journal, Vol.4, No.1, Jan.-Mar. 1999

18.L. Wei (China Academy of TelecommunicationsResearch and Planning), Y. Chen and G. Wong,"The Evolution of China's Optical FiberNetworks," Bell Labs Technical Journal, Vol.4,No.1, Jan.-Mar. 1999

19.S. Johnson, "MONET DC Network," Bell LabsTechnical Journal, Vol.4, No.1, Jan.-Mar. 1999

20.Y. Chen, M. Fatehi, H. La Roche, and J. Larsen,"Metro Optical Networking," Bell Labs TechnicalJournal, Vol.4, No.1, Jan.-Mar. 1999

21.R. Giles, "Wavelength Add/Drop Technologies,"Bell Labs Technical Journal, Vol.4, No.1, Jan.-Mar. 1999

22.Y. Sun, "Optical Fiber Amplifiers," Bell LabsTechnical Journal, Vol.4, No.1, Jan.-Mar. 1999

23.A. Chraplyvy, "High-Capacity LightwaveTransmission Experiments," Bell Labs Technical

Journal, Vol.4, No.1, Jan.-Mar. 1999

24.J. Refi, "Optical Fibers for Optical Networking,"Bell Labs Technical Journal, Vol.4, No.1, Jan.-Mar. 1999

25.N. Jackman, "Optical Cross-Connects forOptical Networking," Bell Labs Technical Journal,Vol.4, No.1, Jan.-Mar. 1999

26.D. Buchholz, C. Byers, S. Hinterlong, G.Huensch, J. Runyon, T. Nanke, and H.-C. Y ,"Broadband Fiber Access (BFA): A Fiber-to-the-Customer Access Architecture,"Bell Labs Technical Journal, Vol.4, No.1, Jan.-Mar. 1999

27.T. Wood, "Fiber Vision System," Bell LabsTechnical Journal, Vol.4, No.1, Jan.-Mar. 1999



GLOSSARY

Abbreviations Used:

ADM Add/Drop Multiplexer

APS Automatic Protection Switching

ATM Asnychronous Transfer Mode

DCS Digital Cross-Connect System

DWDM Dense Wavelength Division Multiplexing

EDFA Erbium Doped Fiber Amplifier

FR Frame Relay

IOF Interoffice

IP Internet Protocol

ITU International Telecommunication Union

kbps kilobits per second

Mbps Megabits per second

NNI Network Node Interface

O-SNCP Optical SubNetwork Connection Protection

OA Optical Amplifier

OADM Optical Add/Drop Multiplexer

OAM Operations, Administration, and Maintenance

OCh Optical Channel

OLS Optical Line System

OSC Optical Supervisory Channel

OTN Optical Transport Network

OXC Optical Cross Connect

PDH Plesiochronous Digital Hierarchy

PSTN Public Switched Telephone Network

SDH Synchronous Digital Hierarchy

SNCP SubNetwork Connection Protection

SONET Synchronous Optical Network

TDM Time-Division Multiplexing

TM Terminal Multiplexer

Tbps Terabits per second

VPN Virtual Private Network

WDM Wavelength Division Multiplexing

This document is for planning purposes only, and is not intended to modify or supplement any Lucent Technologies specifications or warranties relating to these products or services. Performance figures and data quoted in this document are typical and must be specifically confirmed in writing by Lucent Technologies before they become applicable to any particular order or contract. The company reserves the rightto make alterations or amendments tothe detailed specifications at its discretion.

The publication of information in this document does not imply freedom from patent or other protective rights of Lucent Technologies or others.

WaveStar and AllWave are trademarks ofLucent Technologies Inc.

To learn more about our comprehensiveportfolio and the new WaveStar™ ONGSeries, please contact your LucentTechnologies Sales Representativeor call 1-888-4 LUCENT. Visit our web site at http://www.lucent-optical.com

Copyright © 1999 Lucent TechnologiesInc.All rights reservedPrinted in Holland / USA

Lucent Technologies Inc.Marketing CommunicationsOrder Number: WP-004/9905028WP Issue 2 MCO/BAP 05/99

If you enjoyed this publication and are interested in our other White Papers, please visit the Optical Networking web site at http://www.lucent-optical.com/resources/ for a complete listing of currently available titles.

a practical vision for optical transport networkinggrover/mesh_networking/wp004[1].pdf · a...

Documents