INTELLIGENCE IN OPTICAL NETWORKS

Architecting the Services Optical Network

Eve L. Varma, Sivakumar Sankaranarayanan, George Newsome, Zhi-Wei Lin, and Harvey Epstein Lucent Technologies

ABSTRACT 80 percent per year. Reliance on silicon alone cannot match this traffic growth. In this new millennium, the most valuable Demand for access technologies such as cable commodity will be information, as seen by the modems, digital subscriber line (xDSL), and surge in demand for faster access technologies data is rapidly increasing. Indeed, some such as cable modems, xDSL, and wireless data. analysts consider that data will eventually make The tremendous growth in data traffic, particu- up 99 percent of the content sent over telecom- larly that associated with the , is chang- munications networks. Network operators have ing the way it is carried over public and private been finding that their more aggressive projec- networks. Together with the rapid advances in tions of traffic growth have tended to be wrong, optical networking technology and the spawning and furthermore, that this demand for traffic of a new category of wavelength services stimu- occurs in often unexpected locations. The sup- lated by new high-speed data requirements, this porting network infrastructure needs to be inher- is dramatically changing network architectures ently scalable to handle this traffic, and and the relationship between network service sufficiently flexible to respond to forecast uncer- providers and their customers. Incumbent carrier tainties. At the same time, it is worth bearing in globalization and new carrier entry require sup- mind that the exponential growth in IP-based port for a broader set of business models and data traffic has not yet translated into a corre- range of service interfaces. The growth and sponding increase in revenue [1]. Higher-volume expansion of networks imply a greater need for lower-revenue service mixes are driving the need more scalable solutions and more automation for increased profitability reflected in near-term for network maintenance. An intelligent optical concerns with cost per bit, high-capacity fiber, core optimized for service optical networking, and cost of spare equipment. It is clear that net- the services optical network, needs to take the work operators will need to dramatically best aspects of both the transport and data net- increase, and maximally share, backbone net- working domains to meet the above needs. This work infrastructure capacity. Moreover, lowering article focuses on architecting the services opti- the cost per bit is not just equipment first cost, cal network to meet the challenges of optical but requires efficient and optimized network data networking, and includes a snapshot of sup- design to reduce operation cost. porting standardization activities. At the same time, the services offered, how quickly these services can be provided, and effi- NTRODUCTION ciency in supporting those services are key differ- I entiators for network operators. Within the Transport networking has steadily grown more backbone/core network in particular, aggressive complex as a consequence of more sophisticated business models and limited network operator customer needs, the convergence of data and staff are driving the need for immediate availabili- transport communications domains, the rate of ty, rapid deployment, and automated provisioning. traffic growth, and conditions imposed by exter- This growth of and need to pro- nal market and regulatory forces. Telecommuni- vide new sources of revenue to network/service cations increasingly relies on a diverse network providers have led to two major thrusts: of networks, with widely varying topologies, • Development of the infrastructure for the deployed services, and underlying business mod- optical transport network (OTN), a scalable els. At the same time, traffic is growing at an multiservice-capable and client-transparent unprecedented rate of 200 percent per year, transport infrastructure based on dense while the capacity of silicon is only growing by wavelength-division multiplex (DWDM)

80 0163-6804/01/$10.00 © 2001 IEEE IEEE Communications Magazine • September 2001 Carrier Carrier's Carrier Carrier Carrier's Carrier CustomerA carrier C A Customer A carrier C A Customer Or

Number 1: end-to-end monitoring Number 2: section monitoring

Figure 1. A global multicarrier connection.

technology optimized for ultra-high growth, • Inability to carry SONET/SDH clear chan- data-driven capacity demand (gigabit-level nel (transparency) bandwidth granularity required to scale and • Nonscalable maintenance scenarios manage multiterabit networks). • Intelligent optical networking (service opti- SLA Verification and Fault Sectionalization cal networking) — an optical control plane — Referring to Fig. 1, consider a global multi- enabling a network that is self-managed carrier connection between end customers with regard to topology/inventory discovery, involving an end-to-end global carrier network route discovery, and connection setup. Such (e.g., with subnetworks within the a network will support “on-demand” and the United Kingdom) interconnected by a switched transport connections and service carrier’s carrier network. management that can support the range of Existing SONET/SDH network monitoring operator business models. capabilities, called tandem connection monitoring This article focuses on architecting the ser- (TCM), support one level of monitoring. This vices optical network to meet the challenges of means that carrier A can monitor the end-to-end intelligent optical networking, and includes a connection (case #1), or each carrier can moni- snapshot of supporting standardization activities. tor their individual parts separately (case #2). However, it is not possible to perform both types of monitoring. Thus, in addition to limiting what OTN: THE FOUNDATION FOR THE can be monitored, SLA verification requires SERVICES OPTICAL NETWORK tight intercarrier cooperation. For example, if the decision is made to provide end-to-end mon- MOTIVATORS itoring (#1), additional processes are required As discussed in the introduction, incumbent car- for fault localization. If the decision is made to rier globalization and the entry of new carriers provide section monitoring (any cascading of are creating market forces with a major impact #2), an end-to-end carrier serving as the prime on backbone infrastructure requirements. These contractor of the service cannot determine over- factors also foster greater market segmentation all signal quality. This carrier would therefore and greater reliance on carrier cooperation. Ser- need to rely on customer complaints, or work to vice differentiation becomes an increasingly key ensure tight coupling across carrier boundaries. factor, which requires verification of service level agreements (SLAs). Finally, the sheer size of Transparent Carriage of SONET/SDH Ser- networks implies a greater need for more scal- vices — Referring to Fig. 1, suppose the cus- able and automated network maintenance. tomers are IP routers, and the service provided Hybrid solutions involving synchronous digi- by carrier A is data transport between them. As tal hierarchy/optical network (SDH/SONET) an example, assume carrier A makes use of net- integration with DWDM technology have been work facilities provided by carrier C, and desires increasingly deployed to tap into the full capaci- to provide service protection via a SONET ring, ty of fiber plant, thus maximizing the return on as illustrated in Fig. 2. existing facilities. This additionally allows service For the SONET ring capability to be support- providers to flexibly add new services onto the ed, nodes within carrier C’s network must not existing fiber to accommodate new capacity terminate the essential SONET overhead infor- demands. In these applications today, most of mation (the entire OC-N payload including over- the transport networking functionality is provid- head must be transparently carried). This is not ed by SONET/SDH systems that use DWDM possible if carrier C employs SONET multiplex- systems. As network traffic grows and DWDM ing, since carrier C would then terminate the deployment continues, utilization of this SONET overhead information needed by carrier approach for networked DWDM applications A. Thus, carrier A cannot build a ring across results in limitations in supporting the new mul- carrier C’s SONET network. ticarrier and multiservice networking require- If carrier C uses a passive all-optical solution, ments. Specifically, supporting networked DWDM carrier A has the needed SONET overhead, but applications using SONET/SDH layer functional- then there is no overhead for carrier C to main- ity runs into three barriers: tain its own network. • Limited support for SLA verification and fault sectionalization across multicarrier Maintenance Scalability — An interesting connections, even with operations systems example of maintenance scalability issues is pro- involvement vided in the context of support for optically

IEEE Communications Magazine • September 2001 81 Carrier A SONET Carrier C SONET overhead Ring

SONET SONET NE NE

Carrier A SONET Carrier A SONET frame payload

Carrier C OC-192 mux

Figure 2. Service protection via a SONET ring.

transparent subnetworks. Current SONET/SDH communication network. Localizing the fault to networks control faults by providing a specific the specific DWDM line system becomes very alarm indication signal (AIS) indicating that the tedious, and requires that the network manage- fault has been detected, and that downstream ele- ment system handle and correlate huge numbers ments need not raise an alarm. Since it was of LOS alarms. assumed that SONET/SDH would always serve as the lowest network layer, no provision was made OTN Architecture — Recognizing that DWDM for an AIS between SONET regenerators. A new offered the means to support terabit per second generic AIS signal has been defined in standards line capacities, with associated gigabit per sec- for use by DWDM systems to prevent down- ond path capacities, an aggressive work program stream SONET/SDH network elements from was initiated to define the OTN within the Inter- alarming based on a DWDM line system failure. national Union — Telecom- However, since a generic AIS can only be insert- munication Standardization Sector (ITU-T). The ed at 3R points, there is no means of preventing OTN represents a new transport networking alarm storms within all-optical subnetworks. This layer that is the next step beyond SDH/SONET is illustrated in Fig. 3, which shows the effects of a in supporting data-driven needs for bandwidth cable cut causing many downstream nodes in vari- and the emergence of new broadband services. It ous locations to generate loss of signal (LOS) provides a multiservice-capable core infra- alarms within an all-optical subnetwork. In a net- structure that leverages lessons learned from the work with several cables per duct, dozens of fibers SONET/SDH experience and adds optical tech- per cable, and hundreds of wavelengths per fiber, nology to meet the challenges of the evolving hundreds of thousands of LOS indications can networking environment. occur, which will flood onto the management This new layer is called the optical channel

Detect SDH Detect loss AIS cross-connect Detect loss of signal of signal Detect loss of signal AIS Detect signal SDH SDH loss generated NE NE of signal

SDH … … … … … cross-connect

… … … … SDH cross-connect … … Detect loss of signal … … … AIS signal Optical generated line system (NO 3R) Photonic cross-connect

Figure 3. The effects of a cable cut..

82 IEEE Communications Magazine • September 2001 Customer Customer Carrier A Carrier's carrier C Carrier A

#5 #4 #3 #2 #1

Figure 4. OTN connection monitoring in a multicarrier environment.

(wavelength), and it provides the gigabit-level bandwidth granularity required to scale and manage multiterabit networks and: • Maximizes nodal switching capacity, which is the gating factor for reconfigurable net- work capacity • Avoids very large numbers of fine-grained pipes that stress network planning, adminis- tration, survivability, management systems, Carrier C OTN ring and control protocols The logical structure of the OTN networking interface, the optical transport module (OTM) [2], is described further below. Carrier B Carrier C The optical channel (OCh) comprises an SONET frame OTN frame OCh data unit (ODU) and OCh transport unit OChOH FEC (OTU). The ODU is a networkwide transport entity that can transparently transport a wide Figure 5. Supporting an OTN ring. range of client signals including STM-16/64/256 signals, ATM streams, and IP/PPP and signals after encapsulation with Generic Framing SONET/SDH, these capabilities can be deployed Protocol (GFP). It currently has three rates of on a per-carrier basis (#4), end to end (#1), approximately 2.5 Gb/s, 10 Gb/s, and 40 Gb/s, over several carriers (#2), at the same time. respectively, and is referred to as ODUk (k = 1, The issues associated with supporting SONET 2, or 3). ring capability in the multi-operator environment The ODU adds the overhead to support described previously evaporate when carrier C managed services in multi-operator DWDM- supports an OTN infrastructure. As illustrated in based optical networks in the client-indepen- Fig. 5, when carrier C supports an OTN ring, the dent manner essential for operating such entire carrier A SONET frame is mapped into networks. This overhead enables monitoring to the ODU payload and transparently carried by support end customer, service provider, and carrier C, which uses ODU overhead in manag- network operator needs, providing for multiple ing its network. levels of nested and overlapping connection Maintenance scalability issues are also allevi- monitoring. This overhead can also be applied ated by the OTN OAM capabilities. Within the to support protected domain monitoring, test- all-optical subnetwork illustrated in Fig. 3, OTN ing, and optical link connection monitoring. To nonassociated overhead carried within the opti- condition the ODU for transport over a wave- cal supervisory channel prevents alarm propaga- length in a backbone network, it is transported tion. Thus, only the OLS adjacent to the fault within a frame that includes a forward error will report a LOS alarm; forward defect indica- correction (FEC) code. tion (FDI) maintenance signals propagated to 1 Note that while this arti- Overhead is also provided for single-channel downstream nodes prevent LOS alarm flooding. cle focuses on OTN infra- and multiwavelength optical signals to support structure and management of the “all-optical” parts of the automatically switched OTN network. This overhead is typically trans- SERVICES OPTICAL NETWORKING optical networks, the ported via a separate optical supervisory channel SERVICES OPTICAL architecture framework wavelength. described is applicable to NETWORKING OPPORTUNITIES any switched transport OTN OVERCOMES THE BARRIERS Services optical networking enables us to build network. In fact, the term Now we revisit the DWDM networking barriers on the optical transport infrastructure to provide automatically switched described in an earlier section, and see how they switched connections and service management.1 transport network are resolved by OTN capabilities. It enables an automatically switched optical net- (ASTN) was introduced As illustrated in Fig. 4, OTN connection moni- work (ASON) that is self-managed with regard to describe architecture toring capabilities are optimized for a multicarrier to topology/inventory discovery, route discovery, and control plane aspects environment, enabling fault localization and a and connection management. This has major that are generic to any cir- wide range of SLA verification capabilities. Unlike implications for carriers: cuit-switched network.

IEEE Communications Magazine • September 2001 83 • Self-management will enable the introduc- es, which in turn lead to explicit requirements Switched optical tion of new services and billing options for for particular interfaces and information flows carriers. over those interfaces. Following are four exam- networking • Greater automation of the service request ples of currently existing business models that capabilities can process will reduce the delay between the can be used to facilitate the architecture and time a service is requested, and when the interface discussions of a later section [4, 5]. provide an connection is set up and billing can com- Note that one or more of these models might be mence. used by various organizations of the same net- accurate, • The ability to support on-demand “Internet- work operator. real-time view of time” connections will mean that carriers The first business model to consider is that of can provide short-lived connections with an Internet service provider (ISP) that owns the the available usage-based billing for broadband services. network from the “ground up” (i.e., to the duct) bandwidth in the This enables provisioning of services such and only delivers IP-based services. Here, there as bandwidth retailing by the hour, minute, is no need to offer any other connection services network as well or even second (bandwidth on demand) as on the infrastructure, and since it is fully owned well as bandwidth trading/reselling. there are no trust issues between the infra- as bandwidth • By itself, automated route discovery can structure provider and IP layers. This case is utilization over eliminate a manual step or handoff that can actually very unusual, since the rapid installation cost $50–$100 a connection. of capacity makes it unattractive for an ISP to time. This data • Route discovery combined with topology actually own its entire infrastructure. can be used to discovery can prevent connection route fail- The second business model considered is an ures that occur frequently in current net- ISP that leases fiber or transport capacity from a fine tune network works. third party, and only delivers IP-based services. • Clients using a carrier’s network will be able As in the previous case, there is no need to offer growth plans. to choose a class of service level appropri- any other connection services on that infra- ate for their needs (and pay just for what structure. However, since the infrastructure is they need). leased, there is a trust issue between the infra- While the OTN will provide a transport infra- structure provider and the ISP. structure that supports self-healing networks, the Now consider the business that provides this switched optical network enables support for transport capacity to the ISP. In this third busi- various types of survivability mechanisms, includ- ness model, the business owns the layer 1 infra- ing fast mesh-based restoration of connections in structure and sells services to customers who addition to conventional options for unprotected may themselves resell to others. There is a defi- and 1+1 protected connections. The shared nite trust issue between this business and the mesh restoration option provides an opportunity client businesses. to reduce bandwidth reserved for protection and Since the transport business cannot make any provide more services on existing facilities. assumptions about the business of customers, it Automatic topology, inventory, and route dis- is highly likely that several customers may be covery provided by the switched optical network engaged in the same business. If the customers will enable networks to more fully utilize are engaged in the IP business (as previously), resources. Each facility will be discovered and may there will be several instances of IP networks be used for connections as soon as it is deployed. running over this network. However, customers Furthermore, since the network is the source of may also engage in other data services as well topology data, inventory cannot be lost due to (e.g., ATM); thus, a layer 1 infrastructure owner administrative or reporting error as happens with cannot make assumptions about what the client current inventory systems. Anecdotal reports indi- layer service may be without limiting their busi- cate errors in inventory databases relating to ness opportunities. It may also play the role of a 15–30 percent of facilities being misconnected or carrier’s carrier, where the customer network is missing. This results in a large percentage of con- likely to be a circuit network operated by anoth- nection routes failing to be usable. Switched opti- er network operator. cal network self-management allows the carrier to The fourth business model that needs to be avoid these problems and allocate these resources supported is that of a bandwidth broker. The to revenue-producing connections. subtle difference between this and the previous Finally, switched optical networking capabili- example is that the service provider may not own ties can provide an accurate, real-time view of any transport infrastructure(s) that support those the available bandwidth in the network as well as services, and the connection is actually carried bandwidth utilization over time. This data can be over third-party networks. Here there are defi- used to fine-tune network growth plans, which nite trust issues among the involved networks. are prone to be inaccurate in networks growing at 200 percent a year. THE OPTICAL CONTROL PLANE ARCHITECTURE FRAMEWORK SERVICES OPTICAL NETWORKING BUSINESS ISSUES The business models discussed above can now be used to derive the architecture requirements and It is essential that the switched optical network set of interfaces to be supported by the switched be architected in a manner that enables opera- optical control plane. tors to bill for services they provide [3]. Thus, all The most stringent architecture requirements reasonable business models must be supported. are imposed by trust (or security) of the third The business decisions associated with these and fourth business models in the preceding sec- models translate into network partitioning choic- tion. In order to realize the opportunities

84 IEEE Communications Magazine • September 2001 described earlier, the switched optical network tion in terms of capacity between the switches, must have a means of discovering connectivity as well as exchange signaling messages that will Trust issues and capacity, a means of disseminating this infor- control the switches during connection setup. mation between peers, and a means of setting up These actions occur at separate times, and can mandate that 2 a connection across the entire network. If there best be described as different service interfaces. detailed topology are no trust boundaries in the network, the net- The overall interface term used is a network-net- work user will be able to “see” the connectivity work nterface (NNI). The same caveat applies: information and capacity of each link in the entire network. this interface is considered purely logical, and to From an operator perspective, this would involve comprise several distinct services. Since routing cannot be a significant loss of business secrecy [5]. Indeed, decisions are made on the basis of connectivity exchanged across with this information the user could set up end- knowledge among all the switches within the to-end connections at will by commanding the domain, consider this an interior NNI (I-NNI). the interface individual network switches, and network opera- Trust issues mandate that detailed topology between two tors would find it very difficult to bill for the ser- information cannot be exchanged across the vice. To protect their business secrets, avoid loss interface between two operator domains. operator of control over the usage of their resources, and Instead, the information exchanged is the set of allow the creation of billable services, it is neces- users reachable across this particular link. Since domains. Instead, sary to restrict the amount of visibility and control user names do not have routing semantics, and the information granted to an end user. Similarly, it is also neces- we must have a bounded solution to an (in prin- sary to restrict the visibility and control granted to ciple) unbounded network, we take advantage of that is exchanged a peer operator. the routing properties of network addresses and is the set of users These trust issues lead to significantly differ- exchange reachability information in terms of ent interfaces: sets of aggregated reachable addresses. When that are reachable • Between user and provider the addresses are well formed, this reduces to • Between adjacent switches within a single exchanging the set of subnetworks that are across this provider domain reachable from a link, rather than the set of all particular link. • Between providers users that are reachable. The observant reader This is also the case for the Internet, as well will notice that this information is nothing more as the current public switched telephony network or less than a topology description of the con- (PSTN). In the PSTN network, topology infor- nectivity between the domains, rather than that mation is largely manually entered. In the Inter- between the individual switches. This interface net, topology information is kept within a routing has been called an exterior NNI (E-NNI), to dis- (Open Shortest Path First, OSPF) domain that tinguish it from the I-NNI discussed above. usually corresponds to a portion of a service The aforementioned switched optical control provider network. A different protocol (Border plane interfaces are illustrated in Fig. 6. Gateway Protocol, BGP) interconnects domains (or operators). OSPF does not distribute routing THE STANDARDS ROADMAP information to IP clients; the Address Resolu- tion Protocol (ARP) isolates clients from many OTN INFRASTRUCTURE routing decisions. The ITU-T initiated work on OTN standardiza- But how different are these interfaces? To tion by chartering a series of OTN Recommen- answer this question, consider several issues relat- dations in 1997. This work commenced with ed to naming and addressing, since these issues G.872, “The Architecture of Optical Transport will lead to an understanding of which objects Networks,” which was stabilized during 1998, need to be transferred across each interface. The and formally approved February 1999. Work terms name and address are frequently used inter- proceeded rapidly on G.709, “Interface for the changeably, but there are specific and subtle optical transport network (OTN)” (mentioned meanings implied by these terms. The fundamen- earlier), which was stabilized during 2000 and tal difference between a name and an address is approved at the February 2001 meeting of ITU- that names do not have semantics of any kind, T Study Group 15. Equally important is G.959.1, while addresses allow the object to be “located” “Optical Transport Network Physical Inter- in some space. Observe that a name has a longer faces,” which defines optical characteristics, and lifetime than an address, a property that allows was also approved in February 2001. Work is names to be portable over some useful space. progressing on Recommendations dealing with Not allowing users access to the addresses management and equipment. used by the switched optical network has value, since it allows the operators to organize their net- THE SWITCHED TRANSPORT CONTROL PLANE works as they see fit, without affecting their users. Whereas OTN infrastructure standardization is The interface between the user and the provider primarily taking place within the ITU-T, this is therefore needs to talk about user names, not not the case for control plane standards. Increas- network addresses. This interface is frequently ingly overlapping efforts among standardization called a user-network interface (UNI), a term that bodies and industry fora are occurring because of may cause confusion because it is often unclear the need for interdisciplinary expertise; for exam- whether it means the physical interface between ple, data networking protocol expertise of the user and provider equipment, or a logical inter- Internet Engineering Task Force (IETF) and face with properties as discussed above. transport networking expertise of the ITU-T. To 2 The entire network is In a distributed connection management sce- effect progress toward a coherent set of specifica- defined as comprising the nario, the interface between adjacent switches tions that support operator needs, with require- individual networks of within a single operator (or trust) domain allows ments and architecture driving protocol solutions, several operators working switches to exchange informa- a cooperative atmosphere is needed among the in concert.

IEEE Communications Magazine • September 2001 85 Management plane

Management MS system MS

Control plane

CC CC CC Connection Request controller I-NNI agent UNI CC CC CC RA RA CC E-NNI

OXC OXC

OXC OXC OXC OXC OXC Optical cross-connect OXC

Transport plane

Figure 6. Switched optical control plane interfaces.

various standardization bodies and industry fora. specifications (e.g., [7]) for application to the This is essential to avoid undermining the vision control of circuit-switched networks. of a global switched optical network by inadver- • Common control and measurement plane tently creating interoperability issues via conflict- (CCAMP) ing requirements and specifications. – Signaling protocols and measurement pro- The current set of players involved in various tocol specifications to support various tech- aspects of switched transport networking specifi- nologies for connection control and cations include the ITU-T, IETF, and Optical discovery Internetworking Forum (OIF). – Metrics, parameters, and properties that Within the ITU-T, work on the switched can be carried in protocols transport control plane as the ASTN umbrella of – Relationship between routing and signal- specifications (e.g., [6]) is focused within the fol- ing protocols lowing Study Groups: • IP over optical (IPO) • ITU-T Study Group 13, Question 10 – Switched optical control plane applica- – Generic transport control plane architec- tions and requirements tural and functional requirements for global Within the OIF, activities have been focused provider networks on establishing UNI signaling specifications based • ITU-T Study Group 15, Questions 12 and on modifications to Resource Reservation Proto- 14 col with Traffic Engineering (RSVP-TE) and Con- – Optical control plane architecture require- straint-Routed Label Distribution Protocol ments (CR-LDP) protocols, for implementation as part – Distributed connection management spec- of multivendor interoperability demonstrations. ifications – Neighbor and service discovery require- ONCLUSIONS ments, processes, and attributes C Within the IETF, two new Working Groups The growth in demand for bandwidth and expecta- have been created to focus on the switched tion of service fulfillment in “Internet time” require transport control plane, which has involved a different core network design for scalability and restructuring of related work efforts among on-demand services. Within this article, we present other Working Groups and generalizing the mul- an overview of a new broadband networking tiprotocol label switching (MPLS) protocols as paradigm, services optical networking. The promise the generalized MPLS (GMPLS) umbrella of this paradigm offers is motivating industry momen-

86 IEEE Communications Magazine • September 2001 tum toward establishing supporting specifications, naling protocols in ASTNs, and the simulation of network reflected in rapidly accelerating OTN and optical performance under various fault restoration scenarios in MPLS networks. His previous research experience includes Realizing the control plane external standardization activities. the development of protocols for synchronization distribu- Realizing the services optical networking promise tion in SONET/SDH networks, systems modeling, robust Service Optical and adaptive nonlinear control design, and the analysis of requires unilateral commitment to a vision of mul- Networking tivendor and multi-operator interoperable net- transport network architectures using formal specification techniques. He is actively contributing to the development working that enables end-to-end switched services of several key international standards in the area of ASTN, promise requires in a global environment. And a great deal of hard also serving as editor for the draft ITU-T Recommendations work! on auto-discovery, G.disc. In addition, he is a prolific con- unilateral tributor to ANSI T1, IETF, and OIF. commitment ACKNOWLEDGMENTS GEORGE NEWSOME ([email protected]) is a distin- We wish to thank our colleagues Carmine Daloia guished member of technical staff with Lucent Technolo- to a vision of gies, part of Lucent’s Optical Networking Group. He and Maarten Vissers for their technical review multi-vendor and and feedback, which has improved the quality of received his B.Sc. degree in electrical engineering from Uni- versity College, , United Kingdom. He has over 30 this article. years of multidisciplinary expertise encompassing software multi-operator and system architecture analysis and definition, systems REFERENCES engineering and development of transport equipment and interoperable network software, CMOS customer ICs, application of [1] A. Kalro, H. La Roche, and C. Roxlo, “Optical Networks microprocessor technology, technologies underlying design networking that for Data Applications: Market Drivers, Technology and composition of multi-usable software components, Trends, Network Architecture and Applications,” ITS- and methods of domain analysis. His primary research enables end-to- 2000, Sept. 2000. activities currently encompass optical networking architec- [2] ITU-T Rec. G.709, “Network Interface for the Opti- tures and automatically switched transport networking, end switched cal Transport Network (OTN),” Feb. 2000. including global standards development. Previous research [3] OIF Carrier Group, “Carrier Optical Services Framework activities include the establishment of formal specification services in a and Associated Requirements for UNI,” T1X1.5 liaison techniques for transport, information modeling for net- doc., Oct. 2000, work in progress. work management based on open distributed processing, global environ- [4] C. Brownmiller et al., “Requirements on the ASON UNI,” and systems and software engineering. He was a key soft- T1X1.5/2000-194, Oct. 2000. ware architect in the design and prototyping of a distribut- ment. And a [5] “Considerations on the Development of an Optical Con- ed optical network management system as part of the trol Plane,” Internet draft, draft-freeland-octrl-cons- MONET Consortium. He is currently an active contributor great deal of 01.txt, Nov. 2000. to global standards on the ASTN and optical transport [6] ITU-T Rec. G.807, “Requirements for Automatic infrastructure. He has authored over 50 technical standards hard work! Switched Transport Networks (ASTN),” May 2001. contributions during his standards tenure. [7] “Generalized MPLS — Signaling Functional Description,” draft-ietf-mpls-generalized-signaling-04.txt, work in ZHI-WEI LIN ([email protected]) is a member of Lucent progress. Technologies technical staff, part of Lucent’s Optical Net- working Group. He holds an M.S.E.E. degree in electrical BIOGRAPHIES engineering from Cornell University, Ithaca, New York. He has multidisciplinary expertise on network survivability EVE L. VARMA ([email protected]) is a technical manager strategies, and network planning, architecture, and design. with Lucent Technologies’ Optical Networking Group. She This encompasses a wide range of areas from ATM OAM received her M.A. degree in physics at the City University work, to SDH MS-SPRING protection protocol design, to of New York. She has over 20 years of research experience the G-MPLS/ASTN control plane specifications. From 1995 within the telecommunications industry. Her primary he worked at Telcordia as a senior consultant, where he research areas are currently focused on the automatically was responsible for multitechnology network design and switched transport network (ASTN/GMPLS) and the optical deployment, evolution planning, business plan develop- transport network infrastructure (OTN). She is actively ment, and associated analysis and modeling. In addition to engaged in supporting the development of the associated leading teams in these areas, he also played a lead role in specifications within global standards and industry fora representing Telcordia and the regional Bell operating com- spanning ITU-T, IETF, ANSI T1, and OIF. Between 1995 and panies within national and global standards. Since joining 1997, she led the team responsible for designing and pro- Lucent July 2000, he has expanded his standards activities totyping a distributed optical network management system to IETF and OIF as well. During the past 5 years, he has as part of the Multiwavelength Optical Networking provided over 50 technical contributions to these standards (MONET) Consortium (partially supported by DARPA). Her and industry fora, many of which have been used as input previous research experience includes characterization, toward approved international Recommendations. He is analysis, and development of transmission jitter require- currently leading efforts in both ITU-T and IETF in specify- ments; systems engineering and standards for SDH/SONET ing the G-MPLS/ASTN signaling protocol, serving as editor transport and network management applications; as well of draft ITU-T Rec. G.dcm (Distributed Connection Manage- as associated enabling technology and methodology devel- ment). opment/assessment. HARVEY EPSTEIN ([email protected]) is a technical staff SIVAKUMAR SANKARANARAYANAN ([email protected]) is member with Lucent Technologies, part of Lucent’s Optical currently a technical staff member with Lucent Technolo- Networking Group in North Andover, Massachusetts. He gies, part of Lucent’s Optical Networking Group. He holds holds a B.A. degree from Brooklyn College, New York, and an M.S. degree in electrical engineering from Polytechnic M.A. and Ph.D. degrees in mathematics from the University University, Brooklyn, New York. His primary research areas of Wisconsin. He was responsible for operations architecture are focused on the development of the architecture, and planning for optical networking, including optical UNI inter- associated neighbor/service discovery and distributed con- face definition and contributed to development of ODSI Sig- nection management protocols, for ASTNs. His active naling Control Specification, which was the basis of the research areas also include the network simulation of sig- ODSI interworking demonstration in December 2000.

IEEE Communications Magazine • September 2001 87