T o w a r d a n a l l - o p t i c a l n e t w o r k

By: John McFarlane, Philippe Morin, Peter Roorda, Mike Scott

Originally published in Telesis August 1996.

While no one can say for sure what new services will travel the information superhighway of tomorrow, new users and new applications – as evidenced by the accelerating demand for Internet and local area network (LAN) connectivity – will bring many challenges. Network capacities will be stretched to the limit, network flexibility will be tested to deliver a range of applications with different performance and bandwidth requirements, and network costs will have to be contained to deliver services that are profitable to service providers yet affordable for users. The solution may lie with all-optical networks, which carry information on beams of light from one end of the network to the other. Unlike today’s fiber pipes, which typically carry single wavelengths point-to-point across the network through intermediate stages of optoelectronic conversion, all-optical networks will multiplex, amplify, and route multiple wavelengths entirely in the optical domain without the need for conversion. (see figure 1.0)

These all-optical networks offer the potential to deliver the huge capacities, networking flexibility, and low network costs that will be required to support volume deployment of high-speed data services for Internet access and LAN interconnection, as well as emerging high-bandwidth video distribution applications and broadband multimedia services.

Before all the elements of an all-optical network can be deployed, however, a number of technical, performance, and network management challenges must be resolved. To address these challenges, Northern Telecom (Nortel) is leveraging its expertise in such core competencies as optoelectronic components, software design, systems integration, and network planning to develop the architectures and products for all-optical networks. This leadership extends across the key elements of an all-optical network, including optical amplifiers, wavelength division multiplexing (WDM), optical cross-connects, and network management and control systems.

Although migration to all-optical networks will be gradual, some of its elements are already being deployed. Optical amplifiers, for example, are being used to boost the power of optical signals, enabling them to travel farther without regeneration. Increasing the distance, or span, between regenerators allows network operators to extend the reach of their high-speed Synchronous Optical Network/Synchronous Digital Hierarchy (SONET/SDH) networks and significantly reduce their regenerator equipment and site costs. (see figure 2.0)

Nortel, in fact, has shipped more than 6200 of its optical amplifiers to customers since the product line was introduced in February 1995. Nortel is now one of the world’s leading suppliers of optical amplifiers, and is the first to successfully use its optical amplifiers with a 10-gigabit-per-second (Gbit/s) transmission system in a commercial network carrying live traffic.

More recently, network operators have started volume deployment of WDM technology to increase the capacity of their existing fiber routes, thereby avoiding the cost and delay of adding new fiber-optic cable to upgrade capacity. Although WDM has been used in niche applications for years, the introduction of such technologies as bidirectional, multiwavelength optical amplifiers is now making WDM deployment economical.

WDM enables multiple wavelengths to be carried on the same fiber –dozens of different wavelengths in the near future – potentially breaking the terabit barrier by pushing fiber capacities to 1,000 Gbit/s and beyond. In WDM systems, lasers produce light of different wavelengths, which are then multiplexed onto a fiber to carry information across the network. Each wavelength occupies a different position in the optical spectrum, expressed in nanometers (nm), such as 1533 nm or 1557 nm.

While making multiwavelength WDM technology available on its industry-leading 2.5-Gbit/s and 10-Gbit/s SONET/SDH TransportNode products, Nortel continues to push the capacity envelope of its single-wavelength systems. Nortel, in fact, is the world’s leading supplier of high-capacity 2.5-Gbit/s systems, with more than 28,000 systems shipped in North America alone, providing an aggregate capacity of 24 million equivalent DS-1s, or 37 million megabits per second. To push capacities to the next level, Nortel is now exploring product solutions capable of operating at 400 Gbit/s and beyond.

For many applications, single-wavelength systems (such as Nortel’s 10-Gbit/s system) are more cost-effective and easier to manage than multiple-wavelength systems of an equivalent bit rate (such as four-wavelength 2.5-Gbit/s systems, which also provide 10 Gbit/s). For example, to achieve the same level of bandwidth utilization as single-wavelength systems, multiwavelength systems require more extensive use of large cross-connects at the end-points in the network, which increases costs.

Taking single-wavelength systems to the next capacity level, however, normally requires technology breakthroughs. WDM technology can be effectively applied to enable network providers to flexibly and economically add capacity to their existing highest-bit-rate single-wavelength systems. Furthermore, this capacity can be added incrementally – to provide 5 Gbit/s or 7.5 Gbit/s, for example – by increasing the number of wavelengths on a fiber in step with growing traffic demands. To advance WDM technology, Nortel has already deployed 80-Gbit/s WDM transmission that are scalable to 160 Gb/s in-service (sixteen 10-Gbit/s channels).

The other main element of an all-optical network – optical cross-connects, which switch and route traffic from one fiber link to another in the optical domain – is still in the experimental stage. Working with a customer, Nortel in 1995 built one of the world’s first prototype systems, and is extending that work with a testbed network containing multiple cross-connects to investigate such functions as network management and mesh restoration.

Optical cross-connects, together with interconnection at optical line rates and bandwidth management at line-terminating equipment, could offer network providers significant cost benefits by eliminating the need for costly optoelectronic conversion and by reducing the number of low-speed ports. By contrast, optical traffic in today’s long-haul or metropolitan-area networks is typically terminated and demultiplexed at intermediate nodes for electronic cross-connection at line rates of 50 Mbit/s or lower, then remultiplexed and retransmitted onto the fiber at rates of up to 2.5 Gbit/s.

Line rates in all-optical networks can also be upgraded at a low cost because, to some extent, optical cross-connects and amplifiers operate independently of the optical line rate; that is, any wavelength can carry signals of any bandwidth. If their optical links have been designed at the outset for the upgrading of optical line rates, network providers can replace existing end equipment in the network with higher-line-rate equipment (moving from 2.5 Gbit/s to 10 Gbit/s, for example) as traffic demands increase – without having to upgrade all the intermediate equipment in the network.

All-optical networks may also be able to carry and optically switch signals independently of the format (whether SONET/SDH, ATM, IP, digital, analog, or video) – a capability service providers could exploit to flexibly offer new revenue-generating services, such as the transport of cable television signals. All-optical networks could support new services by carrying – on separate wavelengths – non-standard formats end-to-end over the same fiber network that carries SONET/SDH signals. By contrast, in today’s networks, all traffic must first be converted into SONET/SDH formats and then transported, which increases costs. On the other hand, SONET/SDH is a proven, reliable, and widely deployed technology that offers integrated operations, administration, maintenance, and provisioning (OAM&P) capabilities.

Nortel foresees optical networks interworking with SONET/SDH, IP and ATM networks in a multilayered network architecture. SONET/SDH, bandwidth management, IP routing and ATM switching will continue to be performed in the electronic domain, while the optical layer will provide gigabit service carriage and wavelength networking without the need for costly intermediate optoelectronic conversion.

To address the diverse challenges associated with the deployment of WDM optical networks, Nortel is taking a leading role in evaluating the potential of a complete range of optical technologies and applications, both in the lab and in demonstration systems with leading customers and partners.

Optical WDM testbed

A key project is the building of an all-optical WDM demonstration network with nine other members of the National Transparent Optical Network Consortium (NTONC). Other members include Pacific Bell, Sprint, the Lawrence Livermore National Laboratory, Hughes Research Labs, Rockwell Science Center, the University of California at San Diego, Columbia University, Uniphase Telecommunications Products (UTP), and Case Western Reserve University (CWRU).

Located in the San Francisco Bay area on the west coast of the United States, the NTONC network is now operational and has been used to test a variety of optoelectronic and optical components, network management and control strategies, and customer broadband multimedia applications. Testing has been and will continue to be carried out with user-application traffic using commercial, installed fiber-optic cable from Sprint and Pacific Bell, under operating conditions that are as near as possible to standard.

Running for the past three years, the $17-million program is jointly funded by the Defense Advanced Research Projects Agency (DARPA) of the U.S. Department of Defense, and by NTONC members. The optical network will carry SONET/SDH- and ATM-based traffic from supercomputers, and from systems supporting such broadband applications as distance learning, medical imaging, and multimedia.

As manager of the consortium and a major contributor to the NTONC project, Nortel is involved in a number of key aspects, including:

• network studies, analysis, and project management aimed at developing strategies for commercial deployment of all-optical WDM networks;

• development and fabrication of optical amplifiers with gain-flattening filters;

• integration of a variety of optoelectronic devices into a compact, cost-effective

-channel network access module to support very-high-bandwidth applications, continuing development begun under a previous phase of this DARPA-sponsored program; and

• development of prototype optical network management capabilities and integration with SONET/SDH and ATM network management systems.

Initially, the demonstration network is deployed in a four-wavelength, four-node hubbed configuration linking the Lawrence Livermore National Laboratory, the University of California at Berkeley, the Pacific Bell Broadband Lab in San Ramon, and the Sprint Advanced Technology Lab in Burlingame. The network incorporates two Nortel S/DMS TransportNode OC-48 systems (operating at 2.5 Gbit/s) at each of the four sites, and up to 30 Nortel optical line/post amplifiers and gain-flattening filters deployed throughout the testbed. Optical switching will be performed at each node, using optomechanical and acousto-optical tunable filter (AOTF) cross-connects developed by UTP and CWRU.

The network will also include an eight-wavelength network access link between Livermore and San Ramon. The network access module packaged by Nortel will give users cost-effective access to the network at OC-3 and OC-12 (equivalent to STM-1 and STM-4, respectively) data rates. The network access module performs wavelength translation, converting between electrical and optical signals, and provides flexible add-drop and wavelength- switching capabilities at network nodes. These capabilities are used to route locally dropped signals to the appropriate peripherals, and assign added signals to the correct wavelengths.

Nortel, in fact, has already deployed the network access module in its own Corporate Broadband Applications Network (COBRAnet), and used it to carry a mix of high-speed data, voice-over-ATM, and digital video traffic over the same network without the need for any special grooming, synchronization, or formatting.

The four network sites in the NTONC network are linked in a bidirectional optical ring configuration, which enables the examination of optical reconfiguration issues associated with various WDM ring and mesh configurations. In addition, two Nortel Magellan Concorde ATM switches, each with a capacity of 10 Gbit/s, are used to aggregate testbed traffic using permanent virtual circuits and, eventually, switched virtual circuits. The Concorde switches will support ATM-based user applications, and provide a testbed for examining prototype optical/SONET/ATM integrated network management capabilities.

Optical cross-connect testbed

In a second major project, Nortel is working with a lead customer in the U.S. to develop an advanced testbed for evaluating optical cross-connect technologies and applications. Optical cross-connects promise a cost-effective alternative to electronic digital cross-connect systems, and could eventually be used in core networks and central offices to perform critical networking functions, such as long-haul switching and routing, tributary management, and line protection.

Development of the test network follows Nortel’s successful demonstration in December 1995 of one of the world’s first optical cross-connect systems. A key aspect of the project was the definition and demonstration of three leading-edge applications: tributary management, line switching restoration, and line wavelength restoration.

Tributary management is traditionally performed by electronic digital cross-connects at central offices to groom traffic from lower-level tributaries (such as DS-1s) onto higher-level lines (such as DS-3s). Using the optical cross-connect module, Nortel successfully demonstrated tributary management at much higher levels (OC-12/STM-4 and OC-48/STM-16) and at faster speeds (within a few milliseconds). Tributary management at optical levels allows network providers to cut costs significantly by reducing the number of low-level connections they must make. This cost benefit can be achieved when combined with optical interconnection and distributed bandwidth management at the line-terminating equipment.

Line switching restoration is used to restore service after a failure or cable cut by shifting traffic to a different line. During the trial, cutovers were done rapidly – showing that any service disruption under actual operating conditions would be determined by the response of the transmission equipment rather than the switching speed of the optical cross-connect. Future all-optical protection switching and network restoration is expected to reduce line-protection costs by using a single optical switching layer between central offices and external lines. In contrast, conventional SONET/SDH linear and ring line-protection schemes require redundancy in both terminal and fiber equipment (including repeaters) to withstand the effects of equipment failure without impacting signal continuity. SONET/SDH, however, provides fast restoration times (50 milliseconds) and a simpler OAM&P architecture.

Line wavelength restoration will enable providers to restore service on WDM networks by using optical cross-connects to switch multiple wavelengths from one line to another to fill available wavelength slots. In the demonstration, Nortel added WDM multiplexers/demultiplexers to the basic optical cross-connect configuration, and successfully routed affected wavelengths around a cable break by optically switching them to a different line.

Building on this work, the testbed contains multiple cross-connects to allow the simulation of such operations as network management and mesh restoration. The testbed will also incorporate an “all-optical” monitoring system to detect cable cuts within the network and faults within the cross-connect core, as well as provide a trigger for restoration. By extending optical routing and switching to multiple cross-connects, Nortel is laying a foundation for network providers to significantly reduce the cost of building and operating future optical mesh infrastructures. Cost reductions will result from minimizing the need for high-speed optoelectronic and low-speed electrical equipment. Moreover, restoration of a meshed network could be accomplished much more quickly with optical cross-connects than with conventional electronic cross-connects, because it is carried out optically at higher bit-rate levels (typically 2.5 Gbit/s or above).

Network management

A major thrust of Nortel’s involvement in both the NTONC and optical cross-connect (OCC) demonstration networks is the development of network management techniques and prototype systems for reconfigurable all-optical networks – a key to the successful introduction of optical technologies in future networks. These programs provide the opportunity to work with customers in real-life applications to determine their management requirements, and to address a number of challenges inherent in deploying new optical components and technology. These challenges are not insignificant. Optical networks add a new set of operational issues to those related to SONET/SDH and ATM networks. While the transparency of all-optical networks is desirable from an applications point of view – enabling networks to seamlessly transport and route any combination of SONET/SDH, ATM, and other traffic including digital, analog, and video – it introduces management complexity. Because the optical signal is transparent, the format of the traffic carried on a given wavelength may not be known. Consequently, it is not possible to apply traditional performance measurements, such as bit-error rates used in SONET/SDH systems, to transparent optical signals.

It is therefore necessary to develop new techniques for measuring performance degradation in optical networks – for example, to detect when the wavelength of a signal drifts from its correct value and adjacent wavelengths encroach on each other, causing crosstalk.

All-optical networks also add an additional layer to the network management mix. To avoid introducing further management complexity, techniques must be developed to exchange information and coordinate actions among the optical management layer and existing SONET/SDH and ATM layers to enable end-to-end service delivery.

Compounding these challenges is the current absence of standards for managing optical networks. While a wealth of network management standards exists for SONET/SDH and ATM technologies, development of standards for the management of WDM networks and optical switches is just beginning to be addressed.

As part of ongoing research efforts in the lab and with the NTONC and OCC testbed networks, Nortel is conducting pioneering work to solve the challenges of managing optical networks. The overall goals are to establish network management directions and identify management requirements.

To support the NTONC and OCC testbeds, Nortel is defining network management architectures and developing prototype management systems to perform such functions as fault and performance monitoring, connection management, and provisioning. The prototype systems will include: management software for network reconfiguration, routing, and monitoring; interfaces to controllers for managing optical network switches; and integration of the WDM optical management layer with SONET and ATM management layers.

A key aspect of the research is development of an advanced, easy-to-use, and information-rich graphical user interface (GUI) for managing WDM optical networks. Being developed in close conjunction with Nortel’s Corporate Design Group, this new GUI will allow network management personnel to monitor wavelength paths, faults, and other network control information, and to take the appropriate recovery actions based on system recommendations.

A primary goal of Nortel’s GUI program is to provide an integrated view of the optical, SONET/SDH, and ATM network management layers. For example, to establish a connection between two ATM nodes, a physical SONET/SDH path must be set up between the two nodes. This path, in turn, must have a wavelength available to carry the signal in the optical domain. Establishing the ATM connection, therefore, will require knowing what resources are required and available in all the layers.

Deploying prototype network management systems in the NTONC and OCC networks will enable Nortel to experiment with various optical management techniques. The testbeds, for example, will be used to measure the performance of optical components in a real-world environment, establish the correlation between optical network performance and existing SONET/SDH performance degradation measures, and determine thresholds beyond which the performance of optical components is no longer acceptable.

Based on the testbed results and associated research, Nortel will use the information to help drive optical network management standards with national and international standards organizations in such areas as network performance, network management design, and GUI technology. Nortel’s aim is to make the standards compatible with Telecommunications Management Network (TMN) directions. Defined internationally by the International Telecommunication Union (ITU) standards forum and in North America by the American National Standards Institute (ANSI) T1 forum, TMN is an architecture for interconnecting network providers’ operations systems and network elements through standardized interfaces.

With its leadership role in the NTONC and OCC demonstration networks and its R&D efforts in the advanced optical technologies and applications, Nortel is one of the few suppliers capable of bringing together all the components that will be needed to successfully build an all-optical network. Encompassing high-capacity WDM carriage, optical amplification, optical cross-connection, and network management capabilities, Nortel’s technology programs could spur the building of all-optical highways of light that offer previously unattainable cost, revenue, and service benefits to network providers.