WDM and TDM technologies combine to upgrade fiber networks
Flexibility and cost-effective bandwidth growth can be gained in fiber-optic networks by the use of complementary wavelength-division-multiplexing and time-division-multiplexing technologies
Serge Melle and francis gagnon
The exponentially increasing demand for communications bandwidth as new services are adopted has been driving the construction of new fiber-optic networks and the development of high-capacity transport systems, such as those using dense-wavelength-division-multiplexing (dwdm) or time-division-multiplexing (TDM) technology, to operate over existing, often fiber-constrained, cable routes.
By multiplexing up to eight wavelengths in the 1.55-micron region, dwdm systems deployed to date have helped increase the capacity of existing fiber networks. In high-capacity TDM systems, such as OC-192 systems operating at 10 Gbits/sec, capacity is qua drupled compared to existing OC-48 Synchronous Optical Network (Sonet) systems operating at 2.5 Gbits/sec by using higher-speed lasers and electronics. Like dwdm systems, OC-192 systems have been deployed to increase capacity over existing fiber-optic networks. And some OC-192 equipment has demonstrated the ability to operate over the broad range of fibers already deployed without performance limitations from fiber effects such as polarization-mode dispersion (PMD), even over early-vintage fibers deployed in the late 1980s (see "Demonstrating OC-192 operation over existing fiber networks," this page).
Debate continues in the telecommunications industry over the relative merits of using WDM or TDM technology as the best way to increase system capacity. In the process, little attention has been paid to providing the optimal solution to the carrier customer, who more often than not, combines WDM and TDM as complementary technologies for provisioning bandwidth demand.
The optimal high-capacity transport solution is, therefore, not exclusively tied to WDM-only or TDM-only systems. Rather, the specific solution set of OC-48, OC-192, and WDM systems varies based on several factors, including network capacity, growth, service mix, traffic patterns, fiber availability, and network geography. By considering total network cost and service migration flexibility, carriers should discern that both WDM and TDM technologies offer inherent capabilities, which, when combined, provide an optimal solution for upgrading today`s networks for the future.
When used as complementary technologies and deployed appropriately, WDM and TDM technologies not only offer carriers revenue growth potential over a scalable network, but also provide the operational efficiency and cost savings that make this growth economically viable. And as network interfaces evolve and new services are carried over the network, maximum benefit may be obtained from new capabilities offered on OC-192 systems.
In evaluating high-capacity transport systems, carriers search for fiber- network solutions that impact three key sectors of business operations: revenue growth, operating cost reduction, and service differentiation.
The key factors that drive revenue growth for carriers include maximum capacity for new service velocity and connectivity for new customer services. To achieve operating cost reduction, carriers must consider bandwidth management to minimize network complexity, a common platform for WDM and TDM solutions, and capacity scalability to match service demands. In differentiating services, carriers need to implement Sonet ring survivability, enhanced grade of service, and network-wide performance monitoring.
Deployment of high-capacity networks to meet bandwidth demand remains a priority for increasing revenue, but carriers should consider more than just raw capacity when deploying full- spectrum WDM and TDM solutions. Another important factor is the need to complement capacity growth with corresponding bandwidth management to prevent runaway costs and to minimize network complexity.
Bandwidth management provides the ability to crossconnect Sonet payloads at network nodes to route and groom traffic to make the most of transmission capacity. The need for bandwidth management increases as network capacity grows and as traffic patterns change with network evolution, requiring continuing adjustments to optimize use of bandwidth.
Requirements for bandwidth management can be demonstrated using a typical Sonet ring network, in which OC-48 add/drop multiplexers (ADMs) are located at network nodes. As the network grows, capacity can be increased by overlaying additional Sonet OC-48 rings over the first ring, thus creating a network of stacked rings (see Fig. 1). In many cases, existing long-haul networks are fiber- constrained, and new OC-48 systems are combined at different wavelengths over a common fiber using WDM technology. By deploying WDM systems operating over four or eight wavelengths, carriers have been able to grow network capacity to 20 Gbits/sec over a common set of fibers, overlaying as many as eight OC-48 rings.
Problems arise, however, as more and more OC-48 rings are laid over WDM systems. In some networks, customer traffic growth over time must be accommodated on different rings. In other networks, customer churn, or turnover, frees up bandwidth in some rings that might not be efficiently re-used due to new or different traffic patterns. In still other networks, some traffic does not need to be carried over the network but can be rerouted to other tributaries at a network hub. Lastly, because ADMs are not necessarily deployed at the same nodes on every overlaid ring, some part of ADM traffic capacity must be allocated to support inter-ring hand-offs and used for back-hauling traffic to other nodes, rather than for transporting revenue-generating services.
Consequently, as the number of stacked OC-48 rings increases, bandwidth management must be implemented using stand-alone elements such as digital crossconnect systems and additional Sonet multiplexers. Such network elements are deployed at network nodes to manage customer churn and provide the necessary bandwidth management and tributary grooming of the ring traffic.
As a result, the net advantages of incrementally growing network capacity one OC-48 ring at a time are outweighed by the additional equipment costs required for bandwidth and tributary management when overlaying many OC-48 rings. Although total network capacity can be increased to 40 Gbits/sec or more with dwdm systems of 16 or more wavelengths, the inherent network planning complications and operational inefficiencies of such an architecture with many stacked rings can be cost-prohibitive when considering total network cost.
Two other factors significantly increase total cost in an OC-48 WDM-only architecture. First, as hand-offs from one interexchange carrier to another (and from interexchange carrier to local-exchange carrier) make the transition from DS-3 (44.736-Mbit/sec) speeds to OC-12 (622-Mbit/sec) speeds, additional OC-48 multiplexers must be deployed for the sole purpose of grooming multiple OC-12 lines onto an OC-48 multiplexer for cost-effective transport over the WDM system. Second, the introduction of ATM services with optical connections at OC-3c (149.76-Mbit/sec) and OC-12c (599.04-Mbit/sec) speeds require additional OC-48 ADMs solely to groom such services over a WDM system.
Many of these operational issues can be effectively managed through the complementary deployment of WDM systems operating at OC-48 and OC-192 speeds. Deployment of a common WDM platform that can support both OC-48 and OC-192 channels permits cost-effective network capacity growth to 80 Gbits/sec without sacrificing architectural flexibility. Compared to an OC-48 WDM-only architecture, a mixed OC-48 and OC-192 WDM architecture requires fewer rings (25% fewer in some cases), thus increasing the efficiency of ring bandwidth utilization and simplifying churn management (see Fig. 2, page 30). By using OC-48 systems and optical add/drop modules for "local" traffic and OC-192 systems for "express" channels, carriers can further optimize the architectural efficiency of their network.
More important, some OC-192 ADMs incorporate the ability to provide STS-1 (52-Mbit/sec) level bandwidth management across the entire ring and tributary bandwidth. This capability combines the role of a Sonet multiplexer and the functionality of a built-in digital crossconnect, minimizing the need for stand-alone digital crossconnect system capacity at network nodes. The difficulties associated with managing and interconnecting traffic across many overlaid OC-48 rings are minimized by directly interconnecting OC-192 terminals to each other and dynamically optimizing traffic flows. In addition, OC-192 bandwidth management can be used to groom traffic from OC-48 systems used on local channels to and from express traffic carried over OC-192 systems.
The selection of a high-capacity platform should also enable the migration of network interfaces from legacy DS-3 services to Sonet hand-offs between carriers, and the introduction of new services such as Asynchronous Transfer Mode (ATM); see Fig. 3. If OC-192 multiplexers with direct OC-3 (155-Mbit/sec) and OC-12 tributary interfaces are used, migration to Sonet hand-offs between carriers and direct optical connection of new ATM services are supported without the need for sub-tending OC-48 multiplexers for tributary grooming. Deploying a common platform that supports OC-48, OC-192, and WDM technologies, therefore, provides service evolution and capacity scalability. Choices can be made in the context of network-wide planning to minimize total cost, not just today, but over time.
For example, consider two long-haul, fiber-constrained, ring networks identical in terms of initial capacity, growth rate, traffic patterns, and geography (see Fig. 4). In both networks, assume that capacity doubles every year from an initial capacity of OC-48, and a high-capacity platform is deployed that can support WDM at an OC-48 or an OC-192 rate.
In the first network, service mix is constant over time and consists of 100% DS-3 traffic. The lowest-cost migration strategy favors optically multiplexing up to eight OC-48 channels using WDM to increase capacity to 20 Gbits/sec. Beyond that point, the need for further growth in the number of stacked OC-48 rings favors deployment of OC-192 systems to provide bandwidth management and to minimize the need for additional digital crossconnects at network nodes. Network growth to 80 Gbits/sec is supported via an eight-wavelength, OC-192 WDM network, while retaining a small number of OC-48 channels for local traffic service.
In the second network, assume that ATM services are introduced, and the carrier elects to transition all network hand-offs to OC-12 speeds to increase survivability and minimize use of central-office floor space. When capacity approaches 10 Gbits/sec in this network, the lowest-cost solution for accommodating growth is to deploy an OC-192 system, which provides direct OC-3 and OC-12 tributary interfaces. Continuing network growth to 80 Gbits/sec is then supported by multiplexing multiple OC-192 channels over a dwdm system.
Such nuances in the optimal selection of WDM and TDM solutions extend beyond the second network example. Simple point-to-point networks with primarily DS-3 interfaces at terminals favor OC-48 WDM technology because of its limited need for bandwidth or tributary management. Metropolitan networks favor the deployment of OC-192 technology due to the meshed nature of traffic patterns and short distances between nodes. And long-haul ring networks favor a varying mix of OC-48 and OC-192 WDM technologies to accommodate local and express traffic growth to 80 Gbits/sec and beyond, and provide cost-effective bandwidth management as capacity exceeds 10 Gbits/sec. u
Serge Melle is marketing manager at Northern Telecom in Alpharetta, GA, and Francis Gagnon is marketing manager at Northern Telecom, St. Laurent, QC, Canada.
Demonstrating OC-192 Operation over Existing Fiber Networks
With little data to confirm the polarization-mode dispersion (PMD) tolerance of OC-192 systems, Stentor, an association of Canada`s telephone companies; Bell Canada, Canada`s largest telecommunications provider; and Northern Telecom conducted an OC-192 field trial of non-dispersion-shifted fiber (n-dsf) plant. Field testing was performed on old-vintage n-dsf cables in a realistic "long-range" deployment scenario over a variety of buried, conduit, and aerial cables. Optical amplifiers and dispersion-compensation modules were used to extend system reach. Tolerance to high levels of PMD (higher than 40 psec) was further demonstrated using a PMD simulator, transmitting over the theoretical equivalent of 6400 km. In all cases, results indicate that high levels of PMD beyond those typical of installed fiber will not constrain the development of Northern Telecom`s OC-192 system.