Rajive Dhar and Mark Lowry
Dynamic architectures and network systems can exploit the benefits offered by tunable WDM technology to reduce the overall cost of operations, simplify network scalability, and optimize performance for service providers.
A new optical network is required for the metropolitan market—one that can deliver services to end users at a much lower cost. Services may take the form of commodity bandwidth/ access services or value-added application services. Value-added services, such as bandwidth on demand, have the capability to deliver the higher margins necessary for service providers to survive in an intensely competitive environment in which bandwidth has been reduced to a commodity.
Delivering a network architecture that will enable both commodity and value-added services with existing technology is difficult and very expensive. It is impossible to accurately predict changing traffic demands; this issue is further exacerbated when the required bandwidth needs to be delivered in a short time. Over-provisioning a network partially solves this problem but does not necessarily guarantee that bandwidth will always be available when needed. Throwing more bandwidth at the problem necessitates the use of additional equipment or perpetuates constant network upgrades. Each approach is wasteful and negatively affects return on investment—a fact that is not lost on service providers.
Service providers therefore need network equipment that not only reduces their costs by ensuring operational efficiencies but also increases their ability to deliver bandwidth on demand anywhere in the network.
Tunable lasers offer several options to enable metro network systems. While the operational benefits achieved through lower costs are obvious, service providers are exploring equipment and network architectures using tunable lasers that allow them to not only reduce provisioning times, but also to deliver speed-intensive, higher-margin applications and services.
Innovative use of dense WDM (DWDM) technology removes one of the major bottlenecks in network architecture, where traffic must be converted from optical to electrical form for processing and then back to optical before being sent along the network. This optical-to-electrical conversion is both costly and time-consuming, rendering a network unsuitable for supporting new, speed-intensive applications that service providers must roll out to migrate to high-margin services. Incorporating DWDM into metro network architecture allows intelligent grooming of traffic onto wavelengths within a network, ensuring no unnecessary packet processing at intermediate nodes within a network.
Today, the need for additional capacity requires lighting of additional dark fiber, which also requires securing rights of way to lay a conduit, or, where a conduit is available, purchasing and lighting dark fiber with transmission equipment. Significant engineering time is required to configure the additional crossconnects and optical add/drop multiplexers to integrate into existing networks.
TUNABLE DWDM ARCHITECTURE
A tunable DWDM architecture implies reconfiguration of light paths nearly instantaneously. A light path can be described as a point-to-point wavelength connection between two network nodes. Reconfiguration of light paths can be achieved by two fundamental architectural approaches: tunable transmit and tunable receive or tunable transmit and fixed receive.
In the fixed WDM SONET ring approach, multiple logical rings are constructed in one physical fiber by enabling the members of each ring to transmit and receive on a unique wavelength. The nodes of these rings terminate and originate SONET signals just like an ordinary SONET ring—each logical SONET ring rides on a unique wavelength (see Fig. 1).
The tunable transmit and tunable receive architecture differs from the standard WDM SONET approach in that it enables the rings to be reconfigured through wavelength tuning. In this architecture, each node must have the ability to transmit and receive on any wavelength associated with a ring that it wishes to join. Furthermore, since all nodes are interconnected with the same physical fiber, those wavelengths that are associated with other logical rings must pass through the node unaffected—that is, they are "expressed" through the node without stopping or termination, much like an express train does not stop at local stations.
Bandwidth reprovisioning or rebalancing can occur by reducing the number of nodes in bandwidth-hungry rings. The nodes that leave these rings must retune both their transmitters and receivers to the wavelength associated with other, less-utilized rings. Since both receive and transmit must be retuned to facilitate movement to a new ring, relatively sophisticated wavelength control and management is required. This implementation is also expensive since it requires the use of both tunable transmitters and receivers.
Another approach consists of using a tunable transmit and fixed receive node (see Fig. 2). In this architecture, each node receives a fixed, unique wavelength. Since each node is addressable by its fixed receive wavelength, a logical ring is created by tuning the transmit wavelength of the upstream node to that of the downstream neighbor's receive wavelength. Each logical ring therefore comprises a series of logical links, and each logical link comprises a wavelength and a physical link. The node number at each link's end labels the physical links.
Simple rules can be developed for ring formation:
- If a node belongs to a ring, it is reached by its nearest logical neighbor via transmission from this neighbor at the wavelength the node receives.
- Collisions in lambda-space are avoided at the logical link level by using nonrepeating wavelengths.
Reconfiguration of the rings requires the establishment of a new light path ring, adhering to the ring formation rules.
In one version of an optical-node architecture, the fixed wavelength is dropped from the ring fiber by a basic 3-port filter (on 100-GHz spacing, the filter is relatively cheap and is available from several vendors; see Fig. 3). The add functionality is achieved with a broadband directional coupler. The coupling ratio used will depend on the details of network implementation, which include the use of amplifiers to overcome losses and the available transmitter power. Coupling ratios between 1:10 and 1:1 can be easily achieved with commercial power-coupler technology.
The notion of a reconfigurable DWDM network necessarily implies some degree of transparency in order for any node to be reachable. Reconfigurability that scales implies one has to overcome fiber, patch-panel, and optical add/drop multiplexer losses in network architecture, thereby leading to amplification. In fact, oftentimes cascades of amplifiers often will be required.
Amplifier cascades are difficult in practice because of the accumulation of amplified spontaneous emission along the chain and the accumulation of gain spectrum nonuniformities. If the amplifier chain is allowed to form a closed ring, as the gain approaches unity, amplified spontaneous emission grows and lasing instabilities can easily result. Furthermore, a closed ring can lead to severe multipath interference impairments.
Erbium-doped fiber amplifier (EDFA) cascades have been thoroughly studied by those using DWDM for long haul, and their success using EDFA amplifier cascades has largely driven the optical networking revolution. However, in the long-haul application, cost has not been the technology constraint that it is for the metro and access spaces. In general, many of the solutions used for the long haul are not applicable for the metro networks.
In a reconfigurable ring environment, the signal loading presented to each EDFA can vary over time as the result of reconfiguration events. This variation can lead to significant gain transients that produce burst errors, as well as exacerbating nonuniformity in gain. These transients are amplified through the cascade, and if the chain is closed, local transient control becomes more difficult because of feedback from the closed ring. These challenges can be addressed cost-effectively using a unique architectural approach that "opens" the ring without sacrificing full protection.
As designers build systems using these concepts, attention must be paid to various types of power loss resulting from insertion, coupling, and fiber fusion, in addition to the tuning accuracy and tuning speed of the lasers. Power loss has a direct impact on the maximum reach of the lasers and therefore on the amplification systems that would need to be used otherwise (see table).
BENEFITS OF TUNABLE DWDM
Tunable DWDM systems ensure that service providers are no longer bandwidth-bound because additional channels can be opened up by changing the laser frequency, without scheduling expensive truck-rolls. As the traffic demand on the network changes, the use of tunable lasers allows opening of new channels of communication by simply retuning the lasers to a different receive frequency (see Fig. 4).
The benefit of tunable lasers is further enhanced as the bandwidth demand on the network changes. Network nodes can be reprovisioned and moved from a heavily congested ring to one that is partially filled, thereby ensuring optimum network performance. This change is software-controlled and is achieved without disrupting any traffic on the network (see Fig. 5). This doubling of network capacity and network reprovisioning without tunable lasers would have involved a very expensive network upgrade with truck rolls to every single node on the network.
Tunable lasers also minimize sparing costs since a service provider can use a single laser for sparing, whereas a network using fixed lasers would require a spare laser for each node. A service provider can expect a savings of almost 70% in laser sparing costs in simple network configurations. The savings are significantly higher in larger networks.
Use of innovative architectures and network systems, which exploit the benefits offered by tunable DWDM in terms of overall operations-cost reduction, simplified network scalability, and optimization, hold significant promise for service providers. What is even more appealing: to realize the true benefits of such systems, network architectures do not have to undergo a significant change. Carriers can continue to offer their existing services while taking advantage of the tunable architecture at every opportunity.
Rajive Dhar is director of solutions marketing and Mark Lowry is director of photonics at Atoga Systems, 49026 Milmont Dr., Fremont, CA 94538. They can be reached at email@example.com and firstname.lastname@example.org.