New amplifier technology created with the metro in mind will reduce costs and increase carrier capabilities.
DR. MUSTAFA A.G. ABUSHAGUR and BILL McNEILL, LiquidLight Inc.
Innovation in the area of optical networking has dramatically lowered the price of bandwidth provision. The earliest beneficiaries of the commercialization of optical research have been long-haul carriers. But in metropolitan networks, fiber spans are in the tens or at the most hundreds of kilometers, not the thousands of kilometers typical in long-haul networks. The number of metro wavelength drops typically is far higher than that in the long-haul case, as well. Bandwidth requirements at each of these drops might vary from one node to another. These requirements and others distinguish metro from long-haul networks.
Recently, a large number of companies have targeted the metro transport and access space. Most of these companies are migrating long-haul technologies to metro applications; DWDM add/drops and erbium-doped fiber amplifiers (EDFAs) are typical examples. Obviously, such technologies are functionally suitable, but their costs limit broad metro-access deployment.
The implementation of optical amplifiers in long-haul networks has reduced the number of regenerators used and thus lowered network cost significantly. Similarly, metro-access networks desperately need cost-reducing technological breakthroughs. These breakthroughs can be accomplished by developing new multiplexing techniques and lower-cost components such as optical amplifiers, optical add/drop multiplexers (OADMs), and DWDM terminals.
New sets of multiplexing technologies are being deployed in metro networks to efficiently use the fiber bandwidth. The first can be termed multiband multiplexing; the other is subrate multiplexing.
In multiband multiplexing, DWDM traffic is carried in the C-band (1,528-1,561 nm), which is overlaid on a 1,310-nm SONET network. A multiband system layout is shown in Figure 1. Selectivity is obtained by using WDM couplers in each add/drop to permit wavelengths in the C-band to pass through the nodes where wavelength services are not required, thus avoiding unnecessary add/drop of wavelengths.
Subrate multiplexing is employed to make efficient use of wavelengths by multiplexing traffic between the transponder and customer interfaces. Customers may then share a single wavelength at a node. The subrate multiplexing feature is depicted in Figure 2.
Overall, these techniques can result in a moderate cost reduction of the total system. Yet, as mentioned earlier, attempting this technique with DWDM terminals, OADMs, and optical amplifiers developed for long-haul systems limits these cost reductions. Therefore, service providers utilizing these techniques with current technology are not able to justify the expense of extending rings into metro-access areas where customer density around a node is low or customer locations are too distant from the carrier's point of presence.
While EDFAs and Raman amplification have effectively lowered the cost per bit in long-haul DWDM networks, similar cost reduction in metropolitan networks must be achieved by lowering the cost of optical components. The lower number of channels and shorter fiber spans in metro-access networks alleviates the tight specifications long-haul applications require. The development of components for such relaxed requirements will result in significant cost reductions. Service providers will then be able to obtain competitive advantage by delivering low-cost wavelengths enabled by this new class of metro-access optical-networking equipment.
Optical devices such as circulators, combiners, splitters, wavelength shifters, and filters are passive devices that consume optical power. Combining these devices to create OADMs and DWDM terminals increases losses at nodes. Extending rings further into the access network increases the number of nodes in the ring. The combination of lossy components, an increased number of devices in nodes and the nodes in rings, and greater distances between nodes is driving the need for greater amplifier deployment.
EDFAs have made great advancements in long-haul networks, but the need for amplification in metro transport and access networks is different. In these networks, the high gain of a conventional EDFA is undesirable because of the typical short spans between nodes. Amplification in this case is not needed for increasing the span between adjacent nodes but to balance power between add/drop and express channels at these nodes.
A new class of low-gain amplifier with lower cost of operation and maintenance is under development for metro transport and access applications. Service providers need an amplifier that can deliver only the amount of gain required at a specific point in the network. The amplifier must be low-cost, require minimal real estate, utilize existing fiber, and be commercially available as the metro network evolves. EDFA alternatives such as erbium-doped waveguide amplifiers (EDWAs), continuous amplification, and distributed amplification focus on these requirements.
EDWAs are the result of new developments designed to make amplifiers more compact. They build on the EDFA technology, in combination with integrated optical circuits. The EDFA, of course, is a widely understood and implemented solution for optical amplification. A typical amplifier consists of a pump laser, WDM coupler, isolators, and gain-flattening filters. High-gain amplification is achieved through high-power pump lasers coupled into erbium-doped fiber, where erbium ions are stimulated. Because the gain spectrum of the amplifier is nonlinear, gain-flattening filters are used to equalize the output power of individual channels.
While EDFAs utilize several meters of erbium-doped fiber as a gain medium, EDWAs use a highly doped gain medium as short as a few centimeters. The EDWA comprises the same functional components as an EDFA; however, the pump laser, pump multiplexer, isolator, and gain-flattening filter are integrated into a smaller package.
The EDWA challenge lies in manufacturing. A short-gain medium is achieved by increasing the concentration of erbium dopant. The tradeoff is the lower efficiency of erbium ion exchange as concentration increases. A careful balance must be maintained in manufacturing to achieve a small-gain amplifier over a short distance. By integrating a small waveguide and free-space optics onto a substrate in a single package, manufacturers hope to achieve cost reductions and performance suitable for metro environments.
With EDWAs, small amounts of gain can be placed at many points in the network where needed with lower cost than EDFAs. Essentially, the EDFA has been reduced in size, gain, and cost, enabling EDWAs to be placed where needed.
Questions remain whether EDWAs can be manufactured in quantity, given the difficulty in achieving a balance between doping concentration and erbium ion efficiency. The price of the pump laser with coupler and drive electronics adds to the capital and maintenance costs of EDWAs. When the EDWA is placed in various points in the ring where gain is needed, additional real estate and operational demands are imposed for environmental control of temperature and humidity required by active components. Nevertheless, EDWAs appear to be able to utilize existing fiber deployed in metropolitan networks. Component vendors have begun to commercialize this technology.
Continuous amplification allows very small gain to be placed in a uniform fashion throughout the fiber via Raman amplification. The Raman effect transfers energy from shorter pump wavelengths to longer signal wavelengths. Efficiency depends on pump power and the wavelength separation of the signal and pump.
The pump laser is coupled at the beginning of the fiber link while providing gain throughout the fiber. Because a continuously doped fiber is required, existing fiber is unsuitable for this type of amplification.
Although this technology is in an early deployment phase, continuous amplification does offer a promise in cost savings. The pump lasers do not need to be placed at every point in the fiber where gain stages are located, and the cost of pump lasers with couplers and drive electronics is greatly reduced. The cost of the fiber and the economic issues of the need for special fiber and the inefficiencies of the amplification might prove to be major hurdles in the way of Raman amplifiers in the metro network.
A third technique is called "distributed optical-fiber amplification" (DOFA). Based on a patent-pending technology, DOFA uses a single pump laser while distributing gain at discrete points along a fiber ring. Gain is realized by splicing short segments of erbium-doped fiber into the transmission fiber. This technique resembles EDFAs in the gain generation; but unlike EDFAs that use a pump laser integrated with a single large-gain stage, DOFAs use a single pump laser with multiple gain stages. Each stage consists of a few meters of erbium-doped fiber to provide gain only where it is needed throughout the ring.
An implementation of such a system is depicted in Figure 3. The erbium-doped-fiber (EDF) amplifier blocks can be placed either before the node (preamplifier) or after the node (post-amplifier). The signal as it is received in the central office (CO) is amplified with a low-gain amplifier. One or two pump lasers are coupled into the ring in opposite directions. The initial low-gain amplifier in the CO absorbs some of the pump-laser power. The rest of the pump-laser power traverses the network from both directions to pump the passive gain blocks located along the ring network. As the pump beam propagates along the fiber, its power decreases because of the fiber absorption. The amplifier gain is designed to be only a few decibels at each stage of the EDF to compensate for the signal loss as it propagates between the nodes and passes through the add/drop multiplexers. If the distance between the nodes is increased, then the ring length can be extended. This amplification technique can also be implemented for other network topologies.
DOFA allows the placement of the active devices in the CO, which makes it easier to provide and maintain continuous uninterrupted power. Also, since the EDF blocks are passive components, there are no parts susceptible to failure. The distributed nature of this design avoids the nonlinearities in the fiber that are caused with high-gain fiber amplifiers. DWDM metro-access networks can be deployed using this fiber amplifier.
Since there are only a few pump lasers, the cost of amplification will be greatly reduced, making the all-optical network more feasible and economically competitive. DOFA is good for service providers because it eliminates the need for power balancing, reduces the impact of adding wavelengths to a node and nodes to a ring, and thus saves operations expenses and lowers the capital cost of equipment to allow broader deployment.
Distributed amplification allows the necessary amount of gain to be placed in various points in the ring. Costs savings are realized by implementing one centralized pump-laser source for multiple gain stages throughout the ring. The pump laser and signal are carried on the same fiber. Since the distributed gain stages are entirely passive, the cost of maintenance is lowered while reliability is increased.
Though not commercialized today, distributed amplification promises to meet the requirements of service providers to extend fiber rings at low cost. For a comparison of the different amplification techniques considered for metro-networking applications, refer to the Table on page 112.
Innovations in optical networking are continuing to bring high-bandwidth services closer to the business and residential customer. Optical multiplexing techniques and new lower-cost components will play the major role in the realization of such services.
While EDFAs were the key enabling technology that spawned an evolution in the long-haul DWDM market, newer alternatives will enable that evolution to continue in the metro. A new class of amplifiers is being developed to answer the needs for metro optical networks. These amplifiers are characterized by low cost, small size, minimal real estate demands, compatibility with existing fiber plant, and near-term commercial availability.
EWDAs and DOFAs are prime candidates for these applications. EDWAs still face major challenges in manufacturing and performance for metro-access networks, while DOFAs are built on a mature technology and deemed to be the most promising optical technology to realize the Internet Protocol/optical converged solution for the new generation of optical service providers. DOFAs introduce disruptive technology that allows service providers to deploy and manage optical networks far more cost-effectively than is possible with current technology.
Mustafa A.G. Abushagur, Ph.D., is vice president and chief technical officer and Bill McNeill is product manager at LiquidLight Inc. (Duluth, GA).