Cost savings realized in deploying coarse wavelength division multiplexing (CWDM) systems have helped this transport architecture gain acceptance as a viable alternative to dense WDM (DWDM) in short-haul metro-area and enterprise networks. As CWDM evolves technically, its advantages over DWDM will further secure its place as a flexible, low-cost alternative for increasing system capacity. What will cement CWDM’s place? Reduced hardware costs, low power dissipation and a small device footprint will continue to place the cost-savings advantage squarely in the CWDM camp. The savings should become significant enough in the near future that network designers will opt to use CWDM in new ways, including as a means of delaying more costly upgrades to extend the life of today’s network equipment. In a simple CWDM metro-ring architecture, network engineers can increase capacity requirements with a bolt-on hybrid CWDM solution. The basic difference between DWDM, which is built for long distances, and CWDM, which is ideal for short-haul networks, is the spacing between the wavelengths (channels) transmitted through the same fiber. DWDM operates over a narrow band of frequencies, known as the C- and L-bands (between 1530 and 1620 nm), requiring wavelengths to be tightly crammed together at anywhere from 1.6 to 0.4 nm apart. Using erbium-doped fiber amplifiers to boost the signal over long distances, DWDM can transmit from 32 to 128 channels. CWDM, in contrast, operates at a much wider range (1270- to 1610-nm center wavelengths), separating them at 20-nm intervals, which can transmit a maximum of 18 channels over a distance of 60 km. Today’s CWDM systems are about 40 percent less expensive than most DWDM solutions (based on cost per channel at OC-48 speeds), but even greater savings will be possible as the technology evolves from its DWDM roots. The potential for further cost savings is fueled by work already under way, including developments in subcomponent design and packaging, and in enhanced hardware designs. But the real value to carriers may be the capability to prolong the life of network equipment currently in use by providing scalable, plug-and-play upgrades. Subcomponent advances A key advantage of CWDM over DWDM is that it does not require a thermoelectric cooler to compensate for thermal drift as distributed feedback (DFB) lasers heat up and cool down. This requires the use of less expensive wideband optical filters. With channel spacing at 20 nm, CWDM spacing is wide enough to permit the laser wavelength to change over an operating temperature of 70 °C. Creating high-yield devices that have extended temperature capabilities while maintaining good operating specifications is key to handling broader temperature extremes. One area where significant research is under way is laser sources. Laser suppliers are working on providing the full spectrum of sources required with acceptable dispersion and power. Maintaining tolerable thermal drift (0.1 nm/°C) across the spectrum is another area for improvement. While temperature requirements vary by application, a target range of 125 °C is not unusual, especially for outdoor plant or cable TV applications. Backscatter tolerance Another important packaging issue is backscatter tolerance, which affects the operating distance. Many vendors have developed clever low-isolation techniques (at about 8 dB) based on uncollimated optics that are less expensive to implement. Examples include backscatter isolation using fiber stubs or subwavelength structures. The ideal solution, though, would require lasers that are more tolerant to backscatter, which, in turn, could eliminate the need for isolators in most applications. Development efforts are ongoing. Yet another research area involves development of sources with better divergence symmetry to increase coupling efficiency. The existing problem is that asymmetric divergence forms an elliptical image that does not efficiently couple into a circular mode field. Though coupling efficiency is not a problem unique to CWDM, it is especially important for technology that does not rely on amplifiers (which are being used in some channels). A problem with today’s laser sources is the lack of symmetric divergence, which results in an elliptical image that doesn’t couple efficiently into a circular mode field. Finally, low-power requirements highlight key packaging and usability differences between CWDM and DWDM. The latter consumes significantly greater power and dissipates more heat because of the Peltier coolers designed into its optical subassemblies. The coolers can consume about 4 W per wavelength, whereas a CWDM laser transmitter uses only about 0.5 W. A related issue is avalanche photodiode usability. Today, PIN detectors offer a convenient, lower-cost alternative to avalanche photodiodes, but they do not offer a sufficient link budget for many applications. As a result, researchers continue to work on next-generation avalanche photodiodes, which promise lower voltages, better sensitivity and, in some cases, better response over temperature. Once these devices can be easily integrated into transceivers, they will further reduce CWDM costs. Indeed, hot-pluggable transceivers hold the greatest potential for cost savings. The CWDM industry leveraged the cost/volume advantages of the hot-pluggable transceiver early on, which helped standardize and reduce the platform costs much more quickly than DWDM. Currently, DWDM transceivers can be up to five times more expensive, largely because of the packaging and testing of a DWDM laser (at about ±0.1 nm) relative to a CWDM laser (±2 to 3 nm). The tighter tolerances translate into smaller yields and, thus, higher costs. Additionally, to minimize the cost of equipment designed for a staged deployment, network designers today want hot-pluggable transceivers. Because of the high thermal dissipation of DWDM systems, only one vendor has been successful at bringing a hot-pluggable DWDM transceiver to market. Fundamental problems with such plug-and-play devices, however, include poor rack system densities and the inventory expense of supporting high-channel counts. Furthermore, DWDM is a low-volume/high-mix technology that effectively dismisses the fundamental advantage of transceivers, namely volume-driven manufacturing and standardization. Plug and play CWDM, by comparison, can provide standardized small-form-factor transceiver modules with lower power requirements — i.e., less thermal dissipation — which increases rack density and offers an important advantage in today’s space-constrained central offices. Standardization is important for ensuring that multiple vendors are available for hot-pluggable CWDM transceivers. The ITU has already approved a grid that has been proposed by the Full Spectrum CWDM Alliance. Development work is ongoing on application-specific standards. Hybrid CWDM components illustrate its flexibility in architectural choices over DWDM today. Because of lower device costs, customers can use CWDM as an easy plug-and-play solution to add incremental channel capacity when it is needed. Likewise, customers can convert or upgrade one or more CWDM bands to higher-capacity DWDM as traffic increases dictate. Using a hybrid CWDM bolt-on solution, end users can carve out a wide express band at the 1530, 1550 and 1570 bands. Using a bandpass filter, they can inexpensively tack on five channels to increase system capacity. In low-channel-count networks in metro-area or enterprise networks, customers can use such hybrid technology to “bolt on” expansion channels, which can increase network capacity fivefold. For example, if an existing link uses a standard transceiver, the end user can expect that the 1530-, 1550- and 1570-nm bands will be employed (essentially three of the eight possible CWDM upgrade channels). By adding a hybrid bolt-on, it is possible to carve out a wide express band and then, using a bandpass filter, to efficiently tack on five channels to the system. This can be done without displacing the current signal. Similarly, the strategy also can help upgrade long-wave Ethernet networks. In SONET or Ethernet 1300-nm networks, network designers can express all traffic on legacy signals and still create six to 10 expansion channels. Alternatively, CWDM can provide an elegant solution for Ethernet over SONET that uses a CWDM multiplexer to deliver both services transparently. There is an easy upgrade path to DWDM as traffic grows. To that end, CWDM transceivers can be easily replaced by DWDM equipment to expand only those nodes that require additional channels, while leaving the rest of the system unchanged. Final analysis Amplified DWDM devices are ideal for addressing the needs of long-haul 10 Gigabit Ethernet market requirements, but will always come at a higher price. Many networks simply do not require the horsepower and capacity that DWDM provides, making it an overkill and costly solution in these environments. Comparatively, CWDM technology provides a mix of low cost and low capacity in short-distance networks. As the technology evolves, it will continue to outpace DWDM in these key application areas. This, in turn, will boost the usefulness of hybrid solutions to extend the life of network equipment by adding short-term capacity while providing an easy upgrade path to DWDM when capacity demands rise. Ultimately, CWDM will take its place as a viable and important stand-alone transport architecture that complements DWDM and will play a pivotal role in the communications networks of the future. Meet the author Jenkin A. Richard is chief technology officer for Gigabit Optics Corp. in Sunnyvale, Calif.