In particular, one might wonder, whatever happened to the dense wavelength-division-multiplexing (DWDM) revolution that was on the brink of transforming the Internet into the information superhighway? It appears that after marching through the long-haul networking world, DWDM has encountered a serious case of crosstown traffic as it enters metropolitan-area networks.

"For long-haul networks, the argument for DWDM is compelling," said Paul Polishuk, a longtime optical-networking analyst and president of Information Gatekeepers Inc. (Boston). "DWDM components are cheap; they multiply the capacity of the network; and they are robust, simple and work entirely in the optical domain. You want to have them in your network." Although a lot of clever DWDM switching technology has been targeted at the next step — metropolitan-area networks (MANs) — the argument for using it is less compelling. MANs are a difficult brew of many different clients with different needs, and a mix of old and new technologies. The current, slow economy has not helped satisfy those complex demands.

"We have seen a softening in the demand for DSL services, for example. The incumbent telcos control the metro market, and they are going to move at their own pace," he added. Fiber has yet to reach the office or home in any meaningful numbers, and for many in the optical components business, that will be the point where the technology will really take off. Nevertheless, despite the current slowdown, the metro market — where the most demand for new technology is expected to originate — is still growing, Polishuk said.

And for startups with a good idea, the market is buoyant enough to keep them afloat while they wait for the good times to return.

The essential design considerations are cost reduction combined with flexibility, said Arlon Martin, vice president of marketing at Agility, an optical-component company founded in 1998 with technology licensed from the University of California, Santa Barbara. Agility has targeted tunable lasers as a technology that reduces costs yet provides excellent flexibility.

Tunable lasers offer the flexibility that metro DWDM systems need and reduces costs by eliminating the need for a bank of lasers to handle each wavelength. The components also deploy new flexibility by offering remote capability to provision wavelength channels. "We decided that there was an immediate value to these components in that they reduced our customer's inventory, which results in huge cost savings, but we also realized that there is an upgrade path to offering 'point and click' provisioning," Martin said. Tunable lasers could also eventually create dynamic wavelength routing and reconfigurable optical networks.

Tunable lasers employ solid-state design using an indium phosphide substrate, and the devices are delivered in a 40-pin package. The laser is structured laterally with tunable gain and phase regions sandwiched between mirrors formed by diffraction gratings. The laser device has a 4-milliwatt output and is tunable over the entire C and L bands. Agility has also set up its own highly automated fab line for the lasers.

While flexibility and the delivery of a wide variety of services are crucial capabilities for any technology trying to conquer the metro-area market, certain settings favor DWDM more than others, according to Elizabeth Broihier, senior project management engineer at Progress Telecom (St. Petersburg, Fla.). "DWDM is much more than a fiber multiplier and can be competitive in the right setting," she said.

Progress Telecom has been installing DWDM networks in smaller metropolitan areas that host high concentrations of institutions, such as hospitals, medical and other research centers and universities. The demand for services in such markets is not as fragmented as more residential areas, and the high bandwidth capability of DWDM and its flexibility of provisioning become decisive factors over other mainstream technologies such as Gigabit Ethernet or the synchronous optical network (Sonet). A cost analysis showed that while both DWDM and Sonet begin at about the same cost of installation, DWDM's ability to create additional "virtual fibers" at will soon gives it an economic edge.

Despite the current successes of the technology, it is not clear if it will eventually win the race to the home and office, according to some researchers. "WDM technology is really still at the testbed stage,"said Vincent Chan, director of the Laboratory for Information and Decision Systems at the Massachusetts Institute of Technology. Researchers made considerable progress in building DWDM networks with passive add-drop multiplexers, but the active switching components that would make the networks capable of flexible service provisioning remain at the experimental stage.

Chan has been studying the differences between a DWDM-based network architecture and a more futuristic optical network based on time-division multiplexing (TDM). TDM produces simpler control structures and could potentially be more flexible, but the components that high-speed TDM require are still in the research stage, Chan said. Based on his comparative analysis, Chan said, WDM-based systems may never achieve the level of flexibility that metropolitan nets demand. But he pointed out that all the parts needed to build DWDM-based networks are commercially available, whereas the TDM scheme requires many components that are still in the research stage or have not yet been invented.

The ability to buffer signals is crucial to any network architecture, and thus far, optical nets have been hampered by the lack of simple buffers that are taken for granted in electronics.

Chan envisions single-channel TDM encoding at 100 Gbits/second. Such a development would require a practical robust laser source that could switch in about 1 picosecond. Chan said the best candidate is the harmonically mode-locked erbium-doped fiber ring laser.

The erbium-doped optical amplifier, a simple loop of glass fiber doped with the rare earth element erbium, kicked off the DWDM revolution. Overall, timed domain schemes would be much simpler in completing the huge task that metro networks face: creating a wide variety of channels and services.

See related chart