The cost revolution continues
Just as "increased capacity" was the battle cry of the optical communications market as this decade began, calls for "cost reduction" now echo across the industry. Whether we're talking capital or operating expenditures, testing, network management and provisioning, or manufacturing, if your product or service can't claim to save the customer money, you've got a problem. And vendors around the world are listening. For example, on my visit to Germany last month, Agilent's Steffan Loeffler told me that the next big innovation would be in the area of cost rather than technical performance.
While Loeffler's comment doesn't preclude performance innovation (particularly if it results in cost savings), his statement led me to wonder where "the next big thing" in cost reduction would come from. Optical communications technology has already made significant strides in approaching cost parity with other alternatives. As the Telecommunications Industry Association's Fiber Optics LAN Section has already pointed out, it's now about as cheap to cable a building with fiber as it is with copper—and if you take future requirements into consideration, the long-term bill for an optical approach may turn out to be smaller. WDM technology, of course, allows carriers to add services or customers without adding fiber. Ultra-long-haul (ULH) technologies reduce the amount of 3R equipment in the network. Passive optical networks provide high-speed services with no power requirements between end points.
Yet, there's certainly more work that needs to be done. Most of this effort takes one of two paths. The first is lowering the cost of performing the tasks inherent in carrying traffic over optical transport. Well-known areas of research here include VCSELs and the ULH technologies already mentioned. Long-wavelength VCSELs represent the latest advance in the first of these two avenues. Some companies like Picolight and OCP have been very public about their efforts, while others such as Agilent have worked more quietly. Recent acquisitions of companies with long-wavelength VCSEL technology—the fiber optics unit of Infineon purchased by Finisar and E2O Communications bought by JDS Uniphase—highlight the importance industry players place on the technology as a key for the future.
Still, such light sources must first prove their reliability and in all likelihood improve their performance. The inherent cost savings VCSELs provide in terms of manufacturing efficiency versus edge-emitting lasers will not make much of a difference if they must rely on "boosters" that potentially add cost, such as electronic dispersion compensation or some potentially expensive amplification technology, to match the performance of Fabry-Perots or DFBs.
And in the ULH—and even "regular" long-haul—space, the reduction of 3R requirements provides only a partial solution to the overall cost problem. Amplifier chains and the performance-retarding properties inherent in fiber still mean that carriers must foot the cost of compensators and other devices to maintain the fragile balance of parameters necessary to ensure signal integrity. It's no wonder that companies like Infinera (discussed on last month's front page of Lightwave) feel they have a winning proposition in reducing the cost of optical-electrical-optical conversion to a point where they can suggest that carriers throw expensive all-optical transmission equipment out the window.
The other major area of research focuses on avoiding the intrinsic costs of any network that frequently are beyond the direct reach of technology. Most of these costs involve the labor and time necessary to either install equipment or provision services. The cost to install new networks—and the time and hassle involved in getting permits to dig up streets or roll trucks to set up new customers or services—has long been lamented. Free-space optical systems are an example of innovation in this area. So too are reconfigurable optical add/drop multiplexers.
A subset of this research area is the growing necessity to package new innovations as evolutions, rather than replacements, of existing systems and technology. Next-generation SONET/SDH is the most salient example of this trend. Looking forward, I believe that when 40-Gbit/sec links finally appear in the field, they'll originate from line cards that can easily slip into existing 10-Gbit/sec racks. Furthermore, they'll operate with the same amplifier spacings as the 10-Gbit/sec links alongside which they will ride—or, except in rare instances, they won't be installed.
The factor behind this trend toward cost reduction is that carriers, not vendors, are driving the optical communications bus. While vendors could push technology into the market during the bubble, the customer—whether a carrier or end user—is now in control and dictating the direction of R&D. That is as it should be for the long-term health of the industry. For it is only by meeting the requirements of carriers, enterprise-network managers, and the people who use their infrastructures that the optical communications industry will remain viable.