All-optical networks take another hit
With all due respect to wireless communications, optical-network technology has consistently used copper-based electronic approaches as a measuring stick for success. Even today, potential users weigh photonic economics against electronic economics—how much more expensive is an optical approach versus doing the same thing with copper? This question is rarely posed with the order of the two technologies reversed. Fortunately for optical communications companies, the ability to carry a greater amount of traffic over longer distances frequently has overcome the cost disadvantage of most photonic approaches.
In practice, it seems many users hold onto copper until that technology no longer supports their requirements. It is a classic cost/benefit tradeoff. Meanwhile, optical equipment and components companies have worked to make photonic approaches cheaper, while the developers of electronic equipment have striven to increase the capabilities of their products.
The balance of power, so to speak, tips back and forth depending on the application. Copper continues to hold off optics in the LAN, thanks largely to a combination of inertia, the advent of Category 5a and 6 cabling, and the ability to accommodate 1-Gigabit Ethernet—and soon 10-Gigabit Ethernet—without resorting to optical fiber. Optical technology, meanwhile, has conquered long-haul and most of the metro, thanks to bandwidth and distance requirements well beyond the reach of copper.
The success of optics in these networks has led to the development of "all-optical" network technology—and one of those rare cases where the usual cost/benefit relationship between photons and electrons appears to be reversed. The more one can do optically—and avoid expensive optical-electrical-optical (OEO) regeneration—the better, goes the theory. True, one can't easily access the overhead bits that help determine the health of a signal or even where it's supposed to go, but developers continue to work diligently on optical headers, optical buffers, and other photonic counterparts to electronic store-and-forward methods. Meanwhile, the cost advantages of all-optical approaches more than compensate for the lack of access to overhead bits and the overwhelmingly analog nature of such an infrastructure.
However, advances at the wafer level threaten to unravel such an argument—for users and developers alike. At the network level, a story such as Infinera's (see "OEO is good, says startup Infinera," front page)—provided it's true—would reverse the economic underpinnings of the all-optical argument. As always, economics must go hand-in-hand with benefits—and if the economics of OEO conversion can ever be made rational enough to justify the acknowledged benefits, all-optical approaches will appear even more exotic (and unnecessary) than they do to many now. True, for many network engineers, the optics versus electronics debate can take on religious overtones. But for those who remain agnostic—or who had their religion trampled in a move to optical approaches—the lure of inexpensive OEO conversion should prove very appealing.
But even at the component and subsystem level, once promising photonic technologies such as arrayed waveguides and semiconductor optical amplifiers have been displaced on the "hot list" by forward error correction and electronic dispersion compensation. In other words, even optical equipment companies have to look at OEO conversion to reap the benefits of the latest advances—or, at the very least, pair electrical devices such as electronic dispersion compensation chips with emerging optical advances such as long-wavelength VCSELs to get those newoptical technologies into wide deployment.
Part of what we're seeing is a drive to move complexity—and, in many cases, intelligence—from the optical to electronic realm. That shouldn't be surprising, really. As discussed before, technicians still haven't perfected the optical equivalent of several electronic functions necessary in today's networks. And it certainly doesn't make much sense to have services move one way—to become increasingly digital—and have the network move the other—toward analog.
Optical components and systems vendors will continue to advance the technologies associated with all-optical networking. Optical chromatic and polarization-mode dispersion technology will grow more effective and less expensive. As Infinera's work and that of other photonic-IC teams demonstrate, the benefits of creating and moving photons at the wafer level will spread to an ever-wider variety of applications. Optical signals will travel farther distances without 3R regeneration. Yet all these advances—and all the others that will appear over the next several years and perhaps be applied in the name of all-optical networks—will undergo the same cost/benefit analysis common today. And from the viewpoint of such an analysis, it appears that the drive to remove electrons from the network faces more challenges today than ever before. ..