For the past few years, AT&T Labs has been tinkering with a technology called “light-labeling,” which can help monitor and manage performance in optical networks. As reconfigurable optical add/drop multiplexers and photonic cross-connects proliferate in today's optical networks, the number of electronic management points — where optical signals are converted into electrons, read and converted back — are diminishing, leaving operators increasingly in the dark about signal performance.

Light-labeling collects and reports performance information in the optical domain without costly electronic conversions. In this method, each light path is marked with a label upon its creation. That label is read by low-cost receivers throughout the network, which keep track of the light path's travel and report on its progress, pinpointing faults when and where they occur.

The label comprises two layers of code. The first is called a “complementary constant-weight code,” or CCWC, an original extension of an existing technique wherein overhead bits that contain the label information are inserted in the low-frequency part of the bit stream, spread throughout the bottom of the frame rather than at the header. This ensures that they can be read by low-cost, low-frequency receivers — and that low cost ensures that these receivers can be deployed ubiquitously. The other layer of the code is CDMA, which allows the receiving photodiode to read multiple labels simultaneously without having to de-multiplex, or separate, the light paths.

Light-labeling should not be confused with other optical labeling methods such as G.709, which adds overhead bits for management purposes but also includes other functions. G.709 works in the electronic domain, relying on expensive receivers, and there are some faults that cannot be pinpointed precisely using G.709 in all-optical networks.

“There are certain bits [with G.709] that can be used to carry path-trace information, but you can't get them until the signal is detected,” said Mark Feuer, senior technical specialist for AT&T Labs. “That leaves you kind of helpless in the all-optical core. All you can see is around the edges.”

Labeling light paths adds more bits to the signal and increases the line rate slightly, but AT&T Labs researchers are keeping the rate increase in the low single-digit percentages. Still, attaining consistent data rates is an issue that likely will need to be resolved before the technology is adopted commercially, and Feuer won't predict when that might be.

Meanwhile, AT&T Labs is developing the technique further, inventing ways to “rewrite” labels for future use in, say, packet optical burst networks. And researchers have imagined a “clip-on” version of the technology — a small device that technicians could clip onto any fiber to test its performance. “We haven't proven that that works yet,” Feuer said. “But conceptually it should.”