This is part of IEEE Spectrum's special report: Always On: Living in a Networked World.
The first time an optical-fiber network carries commercial traffic is not the time to discover that a problem exists with a key component or that components from different manufacturers cannot operate together. The stakes are especially high as fatter pipes move traffic at ever faster speeds--those needed to transmit the aggregate traffic of Internet service providers and other companies on a high-speed backbone that is the core of the broadband communications arena. Field trials of 40 Gb/s are under way, and devices to drive signals at 160 Gb/s are appearing in development labs.
But, when talking about data rates in the gigabit regime, testing components is often easier said than done. "No test equipment is specifically designed to test line side characteristics or to test 10-Gb/s devices," Chris Hamilton, chief technologist for optical network switching with Williams Communications Group Inc., of Tulsa, Okla., told IEEE Spectrum. (Line side refers to the data terminal connections to a communications circuit that lies between two data terminals.)
So how are components tested? "By the time a component hits the field," said Lawrence Williams, vice president of technology at component developer Altra Broadband Inc., in Irvine, Calif., "it should only require a final functional test in the buyer's network. It should have gone through multiple simulations while in design, and several rounds of lab tests while in prototype stages, to ensure that it reliably handles gigabit data rates."
While most test equipment has sophisticated protocols to determine if a bit is the 1 or 0 it should be, "that just doesn't exist for the line side of the system or for something as sophisticated as high-density wave-division multiplexing," said Hamilton.
Start in the lab
So what does Williams Communications--a company that builds its own network to lease and sell fiber capacity--do to test the components it installs? "It forces us to go back to the fundamentals to understand if a component will work in our system," Hamilton said. That includes lab time using optical and spectral analyzers to examine signals for jitter, and looking at waveforms in eye patterns to spot anomalies.
Before installing any equipment--a router, a switch, a demodulator, or laser source--in its broadband network, Hamilton's group visits the vendor "to see if the vendor's concerns in developing a device mirror our own--our primary concern is building and maintaining a reliable optical-fiber network," he said, "not building a single component."
After seeing a component verified by the vendor, the real questions start. To figure out if the component makes sense for his company's optical network, Hamilton looks at the numbers. Checks are made on spectral efficiency, cost, line rates, data transfer rates to connected or subtended systems operating at different speeds from the main trunk, the optical power budget, and the length a signal can reach. The objective is to examine the performance abilities of the component and compare it with what's in place in the system already. If everything checks out, Hamilton will bring the component into a company lab for tests.
In the lab, Hamilton looks at interoperability issues--will it work with components from other manufacturers that are already in the network? At 40 Gb/s, where the fiber network is headed until the next advance in transmission speeds, a system is beyond the detailed performance criteria specified in a standard by the International Telecommunication Union, the IEEE, Telcordia Technologies, or some other body, explained Hamilton. "It's really a gray area that we opted to define operationally since we can't match a standard."
Basically, Hamilton looks for the components to function in such a way that expectations of abilities are met. "Think of it as a litmus test defined by how and where a component will be used. It either performs the way it's expected to or it doesn't," he noted. Passing this test gets a component a chance at a field test.
Move to the field
Interoperability is not always proven in the lab. "We have found things in field trials that were not apparent in the lab," Hamilton said. But buying components from just one supplier is not an option. "Multiple vendors allow speedy growth and scaling of the network, but the tough part is the interoperability and integration of vendors into the network," Hamilton observed.
Another advantage provided by field trials is an opportunity for more people to see a technology and get a feel for how it works and how it could work in their networks. Lucent Technologies Inc., Murray Hill, N.J., for example, has been showing its 40-Gb/s technology to customers for several months. Field trials present a "true test" for the technology because they involve buried fiber at a customer location, not the ideal world of the lab, said Olaf Herr, product manager for Lucent Technologies' optical networking group in Nuremberg, Germany.
"It forces us to go back to the fundamentals to understand if a component will work in our system"
In November, Lucent had a successful field trial in Belgium of a 71-km optical-fiber line carrying signals at STM-256 speed (the synchronous digital hierarchy equivalent of OC-768 speed, 40 Gb/s). "The goal of the Belgian trial," explained Herr, "was to prove that 40 Gb/s wasn't just a bullet point on a presentation slide, but a demonstrable reality."
The Lucent field trial was a comprehensive system test, where input was checked against output at both ends of an existing fiber between Antwerp and Brussels, part of the optical network owned and operated by Global Crossing Ltd., of Hamilton, Bermuda. "We looked at the system as a whole," explained Herr, specifically to check that the 2.5-Gb/s input signals were properly multiplexed into a single carrier at 40 Gb/s, and then demultiplexed properly at the other end.
"The technology is new," Herr noted. "The challenge is to get a stable, quality transmission in a long-term test with no bit errors, and get it to bridge the distance as we did in this Belgian trial." Herr was pleased to point out that the length of the fiber line was not much of a challenge. From an optical power viewpoint, he said, the signal could have traveled much farther than 71 km.
In bridging the distance, an optical system has an optical power budget that is spent to overcome splices and connectors in the lightwave's path. Enough power must remain in the budget, which is based on a certain output power requirement and receiver sensitivity, to ensure that the bit-error-free signal reaches its destination.
Another boon for field trials, Herr said, is that the test setup is closely watched by customers to see just how easy or difficult it is. "An easily installed system is a definite plus for a customer," Herr said.