Radical optical solutions demand review

Wed, 03/31/1999 - 7:00pm
Fred Dawson, Contributing Editor

... part II

Opening a new window

Upstream applications also represent one of the main reasons behind emerging interest in the possibility of opening the 1400 nm window to cable use. S-A, working with Lucent Technologies Inc., has taken the lead in this area, with tentative plans to develop an alternative to its own TDM and other strategies for maximizing return efficiency. And the 1400 idea has downstream potential as well.

The ability to use 1400 nm wavelengths rests on Lucent's new "All-Wave" fiber manufacturing process, which eliminates the residual traces of OH molecules left from water that seeps into the glass during traditional fiber manufacturing processes. OH happens to have an absorption impact on wavelengths in the 1400 nm region, even though wavelengths below and above are unaffected, which is why 1400 isn't used in communications systems.

Lucent, along with S-A and two vendors serving markets outside cable, recently demonstrated use of the technology at the Optical Fiber Communications conference in San Diego. Using a single fiber, the parties transmitted a gigabit Ethernet LAN signal at 1310 nm, two wavelengths aggregating to 10 Gbps of data in the 1400 window, and two wavelengths of NTSC TV signals via gear supplied by S-A in the 1550 region, all at a distance of 60 km.

Signal loss in the 1400 nm window over most of the installed fiber in use today would be in the range of 1 dB per km vs. 0.35 dB per km for 1310 nm, the key transmission point in the 1300 window, and about 0.25 for 1550 nm. With the AllWave solution, the loss at 1400 falls right between the 1310 and 1550 levels.

"Lucent's fiber offers us an opportunity to open a whole new territory to use as we look at ways to evolve our networks," says Don Sipes, vice president for advanced transmission technology at S-A's network systems unit. "We see a lot of interesting possibilities."

Figure 1: AllWave vs. conventional fiber: more usable optical spectrum.

While it will take several months of experimentation and further evaluation with S-A customers before any product development decisions are reached, Sipes stresses S-A is taking the availability of the 1400 window seriously as it explores options for evolving network capabilities within budget parameters acceptable to MSOs. "You'll probably see the first prototype devices in about a year," he says.

The 1400 option would not likely be cost-effective if it meant replacing existing fiber with AllWave, but it could have a major impact in new fiber installations, either as a way to cut back on how much fiber has to be installed, or as a means to assure there will be more bandwidth available for a given fiber count, Sipes says. This applies to the distribution as well as the backbone plant, he adds.

The demo offered one view of what could be done with the AllWave capability in cable, where regional fiber linking hubs to master headends could be used to carry multiple types of traffic without requiring that all the wavelengths be narrowly defined within the dimensions set for DWDM at 1550. In upstream applications, the availability of the 1400 window could lower the cost of using DWDM to aggregate multiple incoming coax feeds in the 5-50 MHz zone.

The upstream possibility seems to be especially attractive in cable, notes Kathy Szelag, vice president of optical networking at Lucent. Instead of using relatively expensive lasers conforming to the narrow, carefully partitioned wavelengths in the 1500 region for upstream wavelength multiplexing, operators would be able to use less rigidly crafted lasers at all three wavelength regions for the return, greatly reducing costs, Szelag says. "You don't have to worry so much about the specific wavelength and the amount of separation between wavelengths," she explains.

It also happens that, because data signals operate better with some small amount of dispersion than with none, the slightly higher dispersion characteristics at 1400 vs. 1550 mean that lower-power lasers can be used at 1400 over non-dispersion-shifted fiber like AllWave, Szelag adds. This means that, for cable purposes, availability of transmission devices at 1400 shouldn't be a problem, because there are plenty of low-cost Fabry-Perot lasers operating at 1480 nm that are used as pump lasers in some manufacturers' erbium doped fiber amplifiers.

The higher power distributed feedback lasers, such as those used in downstream applications in cable, are harder to come by in the 1400 window, Szelag acknowledges. She says Lucent is the only supplier, having begun manufacturing the devices for an undisclosed military application.

Down a new road

As a challenge to conventional wisdom, nothing outstrips the technology being touted by a San Diego-based startup, SilkRoad Inc., which has come out of nowhere to make what appears to be a strong impression on at least some engineers at long distance and local telephone companies who have had a chance to look at the patented system. At a recent demonstration in New York, for example, the company used a single, externally modulated DFB laser to transmit 840 6-MHz analog television channels a distance of 100 km with a claimed signal-to-noise output of 60 dB.

SilkRoad officials say the system has been shown to operate at 200 gigabits per second over a distance of 300 km at company facilities. With ongoing gains in electronic processing speeds, the system has the potential to operate at the ultimate capacity limit of fiber at 10 terabits (trillions of bits) per second, without requiring further advances in optics, they say.

Even at the hero edge, DWDM doesn't achieve the levels demonstrated so far by SilkRoad. Last year, for example, Lucent, using 80 wavelengths to achieve 400 Gbps throughput, recorded the highest speed yet achieved over a DWDM system designed for commercial use.

But speed and distance aren't the only factors that SilkRoad touts for its technology. The company also claims that its signal aggregation and modulation technique, known as "refractive synchronization," supports adding and dropping of signals in meshed arrays with use of simple beam splitters, avoiding the complexities of all-optical add/drop multiplexers and cross-connects that are just coming to market in the DWDM domain.

"This looks like a significant new breakthrough in the way you can transmit and the capacity of what you can transmit over fiber," says Hans von Braun, research director for San Francisco-based Creative Strategies, a computer industry analyst and consulting firm. Von Braun, one of a handful of analysts who have seen the system in operation, says his only question regarding the viability of the system is SilkRoad's ability to finance and manage the transition from prototype to high-volume product.

"I'm not sure they have the ability to move to commercial operations without outside financing, but they say they do," von Braun says. SilkRoad officials say they already have prototypes undergoing field tests and will be able to ship field units that are compliant with telecommunications operations management specifications starting this quarter.

SilkRoad's system is being tested by a number of long distance and competitive local exchange carriers, von Braun notes. He says he has talked with officials at Sprint, one of the carriers testing the system, about their experiences with the technology. "They were very impressed," he says.

So, too, are engineers at the IP-based broadband carrier Level 3 Communi-cations, according to Dataquest analyst Ken Kelly, who recently completed a report on the SilkRoad technology. While several other carriers expressed great interest in the technology, only Level 3 would allow its name to be used in Kelly's writeup, he says.

"Level 3 seems to be really high on it," Kelly adds. "I saw the system demonstrated with TV signals, and it lives up to its claims."

While the underlying physics and math are extremely difficult, the actual implementation of the technology is simple and low cost, involving use of off-the-shelf lasers and external modulators in combination with an overlay of intelligence applied to the electronic feed into the modulator. "The main drawback is it hasn't been operated in the field long enough to assure carriers they can meet (the) 'five 9s' requirements for telecommunications applications," Kelly says, in reference to the 99.999 percent reliability factor used in telephony. "So the technology will probably migrate into commercial operations for other types of applications, like data and television, to give people time to see how it performs over time."

A spokesman for Level 3 declines to comment on his firm's reported interest in the SilkRoad system, saying only, "We don't talk about our vendors, even when we actually deploy their products. So we're not going to speculate about anything we're evaluating."

As described by Rob Gorman, vice president of marketing and sales at SilkRoad, the SilkRoad technique represents a variation on coherent system technology which, in the late '80s, was seen as the means by which fiber's ultimate potential could be realized. Coherent systems, which use the frequency range and other dynamics of the optical signal itself to transmit information, proved to be too unstable to put into commercial operation and were superseded by DWDM as the next step to high capacity over fiber.

SilkRoad says it has overcome the coherent system problems through the patented ideas of its Chairman and CTO James Palmer. These concepts have to do with narrowing of laser linewidths to minimize dispersion using what is known as the "Palmer Transform" to stabilize the laser onto its optimal frequency. Equally important, the system employs new ways to modulate RF signals onto the optical stream using very high clock speeds in the 60-100 GHz range to rapidly change the light-passing characteristics of an external modulator.

In the modulation process, the system takes in all the various types of RF signals in their native formats without transposing them to a uniform format, shaving away the upper sideband and assigning a clock sampling frequency value in the lower sideband to each carrier, Gorman says. "Every (RF) signal has its own spectral output, which allows us to coherently mix them onto the beam," Gorman notes. The clock values are assigned to the beam via refractive changes in the light-transmitting properties of the external modulator, so that all the dimensional values of the photons, including space and time as well as frequency and polarity, are put to use in carrying the message, Gorman adds.

To get to this multidimensional information-carrying capability over the optical stream, Palmer reworked the original field equations of James Maxwell, which describe how electric and magnetic fields interact to impart electromagnetic energy. As explained in a paper Palmer is preparing to give at a conference in May, the resulting algorithmic solutions provide exact time dilations for each of the photonic properties within each few nanoseconds of time, allowing more signal information to be added to a given point of light than is possible with conventional two-dimensional multiplexing. By reading the photons at the receive end in sync with the assigned sampling frequencies, the original RF signal destined for a given end device can be readily extracted from the optical signal stream, which is why signals can be "switched" by simply splitting the beam, Gorman notes.

Because the signal input is not altered from its native format in the SilkRoad system, the technology is ideally suited for the AM signaling requirements in cable, Gorman adds. Moreover, he says, the low attenuation and high linearity of the narrow linewidth beam combine to allow the system to deliver AM signals at very high carrier-to-noise levels over long distances.

"We realize this technology would be of great use to the cable industry," Gorman says. "We want to open a dialogue with cable companies as soon as possible." The company is also talking with potential manufacturing partners with the intention of maintaining full control over the design and manufacturing process, he adds.

While declining to discuss specific pricing, Gorman asserts that the SilkRoad system cost is so low in comparison to DWDM systems that its availability will "cause a serious change in the whole capital cost structure" of carriers who deploy it. "People are excited about the design implications where they don't have to replace installed infrastructure or add a whole lot of equipment to achieve this level of performance," Gorman says. The system is indifferent to the type of fiber used, he adds.

Clearly, network engineers have a lot to think about as they consider ways to evolve their companies' facilities. No matter which new paths are taken, there seems no getting around the likelihood that sticking with the tried-and-true optical techniques of the past is not a prescription for success in the future.




Share This Story

You may login with either your assigned username or your e-mail address.
The password field is case sensitive.