What the broadband culture craves >BY DAN O'SHEA, Technology Editor

Judging from the proliferation of wave division multiplexing solutions at Supercomm earlier this summer, one might assume available fiber capacity on public networks has been reduced to a trickle.

At the show, several vendors discussed their WDM technology enhancements, and two vendors, Pirelli Cable and Ciena Corp., launched solutions that respectively offer four times and five times the channelization of most current WDM systems.

In the bandwidth game, the ante has been upped quickly and decisively. The surging growth of Internet-related applications in the last 18 months, along with large growth in all types of network traffic, has put a stranglehold on existing network capacity. The anticipated emergence of asynchronous transfer mode, video and other high-speed data applications will only increase demand.

For telephone companies and other public network operators, the challenge is to provide the desperately needed bandwidth in a timely and reliable manner that is also cost-effective and takes advantage of existing facilities to their fullest potential.

Fiber Sonet networks have long presented a telecommunications ideal but have too often been misrepresented as the best possible solution for bandwidth woes. Theoretically, fiber can handle millions of calls, but the industry has only learned to exploit a small part of its potential. To address and manage the encroaching growth, telcos must increase the traffic-bearing capacity of their fiber.

"WDM is one more mechanism to exploit the inherent bandwidth of fiber," says Steve Alexander, director of system development at Ciena.

WDM enables network operators to efficiently unlock fiber's full potential by dividing fiber paths into four, eight or 16 distinct channels. Pirelli's new system, when available next year, will deliver 32 channels. Alcatel Network Systems also offers 32-channel WDM. And Ciena is expected to raise the bar to 40.

These accomplishments are undoubtedly impressive, but how well do they reflect the actual need for new capacity in a market that has so far only begun to sample 16-channel WDM? The answer may be that the need, though not immediate, is clearly apparent. Driving Demand

So far, WDM has been used mostly by interexchange carriers interested in boosting the call-carrying capacity of their long-haul fiber networks without the use of costly regenerators and other devices. Several emerging and ongoing factors have combined to forge a critical bandwidth demand that is driving this necessity.

Internet traffic has passed along interexchange network routes for many years, but simplification of Internet access methods and creation of new Internet applications-in addition to the unrelenting Internet commercialization that has brought it into the mainstream-have turned the Internet into a sudden and insatiable bandwidth hog.

"Network operators tell us they're having trouble keeping up with that demand," says Tim Krause, director of product marketing and business development for lightwave systems at Alcatel Network Systems.

Growth also is surging across the board in regard to other high-speed applications such as videoconferencing and mass data transfers. With these applications, more users are demanding greater reliability assurances. Committing more fiber to back-up and survivable routing adds to the crunch.

In addition, traditional voice and fax traffic and lower-tier data applications such as e-mail continue to increase annually. Sprint experienced 19% growth in minutes of use on its network between the second quarter of 1995 and the second quarter of 1996. AT&T handles more than 210 million calls on an average business day.

All these types of traffic are pressing fiber to its limits, and bandwidth requirements will balloon even more as entertainment video and high-tier ATM applications such as telemedicine become everyday travelers on telco networks.

IXCs have a handful of options for addressing this wild growth, the most obvious of which may be to install more fiber. Not everyone agrees with this, however.

"Putting in fiber is costly, and lighting up new fiber involves more equipment and maintenance considerations," says Marty Kaplan, chief technology officer at Sprint. "If you already have embedded fiber, the last thing you want to do is put more fiber in the ground," says Krause. In fact, depending on the location and geographic nature of an area, adding new fiber could be logistically impossible.

Carriers can also address bandwidth issues by performing electronic Sonet upgrades to higher time division multiplexing standards. "You can upgrade or change out the electronics in the architecture. For instance, going from OC-48 to OC-192 increases capacity by four times," says Kaplan. However, this involves early retirement of existing facilities, which goes against carriers' every natural instinct and ultimately may not be very cost effective. Also, while OC-48 is a clearly defined standard, joining optical elements at OC-192 is not so well-defined.

WDM may offer the best solution from both a cost and a technical perspective. "By using WDM, you are using the existing fiber and electronics. It's a one-time installation. You can put additional capacity over the same pair of fibers," says Kaplan.

"The ability to reuse fiber is WDM's most obvious advantage. It must satisfy our cost structure, but the technological implications must also be clear," says Laurel Clark, a member of technical staff at AT&T.

"It is more cost-effective than stepping up to the next electronic speed," says John Clark, chief operating officer at Atx, an optoelectronics company recently acquired by Scientific-Atlanta.

As WDM channelization intensifies, the technology's cost-effectiveness becomes more distinct. Attaining 16 channels of capacity through electronic upgrades can add well more than 100 new devices to a 300-mile fiber route, upping the cost and operational and maintenance concerns. Using WDM, only a half dozen new devices would have to be added to a similar route (Figure 1).

Market Readiness Despite the unheralded growth in applications and the consequential bandwidth requirements, the industry has so far been well-served by four- and eight-channel WDM systems. The largest IXCs are only beginning to deploy 16-channel systems in a handful of applications, and with smaller networks, local exchange carriers do not yet have much use for WDM.

This begs the question of why vendors are so hot to talk about next generation 32-channel and 40-channel systems. Some say the future of networks has become clear enough that network operators are delighted to hear vendors talking about increased capacity. Others point out that the sudden explosion of the Internet and other applications have left network operators uncertain about what kind of network capacity they will need and when they will need it. Therefore, they believe carriers will want to boost capacity exponentially at the first sign of the next application trend.

"They'll use all the bandwidth they can get," says Steve Chaddick, vice president of product development at Ciena. "They're a little frightened by the uncertainty of their growth."

Recognizing this uncertainty, Ciena said it will be ready to supply a 40-channel WDM when a network operator is ready to deploy it rather than rather than announcing a specific availability date.

"If you look at the growth demands and how they are escalating, it is foreseeable that we'll be using 32- or 40-channel WDM," says Kaplan, adding that Sprint may deploy at least 32-channel WDM by the end of 1997 or in 1998 (Figure 2).

"[32-channel] WDM will happen in time. A lot of talk about it is hype now, but soon it won't be hype," says Rob Koslowsky, director of transport marketing at Northern Telecom.

AT&T takes a more methodical attitude about the WDM craze. "We are working with our vendors to evolve to 16 [channels] and beyond. When we go beyond what we have now, it will really become a matter of performance and price," says George Gawrys, director of transmission planning at AT&T. "We will not be in a situation where if we didn't have 32 [channels] we'd be up the creek," he says.

"WDM will allow us to gracefully move to the next level when we are ready. We've been planning our WDM deployment since the early '90s," adds Clark.

As noted, WDM has been deployed exclusively by IXCs thus far. At the same time that IXCs are mulling 32- and 40-channel WDM, LECs also are seeing large traffic growth. Still, the LECs have not needed WDM, mostly because the fiber in their networks is newer and their fiber routes are not very long, says Atx's Clark.

This may change, however, as local telephone companies begin to take advantage of new telecom law by becoming long-distance providers. More high-bandwidth applications will be introduced into their networks for long-haul transport, perhaps even outside their traditional regional boundaries.

"We are helping a lot of the [Bell companies] plan their long-distance networks. WDM will allow them to grow their networks incrementally," says Alcatel's Krause.

"The prospect of WDM is very good for anybody who wants to increase capacity," says Fahri Diner, product line manager of digital transport products at Pirelli Cable. Pirelli has ongoing WDM projects at some LECs now, but LEC deployment of WDM will start in earnest in 1997, says Diner.

The same dynamics exist in Europe, where deregulation is creating more potential for long-haul networks than was previously seen, says Terry Unter, vice president of electronic components manufacturing at Alcatel Network Systems in France. WDM will be viewed as a way to enhance fiber capacity between Europe's major cities, he says.

The Optical Ideal As WDM expands its presence in interexchange networks and enters the local exchange network scene, it will be the catalyst to a greater transition from traditional electronic networks to optical networks. Many carriers first deployed Sonet only to satisfy high-bandwidth, point-to-point applications, but the major IXCs now tout almost entirely Sonet networks, and LECs are now following suit.

WDM, most effective when deployed in large doses, is the next step toward the optical ideal.

"The industry will move to all-optical networks in time. Transport is moving there first, and switching will go later," says Sprint's Kaplan.

Interest in WDM waxed and waned over several years as network operators concentrated on increasing bandwidth through upgrades in the TDM world, but the recent renewed interest and increased deployment of WDM also points to the eventuality of the all-optical network environment, says Koslowsky.

"WDM, in its coarsest sense, has been around for the last decade. It is only in the last few years that a number of technologies have come together to bring it value," says Ciena's Alexander, adding that filtering and fiber grating techniques have helped enhance the value of WDM.

"The goal is to get as many channels as possible across the widest bandwidth," says Alcatel's Krause. Indeed, network operators will increase their reliance on optical cross-connects and optical add/drop multiplexing to deliver channels of capacity to specific areas.

Investment in WDM will now have long-term ramifications. The migration path to further fiber capacity will be set, and it will be a much smoother path than adding new electronics or burying new fiber altogether. Even if the sudden success and explosive usage of certain broadband applications surprises network operators, the next level of WDM will be available when they need it.

"Networks should follow a logical evolutionary path. You should not have to go through a revolution every time you want more capacity," says Alcatel's Unter. Notes Ciena's Chaddick: "Carriers were saying [15 months ago] they thought four- and eight-channel WDM would be enough for the foreseeable future. The foreseeable future turned out to be about three months."