Optical schemes pick up steam in LAN, WAN

SAN DIEGO — A record attendance of 10,000 and a flood of venture-backed startups at the Optic Fibers Conference this week betrayed the rising belief that optical networks will become ubiquitous as we head into the new millennium.

Passive wavelength-division-multiplexing (WDM) systems are making tenuous forays into enterprise LANs; active optical cross-connects and add-drop multiplexers are being readied for deployment in public backbones. But enthusiasm at the conference was tempered by the realization that optical alternatives must increasingly demonstrate performance and price advantages over more traditional components and systems if they are to reach the next level of growth.

The next two years will see a steady stream of newcomers tout components and system-level products for both LAN and WAN. Next week, Adva Optical Networking (Munich, Germany) will debut its Fiber Service Platform, which lets carriers provision enterprise-LAN services over 4- and 16-channel WDM. Applied Fiber Optics Inc. (Fremont, Calif.) this week announced a name change, to WaveSplitter Technologies Inc. and broadened its approach from filter components to carrier systems. WaveSplitter rolled out the WaveXpander front-end WDM product, which channelizes existing broadband systems using hybrid fused-fiber/Mach-Zehnder devices.

LaserComm Inc. (Richardson, Texas), formed only a year ago, is launching a spatial-mode transformer that is said to allow more effective use of WDM above 10-Gbit rates via chromatic dispersion control.

The stampede of component-level lightwave startups is eclipsed by system-level newcomers pursuing a variety of paths for transporting packet traffic over WDM. New Access Communications Inc. (Santa Barbara, Calif.), a spinoff of MRV Communications that took a best-of-show award at the recent ComNet, is readying a centralized optical router that performs packet-based add/drop multiplexing, for steering traffic packet by packet.

This summer, more newcomers will exploit the optical multiplexing and routing arena recently tapped by Monterey Networks (Richardson, Texas). Such startups as Sycamore Networks Inc., Chromatis Networks, Arroyo Networks and Sirocco Systems are expected to launch products by the June Supercomm show.

Many of the WAN-oriented startups here were working from concepts for optical cross-connects pioneered by Lucent and Tellium Inc. as part of the Multiwavelength Optical Network (Monet) consortium, an intracity testbed linking intelligence agencies and phone companies in and around Washington, D.C. Multiple papers explored the Monet project, which has now spun off a related testbed called the Optical Regional Area Network, dedicated to regional cross-connects.

Even the old hands at optical networking have new tricks up their sleeves. In a conference paper, Hewlett-Packard Labs described a SpectraLAN development for bringing four-channel WDM into the LAN arena. HP Labs research manager Brian Lemoff revealed that the company's upcoming proposals to the IEEE 802 working group for a 10-Gbit Ethernet standard will involve a physical layer using four channels of WDM at 2.5-Gbits/second each, rather than a single 10-Gbit channel.

HP has demonstrated SpectraLAN interfaces that incorporate all passive WDM components within the confines of the small-form-factor MT-RJ connector. The effort to drive WDM into the LAN will likely pick up support from LAN newcomer Unlimited Bandwidth Inc.

Lucent Technologies Inc., meanwhile, is moving on multiple fronts. Lucent's optoelectronics and optical networking groups are rolling out 2.5-Gbit integrated transponders, erbium-doped fiber amps and passive arrayed waveguides. And Lucent Microelectronics has acquired an Ottawa-based Sybarus Technologies Inc. to develop Sonet- and optical-layer IC technology for Lucent's Detroit simplified-data-layer chips, which enable packet traffic over optical layers.

Signs of maturity
With attendance topping 10,000 for the first tim, the Optical Fibers Conference reflected the optical markets' coming of age. At the technical sessions, papers from such companies as Lucent Bell Labs, NTT and Alcatel's Opto+ division described semiconductor optical-amp arrays and planar lightwave circuits that take optical-layer processing further into the network and tighter into circuit-level design than developers would have thought possible five years ago.

The evidence on the show floor, however, was that costs become paramount as systems move to a mass market. When a carrier builds a backbone system from scratch, the possibility exists for moving an advanced optical system without Sonet protection directly into communications backbones. But OEMs and component suppliers seeking to serve upgrades of existing networks— whether large enterprise LANs or global undersea WDM optical cables— must show clear cost advantages for the bandwidth gains.

Optical growth has thus been slower than expected in metropolitan networks, where traditional time-division-multiplexed Sonet rings remain more important than WDM and photonic-switched equipment. Jesus Leon, senior vice president of products and technology at Ciena Corp. said the problem is not so much a lack of adequate offerings from OEMs as the snail's pace of digital-subscriber-line (DSL) and other broadband-access-service rollouts from incumbent local-exchange carriers. Still, the uncertain state of metro WDM didn't keep Lucent Technologies from launching its WaveStar Metro product line at the show.

Brian McCann, president of Adva U.S., said the thinking behind the new Fiber Service Platform from his company is that metropolitan services cannot depend on traditional carrier-based offerings. Instead, Adva wanted to bring WDM into the enterprise using a variety of protocols, including Gigabit Ethernet, Escon and Fibre Channel.

Playing a strict cost-analysis game will make life tough for the startups at both component and system level. A company with a well-defined integrated silicon and micromachined photonics process, such as Bookham Technology Ltd. (Abingdon, U.K.), can hope for continued proof of concepts and volume production to demonstrate it can ride the cost curve for optotransceivers and WDM modules along with the big players. But in most cases, it's a tough life for startups.

Near Margalit, president of New Access, said his company tossed out an optical add-drop architecture based on tunable lasers after determining that the market needed an optical concentrator for IP-based packet systems. New Access had to demonstrate market advantages for an optical concentrator for gigabit router traffic.

At the same time, the company has had to leverage home-grown talents to challenge the large telecom OEMs. New Access has tapped the laser-component-manufacturing capabilities of its major investor, MRV, as well as borrow concepts from the MRV divisions working on packet routing.

WaveSplitter learned a similar lesson. It turned to an affiliate in Japan for fused-fiber building blocks for its WDM filter components. But the patented Mach-Zehnder hybrids did not provide enough of an edge to compete against the likes of Lucent.

Instead, the company moved upstream to offer wavelength combiner systems directly to carriers. That allowed the system-level product to be used in conjunction with existing dielectric and arrayed-waveguide WDM filters, so that WaveSplitter could even offer system-level products that leveraged its competitors' component advantages, said marketing and sales vice president Rick Eisener.

Developers have stopped looking for the holy grail of all-photonic switching and are aiming at making existing fiber networks the standard form of broadband physical infrastructure.

Michael Zadikian, president of Monterey Networks, said he is under no illusion that startups like his own or Sycamore Networks will convince even new-breed carriers to just toss out Sonet transmission equipment. Optical routers and cross-connects will be able to demonstrate clear-cut performance advantages in network restoration but still must compete on price against a standard Sonet OC-3 or OC12 multiplexer.

Margalit of New Access predicted that the new breed of optical-networking startups will have to collaborate closely with manufacturers of the new IP giga-routers to demonstrate that effective backbones for new Internet GigaPOPs will need to incorporate aspects of both architectures. New Access elected to work on Gigabit Ethernet interfaces before Sonet 2.5- or 10-Gbit interfaces because such I/O ports are more familiar to the giga-router vendors.

For laser and transceiver-module manufacturers at the conference the message was this: Keep costs to a minimum and manufacturing methods standardized. The theme transcends single- and multimode fiber types; Fabry-Perot, distributed-feedback and multi-quantum-well lasers; and SC, MT-RJ and VF-45 connectors.

Andrew Rickman, chief executive officer and founder of Bookham, said the lesson had helped focus the efforts of his company's silicon-based application-specific optoelectronic-circuit process. To show that silicon active parts with on-chip micromachined waveguides really worked, Rickman said, Bookham had to accept being stereotyped as a transceiver company until successive development of bidirectional transceivers and WDM parts proved that the ASOC process was universally applicable to lightwave components.

Siemens Components Inc. (San Jose, Calif.) is betting the world will settle for nothing less than moving all optical subassemblies to surface-mounted packages over time. The company this week launched the OptIC program, under which transceivers will move to devices that can be used in a pick-and-place systems. They will be followed by PIN-diode arrays, integrated transceivers and even laser/driver combos that will be pre-aligned before being placed in surface-mount packages.

In the small-form-factor connector war, Siemens backs VF-45 and HP MT-RJ. Both offer interfaces at the low end for virtually all multimode fiber standards.

Conrad Burke, director of marketing for optoelectronics products at Lucent, foresees a further shakeout in the lightwave-component sector. Those that survive will need to manufacturer their own semiconductor lasers.