All-optical and OEO switching will unleash real network capacity

ROBERT COOPER

ALTHOUGH ALL-OPTICAL SWITCHING WILL NOT TAKE OFF IMMEDIATELY, BY 2005 THE OPTICAL CROSSCONNECT MARKET WILL GROW TO $7 BILLION AND THE OPTICAL SWITCH MARKET WILL HAVE GROWN TO $3 BILLION.

Currently, optical network traffic must be converted back to electrical mode at each network intersection so it can be directed to the next node. This conversion from optical to electrical to optical (OEO) mode and the equipment facilitating it are costly. Depending upon the source, some observers say as much as 70% savings over current practice will be realized with the implementation of new optical networking systems.

We have identified more than 141 venture-capital firms that have invested almost $3 billion in startup all-optical switching companies in recent months (see Fig. 1). Additionally, several large and well-known suppliers and systems integrators have spent almost $30 billion on acquisitions of promising young companies that develop optical-switching products, such as Cisco Systems purchased Monterey; Nortel Networks purchased Xros and Qtera; JDS Uniphase purchased E-Tek Dynamics; Sycamore Networks purchased Sirocco; and Corning purchased Willow Systems, US Precision Lenses, Rochester Photonics, and Intellisense.

PAST AND FUTURE
Dozens of telecom systems companies and suppliers continue to offer OEO systems while keeping an eye on and supporting pure optical-switching technology developments. Most of the technologies adopted by promising candidates come from the integrated circuit (IC) industry. Planar lightwave circuits (PLC), microelectromechanical systems (MEMS), ink-jet bubble technology, liquid-crystal systems, electroholography, and thermoelectric techniques are some of the technologies currently under development in more than 35 startup companies and in the labs of larger established firms (see Fig. 2).

These IC-based systems bring mass production, repeatable quality, and lower manufacturing costs than current practice. Switches and crossconnects based on these technologies will perform transparent switching in which traffic stays in the optical form all the way through the network backbone and down into the metro. These systems will require very complex management software, which will replace current SONET/SDH management and restoration capabilities if the electrical mode is eliminated at the intersections.

ALL-OPTICAL TECHNOLOGIES
Planar lightwave circuits take advantage of IC practice in that layers of material are deposited and etched to create channels for either diverting or passing photons. The wall material of the channels can be reflective on command but there are no moving parts. Azanda, Kymata, Lightwave Microsystems, Lynx Photonics, Nanovation, Network Photonics, OptXcon, and Optical Switch Corp. are some of the startups developing PLC technology.

Microelectromechanical systems, as they apply to optical switching, are based upon IC practices that result in a movable reflective surface or mirror, the angle of which can be changed by the application of electrical power or thermal change. The optical wavelength is directed at the reflective surface, which, upon command, permits the photons to pass, or diverts them to another exit. Astarte, C-Speed, Calient, IMMI, OMM, K2 Optronics, Luxcore, and Onix Microsystems are some of the MEMS-based firms.

Ink-jet bubble systems are also IC-based, with the addition that a microscopic amount of a liquid is placed at each intersection of etched channels. With the onset of an electrical pulse the liquid is instantly heated, creating a bubble that is reflective and diverts the photons to another exit. Agilent and Alcatel are pioneering this technology.

Liquid-crystal systems are also IC-based. Polymeric materials are suspended in special liquids. The materials change their alignment upon the addition of electrical power—either permitting light to pass through or diverting it. Chorum and Spectra-Switch are two of the leading liquid-crystal developers.

Electroholography is based upon special microcrystals that can have a hologram stored in them. The hologram is of such a nature that it allows photons to pass through when it is in the 'off' position and is reflective when in the 'on' position, thereby diverting the light upon command. Trellis is currently developing this technology.

COMBINING TECHNOLOGIES
The business plan of Tellium, a MEMS- and OEO-technology-based startup that recently registered its stock in anticipation of an IPO, includes immediate sales of OEO switching and crossconnect products for all network-switching applications up to 1000 x 1000 port systems. Tellium believes that above that size the OEO technology begins to fall behind the efficiency of the new IC-based, pure optical switches. Like many other startups, Tellium believes the two-dimensional (2-D) MEMS technology will be the solution for crossconnects larger than 1000 x 1000 ports.

Electroholography and 2-D MEMS technology hold promise to satisfy large port-count systems because they appear to efficiently scale to these larger sizes. A single 2-D MEMS mirror will operate on two planes much as an old ship's compass card swings on two planes. Each lightwave passing through this crossconnect will need only two mirrors or 2N mirrors. It appears the same will be true for the electroholographic system as well. All other technologies studied so far will need N2 switches to perform the same tasks. At large port counts these other technologies become inefficient and unwieldy.

However, these other IC-based technologies will find their place in the network where lower port counts are required, such as for restoration. Some network designers estimate there will be a need for two 1 x 2 optical switches for each fiber between every optical intersection in a network.

IF NOT NOW, WHEN?
The question at hand is when will these systems begin to proliferate in the network? Corvis claims they have recently received revenue from the sale of an all-optical system, but they are talking only to customers about the technology at the core of the system. Others such as Tellium and Sycamore report revenues from OEO switching systems.

Smaller port-count, all-optical switches and crossconnects will first begin to appear in the backbone of the network during 2001, according to carriers. Some switches and smaller crossconnects also will appear in the metro networks this year. But the larger switches and crossconnects based upon MEMS and electroholographic systems will be the primary target market and will be approximately three years out.

If MEMS devices improve and are utilized as expected and the manufacturing techniques are in place, economies of volume for medium and large devices will reduce the price so they may be economical in the more price-sensitive metro market. All-optical systems will then proliferate throughout the networks. When widespread installation takes place, the potential provided by DWDM will be fully unleashed.

After a rather slow start, by 2005 the optical crossconnect market will be $7 billion and the optical switch market will be $3 billion.

Robert Cooper is a consultant to KMI Corp., America's Cup Avenue at 31 Bridge St., Newport, RI 02840. He can be reached at [email protected].