Optical switching evolves beyond device market

Switching is the key enabler of the next generation of multiwavelength optical network deployments worldwide. As optical networks based on DWDM move from point-to-point and ring deployments to more-dynamic configurable mesh networks, optical switches will perform many of the tasks now done by complex electrical systems. The market should expand from $428 million in 2000 to $10.3 billion in 2004.

SCOTT CLAVENNA

The term optical switch has long been applied to fairly simple optoelectronic components used to switch optical signals from one light path to another. Quite common today, optical switches are used in a variety of applications and integrated into a number of different optical network elements. A typical application is automatic protection switching, frequently used for restoration in the event of fiber breaks or link failure.

Switch devices are also used for network monitoring, in which a 1 × N switch is used to interconnect multiple fibers to an optical time-domain reflectometer (OTDR) for monitoring at remote fiber test sites. In addition, switches are often used in optical add-drop multiplexers (OADMs), selectively switching out or adding in individual wavelength channels from a multiwavelength signal.

The most important use of optical switching lies well beyond these applications, at the core of large-scale optical networks. Here, high-density optical switches are required to perform the tasks of switching, routing, and crossconnect at major network hubs, where many fiber trunks, each carrying multiple optical channels, may terminate.

This switching function is most often performed today by SONET/SDH equipment, which is inefficient in its handling of high-bit-rate signals and is also quite costly. Given that many of the optical channels passing through these hubs are data signals, voice-oriented SONET gear adds unnecessary levels of complexity, greatly slowing the provisioning on optical carrier circuits throughout a network while increasing the likelihood of failures.

The answer, therefore, is intelligent optical switching. Systems are being developed today that perform switching in the optical layer of networks, rather than the SONET/SDH layer. These switches allow networks to scale in capacity much more elegantly than previously possible. Coupled with intelligent software platforms, they will provide network operators a means of provisioning and managing optical circuits across entire networks.

Though the size of this optical-switching system market is quite small at this time-less than $10 million in 1999-the market will expand rapidly as carriers rush to simplify their complex optical networks. Systems will be introduced in 2000 from vendors such as Ciena (Linthicum, MD), Cisco Systems (San Jose, CA), Sycamore Networks (Chelmsford, MA), and Tellium (Oceanport, NJ).

The optical switching systems market in North America will expand from $427.8 million in 2000 to more than $10.3 billion in 2004 (see Fig. 1).

Stumbling blocks

The limiting factor thus far in the optical-switching systems` market growth has been the switch matrix. Network operators have told switch developers that these intelligent optical-switching systems must support at least 64 × 64 optical switching (2.5 Gbit/s signals or greater) in their first release and provide a path to 1024 × 1024 switching over the next two to three years. Analysis of bandwidth-demand growth and capacity requirements of national and regional fiber networks bears this out (see Fig. 2).

Many carriers have already identified areas within their networks that could support 1024 × 1024 switches, indicating a long-term market opportunity for companies that can bring these systems to market with the required reliability and functionality. Today, optical switching systems most often use electrical switch matrices, meaning incoming optical signals are converted to an equivalent-sized electrical signal, switched, then converted back to an optical signal for placement on a DWDM link. Electrical switches have three advantages over all-optical switches: scale, reliability, and availability.

In terms of scale, all-optical switches have not yet been produced that meet the switching densities required for core optical networks. For reliability, electrically based switches are carrier-grade and ready for production and integration into existing operating systems, while most optical switches remain in the laboratory development phase.

Optical-switch components produced today using traditional methods and materials cannot be reliably manufactured at densities over 8 × 8 or 16 × 16. The front-runner among optical-switching options today is microelectromechanical systems (MEMS), which benefit from development for a variety of applications, including imaging, automotive, industrial, acoustics, and chemistry.

In addition, MEMS use IC fabrication and manufacturing techniques, which provide a platform for low-cost production and a high degree of potential integration of analog and digital microelectronic devices on the same chip The ability to use MEMS technology to accommodate a 256 × 256 switch matrix on a one-inch-square chip has already been demonstrated by Lucent Technologies (Murray Hill, NJ).

Many companies that had been developing MEMS technology for other industries are now looking to the telecommunications industry. MEMS-based optical switches are currently being developed by Cronos Integrated Microsystems (Research Triangle Park, NC), Lucent Technologies, Optical Micro-Machines (San Diego, CA), Texas Instruments (Dallas, TX), and Xros (Sunnyvale, CA), to name a few.

Pushing GROWTH

There are several competitors to MEMS for the creation of large-scale optical-switch matrices. One is liquid-crystal switches, which are being developed by Corning Incorporated (Corning, NY), SpectraSwitch (Santa Rosa, CA), and Chorum Technologies (Richardson, TX). Others are frustrated total internal reflection (FTIR) switches, produced by Optical Switch Corp. (Richardson, TX), and thermo-optic switches being produced by companies such as Photonic Integration Research (Columbus, OH). The momentum today is behind MEMS-based switches, but the projected size of the optical-switching market and its rapid expansion over the next decade will drive developments in technology and commercial introduction of many varieties of optical-switching matrices.

The optical add-drop multiplexer is one application that is already becoming common in both long-haul and metropolitan DWDM networks and will benefit from the dynamic and reconfigurable capabilities that these new smaller-scale optical switches will provide. The expansion of the optical-networks market, therefore, will create an commensurate expansion of the optical-switch market, both in size and diversity. At the dawn of this optical networks revolution, opportunities for existing optical-switch manufacturers and startups are plentiful and should create a robust, competitive market for the foreseeable future.

FIGURE 2. Demand for optical-switching density at backbone nodes in carrier networks will rise quickly by 2004.