Optical: Undisputed King Of High-Speed Data Transmission

April 29, 2002
New components and techniques push aggregate speed per fiber above 1-Tbit/s rates.

As telephone and computer networking have evolved over the years, they have steadily moved toward an all-fiber optical physical layer. Today, virtually all long-distance telephone service is by fiber. The Internet's development and growth have boosted optical transport into the stratosphere. Optical dominates simply because nothing else provides the speed and overall data capacity at a reasonable cost.

But there's a downside to too much of a good thing. With the past decade's over-building in the long-haul backbone, we now have a bandwidth glut.

Due to this overcapacity and the recent economic downturn, the optical industry has been one of the hardest hit financially. But it also has experienced threefold revenue growth over the past five years. Thousands of startups have emerged during this period, each seeking venture funding and a position in the market. Thanks to consolidation through mergers and acquisitions during this downturn, most of their new optical-transport killer technologies will survive in some form.

These new developments let fiber optical systems quickly move into the LAN and MAN space previously occupied only by twisted pair and coax. Projections for market growth in North American optical transport equipment show how optical is progressing even during bad times (Fig. 1).

Legacy Systems Still Rule And Benefit: Synchronous optical network (Sonet) and asynchronous transfer mode (ATM) are the traditional legacy architectures still carrying the bulk of the data load worldwide. The most significant change in these systems has been their steadily increasing speed, thanks to new lasers and digital chips. Newer 2.5-Gbit/s (OC-48) systems have replaced older 155- and 622-Mbit/s (OC-3 and OC-12) systems. Today, carriers are gradually upgrading to OC-192 10-Gbit/s systems.

Semiconductor manufacturers are finally ready to release OC-768 systems that run at 40 Gbits/s. Figure 2 shows how the upgrading of the Sonet long-haul networks finally converges with the steady upgrading of Ethernet networks at the 10-Gbit/s level. The rate of speed increase for Ethernet and Sonet has been significantly faster than Moore's Law, forcing semiconductor makers to find new materials and processes to catch up with the technology's potential. We're finally there!

For years, many expected fully packetized Internet Protocol (IP) systems to eventually replace existing long-distance and Internet backbone systems. But that just hasn't happened, de-spite the existence of the new IPv6 protocol and killer technologies to facilitate it. One reason is be-cause the massive investment in the entrenched telecom infrastructure won't be re-placed overnight.

Carriers will move in that direction, but meanwhile, they'll rely on upgrades to keep their systems viable as they attempt to expand their services and make a profit. In a flash, designers will adopt any new technology that helps squeeze costs out of the system while improving performance.

Currently, we have more than enough long-haul and backbone fiber for the immediate future, but there are bottlenecks in the metropolitan and local access networks. Information can travel very fast over long distances, but locally it's restricted to slow or nonexistent metropolitan-area networks. This is gradually being solved because the metro network remains the hottest, fastest growing segment of the optical-fiber space, even during the downturn. Slowly but surely, new metro networks are being put into place, again thanks to the new optical technologies that make high speeds available at reasonable costs.

Although many metro networks are being built with Sonet to maintain compatibility with long-haul networks, newer technologies are being adopted. Optical versions of 1-Gbit Ethernet (1 GE) are already finding their place in the metro space. The forthcoming 10 GE will find application in the metro regions and even in new long-haul WANs. Fresh technologies like mesh metro networks and resilient packet-ring architectures promise faster and more flexible packet MANs with quality-of-service (QoS) guarantees.

To completely benefit from the present awesome long-haul capacity and forthcoming metro capabilities, the first/last mile of access must soon be replaced or upgraded. Full convergence of voice, data, and multimedia with all hot new applications won't happen until everyone can access a broadband connection.

Recently, the telecom industry lobbied the U.S. Congress and the President for regulations to facilitate and expedite a broadband rollout. The killer technology to solve the problem—inexpensive fiber to the home (FTTH)—already exists. For example, 1 GE is the near-perfect technology for home Internet access. Even fiber to the curb and twisted pair to the house will give us a bandwidth far beyond DSL or even cable.

DWDM Shines Brightly: The impressive speed increases over the years have greatly expanded the capacity of fiber optical networks. Coupled with that is a technology that further multiplies existing fiber's data capacity. Known as dense wavelength-division multiplexing (DWDM), this technique has already lit an enormous amount of dark fiber with a rainbow of IR light, each wavelength representing a separate channel of high-speed data.

DWDM has given new life to older fiber and enabled significant increases in total bandwidth and transmission capacity. Today, it's possible to transmit over 160 wavelengths of light on a single fiber, each carrying a data rate of 10 Gbits/s. That's terabits per fiber. Plus, 40-Gbit/s OC-768 systems are coming. With OC-3072 next on the horizon, can petabits per second on a fiber be far behind?

A mix of technological breakthroughs like multiwavelength tunable lasers and narrow-band optical filters made DWDM possible. New optical splitters, filters, and array waveguides (AWGs) permit channel spacing of 12.5 GHz now, to go to 25 GHz later, bringing even higher capacity.

Tunable lasers give DWDM systems more flexibility and lower cost. They also enable the design of new network architectures that incorporate dynamic wavelength provisioning, addressing, and routing. And, new vertical-cavity surface-emitting lasers (VCSELs) are making short-haul LAN or metro systems cheaper and faster.

DWDM also offers a potential solution to the multiprotocol management problem, especially in the metro space. By isolating each protocol to a different wavelength and employing wavelength switching, this mix of protocols can be managed quite readily.

In addition to tunable lasers and VCSELs, there are hundreds of new, improved optical parts like isolators, power dividers, attenuators, filters, dispersion compensators, and modulators. Ad-vances in Raman amplifiers and erbium-doped fiber amplifiers (EDFAs) are be-ginning to eliminate the need for costly optical-to-electrical-to optical (OEO) conversions. New optical switches, including microelectromechanical system (MEMS) devices, provide speed as well as protocol-independent switching to make networks more flexible.

All of these new and affordable parts are gradually bringing about two more killer technologies—integrated optics and the all-optical network (AON). Companies are finding ways to fabricate optical components using standard silicon wafers and processing technology. This has led to large-scale integration of optical components. One recent example is a full-optical add-drop multiplexer on a chip.

Soon we'll see the AON and passive optical networks (PONs), which both eschew electronics and lower costs, eventually leading to new optical networking applications. Metro networks are a target for AONs, whereas PONs are aimed at the first/last-mile access with FFTH or fiber to the curb (FTTC).


To join the conversation, and become an exclusive member of Electronic Design, create an account today!