Higher bandwidth demands lower fiber-optic component costs
Low-cost optoelectronics will enable subscribers to meet future bandwidth demands through fiber-optic connections.
Jeff D. Montgomery
Tens of millions of subscribers who now have copper telephone wires and coaxial cable-television connections will demand higher bandwidth capabilities in the future. This bandwidth will be available on fiber-optic networks at little increase in cost over current copper connections, making the switch to fiber an attractive choice. A major reason is that enabling fiber-optic components have advanced rapidly in performance and functional cost, lowering the cost of equipment and ultimately subscriber services.
Why more bandwidth?
The demand for increased bandwidth (assuming that cost is reasonable) will arise from a host of emerging applications:
Business: Internet and intranet video, voice, and data; massive, global, frequent data transfers; videoconferencing; and video-interconnected, geographically separated expert workgroups.
Residential: HDTV; 3-D, advanced-capability receivers; video phones; distance-learning; Internet video; and video-on-demand.
Many of these applications have been envisioned for more than a decade. But broad commercial acceptance, in most cases, has not been realized as rapidly as many had forecast.
Applications such as the Internet and intranets have already taken off, how ever, leading subscribers to demand higher capacity than is available from voice-grade copper lines. Return on investments--or profit--is the key driver behind this accelerating use of broadband services. Other major drivers propelling the demand for more communications bandwidth include new government regulations, technology advancement, and available capital. Without the return on investment, available capital would dry up, and the other drivers could not prevail. All of these forces are very strong, especially in North America.
A proliferation of broadband access providers has coincided with the rapid expansion of broadband service offerings. This situation has correlated with an explosive growth of subscriber connections and an expansion of Internet-based services from new competition: Internet service providers.
The shift toward data communications has led to an evolution in routers and other data-network equipment, at the expense of traditional telecommunications equipment. This trend will become much more significant over the next decade.
Bandwidth benefit worth the cost?
Global demand from both business and residential subscribers for greater communications bandwidth will continue only if the benefits of increased capa city outweigh the costs. Thus, greater de ployment of bandwidth de pends not only on the services that could become available, but also on their cost.
For the cost/performance ratio to be attractive, the cost of bandwidth-per-megabit must be greatly reduced from its current level. It is useful to realize, however, that the cost-per-megabit of fiber-delivered communication already has fallen exponentially over the past decade. This reduction has made possible, for example, the drop in charges for transcontinental telephone communication from dollar-a-minute levels to dime-a-minute levels.
It is also useful to note that a large share of the total cost of providing local, regional, and global communications links is related to right-of-way costs plus the labor-related costs of installing cable, rather than the cost of the cable and the related signal-processing components. These right-of-way and installation costs are comparable for twisted-pair copper, coaxial copper, and fiber cables on a per-mile basis. On the basis of amount of data transported, how ever, fiber costs much less.
Over the next decade, the cost of singlemode-fiber cable on a per-fiber-km basis is expected to drop slightly, while the cost-per-mile of twisted-pair and coaxial cable is expected to increase moderately. The amount of data transported by the fiber cable, however, will increase by more than 100 times over the next decade, while there will be little change in the throughput of the copper cable. As the number of subscriber fiber connections explodes to hundreds of thousands per central office over the next decade, reduction of the cost and size of optical transmitters and receivers will be key elements of total communication expansion. The transition to optoelectronic integrated circuits (OEICs) will be a major factor in this advancement.
The typical processing of a subscriber`s optical signal through the local- exchange central office is illustrated for a single wavelength by Figure 1.
Costs will continue to drop
By the mid-1980s, after fiber communications had been broadly accepted for long-haul transport and numerous long-haul links had been installed, the cost of a 140-Mbit/sec transmitter/ receiver pair with its full complement of electronics was about $10,000. The trend toward higher volume production, as well as continuing technical advances, dropped the cost below $1000 for a 155-Mbit/sec Synchronous Optical Network (SONET) transceiver by 1990 to 1991. With continuing volume growth and production technique automation, these transceivers are now in the $200-to-$400 range.
The transition to integration, and the related automated assembly and test of these transceivers, will drive their price below $100 each (for production lots of 100,000) by 2001. One benefit of this trend is that the 622-Mbit/sec tran sceiver will be available at ap prox imately the same price. By 2008, the average cost of these 155/622-Mbit/sec transceivers will drop below $50 with a proliferation of multiple transceivers integrated within a single module. The 2.5-Gbit/sec transceivers will follow the 155/622-Mbit/sec transceivers into hybrid OEIC format, starting in 2002; the cost of these components will drop precipitously within 10 years.
Meanwhile, the global production of OEICs will accelerate from $61 million in 1998 to $3.33 billion in 2007. This production currently is nearly all hybrid, but monolithic ICs will expand their production share through 2007 (see Fig. 2).
The key element in the evolution of dense wavelength-division multiplexing (DWDM) technology and deployment was the achievement of the long-life pump diode, which initially was aimed at pumping high-power solid-state and gas lasers. The pump diode made the optical-fiber amplifier (which had been conceived many years earlier) operationally feasible. The remaining stumbling block--high cost of the early production units--was wiped out when that cost was compared to that of the regenerators used in undersea fiber cables. Kokusai Denshin Denwa-AT&T Submarine Cable Consortium took advantage of the opportunity to develop, life-test, produce, and deploy substantial quantities of high reliability optical-fiber amplifiers (OFAs), which in turn, led to terrestrial, long-haul network use. Finally, DWDM became economically attractive, which explains the cascading of capabilities and deployment we now see.
The most expensive component of the OFA is the pump diode. Early OFAs used one pump diode, but, as DWDM moves toward an ever-increasing wavelength count--and each wavelength must have enough power to permit its detection after tens of kilometers transit--two or more pump diodes are needed to achieve the needed OFA power output. Different-wavelength pump diodes also are increasingly being combined in OFAs to optimize power and noise and to gain curve characteristics. Figure 3 presents cost element trends between 1998 and 2008 for a fixed-performance OFA.
Emerging OADM market
In 1998, optical add/drop multiplexing (OADM) was in an early stage of introduction and production. The global consumption of OADMs in 1998 (excluding electronic add/drop multiplexing) is estimated at $23 million. But this will expand dynamically to $1.64 billion by 2003. Very impressive growth will continue, averaging 34% per year, to reach $7.2 billion in 2008. The leading share of this consumption will be by North American fiber networks, holding a 71% share into 2003, but dropping to 64% or $4.61 billion by 2008 (see Fig. 4).
An accelerating demand for higher bandwidth combined with the decreasing cost of fiber-optic components and services will cause a significant number of copper-cable users to switch to optical communications over the next decade. u
Jeff D. Montgomery is chairman and Stephen Montgomery is president of ElectroniCast Corp. (San Mateo, CA). They can be reached at (650) 343-1398 or www.elec tronicast.com.