Dispersion-shifted optical fibers adapt to network capacity and data rate changes
Serving as flexible platforms, dispersion-managed singlemode fibers permit short- and long-term migration network strategies that absorb optical technology advances
George F. Wildeman and
To incorporate emerging optical amplifier, multiplexing and transmission technologies into reliable and flexible high-capacity, long-haul links, network planners are increasingly using chromatic-dispersion-controlled optical fiber.
In fact, standard singlemode fiber, dispersion-shifted fiber and non-zero dispersion-shifted fiber are available for building new networks and upgrading existing networks to handle the higher data rate demands of 2.5 to 10 gigabits per second. The proper fiber selection for a network application relies on cost, capacity, reliability, flexibility and complexity tradeoffs.
Local exchange carriers and interexchange carriers, which already have an installed base of tens of millions of fiber-kilometers of singlemode fiber, are expected to implement the latest transmission technologies, such as wavelength-division multiplexing, to boost bandwidth capacity over existing routes. However, for new routes, dispersion-managed optical fibers offer network planners flexibility and an order of magnitude increase in bandwidth capacity.
The chromatic-dispersion property of optical fiber causes transmitted light pulses to travel at different velocities through the fiber. The net effect spreads out the pulsewidths over a long travel distance, resulting in signal degradation. Special dispersion-influencing materials are infused into the fiber during the production process to counteract this flaw.
Interactive market growth factors are driving new fiber-optic network designs. They include the large number of new companies entering the high-capacity transport business; new business opportunities in developing busy regions; the capacity of existing networks to pick up new traffic; more survivable network architectures; and construction upgrades to relieve bottlenecks in fiber-constrained and fiber-exhausted routes.
Selecting the proper optical fiber type for a new route design depends heavily on network parameters. For example, dispersion-shifted fiber or non-zero dispersion-shifted fiber proves effective in long-haul, high-capacity network applications, but might not be effective in short-haul, high-capacity network applications.
Similarly, standard singlemode fibers optimized at a 1310-nanometer wavelength are appropriate for new, short-haul local exchange carrier applications, but might not be cost-efficient for long-haul, high-capacity routes.
The long-standing debate over whether narrowband wavelength-division multiplexing should be used in telecommunications networks appears finished. In fact, two-channel, 1550 nm-based wavelength-division multiplexed technology is being installed. Presently, the debate centers on the appropriate mixture of wavelength-division and time-division multiplexing technologies required to meet network strategic needs.
Proponents of wavelength-division multiplexed networks are investigating the use of 16 or more channels on singlemode fibers. Similarly, time-division multiplexing supporters are reporting successes for 40-Gbit/sec systems on a single channel. But the short-term outlook for high-capacity networks is seen by industry analysts to involve small numbers of wavelengths (1 to 8), each transporting the basic synchronous optical network, or Sonet, building blocks of 2.4 or 10 Gbits/sec.
Prospects for higher-rate time-division multiplexing technology are favorable as next-generation equipment nears introduction. New 10-Gbit/sec systems are anticipated to be deployed within a year. According to industry analysts, the projected share of 10-Gbit/sec equipment as part of the high-rate Sonet equipment market should reach approximately 10% by 1996 and 25% by 1998. Optical fibers slated for best use in emerging 10-Gbit/sec time-division multiplexed transport equipment include conventional dispersion-shifted fibers and advanced non-zero dispersion-shifted fibers.
The introduction of higher-rate time-division multiplexing technology is predicted to face some technical challenges in the form of fiber non-linear effects, chromatic dispersion and polarization-mode dispersion, among others. However, for the near-term, the cost-effective method for accessing fiber bandwidth depends on the integration of 2.4- and 10-Gbit/sec time-division multiplexing technologies with emerging wavelength-division multiplexed components.
Wavelength-division multiplexing technology, used with 2.4- and 10-Gbit/sec channels, holds the promise of offering incremental capacity upgrades, flexible routing and a cost-effective migration path to 10 Gbits/sec and beyond on a single fiber.
The downside to this technology is network complexity in the form of multivendor equipment arrangements, additional equipment/components and intricate network switching/redundancy approaches. But this added complexity is not likely to delay the widespread use of this technology in transport networks. Advantages to network planners for several applications seem too overwhelming to ignore.
Three fiber types
Several options exist for building new routes and upgrading existing routes using 2.4- and 10-Gbit/sec transmission rates over three different fiber types. Standard singlemode fiber, dispersion-shifted fiber and non-zero dispersion-shifted fiber architectures accommodate both "power amplified" and "line amplified" networks. The best fiber choice for an application depends upon an evaluation of costs, projected capacity requirements, and network reliability and system complexity.
In network evaluation, a key technology parameter is the method used for managing fiber dispersion effects. "Dispersion management" represents several options for both extending the dispersion-limited length of time-division multiplexed systems and simultaneously avoiding the channel-limiting effects of four-wave mixing in wavelength-division multiplexed networks. The three major dispersion-management techniques include:
Dispersion-compensating fibers, which equalize the chromatic dispersion of the outside plant fiber
Externally modulated transmitters with narrow spectral widths, which reduce pulse broadening because of chromatic dispersion
Non-zero dispersion-shifted fibers, which move the zero dispersion point of the fiber outside the 1550-nm window, thereby keeping dispersion high enough to suppress four-wave mixing effects, but low enough to allow high-capacity, long-haul system operation.
The first two techniques are mainly used for network distances of less than 90 kilometers. However, planning for longer-distance and higher-data-rate singlemode fiber systems requires the introduction of dispersion-management technologies--separately or in combination.
To meet operating requirements, network planners should carefully map their strategies for network upgrading. They must consider the appropriate network topologies for managing the future mix of 2.4- and 10-Gbit/sec systems. The near-term goal to build long 2.4-Gbit/sec spans (120 to 140 km) using any of the three major fiber types has to be balanced by future plans to install 10-Gbit/sec line-amplified systems. The higher rate might have to be optimized for some applications with shorter (approximately 70-km) optical amplifier spacings.
Although singlemode and dispersion-shifted fibers are appropriate for upgrading installed plant, non-zero dispersion-shifted fiber is considered better for new-build applications.
With advances in high-speed optical communications technology shifting from 1310 nm to 1550 nm, and with the introduction of high-power erbium-doped fiber amplifiers, optical fiber systems are evolving rapidly. This evolution in system technology has, in turn, mandated changes in the design of dispersion-shifted fiber.
Dispersion-shifted fibers are designed with lo (the zero-dispersion crossing wavelength) centered near the 1550-nm wavelength to optimize transmission signal attenuation and dispersion performance. For many years, this approach was classified as the best bandwidth design point for dispersion-shifted fiber. Unfortunately, recent studies revealed that this characteristic (lo at 1550 nm) is the primary source of potential non-linear effects that appear as mixing-induced noise in wavelength-division multiplexing applications.
Further investigations have explored the limits of these fibers. These studies indicate that wavelength-division multiplexing systems are possible on conventional dispersion-shifted fibers, but undesirable system tradeoffs might be needed in the form of lower input power levels into the fiber and wider and uneven spacing of signal channels. Both tradeoffs are attempts to minimize the four-wave mixing interference, the major nonlinearity effect of concern.
Fiber design considerations
Conventional dispersion-shifted fibers differ from non-zero dispersion-shifted fibers in that lo has been moved outside the 1530- to 1560-nm signal band of erbium-doped fiber amplifiers. The four-wave mixing condition emerges where multiple wavelengths mix to produce new, unwanted wavelengths. These unwanted wavelengths can degrade system performance by interfering with, or sapping signal power from, the original wavelengths. Because of four-wave mixing, optical fiber designers concluded that new designs optimized for wavelength-division multiplexing applications were necessary. The result is non-zero dispersion-shifted fibers designed with finite levels of dispersion that mitigate the effects of four-wave mixing.
Other non-linear effects of concern to optical fiber designers include self-phase modulation, cross-phase modulation and modulation instability. Instability is caused by interaction between the transmitted signal and noise generated by the optical amplifiers that drive wavelength-division multiplexing systems. This phenomenon is driven by the polarity of the dispersion effect.
The polarity of chromatic dispersion for non-zero dispersion-shifted fibers is designated as either positive or negative, depending upon whether lo has been shifted to shorter or longer wavelengths, respectively. Research shows that when the polarity of dispersion is positive (+D), modulation instability can degrade system performance and limit system reach. However, modulation instability is not an issue for negative dispersion (-D) fibers.
The magnitude of dispersion in the optical transmission window is a tradeoff between providing too much or too little dispersion. Greater amounts of finite dispersion allow more tightly spaced, or dense, wavelength-division multiplexed channels to be used at the expense of overall 10-Gbit/sec system distances. For this reason, the latest singlemode fiber design is dispersion-optimized for 10-Gbit/sec system distance while supporting medium density (4 to 8 channel) wavelength-division multiplexed undersea and terrestrial applications.
Specifically targeted at new route designs, the new dispersion-shifted fiber designs are inherently "dispersion-managed," thereby providing long 10-Gbit/sec, multichannel system spans.
The advantages gained by introducing a small (but non-zero) amount of controlled dispersion in singlemode fibers for the purpose of suppressing four-wave mixing has been verified. Slightly higher dispersion provides a benefit by increasing the wavelength-division multiplexed capabilities of these fibers. But the dispersion is controlled at low levels to allow individual 10-Gbit/sec channels to propagate more than 300 km without the need for signal regeneration.
The non-zero dispersion-shifted fiber design allows flexible upgrade paths, reduces the number of electronic regeneration sites, thereby lessening network complexity and lowering operating costs when conventional 2.4-Gbit/sec transmission equipment is used.
As usual, a new fiber design means tradeoffs are involved. The decision to place these fibers depends upon several factors, including network capacity projections and short- and long-term network goals.
Although these non-zero dispersion-shifted fibers are optimized for use in the 1550-nm window, the cable-cutoff wavelength for terrestrial applications needs to be less than 1260 nm if singlemode operation in the 1310-nm window is desired. The 1310-nm window is characterized by high chromatic dispersion, but the window is available for operation. The 1310-nm window of non-zero dispersion-shifted fibers is not likely to be used significantly for high-volume transport, but might find use in short-distance add/drop applications.
Installation of non-zero dispersion-shifted fiber-optic cables in many North American telecommunications networks has been completed. No special handling craft-related procedures were needed. These cables proved strippable, cleavable, sliceable and compatible with optical time-domain reflectometer testing. Many installations used conventional singlemode fiber for fiber pigtails, jumpers and restoration/repair kits. u
George F. Wildeman is senior market development engineer and Curt Weinstein is market manager, telephony applications, at Corning Inc., Corning, NY.