Optical-fiber designs evolve
High-speed, multiple-channel systems have pushed the development of new fiber designs.
Kevin M. Able Corning Inc.
As demands for bandwidth increase, new types of fiber must be developed to keep pace. Non-zero dispersion-shifted fiber is the latest advance.
Twenty years after the first optical-fiber installations, fiber development shows no signs of slowing down. In fact, fiber-optic technology is experiencing a growth spurt, thanks to two simultaneous influences:
multiple-channel transmission (bund-
ling many transmission channels onto a single fiber), which was made possible by advancements in transmitter and multiplexing technology, and especially the introduction of optical amplifiers,
increasing demand for information-
carrying capacity. Factors driving this growth include everything from deregulation to an explosion in Internet use.
The new high-speed, multiple-channel systems are presenting challenges to fiber manufacturers, and development scientists are working on new fiber designs to facilitate changing transmission techniques. This article takes a look at the driving forces behind new fiber designs and examines some current advances, including dispersion-compensating fiber, non-zero dispersion-shifted fiber, and large-effective-area fiber.
Overworked workhorse
The workhorse of optical fibers since 1983 has been unshifted singlemode fiber. Optimized for operation at the 1310-nm wavelength, it accounts for an installed base of tens of millions of kilometers of fiber. Early installations operated at bit rates in the hundreds-of-megabits-per-second range. Today, system upgrades over the same fiber are approaching 10 Gbits/sec.
Standard singlemode fiber was proven to be so capable that it was believed to have unlimited bandwidth, and all that was required to increase the information-carrying capacity was a change in electronics. Several factors came together in the early 1990s to revise that view.
Historically, bandwidth requirements have doubled every 24 to 30 months (see Fig. 1). Installed optical fiber easily met those needs. However, the 1990s ushered in an era of unparalleled growth in the telecommunications industry. The lure of providing true broadband service began to blur the distinction between voice, video, and data providers. The combination of broad deregulation and interactive services began to strain the capacity of the world`s communications infrastructure.
In addition, insufficient fiber counts in early installations have led to bottlenecks for some carriers who have literally run out of fiber in critical routes. As these carriers struggle to add more capacity, new entrants into the high-capacity transport business are constructing new routes. This demand for bandwidth has taxed the capabilities of unshifted fiber at a rate that cannot be met economically merely by installing more fiber.
At about the same time, the erbium-doped optical-fiber amplifier (edfa) was introduced to increase fiber information-carrying capacity commercially. It offered a tremendous advantage over electrical-signal regenerating and amplifying methods, which require signal conversion from optical to electrical and back to optical--and must be designed for specific coding schemes and bit rates. edfas are all-optical and will amplify whatever signal is input, regardless of structure or bit rate. A broad amplification band means multiple wavelengths can be transmitted simultaneously, effectively increasing available bandwidth by factors of 8, 16, 32, or more.
Because edfas operate in the 1550-nm window, they promised to be a boon for dispersion-shifted fiber (dsf), a Corning technology first introduced in 1985. dsf combined the low attenuation inherent at longer wavelengths with a zero-dispersion wavelength matched to the source wavelength. This overcame the dispersion penalty incurred when unshifted fiber was employed in this window. With the advent of practical optical amplification in the early 1990s, dsf seemed poised at last to meet the requirements for high-data-rate, long-distance applications.
Nonlinear effects
However, the increased output power inherent in optical amplifiers, combined with the simultaneous transmission of multiple wavelengths, raised the importance of phenomena that had until then been of academic interest only. Under these new conditions, optical fiber exhibits a nonlinear response, and an entirely new set of issues arose to make the fiber--for the first time--the limiting factor to increased transmission capacity. In addition, a new vocabulary--including terms such as "self-phase modulation," "cross-phase modulation," "modulation instability," and "four-wave mixing"--was added to the industry.
One of the most troubling of the nonlinear effects is four-wave mixing. When multiple signals co-propagate, they mix to produce additional channels that can sap power from and overlap with the original signals. Figure 2 illustrates this process for three evenly spaced channels--l1, l2, and l3. The mixing components occur at lxy¥= lx + ly - lz. Because of the even spacing of the original wavelengths in this example, some of these newly generated signals occur at the original channels.
The total number of mixing components generated, m, is calculated as m = 1/2 (N3 - N2) where N is the number of original channels. For a 3-channel system, this means there are 9 additional signals to contend with. For an 8-channel system, this number increases to 224.
One obvious means of minimizing the impact of four-wave mixing is to employ uneven channel spacing. However, although this is relatively straightforward for three channels, the task becomes significantly more complicated for a 32-channel system and 15,872 potential mixing components.
The four-wave mixing process is most efficient at the zero-dispersion wavelength, in direct conflict with the need to keep fiber dispersion to a minimum to optimize transmission capability. Because standard dispersion-shifted fiber has its zero-dispersion wavelength within the operating band of edfas, these conflicting requirements place limits on the capability of dsf for high-data-rate long-haul networks using wavelength-division multiplexing (wdm). In response, a new category of optical fiber has been developed--non-zero dispersion-shifted fiber (nz-dsf).
Non-zero dispersion-shifted fiber
The concept behind nz-dsf is simple. The zero-dispersion wavelength of dsf is further moved such that it resides outside the edfa`s operating gain band, effectively re-introducing a controlled amount of dispersion into the system. This is depicted in Fig. 3, where the dispersion curves for Corning smf/ds and smf-ls fibers are compared. The resulting dispersion for the smf-ls fiber is low enough to provide for long routes, yet not so low that four-wave mixing leads easily to system impairment. Using nz-dsf, 8 x 10-Gbit/sec data rates over 360 km without compensation have been demonstrated.1
The ability to combine many data channels onto a single low-loss, low-dispersion fiber can have a significant cost benefit over using standard singlemode fiber. Regenerator/amplifier spacing can be extended, and there is no need to add equipment to compensate for dispersion in typical current systems. Estimated savings of as much as 30% to 50% can be realized on the cost of equipping the fiber.2
Dispersion compensation
A significant base of installed unshifted fiber already has been deployed. To use this fiber effectively in transmission systems employing optical amplification, a means of reducing the accumulated dispersion resulting from 1550-nm operation over long distances is necessary. To meet that need, manufacturers have introduced dispersion-compensating fiber (dcf). At 1550 nm, an unshifted singlemode fiber will have dispersion on the order of +17 ps/nm/km. Although this high dispersion eliminates four-wave mixing as a concern, the maximum transmission distance for a given data rate is limited by chromatic dispersion.
By the nature of their design, dispersion-compensating fibers have high negative dispersion. When the fibers are placed appropriately within a system link, the large negative dispersion of dcf brings the overall dispersion for the link back to nearly zero, reversing the pulse spreading that occurred as the signal propagated.
This technique has allowed the use of unshifted singlemode fiber at 10 Gbits/sec over hundreds of kilometers. A typical upgrade scenario is illustrated in Fig. 4.
Nevertheless, the need for increased bandwidth has been steady, and the capabilities of dispersion-compensated standard fiber installations, and even newer nz-dsf fiber, eventually will be strained. To meet the drive for even greater numbers of operating channels on new builds, several methods have been proposed.
One of these uses a variation of dispersion compensation. Generally referred to as "dispersion management," alternating lengths of positive- and negative-dispersion fiber are combined in a link in a planned manner. In this way, a finite local dispersion is maintained while the overall dispersion is limited to a near-zero level. Recent experiments have shown capability reaching 32 channels at 10 Gbits/sec each over 640 km.3 Although effective, this technique requires careful planning to ensure a low overall dispersion. Restoration or other unanticipated reconfiguration could impair the system unless sufficient margin is included in the design. Clearly, the most desirable option is to install a fiber that can not only accommodate today`s range of wdm systems, but also provide the flexibility for future upgrade.
nz-dsf with large effective area
By increasing the light-carrying cross section of the fiber, the path average intensity can be lowered for a given total power. The advantages to this increase in effective area include higher power-handling capability, higher signal-to-noise ratio, lower bit-error ratio, longer amplifier spacing, and most important, higher information-carrying capacity.
Typical dsfs have effective areas of approximately 50 sq microns. Large-effective-area fibers with areas as large as 92 sq microns have been reported.4 Corning recently introduced ls-leaf, a nz-dsf with large effective area, to meet the increasing demands of amplified high-data-rate systems being introduced or planned for the future.
As shown in Fig. 5, an immediate benefit to this increased effective area is a reduction in the amount of power funneled into four-wave mixing components. This implies higher power handling capabilities, and consequently longer amplifier spacing. This is further illustrated in Fig. 6, which plots optical amplifier spacing as a function of effective area.
Another advantage is the capability for future upgrade to very-high-bit- rate dense wdm. Systems capable of 10 Gbits/sec per channel over multiple channels are now being deployed in commercial volumes. nz-dsf large-effective-area fiber will provide the platform capable of handling future 40-Gbit/sec-per-channel information rates over the entire edfa gain band. Both of these benefits--reduced amplifier counts and increased spacing--and the ability to easily upgrade to higher data rates translate directly into reduced installation and operating costs.
Summary
Deregulation, competition, and expanded services have played a part in the hunger for bandwidth. A continued increase in this requirement has been responsible for the development of cost-effective technologies to meet that demand. Foremost among these advancements has been the successful simultaneous deployment of optical amplifiers and wdm. Moreover, the move to full use of the 1550-nm operating window has spurred activity to introduce other enabling technologies. Dispersion-compensating fiber has facilitated a means of using the vast installed base of standard singlemode fiber at bit rates of 10 Gbits/sec per channel over distances of several hundred kilometers. However, the development of nz-dsf with large effective area for new builds provides significant advantages in system performance and cost.
Next-generation fibers must also be capable of minimizing the impact of nonlinear effects and accommodating the transport of hundreds of information channels. Fibers with large effective area will fill this need by providing a platform that encompasses the flexibility to accommodate a range of wavelength plans, information-carrying capacity for now and into the future, and cost-effectiveness. u
References
1. V. da Silva et al., "Error free 8 x 10 Gb/s wdm transmission over 360 km of non-zero dispersion-shifted fiber without dispersion management," Optical Fiber Communication Conference, Dallas, TX (Feb. 1996), post-deadline paper.
2. P. Palumbo, "Bandwidth needs spur fiber diversity," Lightwave, Nashua, NH: PennWell Publishing Co. (Nov. 1996), p. 1.
3. A. Srivastava et al., "32 x 10 Gb/s wdm transmission over 640 km using broadband, gain-flattened erbium-doped silica fiber amplifiers," Optical Fiber Communication Conference, Dallas, TX (Feb. 16-21, 1997), post-deadline paper.
4. Y. Liu, "Dispersion-shifted large-effective-area fiber for amplified high-capacity long-distance systems," Proceedings of the Optical Fiber Communication Conference, Dallas, TX (Feb. 16-21, 1997).
Kevin M. Able is applications engineering project leader in the Telecommunications Products Div. of Corning Inc., Corning, NY.