Nearly 30 years ago, the first optical fibers were multimode and produced in limited quantities. The first deployment of single-mode fibers in 1983 marked a revolution in the telecommunications industry. These fibers were designed for operation in the 1310-nm window and, over time, evolved into today's standard single-mode fiber, also known as ITU G.652-compliant fiber.
While an industry workhorse at 1310 nm, fiber and transmission scientists appreciated the nearly 40% attenuation reduction available by moving the transmission window to the low-loss 1550-nm region of silica. In 1985, this opportunity was addressed by a fiber with low chromatic dispersion at 1550 nm: dispersion-shifted fiber (DSF or ITU G.653).
The introduction of erbium-doped fiber amplifiers (EDFAs) provided the ability to amplify light signals in the 1550-nm region where silica fiber has its lowest attenuation, enabling the extension of electronic regeneration from approximately every 40 km to approximately 500 km. The development of EDFAs in addition to advances in optical-filter technology provided the capability to perform wavelength- division multiplexing (WDM).
Since adjacent channels interact and create nonlinear noise such as four-wave mixing (FWM) in a near-zero-dispersion fiber, the move to WDM posed a challenge for dispersion-shifted fiber. The advent of WDM systems created the need for new fiber types that considered the impact of dispersion on nonlinear interchannel crosstalk penalties in the 1550-nm window. Fiber designers shifted the zero-dispersion point out of the operating band, ensuring sufficient chromatic dispersion would be present to prevent nonlinear penalties. Thus, nonzero-dispersion-shifted fiber (NZDSF or ITU G.655) was invented (see Fig. 1).Current metropolitan optical networks consist of a mix of rings, point-to-point, star, and mesh architectures. Additionally, the variety of signal transport protocols is even more diverse: SONET/SDH, Fiber Channel, ATM, FICON, Gigabit Ethernet and 10 Gigabit Ethernet, to name a few. In addition to system loss, the overriding performance of this wide-ranging suite of old and new technologies is governed by the interaction between laser spectral width and chromatic dispersion.
Driven by the dominance and low cost of 1310-nm lasers, the most widely deployed fiber in metropolitan networks has been standard single-mode fiber. While these two technologies have coexisted well, laser technology is evolving. In the 1310-nm region, lasers with very narrow spectrums such as distributed-feedback (DFB) lasers are providing linewidths on the order of 0.2 nm and are becoming more competitive with higher linewidth, multilongitudinal Fabry-Perot devices. In addition, vertical-cavity surface-emitting lasers (VCSELs) have narrow linewidths and are cost competitive with Fabry-Perot lasers.
In addition to the 1310-nm linewidth trajectory, there are a number of other technologies enabling metropolitan networks to mirror the steep technological curve experienced by the long-haul market in the late 1990s. As the bandwidth bottleneck at the edge of metro networks is opened up by network deployments, bandwidth requirements in metropolitan networks are increasing significantly.
To fulfill these increased bandwidth needs operators will look toward the 1550-nm region, which—with its faster laser technologies, wideband transmission, and optical amplification—will facilitate the high-data-rate, multiple-channel transmission systems required to meet such bandwidth demands. The confluence of technologies pointing to 1550-nm centric networks is very compelling: lowest loss, optical amplification, and the fastest available lasers. It is no surprise that next-generation SONET/SDH will operate in this window.
Optical fiber attenuation at 1550 nm necessitates that signals be amplified at intervals of approximately every 80 to 100 km along the optical path. Some metro-regional networks extend beyond these distances and as a result many metro equipment vendors offer options for erbium-based amplifiers to extend system reach. However, some of the low-cost 1550-nm laser devices used in the metro systems, such as directly modulated lasers (DMLs) have a restricted dispersion-limited reach when compared to more expensive externally modulated lasers. High local fiber dispersion at 1550 nm combined with low-cost DML transmission systems may require additional costly regeneration cards or use of cost-added dispersion-compensation modules.
A long list of emerging technologies are intended to reduce cost in current long-haul networks: new and super-effective forward-error-correction coding, unique modulation formats such as carrier-suppressed return-to-zero and differential phase-shift keying, polarization-mode-dispersion (PMD)-compensation devices, optical crossconnects, distributed Raman amplification techniques, and dynamic dispersion-compensation modules. Some of these technologies—such as polarization-mode dispersion compensation and distributed Raman amplification—improve optical performance, while others such as optical crossconnects push existing fiber types closer to the edge of practical capacity or distance.
In recent years, the market has demanded optimization of transmission in the conventional and long bands (C- and L-bands). This demand has led to development of fibers designed with specific capabilities such as higher-power handling and optimized dispersion values.
Dispersion mapping and dispersion-compensation accuracy is one area in which new fibers and dispersion-slope-compensation modules can deliver significant value in long-haul networks. Since dispersion tolerance decreases as the square of the bit rate, dispersion-matched compensation devices must be better than ever. For example, if the receiver's dispersion tolerance is 10,000 ps/nm at 2.5 Gbit/s, the allowable dispersion is under 40 ps/nm at 40 Gbit/s, assuming an externally modulated source.
Another key enabler for long-haul networks is Raman pumping. This rather simple technology turns the fiber itself into a distributed amplification medium by exploiting one of the well known nonlinear effects, stimulated Raman scattering. By transmitting laser energy in the 14xx-nm region, a gain spectrum is developed over a portion of the fiber span at approximately 100 nm higher in wavelength (15xx nm). The transmission advantages realized from Raman pumping are largely due to lower amplifier noise figures.
Optical-fiber design involves tradeoffs between the most desired attributes. For example, dispersion, dispersion slope, and effective area are inherently linked—optimizing one of these attributes will have a negative effect on the others and vice versa. This limits the designer's ability to maximize each attribute. Sometimes, though, a number of design parameters line up such that each is both beneficial and attainable.
An important current trend is to move toward fiber designs that support a multitude of launch powers, amplification techniques, modulation formats, channel spacing, and bit rates across the entire usable transmission window. Next-generation fibers must consider critical attributes such as attenuation and dispersion in multiple transmission windows. These critical parameters will significantly impact the cost and performance of the network.
For next-generation NZDSF, the flattened dispersion profile allows for low dispersion at 1310 nm for single-channel applications in metropolitan networks. In addition, moderate dispersion at 1550 nm enables DWDM systems in metro, regional, and long-haul networks for both 10- and 40-Gbit/s transmission. Next-generation fibers intended for high-bit-rate applications will also require co-designed dispersion-slope-compensation modules to minimize penalty inducing residual dispersion, which will be especially critical at 40-Gbit/s line rates (see Fig. 2).
The dispersion-limited reach is inversely proportional to the fiber dispersion and the laser linewidth. Therefore, near-zero fiber dispersion at 1310 nm facilitates greater flexibility on the linewidth of the chosen laser device at 1310 nm. However, lower linewidth devices are becoming commercially available at competitive prices. As a result, optical fiber with higher local dispersion at 1310 nm may provide the capability for sections of metropolitan networks to be operated at 1310 nm.
For high-bit-rate signals, the dominant dispersion-induced penalty is the single-channel impairment arising from overlap of optical pulses at the same wavelength. Because higher chromatic dispersion causes these pulses to overlap more, keeping the fiber-dispersion low for high-bit-rate systems is better. For terrestrial telecommunication distances, a range between 4 and 8 ps/nm.km at 1550 nm for optical fiber will enable larger distances without the use of dispersion compensation. Values closer to the lower edge of this limit will provide longer uncompensated reach to satisfy regional networks.
In addition to considering requirements for 1310- and 1550-nm transmission, next-generation nonzero-dispersion-shifted fiber should be equipped with low 1383-nm attenuation. Because this lower-water-peak feature also lowers the attenuation in the 14xx region, Raman pump energy travels further into the fiber and greatly increases the benefits of Raman amplification. In addition to increased Raman performance, a low water peak will ensure operators are prepared for future wide-band, high-capacity technologies (see Fig. 3).
Current transmission fibers require a separate dispersion-slope-compensation module for each transmission band (C- and L-band) to achieve acceptable levels of slope and dispersion matching. By co-designing the dispersion-compensation fiber in conjunction with the transmission fiber, the resulting residual dispersion can be closely controlled or matched. This close match is attractive to system vendors who normally accommodate worst-case dispersion-match penalties in their design rules. The regained margin can be redistributed in the system design as distance, engineering margin, or less-expensive components.
The rate of capital expenditure for building optical fiber networks has slowed considerably, although future optical network technologies must continue to anticipate bandwidth growth. New and improved fibers are on the technology horizon and will offer service providers greater flexibility across both long-haul and metropolitan networks.
Joel Orban is product line manager at Corning Optical Fiber, Alan Dowdell is a telecommunications consultant, and Merrion Edwards is systems engineering manager, at Corning Optical Fiber, Europe. Joel Orban can be reached at [email protected].