Managing chromatic dispersion increases bandwidth
Three significant trends are evident in long-haul transmission system development as the industry makes the transition to advanced optical networks. First, there are continual increases in the number of dense wavelength-division-multiplexed (DWDM) channels being transported over a single fiber both in the C band (1535 to 1565 nm) and the L band (1570 to 1630 nm). Second, data rates are increasing rapidly from 2.5 Gbit/s to today`s 10 Gbit/s, then to tomorrow`s 40 Gbit/s and beyond. Finally, distances between electrical regeneration sites are increasing from approximately 400 km to more than 2000 km.
As these trends have become key enablers to increasing optical network capacity, chromatic dispersion has become a major inhibitor to each of them. This is because each of the trends increases sensitivity of transmission systems to the effects of chromatic dispersion-the spreading of optical pulses as they travel through the transport fiber.
With the number of DWDM channels increasing across the C and L bands, the problem of chromatic dispersion becomes more acute. The amount of dispersion experienced by each narrow wavelength band, or channel, is unique, creating dispersion slope across the transmission band (see Fig. 1). The extent of this effect grows with the increase in the number of channels transmitted per fiber, which makes it increasingly difficult to properly correct dispersion for all channels simultaneously.
The problem created by dispersion slope is further complicated because the effect varies by fiber type. Today`s transmission fiber of choice, nonzero-dispersion-shifted fiber (NZDSF), has higher dispersion slope than standard single-mode fiber. In some NZDSF, a channel at the high end of the C band will experience three times as much dispersion as a channel at the low end of the band.
As data rates in optical networks increase, the amount of tolerable dispersion dramatically decreases. Higher data rates mean that optical pulses are closer together and begin to overlap sooner as they experience chromatic dispersion. The amount of dispersion an optical network can tolerate is inversely proportional to the square of the transmitted bit rate, so as data rates increase, dispersion tolerance drops dramatically. While 2.5-Gbit/s networks can tolerate up to 16,000 ps of dispersion 10-Gbit/s networks can tolerate only 1000 ps of dispersion, and 40-Gbit/s systems can tolerate only 60 ps of dispersion.
Increased distance between electrical regeneration sites presents its own problem. Since chromatic dispersion accumulates at a linear rate along the transmission fiber, longer distances between electrical regenerators mean that more dispersion accumulates and must be precisely managed for proper system operation.
The ideal solution
For advanced optical networks to fully enable more DWDM channels, faster data rates, and longer distances, an ideal chromatic-dispersion management technique should exhibit several characteristics.
First, the technique should correct for dispersion slope (defined as the rate of change of the dispersion as a function of wavelength). As DWDM channel counts continue to increase, effective management of dispersion slope becomes critical in order to fully utilize the entire band for transmission.
The ideal solution also should provide continuous dispersion management, which means that dispersion is properly corrected across the entire operating-wavelength range independent of DWDM channel spacing. Continuous solutions provide flexibility to accommodate network changes and upgrades (for example, migrating from 8 to 32 channels), and are not susceptible to troubling phenomena such as laser drift. Additionally, continuous solutions tend to exhibit low dispersion variation, or ripple, within the operating band.
The solution should also be capable of handling high optical power without stimulating nonlinear effects. As networks migrate from long-haul to ultralong-haul distances, individual optical components and modules must be able to tolerate higher levels of optical power. If components are exposed to more optical power than they were designed for, nonlinear effects are stimulated, which decrease overall system performance. In fiber-based optical components, the ability to handle high optical power is directly related to the fiber`s effective area (Aeff), so devices with large Aeff fibers are desirable.
Finally, the transmission-cost per bit-km must be lowered. As with any component or module being supplied to optical systems, dispersion-management solutions must help system providers lower the cost required to transmit data over long distances. An ideal solution for long-haul transmission can lower overall system cost by reducing the number of signal regeneration sites required.
There are several technologies, both existing and in development, that are directed at solving the chromatic-dispersion problem. For simplicity, they can be divided into two main categories: channelized and continuous solutions (see table, p. 71).
Channelized solutions provide dispersion management only at certain wavelengths and WDM channel spacing. The two main technologies that provide channelized dispersion management are fiber Bragg gratings and virtual-image phased array (VIPA).
Dispersion solutions based on fiber-Bragg-grating techniques correct dispersion by using wavelength-selective mirrors that result from impressing index-of-refraction changes into a fiber core in a pattern by ultraviolet light. Bragg gratings in dispersion-management applications yield narrowband, channelized solutions that typically correct dispersion over 2 to 6 nm. Therefore, many devices must be multiplexed together to correct for the entire operating band. Additionally, fiber Bragg gratings exhibit great deviation from the required dispersion levels in each channel. This deviation, called dispersion ripple, distorts portions of each pulse so that they shift in time with respect to the rest of the pulse train, negatively impacting system performance.
Virtual-image phased-array technology uses a combination of lenses and mirrors to vary the optical propagation distances of individual wavelengths to correct for the effects of chromatic dispersion. The individual components are designed to allow the longer wavelengths-the ones that fall behind during transmission-to propagate over shorter distances, while the shorter wavelengths must propagate over longer distances. As with fiber-Bragg-grating-based products, the channelized nature of VIPA-based products is susceptible to dispersion ripple within each channel.
The main technologies that achieve dispersion management across a continuous, broad wavelength range are based on either dispersion-compensation fiber (DCF) or high-order-mode (HOM) technology.
Dispersion-compensation fiber is a special type of fiber that exhibits a standard amount of negative dispersion per unit length, typically in the range of -70 to -90 ps/nm*km. Since DCF yields nearly constant dispersion across the C band, it is effective at correcting chromatic dispersion for a single wavelength-typically 1545 nm. However, the high dispersion-slope characteristics of the various NZDSF transmission fibers mean that DWDM channels away from this wavelength receive either too little or too much negative-dispersion correction. This inadequate slope-matching limits transmission distance and use of the full band for transmission in advanced optical networks.
Another problem with DCF is its inability to handle high optical power without eliciting nonlinear effects. A negative dispersion per unit length is achieved for DCF by decreasing the Aeff of the fiber core to the 30- to 40-µm2 range. This decrease limits the amount of optical power that can be transmitted through the fiber before nonlinear effects manifest themselves.
As management of dispersion slope has become more critical, DCF vendors have worked to improve the dispersion-slope-correction capability of DCF, and have developed a fiber referred to as enhanced DCF, or eDCF. The somewhat-improved slope-matching performance of eDCF is accomplished by further reducing the Aeff of the compensating fiber to less than 20 µm2. While this does slightly improve slope-matching capability, the power tolerance of devices employing eDCF is further decreased due to the smaller Aeff.
High-order-mode fiber is designed so that it propagates one or more selected high-order modes in addition to the basic mode. Modes can be thought of as guided optical waves, propagating with unchanging lateral distribution along the optical fiber. Such fibers exhibit many desirable characteristics necessary for effective dispersion management, including continuous, broadband dispersion-slope matching and high optical-power tolerance.
A key benefit of the HOM-based approach is that it yields a dispersion response that very accurately matches the amount of negative dispersion required for any NZDSF across the C and/or L band. The fiber thus achieves excellent dispersion-slope matching. Additionally, HOM fiber has a large Aeff, which translates to a high optical-power tolerance without eliciting nonlinear effects. As with other fiber-based techniques, the HOM technology results in a continuous response across the operating wavelength range, allowing full use of the band.
Chris Bradford is product manager at LaserComm Inc., 2600 Technology Drive, Suite 900, Plano, TX 75074. He can be reached at email@example.com.
Figure 1. Dispersion as a function of wavelength varies for different fiber types.