High-capacity fiber-optic networks to transport diverse multimedia services
Capable of delivering versatile voice, interactive video, high-speed data and high-resolution image signals, advanced optical-fiber-based communications systems technologies are expected to unfurl long-haul and interoffice transmission networks for transporting broadband and multimedia services
bell communications research
During the next decade, broadband networks based on optical fiber backbones are projected to transport all types of information services, including voice, high-resolution images, high-speed and color facsimile, videophone, desktop videoconferencing, video-on-demand and multimedia communications for business, education and medical applications. To deliver these varied services, future interoffice and long-haul trunking systems are anticipated to incorporate advanced lightwave systems operating at 10- to 40-gigabit-per-second capacities and beyond. The key lightwave technologies needed to achieve these broadband fiber systems include optical amplifiers; wavelength-division multiplexing, or WDM, devices and components; high-speed electronics and opto-electronics; and dispersion-compensation and non-linearity fiber control techniques.
Fortunately, technology trends in advanced high-capacity lightwave systems involve fiber-optic network upgrades that can handle increasing information-transmission rates. For example, the development of synchronous optical network, or Sonet, and synchronous digital hierarchy, or SDH, optical fiber transmission systems during the past six years has progressed from OC-12 systems at 622 megabits per second to large-scale deployment of OC-48 at 2.5 Gbits/sec in metropolitan area networks. Other developments are focusing on upgrading OC-48 systems to OC-192 systems at 10 Gbits/sec for even higher capacity.
Over the past two years, the development of 10-Gbit/sec opto-electronics and electronic add/drop multiplexer subsystems for Sonet/SDH OC-192 systems has attracted worldwide attention. In addition to 10-Gbit/sec system field trials by major communications companies, external-modulator-based 20-Gbit/sec time-division multiplexing optical transmission systems have also been demonstrated in several research laboratories.
Essential to the operation of high-speed (2.5- to 10-Gbit/sec) lightwave transmission systems are optical fiber amplifiers. Many industry analysts consider these amplifiers to be the most significant advance in lightwave systems technology development in the past decade. Optical fiber amplifiers provide network designers with a vastly increased optical power budget for overcoming fiber link-loss limitations in long-haul transmission and local-loop signal distribution.
In long-distance optical fiber networks, the three primary transmission limiting factors are fiber link loss, dispersion and nonlinear fiber characteristics. Fiber nonlinearities, such as stimulated Brillouin scattering, stimulated Raman scattering or Raman amplification, four-wave mixing and self-phase modulation, can either become limiting factors or be used to help improve transmission at the high optical power levels needed for long-distance transmission. Their impact on lightwave systems depends mainly on nonlinear transmission processes, transmission system configurations, phase-matching and dispersion.
Nonlinear optical interactions
Although optical fiber amplifiers provide the needed signal gain and high power needed to overcome fiber loss, their use in long-haul and WDM systems raises the probability of nonlinear optical interactions occurring in the transmission fibers; these interactions tend to limit the system`s overall performance. This condition generally takes place when optical amplifiers are used as power (or booster) amplifiers after the optical transmitters to increase the optical power launched into the transmission fiber to the 13- to 20-decibel relative to milliwatt (20 to 100 mW) range.
In long-haul transmission, a cascade of inline optical amplifiers, typically located approximately 50 to 80 kilometers apart, can be used to replace conventional opto-electronic regenerators. The use of an optical power amplifier at the beginning of the transmission link (to boost the first optical transmitter`s output power) and the use of an optical preamplifier at the receiver end (to improve receiver sensitivity) contribute to expanding the optical power budget. Of course, two amplifiers do not contribute markedly to the total optical power budget as tens to hundreds of inline optical amplifiers, which are generally needed in ultra-long links.
The amplification situation is different for long-span repeaterless transmission (between distant telephone central offices or between landing points for undersea links) in 100- to 300-km links. In these networks, network designers typically prefer to install a maximum fiber transmission link length without the use of inline repeaters or inline amplifiers to avoid the construction and maintenance costs of mid-span regenerator sites.
In island-hopping undersea fiber links, network designers also opt to achieve long-span transmission without undersea repeaters or optical inline amplifiers. They favor a maximum power budget with only two end-point optical fiber amplifiers. However, in some undersea networks, the link distance can be stretched to 400 or 500 km by remote pumping (at 1480 nanometers) additional erbium-doped fiber amplifiers near both ends of the link. Long-span links that employ various configurations of optical fiber amplifiers (including remote pumping of erbium-doped fiber amplifiers and Raman fiber amplifiers) are of great interest in research and development.
Semiconductor optical amplifiers
Since l990, significant technology and application developments have been accomplished for both optical fiber amplifiers and semiconductor optical amplifiers. Erbium-doped fiber amplifiers work best in the 1530- to 1560-nm spectral region. Praseodymium-doped fiber amplifiers are targeted at 1310-nm system applications. Semiconductor optical amplifiers have not gained popularity as power or inline amplifiers because of low gain and saturation-induced signal distortion; however, these problems are being studied.
Erbium-doped fiber amplifiers have outperformed their semiconductor counterparts for a range of systems applications, such as low-distortion optical power amplifiers for analog video transmission and multi-Gbit/sec digital optical signals. On the debit side, though, erbium-doped fiber amplifier operation is limited to the 1530- to 1560-nm range. For their part, praseodymium-doped fiber operating in the 1310-nm region possess lower pump efficiencies and require higher pump powers. Presently, commercial praseodymium-doped fiber prototypes are available but the needed pump power remains high. Field demonstrations have been reported, but practical deployment has not been initiated.
Another technology trend concerns semiconductor optical amplifiers made with quantum-well structures that have exhibited high-gain (20 to 25 dB) and high-saturation output power. Recent high-capacity long-distance lightwave system experiments with 1310-nm semiconductor optical amplifiers have demonstrated their effectiveness in high-capacity systems. This application represents a new role for these amplifiers for upgrading existing systems to higher-speed, higher-capacity systems at 1310 nm using embedded standard singlemode fiber.
On another technology front, WDM is serving as an alternative to time-division multiplexing, as well as providing an additional layer of multiplexing complementary to time-division multiplexing for increasing transmission capacity. For example, four WDM channels operating at 2.5 Gbits/sec serve as an alternative to a 10-Gbit/sec time-division multiplexing system, but can function with 2.5-Gbit/sec opto-electronics and electronics. Likewise, network designers can achieve a 40-Gbit/sec time-division multiplexing equivalent system by employing sixteen 2.5-Gbit/sec WDM systems using 2.5-Gbit/sec technologies, or by employing four 10-Gbit/sec WDM systems using 10-Gbit/sec technologies, with different consequences in dispersion limitation.
Although these systems have been demonstrated, practical WDM systems have not yet been widely deployed. The slowdown occurs because WDM requires separate optical transmitters and receivers at each wavelength. To date, this requirement has proven more expensive than adding additional stages of time-division multiplexing and operating at higher speeds to achieve increased capacity. However, the use of multichannel WDM is becoming more practical. It may be more economical to deploy 2.5-Gbit/sec transmission if a 10-Gbit/sec or higher bit-rate time-division multiplexing system becomes limited by the high dispersion of a 1550-nm system using embedded conventional singlemode fibers. Furthermore, the successful research and development of high-speed WDM laser arrays and receiver arrays, as well as integrated multiplexer and demultiplexer devices, are accelerating interest in deploying WDM system technologies.
Experimental demonstrations have verified the operation of multichannel WDM systems operating at an OC-48 or OC-192 rate for each channel. WDM systems operating at 10-, 20-, 40-, 80- and 160-Gbit/sec total transmission capacity have been demonstrated with 4-, 8- and 16-WDM optical channels at 2.5 and 10 Gbits/sec per channel. These results show reliable WDM system performance and represent alternatives to straight high-speed time-division multiplexing systems for very-high-capacity upgrading of broadband fiber backbone networks.
WDM systems were initially considered for increasing the number of optical channels for higher-capacity transmission over a single fiber. Renewed interest in WDM is due to high-capacity upgrading considerations and because multiple-wavelength WDM optical signals are transparent to each other in optical fiber. This characteristic allows different wavelengths to carry different signal formats or services over the same fiber network. It also permits flexible upgrades by adding only the optical channels needed for the desired services.
Erbium-doped fiber amplifiers
Erbium-doped fiber amplifiers have proven useful as optical amplifiers for multi-optical-channel WDM distribution systems. For long-haul and long-span systems, though, only traditional optical amplifiers have been used for single-channel applications. However, with the increased interest in multichannel WDM systems, whether at 2.5 Gbits/sec or 10 Gbits/sec per optical channel, WDM optical amplifiers are capable of multi-wavelength (4 to 16 channels typically) simultaneous amplification over the 30-nm erbium-doped fiber amplifier band but require special network design considerations.
Whereas erbium-doped fiber amplifiers suit the 1530- to 1560-nm spectral region for single-optical-channel signals, they need gain equalization among the different channels for the multiple-optical-channel signals used in WDM systems. Network designers must evaluate gain non-uniformity and gain-saturation induced crosstalk in these systems. Overall WDM system configuration and network architecture greatly influence these issues. Active or passive gain-equalization or optical-power-equalization techniques might have to be deployed, depending on the characteristics of the signals, gain, number of optical fiber amplifiers, WDM multiplexer and demultiplexer devices and components, and the optical receiver dynamic ranges. These issues are currently being addressed by research and development groups for practical WDM systems applications.
A proposed WDM optical fiber amplifier has demonstrated effectiveness in gain equalization using a fiber optical amplifier followed by an optical fiber amplifier array, a WDM demultiplexer for the WDM signals and an optical fiber amplifier for each optical channel. The gain-saturated amplifier array provides the self-equalizing property. This scheme calls for each optical channel to have its own optical amplifier, thus requiring an amplifier array, which can be pumped by a single high-power laser diode.
Numerous 1310-nm optimized (minimum dispersion near 1310 nm) standard singlemode fibers have been installed. To upgrade such embedded fiber networks without installing new fiber cables, network designers are replacing the existing 1310-nm transmission system equipment in the central office with new Sonet/SDH-based transmission system equipment that can operate at higher bit rates. With this approach, they can keep the existing 1310-nm system and add higher-bit-rate systems at 1550 nm by incorporating a WDM system. However, the dispersion of the existing fiber (regular, non-dispersion-shifted fiber) at 1550 nm gives rise to bit-error-rate penalties for long-distance high-bit-rate transmission. However, special low-chirp lasers, external modulation and special pre-chirp techniques are available to reduce the chirp-induced dispersion penalty.
Another approach is to use a passive dispersion-compensating fiber based on an optical dispersion equalizer fiber, which allows nearly dispersion-free transmission with regular laser transmitters. The use of a passive dispersion-compensating fiber can, in principle, achieve nearly dispersion-free transmission at the bit rates (2.5 to 40 Gbits/sec or higher) needed in future interoffice high-capacity upgraded networks. Dispersion-compensation for multiple-optical channel WDM systems (tested at 2.5 Gbits/sec per channel) over more than 20 nm within the erbium-doped fiber amplifier gain bandwidth has also been demonstrated. The effectiveness of dispersion-compensating fiber has also been demonstrated in 10-Gbit/sec system experiments.
Several active and passive dispersion-compensation schemes have attracted widespread attention. These include a mid-span spectral inversion technique based on phase-conjugation by four-wave-mixing in either dispersion-shifted fibers or semiconductor optical amplifiers, and passive dispersion-compensators based on filters or fiber gratings. The passive dispersion-compensating fiber grating and techniques appear more practical than the mid-span spectral inversion techniques. Moreover, commercial dispersion-compensating fibers are available.
Because of nonlinear optical transmission limitations, such as simulated Brillouin scattering, self-phase modulation, simulated Raman scattering and four-wave mixing, network designers must incorporate special system methods to minimize these effects. They can suppress simulated Brillouin scattering relatively easily by frequency dithering the laser transmitter source. Moreover, they can usually tolerate or manage self-phase modulation. In WDM systems, four-wave-mixing effects can deplete WDM signal energy and generate undesirable spectral components. Because four-wave mixing readily occurs in the zero-dispersion region resulting from easier phase-matching, network designers must operate the WDM system in a spectral region that has finite, non-zero dispersion to minimize or eliminate its effects. On the other hand, the dispersion must not be large enough to inflict a dispersion penalty for long-haul transmission. Consequently, network designers face a nonlinearity/dispersion tradeoff. New non-zero dispersion-shifted fibers provide a finite dispersion of approximately ۫ to 4 picoseconds/nm-km over the 1540- to 1560-nm region of erbium-doped fiber amplifier gain bandwidth for long-distance high-capacity WDM systems applications. u
Chinlon Lin is director, lightwave video systems, at Bell Communications Research in Red Bank, NJ.