Helsinki University of Technology
Department of Electrical and Communications Engineering
S-38.128 Telecommunications Technology, Special Assignment
Liisa Peltonen, 37881S
22 November, 1998
Wavelength division multiplexing;
an overview
1Acronyms
3RRegeneration with retiming and reshaping
AOTFAcousto-optic tunable filter
APDAvalanche photodiode
AWGArrayed waveguide grating
BERBit error rate
CPMCross-phase modulation
DCFDispersion-compensating fiber
DSFDispersion-shifted fiber
DFBDistributed feedback
EDFAErbium-doped fiber amplifier
FWMFour-wave mixing
ISIIntersymbol interference
ITUInternational Telecommunication Union
LASERLight amplification by stimulated emission of radiation
LEDLight-emitting diode
MANMetropolitan area network
MZIMach-Zehnder interferometer
MLMMultilongitudinal mode
NDFNonzero dispersion fiber
NNINetwork node interface
OADMOptical add-drop multiplexer
OOKOn-off keying
OSNROptical signal-to-noise ratio
OTDMOptical time division multiplexing
OXCOptical cross-connect
PDFAPraseodymium-doped fiber amplifier
PDHPlesiochronous digital hierarchy
PDLPolarization-dependent loss
PMDPolarization-mode dispersion
POTSPlain old telephone service
QoSQuality of service
SBSStimulated Brillouin scattering
SDHSynchronous digital hierarchy
SOASemiconductor optical amplifier
SONETSynchronous optical network
SLMSingle-longitudinal mode
SMFSingle-mode fiber
SPMSelf-phase modulation
SRSStimulated Raman scattering
TFFThin-film filter
TFMFThin-film multicavity filter
TDMTime division multiplexing
WDMWavelength division multiplexing
2List of Figures
Figure 1: The structure of an optical fiber......
Figure 2: Services offered by a second-generation optical network......
Figure 3: Optical layer networks......
Figure 4: Evolving broadband network layers......
Figure 5: A unidirectional WDM link......
Figure 6: Relationships between the ITU-T optical networking recommendations......
Figure 7: A WDM broadcast and select network......
Figure 8: A WDM wavelength routing network......
Figure 9: A simple optical filter......
Figure 10: A wavelength multiplexer......
Figure 11: A wavelength add/drop multiplexer......
Figure 12: A wavelength router......
Figure 13: A splitter......
Figure 14: A combiner......
Figure 15: A coupler......
Figure 16: A Fabry-Perot filter......
Figure 17: A multilayer dielectric thin-film filter wavelength (de)multiplexer......
Figure 18: A Mach-Zehnder interferometer......
Figure 19: An arrayed waveguide grating (AWG)......
Figure 20: A static routing pattern of a 4x4 arrayed waveguide grating......
Figure 21: Circulators: (a) three-port (b) four-port......
Figure 22 : Optical fiber transmission with electrical regenerators......
Figure 23: Block diagram of an Erbium-doped fiber amplifier (EDFA)......
Figure 24: A possible evolution scenario for optical network architecture......
3Introduction
Optical fiber transmission has played a key role in increasing the capacity of telecommunication networks. The large, low-loss transmission capacity of fiber has made optical transmission to become the preferred means of high bit rate data transmission over long distances. However, as the demand for network capacity is rapidly increasing, even the current optical backbones are fast proving inadequate.
In many cases, it is relatively expensive to lay new fiber in order to increase network capacity. Wavelength division multiplexing (WDM) is a transmission technology which allows the capacity of existing optical links to be increased without installing new fiber, thereby enabling significant cost savings. In WDM, multiple signals at different carrier wavelengths are transmitted simultaneously over a single fiber. Today, point-to-point WDM links are already widely deployed in the US long distance networks, and the deployment has also started in Europe and Asia. The purpose of this paper is to give an overview of optical networks, wavelength division multiplexing and its key enabling technologies, and to briefly describe the fast-changing WDM deployment and standardization situation.
4Optical networks
Optical networks use optical fiber as a transmission medium. This chapter serves as an introduction to optical networks, covering the basics of optical fiber transmission, the evolution of optical networks, and the position of the optical layer in the communication network infrastructure.
4.1Light propagation in an optical fiber
Basic knowledge of light transmission in an optical fiber is key to understanding both the significant advantages of using fiber as a propagation medium and the system limitations which need to be considered when designing optical communication systems. The basics of light propagation, pulse broadening effects caused by dispersion and nonlinear effects constraining the design of higher bit rate systems, are described in this section.
To begin with, light in a strict sense means the region of the electromagnetic spectrum that can be perceived by human vision. This visible spectrum contains approximately the wavelength[1] range of 0.4 m to 0.7 m. However, in the laser[2] and optical communications fields, custom and practice have extended the usage of the term light to include a much broader portion of the electromagnetic spectrum, extending from the near-ultraviolet region of approximately 0.3 m through the visible region and into the mid-infrared region of approximately 30 m. [FED]
The medium used to guide the light signals in optical networks, the optical fiber, consists of a cylindrical core, which is surrounded by a cladding, as shown in Figure 1. Both the core and the cladding are primarily made of silica (SiO2). The refractive index[3] of the core is made slightly higher than that of cladding by introducing certain impurities, or dopants, into the core and/or the cladding.
Figure 1: The structure of an optical fiber
A simplified understanding of the propagation of light in the fiber can be described with the help of ray theory. From the ray theory viewpoint, light propagates in the fiber due to a series of total internal reflections that occur in the core-cladding interface. However, ray theory is an approximation that holds only when the signal wavelength is much smaller than the radius of fiber core.
A more general theory, applicable for all values of fiber radius, is the wave theory, which treats light as an electromagnetic wave, the propagation of which is governed by Maxwell’s equations. In wave theory, the propagation of light in any medium can be described by specifying the evolution of the associated electric and magnetic field vectors, denoted by E(r,t) and H(r,t), respectively, in space and time. [RS98] A more detailed look into the theory of electromagnetic waves is given e.g. by Cheng [Che89].
The electromagnetic waves travel partly in the core and partly in the cladding. The electric and magnetic field vectors in the core and the cladding must satisfy wave equations, which are second-order, linear partial differential equations:
2E + 2E = 0
2H + 2H = 0
Here is the angular frequency (rad/s) related to the light frequency f by = 2f, and are the magnetic and dielectric constants of the medium, respectively, and 2 is the Laplacian operator .
The solutions in the cladding and core, however, are not independent but are related by boundary conditions in the core-cladding interface. A fiber mode is every pair of solutions that satisfies these boundary conditions of the wave equations. Multi-mode fibers with core diameters of about 50 to 85 m can support more than one mode, and single-mode fibers, in which the radius of the core is of the order of the operating wavelength, can support only one mode. Single-mode fibers can carry more information than multi-mode fibers and are therefore the preferred guiding medium in high-bit rate optical communications over long distances. A more quantitative description of single- and multi-mode fibers is given e.g. in [KBW96].
The physical explanation of light propagation in a single-mode fiber follows from the difference of the refractive indices in the core and cladding. In any medium with a constant refractive index, a narrow light beam tends to spread due to a phenomenon called diffraction. The spreading can be counteracted by using an inhomogeneous medium in which the refractive index near the beam center (fiber core) is larger than at the beam periphery (fiber cladding), so that the beam center travels slightly slower than the beam periphery. This effectively provides continuous focusing of the light to counteract the spreading effect, and allows light to be guided in the medium and travel long distances with low loss. [RS98] A more quantitative description of light propagation in single-mode fibers using the wave theory approach is given e.g. in [KBW96].
4.2Capacity limits of optical transmission
Optical fiber offers low-loss transmission capacity over an enormous frequency range of about 25 THz. [Bor97] Compared to the bandwidth available in other transmission media such as copper cable or free space, this is orders of magnitude more. Also the attenuation of silica is very low in wavelength regions dedicated to optical communications. These properties allow the transmission of signals over long distances at high bit rates before they have to be amplified or regenerated. Here lies the reason for the fact that optical communication systems are so widely deployed today.
Like mentioned above, the intrinsic attenuation of silica is very low. With today’s technology, it is possible to fabricate optical fibers in which the attenuation of the signal traveling in the fiber is close to the theoretical limits due to scattering and absorption of light by silica molecules, less than 0.5 dB/km [Hew97]. The two low-loss regions are around the 1.3 m and 1.55 m wavelengths, the 1.55 m region having the lowest attenuation. Both low-loss regions, or optical windows, are used for communications. In some short-distance applications, such as computer interconnects, other wavelengths can be used as well [KBW96], [RS98]. Signal attenuation in optical fiber is therefore not considered a major limiting factor of optical transmission. Instead, two major effects which set limits on the feasible bit rates and transmission distances of today’s optical communication systems are dispersion and fiber nonlinearities, which are described in the following.
4.2.1Dispersion
When a light pulse travels in an optical fiber, its different components (different modes and/or different frequencies) propagate at slightly different velocities. This distortion in general is called dispersion.As a result of dispersion, the pulse becomes broadened, and the signals in adjacent bit periods may overlap, a phenomenon called intersymbol interference (ISI).
Chromatic dispersion
In a single-mode fiber, the dominant dispersion mechanism is chromatic dispersion, caused by different light frequencies traveling with different velocities. The wider the spectrum of the transmitted pulse, the greater the effect of chromatic dispersion. The early light transmitters such as light-emitting diodes (LEDs) or multilongitudinal mode[4] (MLM) Fabry-Perot lasers emitted light over a fairly large spectrum of several nanometers (hundreds of GHz). Today, chromatic dispersion is significantly reduced with the use of narrow spectral-width single-longitudinal mode (SLM) distributed-feedback (DFB)[5] lasers.
Chromatic dispersion is a characteristic of a fiber; different fibers have different chromatic dispersion profiles. It turns out that a silica-based, standard single-mode fiber (SMF) has essentially no chromatic dispersion in the 1.3 m optical window, but has significant dispersion in the 1.55 m window, which on the other hand has the lowest attenuation. However, fiber dispersion is a linear phenomenon and can therefore be compensated for by means of the transmission medium, and dispersion-shifted fiber (DSF), which has the zero-dispersion wavelength shifted to the 1.55 m window, has been developed for this purpose. DSF is suitable for single-channel systems operating at high bit rates (10 Gb/s and above) over long distances. However, DSF is not well suited to WDM systems, mainly due to the detrimental effects of four-wave mixing and other fiber nonlinearities.
The accumulated chromatic dispersion penalty increases with the link length. When the distances and bit rates increase, chromatic dispersion can be compensated for e.g. by using nonzero dispersion fiber (NDF), which has a small amount of dispersion in the 1.55 m window, thereby reducing the penalties due to nonlinearities but retaining most of the advantages of DSF. Also special dispersion-compensating fibers (DCFs) that provide negative dispersion in the 1.55 m range to enable a zero net dispersion are commercially available. However, a drawback of using DCFs is the additional loss they introduce to the system. [RS98]
Modal dispersion
In a multi-mode fiber, the energy of a pulse travels in different modes, each with a different velocity. The resulting dispersion mechanism is called modal dispersion. This was a problem especially in early telecommunication systems, which used multimode fibers along with light-emitting diodes (LEDs) or multilongitudinal-mode (MLM) Fabry-Perot lasers as transmitters.
Polarization-mode dispersion
Finally, polarization-mode dispersion (PMD) arises because the fiber core is not perfectly circular. This causes different polarizations[6] of a signal to travel at different group velocities. PMD becomes an impediment in high-bit-rate systems operating at 10 Gb/s and above. [RS98]
4.2.2Fiber nonlinearities
As long as the optical power of a signal traveling in an optical fiber is relatively small, the fiber can be considered a linear medium. However, when the signal levels get higher, fibernonlinearities start imposing limitations on link length and/or bit rate of the system. The nonlinearities arise because the loss and refractive index of the fiber have a component dependent on optical power. In many cases, chromatic dispersion plays a key role in reducing the effects of nonlinearities: when a little chromatic dispersion is present in the fiber, the different interacting waves then travel with different group velocities. Nonzero dispersion fiber (NDF) is being installed to new WDM systems for this purpose.
The nonlinearities can be classified into two categories. The first is due to the scattering effects owing to the interaction of light waves with molecular vibrations in silica medium. Examples of this category are stimulated Brillouin scattering (SBS) and stimulated Raman scattering (SRS).
Effects in the second category arise because the refractive index of the fiber has an intensity-dependent component. Effects in this category include four-wave mixing (FWM), self-phase modulation (SPM) and cross-phase modulation (CPM).
Stimulated Brillouin scattering (SBS)
Stimulated Brillouin scattering (SBS) depletes the transmitted signal by producing gain in the direction opposite to signal propagation, that is, back toward the source. This often calls for the shielding of the transmitter with an isolator. SBS does not cause interaction between different wavelengths as long as the wavelength spacing is greater than 20 MHz, but can cause significant distortion within a single channel.
Stimulated Raman scattering (SRS)
Stimulated Raman scattering (SRS) causes power to be transmitted from lower-wavelength channels to higher-wavelength channels, the gain coefficient being a function of wavelength spacing. Coupling occurs between two channels only if power is present both channels, that is, is both channels are sending 1 bits. The direction of coupling is both in the direction of propagation and the reverse direction. The penalties due to SRS are reduced when dispersion is present, because the signals in different channels then propagate at different velocities and the probability of pulses at different wavelengths overlapping at any point in the fiber is reduced. Considering that the channel spacing is fixed, the impairments due to SRS grow with the number of wavelength channels and the resulting total system bandwidth.
Four-wave mixing (FWM)
Four-wave mixing (FWM) induces signals at new frequencies that appear as crosstalk to the existing signals. The FWM effect is independent of the bit rate but is highly dependent on frequency channel spacing and is reduced when dispersion is present.
Self-phase modulation (SPM) and cross-phase modulation (CPM)
SPM and CPM arise when fluctuations in the optical power of a signal cause changes in signal phase. Thus, different parts of a pulse undergo different phase shifts, which causes pulse chirping[7]. This causes spectral broadening, which in turn increases dispersion penalties. The impairments due to SPM are significant mainly in high bit rate (over 10 Gb/s) systems. CPM becomes a problem if the wavelength channel spacing is tight (a few tens of GHz). [RS98]
4.3Evolution of optical networks
The starting point of optical fiber communications technology can be dated back to the 1960s when the increasing voice traffic started to exhaust the wire pair circuits between the central offices of the telephone network, and a new, higher-capacity transmission medium was needed. Early experiments demonstrated the capability of waveguides to transport information encoded in light signals but it was not until the invention of the low-loss silica-based fiber in the 1970s that optical transmission really took off.
The early fibers were multi-mode fibers and were used along with LEDs or multilongitudinal mode (MLM) Fabry-Perot laser transmitters operating at 0.8 m and 1.3 m wavelength bands. The resulting system was thereby heavily degraded by modal dispersion and had to have electronic signal regenerators every few kilometers. The primary focus was on multiplexing digital voice circuits and the infrastructure was based on plesiochronous digital hierarchy (PDH).
In the next generation of optical fiber communications systems deployed in the early 1980s, standard single-mode fiber (SMF) was used to eliminate modal dispersion. This enabled a substantial increase in the bit rates and distances between regenerators. MLM Fabry-Perot lasers in the 1.3 m band were used as transmitters. Modal dispersion was effectively eliminated and the distances between regenerators were primarily determined by fiber attenuation. Typically, the spacings between regenerators were about 40 km and the systems operated at bit rates of a few hundred Mb/s.
To attain longer spans between regenerators, the lower loss of the 1.55 m wavelength band motivated the deployment of systems operating in this optical window in the late 1980s. At this point, chromatic dispersion started to become a problem. To overcome these limitations, dispersion-shifted fiber (DSF) was developed. However, there already existed a large installed base of standard single-mode fiber for which it was not possible to apply this solution. Another solution to the problem was found by narrowing the spectrum of the transmitted pulse. Narrow spectral-width single-longitudinal mode (SLM) distributed-feedback lasers were deployed as transmitters, and it was possible to put 1-2 Gb/s transmission systems into use. In parallel, synchronous digital hierarchy (SDH) was standardized by ITU-T in 1988.