Chapter 2

OPTICAL FIBER CHARACTERISTICS AND SYSTEM

CONFIGURATIONS
One attractive aspect of optical fibers is their enormous bandwidth compared to
other media, such as radio waves and twisted-pair wires. Still, an optical fiber is not ideal;
it possesses some unwanted properties. Dispersion and nonlinearity are the major limiting
factors in lightwave communication. Fiber dispersion causes different spectral
components of a signal to travel at different speeds. Hence, for a given transmission
distance different spectral components arrive at the destination at different times. This
results in a pulse broadening effect when a pulse propagates along an optical fiber.
However, the situation is different when the nonlinearity and dispersion are considered
together. In some circumstances, the nonlinearity could counteract the dispersion. In
addition, when multiple channels are considered, the fiber nonlinearity results in
interactions among channels.
The purpose of this chapter is to discuss the effects of dispersion and nonlinearity
in terms of their origins and corresponding impairments. Those impairments lead to
various system designs intended to minimize deleterious effects caused by dispersion and
nonlinearity. This chapter begins with a discussion of dispersion in single-mode fibers,
and types of optical fibers based on the value of dispersion. It is then followed by the
effects of nonlinearity and approaches to minimize such effects, in Section 2.2. The
discussion of system configurations based on the types of transmission optical fiber is
presented in Section 2.3. The benefits and trade-offs on different system configurations
are also explained.

2.1 FIBER DISPERSION

When one considers an optical fiber, the first parameter of interest is the value of
dispersion. This is simply because different types of optical fibers have different
dispersions. For a single-mode optical fiber, the only source of dispersion is due to group 23
Chapter 2: Optical Fiber Characteristics and System Configurations

velocity dispersion (GVD), or intramodal dispersion where the dispersion is the result of
the wavelength dependence of the group velocity vg . This causes different spectral
components to propagate along an optical fiber with different group velocities. For an
optical fiber of length z , the spectral component at a frequency would exit the fiber at
a time delay of T = z / vg [1]. The group velocity vg can be related to the phase
(propagation) constant by
vg1 = d d .

(2.1)

Due to the frequency dependence of , one can show that a pulse having the spectral
width of is broadened by
T = 2 z ,

(2.2)

where 2 = d 2 d 2 , which is generally called the GVD parameter, with units of

ps2/km. The GVD parameter 2 can be interpreted as the dispersion per unit transmission
distance per unit frequency spread of the signal. In optical fiber communication systems,
the wavelength unit is more commonly used than the frequency unit, and (2.2) can be
rewritten as
T = Dz ,

(2.3)

where is the signal spectral width in wavelength units, and D is the dispersion
parameter in the units of ps/(kmnm), which can be related to 2 by
D=

d 1

d vg

2
= ( 2 c ) 2 ,

(2.4)

where c is the velocity of light in vacuum, and is the wavelength. It should be noted
that the dispersion parameter D rather than the GVD parameter 2 is generally used to
indicate the amount of dispersion in fiber specifications.
When the fiber loss and nonlinearity are neglected, a single-mode optical fiber
can be considered as an all-pass filter with nonlinear phase response, and the
corresponding transfer function can be written as

1
1

H o ( ) = exp jz 0 + 1 ( ) + 2 ( ) 2 + 3 ( ) 2 + ,
2
6

(2.5)

24
Chapter 2: Optical Fiber Characteristics and System Configurations

where m = ( d m d m )

=0

, and 0 is the operating frequency. It should be noted that

(2.5) is derived under the assumption that << 0 , so ( ) can be expanded about 0
by using a Taylor series. In terms of communication theory, the all-pass filter whose
transfer function is given by (2.5) would cause waveform distortion due to the nonlinear
phase response. This is commonly called dispersion in the fiber optic communication
areas. When (2.5) is considered, the first term in the exponent causes only a constant
phase shift while the second term results in a time delay. The third and following terms
(the second-order and higher-order terms when 2 is considered) in (2.5) are the sources
of dispersion. The second-order term is related to the dispersion parameter D by (2.4)
whereas the third and higher-order terms are a result of the phase constant not being a
quadratic function of frequency. For the third-order term, it can be related to the
dispersion slope S by

S=

2
dD
= ( 2 c 2 ) 3 + ( 4 c 3 ) 2 .
d

(2.6)

In practice, the third and higher-order terms in (2.5) can be safely neglected as long as the
operating wavelength is sufficiently far from the zero-dispersion wavelength so that the
contributions of those terms are negligible. Thus, the quadratic dependence of the phase
response on the frequency in (2.5) is the major source of dispersion.
2.1.1 Sources of Dispersion

One source of dispersion in a single-mode optical fiber comes from the fact that
the refractive index of the material used to make an optical fiber is a function of the
wavelength. This is commonly referred to as material dispersion DM or chromatic
dispersion [58]. Generally, an optical fiber consists of a core and cladding. The refractive
index profile of core and cladding also gives rise to dispersion, which is called waveguide
dispersion DW . Shown in Fig. 2.1 are the material and waveguide dispersions as a
function of wavelength for a single-mode optical fiber. It is clearly seen that the material
dispersion increases with wavelength for silica based optical fibers in the wavelength
range of interest. The wavelength ZD at which the material dispersion becomes zero is in
the range of 1.27-1.29 m depending primarily on the materials used to control the index
25
Chapter 2: Optical Fiber Characteristics and System Configurations

of refraction of the core of the fiber [1]. Beyond this wavelength ZD , the material
dispersion becomes positive. On the other hand for typical fiber designs the waveguide
dispersion is negative in the entire wavelength range 0 1.6 m as shown in Fig. 2.1.
Theoretically, the material dispersion and waveguide dispersion are interrelated;
however, a good approximation of the total dispersion is simply the sum of both
dispersions [58]. The effect of the waveguide dispersion is to shift the zero-dispersion
wavelength to the longer wavelength. For a standard single-mode fiber the dispersion D
is zero near 1.31 m. The typical value of D for a standard single-mode fiber is around
15 to 18 ps/(kmnm) in the 1.55-m wavelength window where the fiber attenuation is
lowest. Such high values of D poses the major limitation on 1.55-m systems.
Although the material dispersion can be altered, significant changes are difficult
to achieve in practice. On the other hand, waveguide dispersion can be modified fairly
easily by changing the refractive index profile of core and cladding. Fig. 2.2 shows the
refractive index profiles of a standard single-mode optical fiber and a dispersion-shifted
fiber. By changing the refractive index profile of an optical fiber the variation of
waveguide dispersion as a function of wavelength can be tailored so that the total
dispersion is zero at 1.55 m. This type of optical fiber is generally called dispersionshifted fiber. The dispersion-shifted fibers were particularly deployed in single-channel
systems. Its minimum dispersion at 1.55-m wavelength region allows the effective use
of the wavelength region where the fiber loss is minimum. Still, there is a problem
associated with the nonlinear interaction between the signal and the noise as discussed
below.
By employing dispersion-shifted fibers, the distance between the transmitter and
the receiver can be enhanced considerably due to the small effect of dispersion. However,
the distance cannot be extended indefinitely due to the fact that the optical fiber is not a
lossless medium. Fortunately, the fiber attenuation can be compensated by employing inline optical amplifiers to periodically boost up the signal power. Since the optical
amplifiers also emit noise called amplified spontaneous emission (ASE) noise, the phase
matching between the signal and the ASE noise results in the nonlinear four-wave
mixing, which in effect transfers energy from the signal to ASE noise. In general, the
efficiency of the four-wave mixing process decreases with the increase in the dispersion.
26
Chapter 2: Optical Fiber Characteristics and System Configurations

Hence, for single-channel systems employing dispersion-shifted fiber, the operating

wavelength has to be sufficiently away from the zero-dispersion wavelength. It should be
noted that the four-wave mixing process is not the only nonlinear effect in an optical
fiber. The various nonlinear effects prevent the effective use of WDM technique in
systems employing dispersion-shifted fibers. For example, the four-waving mixing
among WDM channels generates new signals whose wavelengths might be collocated
with the WDM signals themselves. This in effect causes a decrease in the signal power
and cross talk among WDM channels.
Similar to the four-wave mixing process, the other nonlinear effects in general
decrease with the increase in dispersion. In order to effectively suppress the nonlinear
effects, the local dispersion of the transmission fibers has to be sufficiently large.
Although standard single-mode optical fibers seem to be the answer for solving the
problems caused by the nonlinear effects to enable the deployment of WDM techniques,
large local dispersion implies that the dispersion compensation scheme is necessary for
combating the large accumulated dispersion. This in turn increases the system
complexity. In order to avoid such complexity, an optical fiber whose dispersion is the
compromise between the standard single-mode fiber and dispersion-shifted fiber is
desirable. This led to the development of a new type of optical fiber known as non-zero
dispersion fiber (NZDF) in the early 1990s [59], [60]. Its purpose is to facilitate WDM
deployment at the channel rate of 10 Gb/s without dispersion compensation. The
dispersion of this type of fiber is designed so that it is sufficiently large to minimize the
nonlinear effects while the effect of dispersion itself is not so strong as to cause severe
waveform distortion. The non-zero dispersion fiber normally has a value of dispersion in
the 1.55-m wavelength region of several ps/(kmnm). It is clearly seen that the fiber
nonlinearity is not negligible, and has to be taken into account seriously. Systems have to
be properly configured in order to minimize nonlinear effects, and in some special
circumstance to cleverly use fiber nonlinearity as an advantage.

2.2 NONLINEAR EFFECTS IN OPTICAL FIBERS

Although silica glass used as material for making optical fibers has very small
nonlinear coefficients, the nonlinear effects in optical fibers are generally not negligible.
27
Chapter 2: Optical Fiber Characteristics and System Configurations

This is because the optical intensity of a propagating signal is high despite the fact that
the signal power is rather low (several milliwatts to tens of milliwatts). The small cross
section of an optical fiber causes the intensity to be very high, which in effect is sufficient
to induce significant effects of nonlinearity. Furthermore, for optically amplified systems,
the distance between regeneration is large, so the nonlinear effects may accumulate over
long distances. The nonlinear effects can be divided into two cases based on their origins:
stimulated scatterings and optical Kerr effects. The latter is the result of intensity
dependence of the refractive index of an optical fiber leading to a phase constant that is a
function of the optical intensity, whereas the former is a result of scattering leading to an
intensity dependent attenuation constant. There are two stimulated scattering phenomena
in an optical fiber: Raman scattering and Brillouin scattering.
The intensity dependence of refractive index results in self-phase modulation
(SPM), cross-phase modulation (XPM or CPM), and four-wave mixing (FWM). Another
difference between stimulated scatterings and the effects of nonlinear refractive index is
that the former is associated with threshold powers at which their effects become
significant. All of these effects are discussed in this section.
2.2.1 Stimulated Raman Scattering (SRS)

Stimulated Raman scattering (SRS) is the result of interaction between incident

light and molecular vibration. Some portion of the incident light is downshifted in
frequency by an amount equal to the molecular-vibration frequency, which is generally
called the Stokes frequency. This in effect depletes the optical power of the incident light.
When there is only a single light wave propagating along the optical fiber, Raman
scattering results in the generation of spontaneous Raman-scattered light waves at lower
frequency. In general, the criterion used to determine the level of scattering effects is the
threshold power Pth defined as the input power level that can induce the scattering effect
so that half of the power (3-dB power reduction) is lost at the output of an optical fiber of
length z . For single-channel lightwave systems, it has been shown that the threshold
power Pth is given by [15]
Pth = 32 Aeff g R ,

(2.7)

28
Chapter 2: Optical Fiber Characteristics and System Configurations

where and Aeff are the attenuation coefficient and effective core area of an optical
fiber, respectively, and g R is the Raman gain coefficient. Note that in (2.7) is the
approximation of effective interaction length when z is large. For a 1.55-m singlechannel system employing a standard single-mode optical fiber, Pth is found to be of the
order of 1 W. Commonly, the transmitted power is well below 1 W; thus, the effect of
SRS is negligible in single-channel systems.
The effect of Raman scattering is different when two or more signals travel along
an optical fiber. When two optical signals separated by the Stokes frequency travel along
an optical fiber, the SRS would result in the energy transfer from the higher frequency
signal to the lower frequency signal. That is, one signal experiences excess loss whereas
another signal gets amplified. The signal that is amplified is called the probe signal
whereas the signal that suffers excess loss is called the pump signal. The effectiveness of
the energy transfer depends on the frequency difference (Stokes frequency) between the
pump and probe signals. Shown in Fig. 2.3 is the Raman gain coefficient for the probe
signal as a function of Stokes frequency for a pump wavelength of 1 m [61]. Since the
gain coefficient scales inversely with the wavelength, the peak gain coefficient shown in
Fig. 2.3 becomes 7 1012 cm/W at the pump wavelength of 1.55 m. It is clearly seen
that the Raman gain bandwidth is very broad. The SRS can still happen even when two
signals are separated by 15 THz (500 cm-1).
In WDM systems, the channels at shorter wavelengths will act as pump signals
and suffer excess loss. On the other hand, the channels at longer wavelengths acting as
probe signals are amplified via SRS. This might cause SNR discrepancy among channels.
Shown in Fig. 2.4 is the schematic explanation of the SRS in the case of two-channel
systems. In this case the pump channel (channel 1) is located at 1 whereas the probe
channel (channel 2) is at 2 ( 2 > 1 ). Here the effect of dispersion is neglected. Shown
in Fig. 2.4 (a) are bit patterns of both channels without SRS. Under the presence of SRS,
the signal in channel 1 is depleted and transferred to channel 2 whenever both channels
carry bit 1 as shown in Fig. 2.4 (b). However, there is no SRS when either channel carries
bit 0. It is clearly seen that the SRS affects the two channels differently. At the receiver

29
Chapter 2: Optical Fiber Characteristics and System Configurations

output, the eye opening of channel 1 would be smaller than that of channel 2 due to the
energy depletion on some of bit 1 caused by SRS, which can be viewed as cross talk.
Practically, the situation is not as severe as the illustrated example. In WDM
systems, the chance that all channels transmit bit 1 simultaneously decreases with the
increase in the number of channels. This decreases the effect of SRS. It has been shown
in [62] that when this statistical consideration is taken into account, the threshold power
of the SRS to cause a noticeable effect is increased by 3 dB compared with that predicted
by assuming all channels are continuous-wave. The effect of SRS is also reduced under
the presence of dispersion [63]. Fiber dispersion causes the signals at different
wavelengths to travel at different speeds, causing walk-off between bit sequences at
different channels. The walk-offs among channels decreases the effect of SRS, hence
increasing the threshold power. In general, SRS is not the limiting factor in lightwave
communication systems compared with the other nonlinear effects due to its high
threshold power.
2.2.2 Stimulated Brillouin Scattering (SBS)

This nonlinear effect is due to the interaction between the incident light and
acoustic vibration in the optical fiber. Similar to SRS, stimulated Brillouin scattering
(SBS) causes frequency down-conversion of the incident light, but the frequency shift in
this case is equal to the frequency of the interacting acoustic wave. However, unlike
Raman-scattered light, the Brillouin-scattered light propagates in the backward direction.
This in effect causes excess loss on the incident light similar to SRS. For SBS, the
threshold power is given by [64]
Pth = 42

Aeff
gB

vp
1 +
vB

(2.8)

where g B is the Brillouin gain coefficient, v p is the signal linewidth, and vB is the
Brillouin gain bandwidth. For a 1.55-m system employing a standard single-mode
optical fiber, Pth can be as low as 2 mW when the signal is continuous-wave and has
ideal zero linewidth ( v p 0 ).

30
Chapter 2: Optical Fiber Characteristics and System Configurations

This seems to suggest that the SBS is the major limiting factor in the allowable
launched power. However, it should be noted that modulation broadens the signal
spectral width considerably. In addition, for silica fiber the Brillouin gain bandwidth vB
is as small as 20 MHz at 1.55 m [65]. When the ratio between the signal spectral width
v p and the Brillouin gain bandwidth vB in (2.8) is taken into account, the threshold power
increases significantly, especially when the bit rate is high. Since the gain bandwidth vB
is very small, there are no nonlinear interactions among WDM channels due to SBS.
Therefore, in order to avoid SBS the power of each WDM channels, not the total power
of all transmitted channels, has to be kept below the threshold. Due to its high threshold
power Pth , the effect of SBS is negligible when the operating bit rate is sufficiently high,
which is generally satisfied in practice.
2.2.3 Self-Phase Modulation (SPM)

Self phase modulation (SPM) arises from the intensity dependence of the
refractive index. SPM results in the conversion of intensity variation to phase variation.
For a silica optical fiber, the refractive index n is given by [15]
n = n0 + n2 I ( ) ,

(2.9)

where n0 is the linear refractive index of the material, n2 is the nonlinear refractive
index, and I ( ) is the optical intensity in units of W/m2. The value of n0 is
approximately 1.5 whereas n2 is around 3 1020 m2/W [15]. Although the value of n2
seems very small, high signal intensity and long transmission distance make the effect of
nonlinear refractive index not negligible. For a propagation distance of z , the phase of
the signal is given by

( z , ) =

2 n0 z

2 n2 I ( ) zeff

(2.10)

where zeff is the effective transmission distance taking into account of the fiber
attenuation, and it is given by

zeff =

1 e z

(2.11)

31
Chapter 2: Optical Fiber Characteristics and System Configurations

Note that (2.10) is derived under the assumption that a plane wave propagates in an
infinite uniform medium. The first term in (2.10) is just a linear phase shift, and depends
only on the transmission distance z . On the other hand, the second term in (2.10)
depends not only on the transmission distance z by way of zeff , but also the intensity
variation of the signal I ( ) itself. Therefore, this effect is called self phase modulation
(SPM). The intensity dependence of refractive index causes a nonlinear phase shift,
which is proportional to the intensity of the signal I ( ) . It should be noted that zeff is
less than z . This implies that fiber attenuation reduces the effect of nonlinear phase shift.
SPM alone broadens the signal spectrum, but does not affect the intensity profile
of the signal. The spectral broadening effect can be understood from the fact that the
time-dependent phase variation causes instantaneous frequency deviation ( ) , which
is given by

( ) =

2 n2 zeff I ( )

=
.

(2.12)

(2.12) suggests that the magnitude of instantaneous frequency deviation increases with
distance and the intensity variation of the signal. The spectrum broadening effect caused
by SPM may result in interchannel cross talk among channels in WDM systems.
Although SPM alone does not affect the intensity profile of the signal, that is no
longer true in the presence of dispersion. The combined effect of dispersion and SPM
strongly depends on the sign of dispersion (the sign of D ). In terms of the sign of
dispersion, the operating wavelength region is divided into two regimes: normal
dispersion and anomalous dispersion regimes. The normal dispersion regime corresponds
to the wavelength range in which dispersion D is negative ( 2 is positive) whereas the
anomalous dispersion regime is the wavelength region in which D is positive ( 2 is
negative). As seen from (2.12) the sign of SPM-induced instantaneous frequency
variation is independent of the sign of dispersion; however, the sign of dispersioninduced instantaneous frequency variation depends on the sign of dispersion. For the
propagation of an optical pulse, the SPM enhances the effect of dispersion when the
operating wavelength is in the normal dispersion regime. That is, the pulse is broadened
more severely than under the effect of dispersion alone in the normal dispersion regime.
32
Chapter 2: Optical Fiber Characteristics and System Configurations

This results from the fact that the sign of instantaneous frequency variation across the
pulse induced by the SPM is identical to that induced by the dispersion [66]. Conversely,
the SPM negates the effect of dispersion in the anomalous dispersion regime.
When the SPM is sufficiently strong, a pulse can be compressed during initial
propagation. In addition, under ideal conditions the pulse shape can be continually
preserved during propagation due to the balancing of the effects of dispersion and SPM in
the anomalous dispersion regime. This type of pulse is called a soliton [8]-[10]. However,
the effect of SPM decreases with transmission distance due to fiber attenuation,
preventing soliton formation [67].
It is seen from (2.10) that the effect of nonlinear refractive index, causing
nonlinear phase shift, is proportional to the signal intensity. When multiple channels are
simultaneously transmitted, the nonlinear phase shift on a given channel depends not only
on the intensity of that channel but also the intensities of the other channels. This is
discussed in the next section
2.2.4 Cross-Phase Modulation (XPM or CPM)

Similar to the SPM, cross-phase modulation (XPM) is due to the nonlinear

behavior of the refractive index on the optical intensity. However, in this case the total
nonlinear phase shift on a given channel is due to the combined intensities of all
transmitted channels, which can result in cross talk among WDM channels. When N
channels are transmitted in a single optical fiber, the nonlinear phase shift on the j th
channel is governed by [1]

j ( z , ) =

2 n0 z

N
2 n2 zeff

+
I j ( ) + 2 I m ( ) ,

m j

(2.13)

where I m ( ) is the optical intensity of the m th channel. The first term in the parentheses
on the right hand side of (2.13) corresponds to SPM discussed in the previous section
whereas the second term is responsible for XPM. The factor of 2 in (2.13) suggests that
the effect of XPM from a neighboring channel is two times stronger than that caused by
SPM itself.
In the case of continuous-wave signals, XPM would dominate over SPM as
suggested by (2.13). However, when pulses at different wavelengths are considered, the
33
Chapter 2: Optical Fiber Characteristics and System Configurations

effect of XPM depends on the relative temporal locations of those pulses. XPM is
strongest when pulses completely overlap one another. Additionally, the probability that
all channels transmit bit 1 has to be taken into consideration, similar to the case of SRS.
The low probability of all channels simultaneously transmitting bit 1 reduces the effect of
XPM on average. Under the presence of dispersion pulses at different wavelengths travel
at different group velocities, which in effect causes pulses to walk off from one another,
thus reducing the effect of XPM [68], [69]. The larger the dispersion discrepancies
among channels, the more rapidly the pulses walk off from one another. In other words,
the effect of XPM is inversely proportional to dispersion discrepancies among channels
in WDM systems. Thus, in order to minimize the impairment caused by XPM, the
channel separation and/or local dispersion have to be properly chosen in WDM systems,
which can be achieved in practice.
2.2.5 Four-Wave Mixing (FWM)

The intensity dependence of refractive index not only causes nonlinear phase shift
but also gives rise to the process by which signals at different wavelengths are mixed
together producing new signals at new wavelengths. This process is known as four-wave
mixing (FWM). The difference between the processes (SPM and XPM) causing the
nonlinear phase shift and the FWM process is that energy transfer occurs in the FWM
process. When signals at frequencies fi , f j and f k propagate along an optical fiber, the
nonlinear interactions among those signals by mean of FWM result in the generation of
new signals at [70]

fijk = f i + f j f k .

(2.14)

The energies from the interacting signals are transferred to those newly-generated signals
at frequencies fijk . In the simplest case when only two signals at frequencies f1 and f 2
are involved, the FWM generates new signals at 2 f1 f 2 and 2 f 2 f1 as demonstrated in
Fig. 2.5 (a). The number of FWM-generated signals grows rapidly with the number of
involved signals as shown in Fig. 2.5 (b), which corresponds to when three signals
undergo FWM generating new signals at frequencies given by (2.14).

34
Chapter 2: Optical Fiber Characteristics and System Configurations

When N signals are involved in the FWM process, the number of FWMgenerated signals is given by [70]

M=

N 2 ( N 1)
.
2

(2.15)

For example, when 10 signals are four-wave mixed together, 450 signals are generated.
In WDM systems, not only does the FWM cause power depletion on the participating
channels, but it also results in cross talk among channels. The cross talk comes from the
fact that some of the FWM-generated signals can have the same frequencies as the WDM
channels when all channels are equally spaced in frequency. One approach to avoid cross
talk is to employ unequally spaced channel allocation (USCA) [28], [70]. In this method,
all channels are allocated so that the FWM-generated signals do not have the same
frequencies as the channels. The illustration of this method is shown in Fig. 2.5 (b) for
the case of three transmitted channels. It should be noted that in this method only cross
talk is avoided, but the signals still suffer excess power loss.
Mathematically, the power of a FWM-generated signal at the frequency fijk given
by (2.14) is proportional to [71]

z
2
1 + 4e sin ( z 2) ,
z 2
2 + 2

1
e
(
)

(2.16)

where is called FWM efficiency and is the phase mismatch, which is given by

2 2
dD 2

f
f
f
f
D
=

+
i
fi f k + f j f k . (2.17)
k
j
k
d 2c

The phase mismatch in (2.17) generally represents the difference of propagation

constants of the original signals and the FWM-generated signals. From (2.16), one can
see that the FWM efficiency decreases with the increase in the phase mismatch .
The phase mismatch is proportional to the frequency spacing between the original
signals ( fi f k and f j f k ), and the dispersion D . Hence, for systems employing
optical fibers whose dispersion is small, the channel spacing has to be sufficiently large
in order to avoid phase matching. On the other hand, channels can be packed closer
together if the local dispersion D is large. Shown in Fig. 2.6 is the FWM efficiency as
35
Chapter 2: Optical Fiber Characteristics and System Configurations

a function of channel spacing f when three interacting signals at frequencies fi , f j

and f k are equally spaced. The solid line corresponds to a standard single-mode fiber
(SSMF) with D = 16 ps/(kmnm). The dash line represents a dispersion-shifted fiber
(DSF) when D = 1 ps/(kmnm). In the case of SSMF, the effect of FWM becomes

negligible when channels are equally spaced apart by several tens of GHz. On the other
hand, the effect of FWM is still not negligible even when the channel spacing is 50 GHz
in the case of DSF. For WDM systems, large local dispersion is the preferred mean to
suppress the FWM because the ultimate goal of employing WDM is to transmit as many
channels as possible for a given available transmission bandwidth [72].
One may notice that nonlinear interchannel interactions discussed so far decrease
with the increase in local dispersion of the transmission fiber. This appears to suggest that
transmission fibers should have very large dispersion to make nonlinear effects
negligible. The accumulated dispersion can be removed by concatenating the
transmission fiber with an optical fiber having the opposite amount of dispersion.
However, practical system implementations have to be taken into consideration. There
exist trade-offs among different system implementations. This is presented in the next
section.

2.3 SYSTEM CONFIGURATIONS

In terms of dispersion, transmission optical fibers can be divided into three
categories: small, moderate, and large local dispersions. Each is designed for specific
system applications. It is seen from the previous section that large local dispersion is
beneficial in term of nonlinearity suppression. However, such large local dispersion
results in large accumulated dispersion, which can make dispersion compensation
difficult. Thus, in terms of practical implementations, every aspect of a system has to be
considered, including suppression of impairments caused by dispersion and nonlinearity,
and feasibility of such implementations.
Three types of transmission optical fibers are considered in this dissertation: a
dispersion-shifted fiber (DSF), TrueWave reduced slope fiber (TRSF), and standard
single-mode fiber (SSMF). Their characteristics are summarized in Table 2.1. Among the
36
Chapter 2: Optical Fiber Characteristics and System Configurations

three the DSF has lowest dispersion whereas the SSMF has largest dispersion in the 1.55
m wavelength window. For the TRSF, its dispersion is the compromise between the
DSF and SSMF, and it is generally referred to as NZDF. In this dissertation, DSF
represents the transmission fiber having small local dispersion whereas TRSF and SSMF
are used to represent transmission fibers having moderate and large local dispersion,
respectively. It should be noted that there are several optical fibers that have moderate
dispersion in the market. All of them are comparable in terms of characteristics.
2.3.1 Small Local Dispersion

The apparent advantage of the DSF is that it has low dispersion in the 1.55-m
wavelength window. Hence, the effect of dispersion is not strong, but the nonlinear
effects are no longer negligible. By choosing the operating wavelength so that it is in the
anomalous dispersion regime ( D > 0 or 2 < 0 ), the pulse compression effect can be
exploited to maximize the system performance. This is due to the fact that in the
anomalous dispersion regime the effect of SPM counteracts the dispersion.
With a small local dispersion, dispersion compensation is not necessary; however,
WDM deployment with this type of optical fibers is not efficient. This is because small
local dispersions implies that the channel spacing between two WDM channels has to be
large in order to suppress the interchannel nonlinear impairments; such as XPM and
FWM. In addition, the operating wavelength of each channel has to be sufficiently far
from the zero-dispersion wavelength in order that the dispersion on that channel is
sufficient to suppress the FWM between the signal and the ASE noise in systems
employing in-line optical amplifiers. This introduces an additional constraint on the
utilization of available bandwidth, hence further limiting the number of channels that can
be transmitted.
2.3.2 Moderate Local Dispersion

Initially, the purpose of having moderate local dispersion in an optical fiber is to

keep the effect of dispersion under control while the effect of fiber nonlinearity is
minimized [59], [60], [73]. The TrueWave fiber enables the transmission of multiple 10Gb/s channels up to several hundreds kilometers without the need of dispersion
37
Chapter 2: Optical Fiber Characteristics and System Configurations

compensation [59]. The specification of the TrueWave reduced slope fiber (TRSF) is
shown in Table 2.1.
The effect of dispersion does not scale linearly with bit rate. In fact, the effect of
dispersion grows quadratically with bit rate [1] since the pulse bandwidth grows linearly
with bit rate. (This assumes, as is true in modern high speed fiber optic systems, that the
linewidth of the signal source is small compared to the modulation bandwidth.) Hence, at
a bit rate of 40 Gb/s the effect of dispersion is 16 times more severe than at 10 Gb/s, and
dispersion compensation is necessary. In the case of single-span systems, the dispersion
compensation is performed at the receiver by placing a negative-dispersion optical fiber,
commonly called a dispersion compensating fiber (DCF), at the end of the fiber span as
shown in Fig. 2.7 (a). On the other hand, in the case of multiple-span systems the
accumulated dispersion is generally compensated periodically along the transmission link
at the in-line amplifier sites as shown in Fig. 2.7 (b). Note that in the last span the DCF is
at the receiver site. Periodic dispersion compensation is also known as dispersion
management. In terms of accumulated dispersion for a given span length, optical fibers
having moderate dispersion also provide another advantage compared with those having
large local dispersion. Smaller accumulated dispersion implies that a shorter length of
DCF is needed.
When DCF is considered, it is desirable that the DCF has large negative
dispersion so that only a short length of DCF is required, assuming the transmission fiber
(e.g. SSMF) has positive dispersion. The advantages of having a short length of DCF are
that it does not take up much space at the in-line amplifier site, and that the additional
loss introduced by the DCF is reduced. Note that the DCF commonly has higher loss than
other types of optical fibers. In this dissertation, we choose the extra-high-slope
dispersion compensating fiber (EHS-DK) as the dispersion compensating fiber due to its
high dispersion. Its characteristics are shown in Table 2.1.
2.3.3 Large Local Dispersion

In general, the larger the dispersion, the lesser are the effects of nonlinearity. This
is the apparent advantage of employing optical fibers having large local dispersion
because ideally a linear transmission fiber is desired. However, large local dispersion in
38
Chapter 2: Optical Fiber Characteristics and System Configurations

transmission fibers results in large accumulated dispersion, which requires long length of
the DCF. Such long length of the DCF may cause problems in the practical
implementations especially when the span length is large. In addition, long length of the
DCF also introduces high additional loss. For example, if the span length is 160 km, the
length of EHS-DK required is 11.72 km to manage the end-to-end dispersion back to zero
when SSMF is employed as the transmission fiber. On the other hand, the length of EHSDK needed is reduced to only 2.76 km in the case of TRSF.
One approach to take advantage of the optical fiber having large local dispersion,
such as the SSMF, is to perform dispersion compensation within the transmission link
instead of at in-line amplifier sites. In this approach the loss associated with the
dispersion compensation is counted as transmission loss. The schematic diagram of this
approach is shown in Fig. 2.7 (c) and (d) corresponding to single-span and multiple-span
systems, respectively. As seen from Fig. 2.7 (c) and (d), there are two transmission fibers
within a given span. When the SSMF is employed as the first transmission fiber, the
dispersion compensation performed by the second transmission fiber is practically
accomplished by using reverse-dispersion fiber (RDF). The RDF is designed for
compensating the accumulated dispersion introduced by the SSMF. The characteristics of
an RDF are listed in Table 2.1. Note that the RDF has larger absolute dispersion than the
SSMF. This is beneficial since the RDF is more nonlinear than the SSMF due to its much
smaller effective core area Aeff . Longer length of the SSMF deployed implies lower
signal power entering the RDF fiber, hence reducing the unwanted effects of nonlinearity
in the RDF fiber. One drawback of this approach is that any given span consists of two
types of optical fibers, which may cause difficulties in installation and maintenance
especially when a fiber cut happens. In addition, this approach cannot be deployed to
upgrade already-installed systems which employ SSMFs as transmission fibers.

2.4 SUMMARY
In this chapter, the properties of optical fibers are discussed. When the
nonlinearity is considered, their deleterious effects decrease with the increase in the local
dispersion of the transmission fiber. Although large local dispersion is desirable in order
to suppress nonlinear effects, large accumulated dispersion makes dispersion
39
Chapter 2: Optical Fiber Characteristics and System Configurations

compensation necessary. However, dispersion compensation introduces additional loss

and complexity to systems. This leads to various system configurations in terms of
transmission fibers deployed. Each configuration has its own advantages and drawbacks.
They are all discussed and explained. The system configuration is not the only key factor
that improves system performance. The optical signal formats also play an important role
in maximizing system performance. That is the subject of the remainder of this
dissertation. The principal signal formats considered in the literature defined and
compared in the next chapter.

40
Chapter 2: Optical Fiber Characteristics and System Configurations

Fig. 2.1: The values of material and waveguide dispersions as a function

of wavelength for a standard single-mode fiber [58].

Fig. 2.2: Index profiles of core and cladding for various types of optical
fiber. Vertical scale is refractive index, and horizontal scale is radius of
optical fiber. Note that the center of fiber is at the middle of index profile.
(a) Standard single-mode optical fiber, and (b) Dispersion-shifted fiber
[58].

41
Chapter 2: Optical Fiber Characteristics and System Configurations

Fig. 2.3: Raman gain coefficient g R as a function of Strokes frequency for

fused silica at the pump wavelength of 1 m. Note that 1 cm-1 = 30 GHz
[61].

Fig. 2.4: Schematic explanation of SRS for 2-channel system. Horizontal

scale represents time, and vertical scale represents signal level. (a) Bit
patterns of channel 1 and 2 without SRS. (b) Effects of SRS on both
channels [15].

42
Chapter 2: Optical Fiber Characteristics and System Configurations

(a)

(b)
Fig. 2.5: Illustrations of FWM process. Horizontal axis is frequency. (a)
FWM of two signals at frequencies f1 and f 2 . (b) FWM of three signals
at frequencies of f1 , f 2 and f3 [28].

Fig. 2.6: Four-wave mixing efficiency as a function of channel spacing

f = fi f k = f j f k at 1.55 m. The solid line corresponds to standard

single-mode optical fiber with D = 16 ps/(kmnm). The dash line
represents dispersion-shifted fiber when D = 1 ps/(kmnm) [15].
43
Chapter 2: Optical Fiber Characteristics and System Configurations

Table 2.1: Summary of fiber characteristics considered in this

dissertation.

Dispersion D
[ps/(kmnm)]

Effective Core
Area Aeff
(m2)

Nonlinear
Refractive
Index n2
(10-20 m2/W)

Standard
Single-Mode
+17
85
2.3
Fiber
(SSMF)
DispersionShifted Fiber
0
70
3.0
(DSF)
TrueWave
Reduced Slope
+4
55
2.7
Fiber
(TRSF)
ReverseDispersion
25
2.3
26.91
Fiber
(RDF)
Extra-HighSlope
Dispersion
19
2.7
232
Compensating
Fiber
(EHS-DK)
Note: For DSF, it is also known as large effective area fiber (LEAF).

Attenuation
(dB/km)

0.2

0.2

0.2

0.22

0.59

44
Chapter 2: Optical Fiber Characteristics and System Configurations

Transmission
Fiber

DCF

Tx

Rx
Preamp
.
Receiver
Site

(a)

Transmission
Fiber

DCF

Transmission
Fiber

Transmission
Fiber

DCF

DCF

Tx

Rx
Preamp
.
In-Line
Amplifier Site

Receiver
Site

In-Line
Amplifier Site

(b)

Fiber #1

Fiber #2

Tx

Rx
Preamp
.
Receiver
Site

(c)
Fiber #1

Fiber #2

Fiber #1

Fiber #2

Fiber #1

Fiber #2

Tx

Rx
Preamp
.
In-Line
Amplifier Site

In-Line
Amplifier Site

Receiver
Site

(d)
Fig. 2.7: Schematic diagrams of considered system configurations. (a)
Single-span system with dispersion compensation at receiver. (b)
Multiple-span (Dispersion-managed) system employing DCF. (c) Singlespan system employing two types of transmission fibers. (d) Multiple-span
(Dispersion-managed) system employing two types of transmission fibers.

45
Chapter 2: Optical Fiber Characteristics and System Configurations