FIBER OPTICS

Prof. R.K. Shevgaonkar

Department of Electrical Engineering

Indian Institute of Technology, Bombay

Lecture: 08

Analysis of Signal Distortion in Optical Fiber

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 1
 The spectral width of an optical source is defined as the range of wavelengths
that the source emits around the main central wavelength for which it was designed.
This range of wavelength is denoted by ∆λ and can be calculated as given below:

(8.1)

In this relation ‘f’ and ‘λ’ are the frequency and wavelength respectively of the
optical source and ‘∆f’ is the frequency domain equivalent of the spectral width of the
source. For example, equivalent frequency domain width for about 1µm wavelength
source (like LASERs) having a spectral width of the order of 1nm is about a few
hundred Giga-hertz. One of the fundamental differences between information
transmission at radio frequencies and that in optical domain is that, at radio
frequencies, the spectral width of the carrier signal is so small that it is neglected and
the frequency spectrum of the carrier signal is represented by a delta function at the
carrier frequency. That is why when we consider amplitude modulated radio
transmission; the spectrum of the modulated signal shows three sharp peak points
represented by delta functions as shown in the figure 8.1 below:

Figure 8.1: Spectrum of Radio Frequency Modulated signal

As the carrier frequency increases to the optical domain the spectral width of
the carrier signal cannot be neglected. That is why, in the optical domain, the carrier
signal cannot be represented by a delta function as in case of radio carrier signal.
The spectrum of its carrier signal is like a distributed frequency function about the
central frequency of the source for which the source was designed. This situation is
shown in the figure 8.2 below which shows the frequency domain equivalent of the
carrier signal source output. The output of a typical optical source like 1µm
wavelength LASER, which has a spectral width of the order of 1nm, when fed to a
Spectrum Analyser, shows an output as shown below in figure 8.2.

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 2
 Figure 8.2: Frequency spectrum of the un-modulated optical carrier signal.

If the optical carrier signal shown in figure 8.2 above is amplitude modulated
by modulating signal having a modulating frequency ‘fm’ , the sidebands in fact would
completely be masked by the intrinsic spectral width of the optical carrier signal due
to which the modulated signal would not be able to carry any useful information. This
causes the basic requirements for distortion-less system to remain invalid. Hence, a
modification in the basic understanding of distortion on the basis of the distortion in
radio frequency communication is needed.

When an optical signal is transmitted on an optical fiber, the signal is distorted

owing to two phenomena. These phenomena are known as dispersion and
attenuation. The following diagram shows the constituents of each phenomenon that
contribute to the distortion of optical signal in the optical fiber.

Figure 8.3: Constituents causing Distortion in Optical Fibers

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 3
 When light is launched into an optical fiber in the form of a light pulse, the
pulse energy propagates by different paths which lead to the broadening of the
actual time-width of the pulse. This phenomenon is called as dispersion and has
already been discussed. (The reader is assumed to have been familiar with pulse
broadening by now). Dispersion may be due to material properties and imperfections
too. It may also be to pulse broadening within a single mode which is referred to as
intra-modal. Intra-modal dispersion occurs in single mode fibers too on account of its
finite bandwidth. The dispersion caused by multipath propagation of light energy is
referred to as inter-modal dispersion and has already been discussed. Inter-modal
dispersion does not occur in single mode fibers.

Another cause of signal distortion in optical fibers is the attenuation of the

optical signal inside the optical fibers. Attenuation can occur due to material
absorption properties too. During manufacture process of the optical fiber, some
imperfections called as micro-centres are created inside the optical fibers which have
dimensions of the order of λ-4. These micro-centres lead to scattering of light inside
the optical fibers and as a result the light signal is attenuated. When optical fiber
cables are laid for communication, most often, it is not possible to align them in
perfect straight lines and some bends remain here and there. These micro-bends
lead to spurious leakage of light energy causing loss of optical energy. Such
leakages can also be witnessed within the laboratory by deliberately bending the
optical fiber into which light is launched. When fibers are laid in the form of large arcs
or ‘macro-bends’, another phenomenon referred to as radiation creeps in causing
loss of light energy and thus leading to attenuation of the optical signal within the
optical fiber.

As an example, let us now consider a case of optical transmission in which we

have an optical carrier signal of frequency of the order of 10 14 Hz and the highest
possible data rate that is possible presently, i.e. about 1010 Hz which is the
modulating signal. The quality factor in this transmission is given by

The fractional bandwidth of this transmission can be calculated as:

The loss profile of an optical fiber with respect to frequency (or wavelength) is
a very smooth variation and is very slowly varying function of wavelength. Thus if we
consider a band equal to the fractional bandwidth (10 GHz) around the carrier
frequency, we would find almost no differential loss in such a tiny band. This means,
the signal amplitude response of the input light signal over this region almost
remains constant and this satisfies the first requirement of a distortion-less system.
Thus the loss on the optical fiber decreases the amplitude of the signal almost

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 4
uniformly over the entire bandwidth. Hence, there would be no distortion in the
optical fiber due to amplitude variation. The only question that remains now is,
whether the phase requirement of a distortion-less system is satisfied by the optical
fiber.

The phase performance of an optical fiber is inherently related to the

phenomenon of dispersion. Let us now try to visualize the dispersion in an optical
fiber. Figure 8.4 below shows two light pulses, launched into an optical fiber, which
propagate as a function of time. Here the shape of the light pulse is not given
importance and just the presence/absence of light indicates 1/0 bit in the
transmission.

Figure 8.5: Pulse broadening in an optical fiber

As is obvious from the above figure, as the pulses progress inside the fiber
they start to overlap and the effective pulse width broadens. Since the light energy
has to be conserved inside the fiber, the amplitude of the pulse reduces as a result
of this broadening. This overlapping or merging of adjacent pulses is also sometimes
referred to as inter-symbol interference (ISI). Thus distortion occurs in the data
transmission due to this pulse broadening inside the optical fiber. After considerable
overlapping, it would be difficult to even predict the presence of two distinct pulses.
Before this happens, we have to regenerate the pulses which are done by optical
repeaters. The important notion to realise here is that, the pulse broadening effect

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 5
poses some kind of a restriction on the maximum length of fiber that can be used
without repeaters. If a transmission is made beyond this length, the pulse broadening
will not allow any useful output to be detected at the receiver end. So when light
pulses propagate in an optical fiber, two significant effects take place; firstly the
pulse broadens and secondly the peak amplitude of the pulse reduces as a result of
this broadening. Since at the detecting end we do not integrate the pulse over the
pulse width but just sample the output at regular intervals of time, the Signal-to-Noise
ratio also gets affected due to the distortion.

Let us now go back to our discussion on the distributed nature of the optical
carrier signal and try to sketch the carrier and the modulating signals in both radio
and optical frequency domains.

Figure 8.6: Frequency spectrum of carrier and modulating signals

In the analysis of a radio system, since the side bands in the modulated signal
are clearly visible, both spectral domain as well as time domain techniques can be
employed in the demodulation (detection) of radio frequency signals. But the
scenario in the optical frequency domain is not so. Due to the wide spectral width of
the carrier signal (as seen from the figure8.6) the modulated signal, which is just a
convolution of the carrier and the modulating signals, has very little trace of
modulation and as such there is a very little change in the nature of the carrier signal
spectrum. This renders the spectral domain techniques to be of no use in
demodulation of optical signals and so time domain approach has be used to decode
optically modulated carriers. The only time domain modulation/demodulation scheme
which does not require any knowledge about the frequency spectrum of the
modulated signal, is the amplitude modulation/demodulation (with carrier). This is
because, for such a signal the modulating signal can be recovered from the
modulated signal by simple envelope detection technique. That is why optical
communication system uses simple amplitude modulation/demodulation for data
communications. The use of simple amplitude modulation/demodulation projects a
picture of the optical communication system as the most primitive system of
communication system. This is because, simple amplitude modulation/demodulation
technique was already used in radio-communication, and that too about hundred
years ago.

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 6
 To visualise the distortion caused by the wide spectral width of the carrier
signal we may assume the wide-band optical carrier to be actually composed of
numerous narrow-band individual carrier signals of small bandwidths which are
transmitted simultaneously. This would be clearer from the figure 8.7 below:

Figure 8.7: Optical Carrier expressed in terms of numerous narrowband carriers.

The task now is to modulate this composite carrier signal according to the
modulating signal. In other words, we have to change the envelope of this wideband
carrier signal according to our modulating signal. Each narrow-band carrier is
modulated by the modulating frequency ‘fm’ and all these modulated carriers are
simultaneously transmitted. This situation can be visualised as a simultaneous or
parallel transmission of numerous individual amplitude modulated signals having
different carrier frequencies. Each of these signals has its individual side bands, but
all such different sidebands are completely merged together and no individual side
bands are visible. From the b-V diagram of an optical fiber, we can observe that
propagation constant is a non-linear function of frequency. This means that different
wavelengths (or frequencies) in an optical fiber travel with different speeds. Hence,
all carrier signals that compose the wide band optical carrier signal, each travel with
its corresponding speed and also the information which it carries. Since the
modulating signal is same for all the carriers, same information takes different times
to travel a given length of fiber; or more specifically, the same pulse takes different
times to reach the output of a given length of optical fiber causing an effective pulse
broadening. And this is what is called as the material or the intra-modal dispersion.
Thus, larger the spectral width of the optical carrier signal more is the dispersion or
pulse broadening. In other words we can say that pulse broadening is related to the
spectral width of the carrier signal.

If the optical carrier signal shown in figure 8.7 is now amplitude modulated by
a pulse and observed in the time domain, we would observe a situation that is
depicted in figure 8.8. In this we figure we consider the optical carrier signal to be
modulated by a pulse of the modulating signal and also assume that the modulation
index to be constant for all the assumed individual carrier frequencies of figure 8.7.
We consider a time domain propagation of the modulated carrier through the fiber
core with the frequency axis, normally outwards to the plane of the paper and
indicate the corresponding pulse amplitudes as shown in the figure. Just before the
modulated optical carrier signal pulse enters the optical fiber core (i.e. at t0) it may be

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 7
considered to be composed of a number of pulses which effectively are the
individually modulated carrier pulses in the wide-band carrier signal, all modulated to
the their corresponding indices. When this carrier signal pulse enters the optical fiber

Figure 8.8: Pulse broadening effect

core, the process of pulse broadening due to intra-modal or material dispersion

starts and the pulses start separating out from each other as shown in the figure 8.8.
In the figure Ai is the amplitude of the pulse correspond to the f ith frequency
component in the optical carrier signal. The fn+rth pulse travels faster than the fnth
pulse and hence separates out faster as shown in the figure above. The question
now is that what would the effective shape of the pulse inside the fiber core be? The
answer to this doubt is rather hidden in the figure 8.8. If we notice carefully, the
shape of the pulse inside the optical fiber core would just be a projection of the
shape of the carrier spectrum at t0 and hence would be dominated by the shape of
the frequency spectrum of the carrier signal. This observation seems weird in the
very first instance. It is because of the fact that we were tuned to the knowledge that
the shape of the modulating spectrum determines the shape of the resultant
modulated spectrum. But in this case the scenario is a bit different because the
carrier spectrum now decides the effective shape of the modulated pulse. Thus in
case of optical communication, though the modulation scheme is as simple as
amplitude modulation, yet there are fundamental differences that exists in it. The
bandwidth required in optical communication is decided by carrier frequency
whereas in radio-frequency communication, the band-width requirement is
determined by modulating signal (which according to Nyquist’s theorem was two
times of the maximum frequency component in the modulating signal) irrespective of
whatever the carrier frequency used. Thus in an optical communication system the
pulse broadening depends mainly on two aspects- first, the spectral width of the
carrier signal and secondly, the distance travelled by the pulse inside the optical fiber
core. Also the bandwidth requirement and the shape of the pulse is determined by
the spectrum of the carrier signal. In other words, the modulating signal almost loses
Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 8
its identity once modulated on to the optical carrier signal and this is the reason why
digital information is decoded on the basis of the presence or absence of a pulse.

On the above backdrop of a basic understanding, let us now proceed to a

rather technical picture of dispersion which helps us to have a quantitative view of
dispersion in an optical fiber. The group velocity of the group of pulses of light (in
figure 8.8) inside the optical fiber is defined by:

( ⁄ ) (8.2)

Hence, the group delay per unit length is given as,

(8.3)

This equation then helps us to calculate the effective pulse broadening which
is the delay between the first and the last pulse. Hence, the pulse broadening can be
written as:

{ } (8.4)

Here ‘σλ‘ is the spectral width of the source and ‘c’ is the velocity of light. We
can now formally define the dispersion as the pulse broadening per unit distance per
unit spectral width of the source. It is denoted by ‘D’ and has a unit of (pico-
second/kilometre/nanometre). That is

{ } (8.5)

Dispersion is rather a weak phenomenon. In other words, any of the sources

of dispersion mentioned in figure 8.3 can be calculated by considering the others to
be negligible small and then after calculating similarly for all the other sources, just
add them linearly to get the net dispersion in the optical fiber. The meaning of this
statement is that none of the dispersive sources are strongly linked to one another
and hence can be independently calculated. Let us now have a better understanding
of these sources of dispersion one by one by talking about material dispersion first.

For calculating material dispersion, we assume that the optical fiber is an

infinite medium and there is no wave guiding of light in the material of the optical
fiber and the refractive index of the material is a function of the wavelength as we
have already seen from the b-V diagram. In such an infinite medium, light would
exist in the form of transverse electromagnetic wave and the phase constant of the
light would be that of a transverse electromagnetic wave, which is given by

( ) (8.6)

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 9
 Here n(λ) is function which depicts the refractive index variation with respect
to wavelength. To find an expression for dispersion, we have to first calculate the
group delay. Group delay can be calculated by differentiating equation (8.6) with
respect to angular frequency ‘ω’.

(8.7)

Now, by differentiating equation (8.7) with respect to the wavelength we would

get the resultant expression for the material dispersion as:

( )
(8.8)

From equation (8.8) , we see that the material dispersion is directly

proportional to the wavelength as well as to the second derivative of the refractive
index function. But without the knowledge of the actual variation of the refractive
index with respect to wavelength, we cannot predict the material dispersion. The
second derivative of any function (if it exists) indicates the curvature of the function
about the point which it is calculated. If we plot an experimentally calculated graph of
the refractive index as a function of wavelength for the material glass, we would get
a plot as shown in figure 8.9.

Figure 8.9: Variation of Refractive index with wavelength

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 10
 The graph shows that the refractive index of glass decreases as the
wavelength of light increases, but this decrease is non-linear and has a change of
curvature at an wavelength of about 1.27µm (1270 nm). Note that this is curve is
experimentally plotted and is of glass which is not moulded into optical fibers. So this
a natural curve for glass as a material. The wavelength 1270nm has a special
importance, not only that the curvature of graph changes at this wavelength but also
the second derivative of the curve at this wavelength is zero. This means that if we
use light of 1270 nm in the optical fiber we would get either very low or even no
material dispersion and hence would have the highest possible data rate for
communication. The material dispersions corresponding to different windows of
optical communication is tabulated below:

Wavelength (λ) (nm) Dmaterial (ps/Km/nm)

850 85
1310 0.1
1550 -20

As we can see from the table, initial optical communications were very
dispersive with dispersions ranging as high as 85 ps/Km/nm. In the second window
the dispersion value is very low and so the 1310 nm window has some special
importance. The third window of operation has a dispersion of -20ps/Km/nm. The
significance of the negative sign would be illuminated later. So it is preferable to
perform optical communications in the 1270nm wavelength or to keep the operating
wavelength as close to it as possible in order to have minimum distortion and high
data rates.

Fiber Optics, Prof. R.K. Shevgaonkar, Dept. of Electrical Engineering, IIT Bombay Page 11