Analysis of Second order Dispersion on Free Space Optical Propagation

Analysis of Second order Dispersion on Free Space

Optical Propagation
David F. W. Yap1, Y. C. Wong1, C. Wasli1, S. P. Koh2,
S. K. Tiong2, M. A. E Mohd Tahir3
Faculty of Electronics and Computer Engineering,
Universiti Teknikal Malaysia Melaka (UTeM)
Hang Tuah Jaya, 76100, Durian Tunggal, Melaka, Malaysia.
1

College of Engineering, Universiti Tenaga Nasional (UNITEN) km 7,

Kajang-Puchong Road, 43009 Kajang, Selangor Darul Ehsan, Malaysia.

Faculty of Engineering & Technology, Multimedia University, Jalan Ayer

Keroh Lama, 75450 Melaka, Malaysia.

david.yap@utem.edu.my, ycwong@utem.edu.my,
johnnykoh@uniten.edu.my, siehkiong@uniten.edu.my, censplit@gmail.com
Abstract
Free space optic (FSO) can be regarded as a
potential and attractive option to fiber optic.
FSO has the ability to go beyond the limit of
fiber optics. Unfortunately, due to the dispersion
effect in the atmosphere, FSO suffers from
signal loss and attenuation. Thus, practical
and detailed research is needed to improve the
system. Simulation on FSO propagation using
measured parameter values is important to gain
better understanding and level of accuracy on
the pulse behavior in free space. Using MATLAB
as the simulation platform and with the help of
experimental parameter values, an accurate
model can be obtained and studied. This will
allow some level of prediction on the behavior
of the propagating light pulse in the atmosphere
and subsequently the FSO performance can be
further improved.
Keywords: Atmospheric turbulence, Binary
Pulse Position Modulation, Bit Error Rate,
Simulation System.

I. INTRODUCTION
Laser communication in free space offers
an attractive alternative for transferring
high-bandwidth data when optical fiber
cable is either impractical or not viable.
Here, wireless optical connectivity can
be used as the last mile to connect fiber
backbone to end users, such as from

ISSN: 2180 - 1843

Vol. 2

building to building, due to the cost

and time-consumption on top of the
impossibility and impracticality in laying
down optic fibers [1]. Other advantages
of adopting the optical wireless
communication systems, also termed as
free space optics (FSO) or lasercom (laser
communications), includes [2]:a) no licensing or tariffs fees required
for its utilization [3];
b) small, lightweight and compact;
c) ease of installation and deployment
(digging up of road is
unnecessary);
d) it offers very high data rates due to
its large bandwidth;
e) high security fears (the extremely
directional, narrow beam optical
link makes eavesdropping and
jamming
nearly impossible);
f) it operates at low power
consumption;
g) there are no rf radiation hazards
(eye-safe
power
levels
are
maintained).
However, random fluctuations in the
atmospheres refractive index can
severely degrade the wave front of a
signal-carrying laser beam, causing the
receiver to suffer from intensity fading.

No. 1

January - June 2010

59

Journal of Telecommunication, Electronic and Computer Engineering

This results in increased system bit error

rates (BERs) particularly along horizontal
propagation paths [4].
Research related to pulse propagation
in both fiber optic and FSO show the
propagating pulse is affected by both linear
and nonlinear elements. The linear effects
include the group velocity dispersion
(GVD) and third order dispersion (TOD),
while the nonlinear effects comprise of self
phase modulation (SPM). Both the linear
and nonlinear effects are responsible for
pulse broadening as well as distortion [5].
Based on the severity of these effects, data
reliability can be compromised and may
lead to the increase in BER. In fiber optic
the extent of these effects can be estimated
and anticipated through numerous
literatures and research. Unfortunately
for FSO the extent of these effects cannot
be estimated easily due to the random
nature of the atmosphere. Thus, it is
important to have an accurate prediction
model to estimate pulse behavior in the
atmosphere.
In this paper, the simulation on FSO is
carried out without the nonlinear effects.
The nonlinear Schrdinger equation is
briefly discussed in Section II while the
type of pulses used in the simulation is
shown in Section III. Simulation results on
the second order dispersion are presented
in Section IV. The conclusion is given in
Section V.

II. NONLINEAR SCHRDINGER

EQUATION
The Nonlinear Schrdinger Equation
(NLSE) is used to mathematically explain
varying pulse envelope propagating
in a medium with linear and nonlinear
elements. Thus, NLSE is suitable for
describing pulse propagation in free
space. Numerical solution for NLSE
can be obtained by applying split step
Fourier (SSF) or beam propagation (BPM)
method.

60

ISSN: 2180 - 1843

Vol. 2

Equation
represents
the
II.
NONLINEAR(1)
SCHRDINGER
EQUATION

generalized
form
of NLSESchrdinger
for complex
envelope
The Nonlinear
Equation
(NLSE) isA(z,t).
used to
mathematically
explain
varying
pulse envelope
in
Equation (2)
is the
linear
part ofpropagating
NLSE. It
a medium with linear and nonlinear elements. Thus, NLSE is
consists
secondpulse
order
dispersion
(SOD),
suitable
forof
describing
propagation
in free
space.
Numerical
solution
for NLSE canEquation
be obtained (3)
by applying
TOD and
attenuation.
is the
split step Fourier (SSF) or beam propagation (BPM) method.
nonlinear
part of NLSE that denotes the
Equation (1) represents the generalized form of NLSE for
II. NONLINEAR SCHRDINGER EQUATION
SPM. The
2 and
are the
quadratic
and
complex
envelope
A(z,t). 3
Equation
(2) is
the linear part
of
The It
Nonlinear
Schrdinger
Equation
(NLSE)
usedand
to
NLSE.
consists
of
secondcoefficient
order
dispersion
(SOD),isTOD
cubic
dispersion
respectively,
mathematically
explain
varying
pulse
envelope
propagating
in
attenuation. Equation (3) is the nonlinear part of NLSE that
medium
is the
attenuation
factor
and

is
the
adenotes
with
linear
and
nonlinear
elements.
Thus,
NLSE
is
the SPM. The 2 and 3 are the quadratic and cubic
suitable
for
describing
pulse [5].
propagation
in free space.
dispersion
coefficient
respectively,
is the attenuation
factor
nonlinear
coefficient

Numerical
NLSE can
and is the solution
nonlinearfor
coefficient
[5]. be obtained by applying
split step Fourier (SSF) or beam propagation (BPM) method.
E3 d 3 AtheDgeneralized 2form of NLSE for
iE2 (1)
dAEquation
d 2 Arepresents
 envelope
 A(z,t).
 A  iJ(2)| Ais| the
A linear part of
complex
Equation
2
dz
2
dt
dt 2 order
2 dispersion (SOD),


NLSE. It consists of 6second
TOD
and
attenuation. Equation (3) is the nonlinear part of NLSE that
denotes the SPM. The 2 and 3 are the quadratic and cubic
dispersion
coefficient
w AL
i
w 2 A respectively,
1
w 3 A isDthe attenuation factor
 E2
 E 3 [5].

A
and
w z is the 2nonlinear
w T 2coefficient
wT 3
6
2


i E d 2 A E3 d 3 A D
dA
 2

 A  iJ | A |2 A
wdzANL 2 dt 2 2 6 dt 2 2


iJ | A | A
wz




frequency dependence
broadening simply beca
disperse during propaga
[5]. Pulse broadening
generated by the GVD i
phase of each spectral c
that depends on the freq
The generated frequenc
frequencycomponents
dependence c
spectral
broadening
simplycompo
beca
velocity.
Spectral
disperse during
propagae
compare
to the trailing
[5]. Pulse
arrival.
Pulsebroadening
broadeni
generatedcorrelated
by the GVD
linearly
withi
phase
each
not
relyofon
the spectral
sign of c2
thatTo
depends
on
the
freq
observe the effec
Thetogenerated
set
zero whilefrequenc
GVD,
spectral components c
velocity. Spectral compo
compare to the trailing e
arrival. Pulse broadeni
linearly correlated with
not rely on the sign of 2
To observe the effec
set to zero while GVD,

i
w2A 1
w3A D
 E3

E2
A
3
w T 2 III.6 PULSE
w T TYPE
2
2

Two types of pulses were used in the simulation. They are
III. PULSE
the
chirped GaussianTYPE
pulse and the chirped hyperbolic secant
wANL [5] as shown2 in (4) and (5) respectively.
pulse
w AL
wz

iJ | A | A

Two
were used in the
wz types of  pulses
1  iC t 2 

simulation.
are the
(4)

A( z , t ) A0 expThey
chirped Gaussian
2
T

0
chirped
pulse and the
hyperbolic
III. PULSE TYPE
secant
pulse [5]
as shown
in (4) and (5)
Two types of pulses
simulation. They are
t were used
iCtin2 the
(5)
A( zchirped
, t ) A0Gaussian
sec h pulse
expand
 the 2chirped

respectively.
the
hyperbolic secant
T
2T
0
and (5) respectively.
pulse [5] as shownino(4)

A0
A( z , t )

2
(6)
P0
1  iC t
(4)
A0 exp 

Equation 6 is the
pulse
initial
amplitude,
while
t
is
time
2
T
0

period, T0 is the half-width at 1/e intensity point, P0 is the

2
initial peak power and
chirp. All of the

t C is the
iCtfrequency
A( z, t ) A0 were
sec h carried
simulations
of,
expout
using2 the parameter values (5)
To C = 0 (unchirped).
2T0
T0 = 2 ps, P0 = 1W and

A0

P0

(6)

IV. SIMULATION RESULTS DUE TO SECOND ORDER

Equation 6 is the pulse initial amplitude, while t is time
DISPERSION
period, T0 is the half-width at 1/e intensity point, P0 is the
Equation
6 dispersion
isand
theC pulse
initial
amplitude,
Second
is
a linear
effect
initial
peakorder
power
is(SOD)
the frequency
chirp.
Alland
of the
the
primary
ofcarried
pulse
broadening.
Eq.
2,values
SOD of,
is
simulations
out
using
parameter
while source
t iswere
time
period,
T0the
isFrom
the
half-width
the group velocity dispersion
governed
2, known
T
2 ps,intensity
Pby
C =as0 (unchirped).
0 = 1W andpoint,
at0 =1/e
P0 isofthe
initial
peak
(GVD).
GVD represents
dispersion
group
velocity
that
determines
the broadening
of the
pulse. All
The
power and
C is thecharacteristic
frequency
chirp.
DUEcarried
TO SECONDout
ORDER
IV. simulations
SIMULATION RESULTS
of the
were
using
DISPERSION
the parameter values
of, T0 = 2 ps, P0 =
Second order dispersion (SOD) is a linear effect and the
1W
and C = 0 (unchirped).
primary source of pulse broadening. From Eq. 2, SOD is

governed by 2, known as the group velocity dispersion

(GVD). GVD represents dispersion of group velocity that
determines the broadening characteristic of the pulse. The

IV. SIMULATION
RESULTS
DUE TO SECOND ORDER
DISPERSION
Second order dispersion (SOD) is a
linear effect and the primary source

No. 1

January - June 2010

Fig. 1. Pulse propagation at z =

unchipred Gaussian pulse and

The broadening exp

observed in Fig. 1, wh
amount of broadening. P
with the propagated d
constant phase shift cau
increase
chirp affect
Fig.
1. Pulseinpropagation
at z =
pulse spectral
compo
unchipred
Gaussian pulse
and
consequentially increase
The
broadening
exp
The
magnitude
of dela
observed
1, wh
effects caninbeFig.
observed
i
amount of broadening. P
with the propagated d
constant phase shift cau
increase in chirp affect
pulse spectral compo
consequentially increase
The magnitude of dela
effects can be observed i

Analysis of Second order Dispersion on Free Space Optical Propagation

of pulse broadening. From Eq. 2,

SOD is governed by 2, known as
the group velocity dispersion (GVD).
GVD represents dispersion of group
velocity that determines the broadening
characteristic of the pulse. The frequency
dependence of the group velocity leads
to pulse broadening simply because
different component of the pulse disperse
during propagation and do not arrive
simultaneously [5]. Pulse broadening
occurs due to frequency chirps generated
by the GVD induced phase shift. GVD
changes the phase of each spectral
component of the pulse by an amount
that depends on the frequency and the
propagated distance [6]. The generated
frequency chirps changes the velocity of
each spectral components causing them
to frequency
traveldependence
in different
velocity.
Spectral
QUATION
of the group
velocity leads
to pulse
broadening simply because different component of the pulse
Equation (NLSE) is used to components
at
the
leading
edge
travel
ulse envelope propagating in disperse during propagation and do not arrive simultaneously
compare
to the
edges. This
[5]. Pulse
broadening
occurstrailing
due to frequency
chirps
ear elements. Thus, NLSE is faster
generated by the GVD induced phase shift. GVD changes the
propagation in free space. causes
a
delay
on
the
pulse
arrival.
Pulse
an be obtained by applying phase of each spectral component of the pulse by an amount
is frequency
dependent
on the
delay
that depends on the
and the propagated
distance
[6].
propagation (BPM) method. broadening
The generated frequency chirps changes the velocity of each
eneralized form of NLSE for and
linearly
correlated
with
distance.
The
spectral
components
causing
them
to
travel
in
different
ion (2) is the linear part of
broadening
does
on faster
the
velocity.
Spectral components
at the not
leadingrely
edge travel
r dispersion (SOD), TOD and pulse
compare to the trailing edges. This causes a delay on the pulse
nonlinear part of NLSE that sign
of
2.
arrival. Pulse broadening is dependent on the delay and

are the quadratic and cubic

y, is the attenuation factor linearly correlated with distance. The pulse broadening does
rely on the sign of 2.
[5].
Tonot
observe
the effect of SOD alone, 3 and
To observe the effect of SOD alone, 3 and in Eq. 1 are
in
Eq.
areGVD,
set to
zero
while
GVD, 2 =
ps2/km
[3].
set to
zero1while
2 = 21

A  iJ | A |2 A



21 ps2/km [3].



3
A D

A
T3
2




E TYPE

in the simulation. They are

e chirped hyperbolic secant
respectively.
2

(4)

Ct 2

T02

(5)
(6)

l amplitude, while t is time

1/e intensity point, P0 is the
frequency chirp. All of the
ing the parameter values of,
chirped).

DUE TO SECOND ORDER

SION

D) is a linear effect and the

ening. From Eq. 2, SOD is
e group velocity dispersion
sion of group velocity that
acteristic of the pulse. The

Fig. 1. Pulse propagation at z = 0;0.5;1;1.5

km with 2 = 21 ps /km for (a) unchipred
The broadening
experienced
by both pulses
can be
Gaussian
pulse and
(b) unchipred
hyperbolic
observed in Fig. 1, secant
where both
pulses show a significant
pulse
Fig. 1. Pulse propagation at z = 0;0.5;1;1.5 km with 2 = 21 ps2/km for (a)
unchipred Gaussian pulse and (b) 2unchipred hyperbolic secant pulse

amount of broadening. Pulse broadening is linearly correlated

with the propagated distance. As the pulse propagates,
constant phase shift cause a constant increase in chirp. The
increase in chirp affects the velocity and the arrival of the
pulse spectral components. The change of velocity
ISSN: 2180 - 1843 Vol. 2
consequentially increases delay and cause further broadening.
The magnitude of delay increases with the distance. These
effects can be observed in Fig. 1(a) and Fig. 1(b).

The broadening experienced by both

pulses can be observed in Fig. 1, where
both pulses show a significant amount of
broadening. Pulse broadening is linearly
correlated with the propagated distance.
As the pulse propagates, constant phase
shift cause a constant increase in chirp.
The increase in chirp affects the velocity
and the arrival of the pulse spectral
components. The change of velocity
consequentially increases delay and cause
further broadening. The magnitude of
delay increases with the distance. These
effects can be observed in Fig. 1(a) and
Fig. 1(b).
From Figure 2(a) and 2(b), the waterfall
plot for both pulses show similar
characteristics in pulse broadening. It is
obvious SOD induced broadening increase
linearly with propagating distance.
Nevertheless, both pulses have displayed
different broadening rates. Hyperbolic
secant pulse reveals a lower broadening
rate compare to Gaussian. This can be
observed as Gaussian pulse exhibits wider
broadening and lower pulse amplitude as
it propagates, in comparison to hyperbolic
secant pulse. This implies that both pulses
have different effect to GVD. There is one
important attribute; hyperbolic secant
pulse shows a faint distortion at both
edges of its pulse. Distortion can be seen
between distances 0.3 km and 0.5 km but
disappears as the pulse propagates; as
can be observed in Fig. 2(b).
Broadening rate for both Gaussian and
hyperbolic secant pulse can be observed
in Figure 3. Hyperbolic secant pulse
shows lower broadening rate at about
34.5% compare to Gaussian pulse. The
difference in broadening rate can be
traced back to the difference in the pulse
shape. The pulse shape is defined by the
pulse equation and both pulses manifest
differently over the same parameters as
can be seen in the Gaussian pulse which is
presented by Eq. 4 and hyperbolic secant
pulse as in Eq. 5. These differences create
variations and rare anomalies

No. 1

January - June 2010

61

traced back to the difference in the pulse shape. The pulse

shape is defined by the pulse equation and both pulses
manifest differently over the same parameters as can be seen
Journal of Telecommunication, Electronic and Computer Engineering
Hyperbolic
in the Gaussian pulse which is presented by Eq. 4 and Fig. 3. Broadening factor for unchirped Gaussian and unchirped
2
hyperbolic secant pulse as in Eq. 5. These differences create Secant pulse over the distance, z = 1 km, with 2 = 21 ps /km,
variations and rare anomalies

propagation in free space was simulated

with the SODV.inCONCLUSION
order to observe the
In this paper the dispersion effects were simulated
pulse
behavior
in
free
space. Simulation
individually in 1D and 2D graphical representation.
result may
serve
as toa observe
prediction
model
Simulations
were done
in order
pulse behavior
and
response
to linear
Pulse propagation
in free space
that can
be parameters.
used to estimate
or predict
to
was
with actual
the SODpulse
in orderbehavior
to observe in
the free
pulse
ansimulated
extent the
behavior in free space. Simulation result may serve as a
space. model that can be used to estimate or predict to an
prediction
extent the actual pulse behavior in free space.
ACKNOWLEDGMENT

ACKNOWLEDGMENT
This work is supported by the Internal Grant of Universiti

Teknikal Malaysia Melaka under grant PJP/2009/FKEKK

This work is supported by the Internal
(20D) S614.

Grant of Universiti Teknikal Malaysia

Melaka under REFERENCES
grant PJP/2009/FKEKK
[1] G. P. Agrawal, Nonlinear Fiber Optics. San Diego, CA: Academic
(20D)
S614.
Press, Chap. 2, 2001.

[2]

M. J. Potasek & S. Kim, Optical Processing in Free Space Using

Ultrashort Pulses, CLEO Tech Digest, Baltimore, MD, pg. 181-182,
2001.
[3] M. J. Potasek, Spatiotemporal Effects in Nonlinear Dispersive and
Diffractive Media Using Ultrashort Optical Pulses, invited paper,
Nonlinear Optics, Vol. 21, pg. 399, 1999.
[4][1]
J. C. Ricklin
F. M. Davidson,Nonlinear
Atmospheric optical
communication
G. and
P. Agrawal,
Fiber
Optics.
with a Gaussian Schell beam, J. Opt. Soc. Am. A, Vol. 20, No. 5, May
2003. San Diego, CA: Academic Press, Chap.
[5] G. P. Agrawal,
Fiber Optic Communication Systems. New York: John
2, 2001.
Wiley & Sons, inc., 3rd edition, 2002.
[6] U. Bandelow, A. Demircan, and M. Kesting, Simulation of Propagation
[2] in Nonlinear
M.Optical
J. Potasek
& S.no. 23,
Kim,
Fibers. WIAS Report
2003. Optical

REFERENCES

Processing in Free Space Using

Ultrashort Pulses, CLEO Tech Digest,
Baltimore, MD, pg. 181-182, 2001.

Pulse propagation
= ps12/km
kmforand
with
Fig. Fig.
2. Pulse2.propagation
at z = 1 km and withat
2 =z21
(a)unchirped
Gaussian
pulse
and ps2/km
(b)unchirpedfor
hyperbolic
secant pulse. Gaussian
2
= 21
(a)unchirped
pulse and (b)unchirped hyperbolic secant
pulse.

plot for both pulses

ening. It is obvious
y with propagating
displayed different
e reveals a lower
can be observed as
g and lower pulse
o hyperbolic secant
different effect to
hyperbolic secant
edges of its pulse.
0.3 km and 0.5 km
can be observed in

hyperbolic secant
rbolic secant pulse
34.5% compare to
ening rate can be
e shape. The pulse
n and both pulses
ters as can be seen
ed by Eq. 4 and
e differences create

[3]

M. J. Potasek, Spatiotemporal Effects

in Nonlinear Dispersive and Diffractive
Media Using Ultrashort Optical
Pulses, invited paper,
Nonlinear
Optics, Vol. 21, pg. 399, 1999.

[4]

J. C. Ricklin and F. M. Davidson,

Atmospheric optical communication
with a Gaussian Schell beam, J. Opt.
Soc. Am. A, Vol. 20, No. 5, May 2003.

[5]

G.
P.
Agrawal,
Fiber
Optic
Communication Systems. New York:
John Wiley & Sons, inc., 3rd edition,
2002.

[6]

U. Bandelow, A. Demircan, and M.

Kesting, Simulation of Propagation
in Nonlinear Optical Fibers. WIAS
Report no. 23, 2003.

Fig. 3. Broadening factor for unchirped

Fig. 3. Broadening factor for unchirped Gaussian and unchirped Hyperbolic
Gaussian
and
unchirped
Secant
Secant pulse over
the distance,
z = 1 km,Hyperbolic
with 2 = 21 ps2/km,
pulse over the distance, z = 1 km, with 2 = 21
ps2/km,
V. CONCLUSION
In this paper the dispersion effects were simulated
in 1D and 2D graphical representation.
V. individually
CONCLUSION
Simulations were done in order to observe pulse behavior and
to linearthe
parameters.
Pulse propagation
free space
Inresponse
this paper
dispersion
effects inwere
was simulated with the SOD in order to observe the pulse
simulated
individually
in
1D
and
2D
behavior in free space. Simulation result may serve as a
graphical
representation.
Simulations
wereto an
prediction model
that can be used
to estimate or predict
extentin
theorder
actual pulse
in free
space.behavior
done
to behavior
observe
pulse

and response to linear parameters. Pulse

ACKNOWLEDGMENT

62

This work is supported by the Internal Grant of Universiti

Teknikal Malaysia Melaka under grant PJP/2009/FKEKK
(20D) S614.

ISSN: 2180 - 1843

Vol. 2

No. 1

REFERENCES
[1]

G. P. Agrawal, Nonlinear Fiber Optics. San Diego, CA: Academic

January - June 2010