Educational Software

An Interactive 1D Matlab FDTD Code

for Education
ALLEN W. GLISSON, ATEF Z. ELSHERBENI
Department of Electrical Engineering, University of Mississippi, P.O. Box 1842, University, Mississippi 38677

Received 11 August 2000; accepted 5 March 2001

ABSTRACT: This paper describes a 1D Matlab finite difference time-domain (FDTD) code
with a graphical user interface for visualization of the time-domain electromagnetic response.
The program animates time-domain reflection and transmission of a Gaussian plane wave
through one or two homogeneous material slabs. 2001 John Wiley & Sons, Inc. Comput Appl Eng
Educ 9: 136147, 2001

Keywords:

finite difference time domain; electromagnetic simulation; plane wave reflection

INTRODUCTION
The finite difference time-domain (FDTD) method
[1] has become widely used for the simulation of
electromagnetic radiation and scattering problems.
The method is relatively simple to understand and
implement, yet it is capable of modeling very complex
geometrical configurations comprising many different
types of materials and it can provide wide frequency
response data from a single calculation.
This paper describes a simple 1D Matlab [2] FDTD
formulation and an interactive computer program, and
illustrates the use of the developed code. The program
has been developed primarily for educational use to
model the time-domain reflection and transmission of
a Gaussian plane wave through one or two homogeneous material slabs. It has a graphical user
interface (GUI) for setting the parameters of the
FDTD simulation that provides the user with the ability to quickly change parameters and see the effects
of the new parameters on the simulation results. The
program presents the results of the FDTD simulation
Correspondence to A. W. Glisson ([email protected]).
2001 John Wiley & Sons, Inc.

136

as an animation of the time-domain reflection and

transmission of the electromagnetic wave. The
program also allows the user to choose between
several different formulation models for the same
problem, such as using either a total field or a
scattered field FDTD formulation.

1D FDTD
Total Field Formulation
For the 1D simulation we consider the case of a
Gaussian plane wave propagating in the x-direction,
and having only a y-component of electric field and a
z-component of magnetic field. In the total field
formulation, the time-domain Maxwells equations
for this special case reduce to


@Ey 1
@Hz
e

 Ey ;

@t
@x
"


@Hz 1
@Ey

m H z ;

@t
@x


1D MATLAB FDTD CODE FOR EDUCATION

where ", , e , and m are the permittivity, permeability, electric conductivity, and magnetic conductivity of the medium, respectively, and Ey and Hz
represent the total electromagnetic fields. These
constitutive parameters may be functions of x in the
general 1D case, but for this work we assume they are
constants within each of the material slabs and in the
surrounding space.
Using central finite differences in time and space
for the derivatives in (1), one can obtain the secondorder accurate FDTD update equations for a total field
formulation as [3]
Eyn1 i Ceye iEyn i Ceyh i
"

#
Hzn1=2 i 12 Hzn1=2 i 12

;

x


1
n1=2
i
2
Hz
2




1
1
Chzh i Hzn1=2 i
2
2

 n

1 Ey i 1 Eyn i
;
Chze i
2

x
where the notation F n i represents F evaluated
at time t n
t and spatial position x i
x,
with
t being the simulation time step and
x the
space increment. The coefficients C in (2) are defined
by
"r i Ce e i
;
"r i Ce e i
2Ce
;
Ceyh i
"r i Ce e i
r i Ch m i
;
Chzh i
r i Ch m i
2Ch
;
Chze i
r i Ch m i
Ceye i

Ey Eys Eyinc ;

Hz Hzs Hzinc ;

137

where the superscripts s and inc denote the scattered

and incident fields, respectively. The incident fields
are assumed to be known and the scattered fields are
the unknowns to be determined via the FDTD update
equations. In the scattered field formulation, the 1D
time-domain Maxwells equations may be written
as [3]
!
@Eys 1
@Eyinc
@Hzs
e s
e inc

 Ey " "0
 Ey ;
"
@t
@x
@t


@Eys
@Hzs 1
@H inc

m Hzs  0 z m Hzinc :

@t
@x
@t
5

Using central finite differences in time and

space for the derivatives on the scattered field terms
in (5), one can obtain the second-order accurate
FDTD update equations for the scattered field
formulation as
Eys;n1 i Ceye iEys;n i Ceyh i
"
#
Hzs;n1=2 i 12 Hzs;n1=2 i 12


x
!n1=2
"i "0 @Eyinc i

@t
"i
e i inc;n1=2
E
i;
"i y


1
Hzs;n1=2 i
2






1
1
1
s;n1=2
i
Chze i
Chzh i Hz
2
2
2
 s;n

s;n
Ey i 1 Ey i


x

n
i 12 0 @Hzinc i 12

@t
i 12


m
1
 i 2 inc;n
1
H
6

i :
2
i 12 z

where "r and r are the relative permittivity and permeability, Ce
t=2"0 ; and Ch
t=20 : It should
also be noted that field averaging has been used
for the undifferentiated field components in (1) to
obtain (2).

Scattered Field Formulation

In the scattered field formulation, the total electromagnetic field is separated into incident and scattered
fields. Thus, the total electric and magnetic fields for
the 1D case are written as

The time derivatives of the incident fields in

(6) have been left in the same form as in (5) since
the incident field is known and can be differentiated analytically. However, (6) can also be recast
using numerical differentiation of the incident
fields.

138

GLISSON AND ELSHERBENI

Excitation Waveform
A Gaussian pulse waveform is used as the excitation
for the FDTD simulation. The expression for the
Gaussian pulse is defined as
 

x xp ct2
inc
;
Ey x; t exp


where xp represents the location of the peak of the

pulse at time t 0 and  is a parameter related to the
width of the pulse. In this work,  is chosen such that
the incident field at x xp has decayed to 0.1% of its
peak value when t is equal to one-half of the pulse
width as defined in the Data Entry window (see
SIMULATION DATA ENTRY). In the total field
formulation, the total electric field at time t 0 is set
equal to the incident electric field, while the total
magnetic field at time t 12
t is set equal to the
corresponding incident magnetic field.

electric fields in (8) are interpreted to mean the

scattered field rather than the total field.

SYSTEM REQUIREMENTS
The user must have access to Matlab to run the 1D
FDTD simulation program. The program was developed and tested under Matlab version 5.3. The program consists of two preparsed pseudocode (pcode)
Matlab files:
gui_fdtd_1d.p
gui_fdtd_dataentry.p

These files are available at the Applied Computational Electromagnetics Society (ACES) web site,
http://aces.ee.olemiss.edu. Both files must be in the
Matlab search path.

PROGRAM USAGE
Termination of the Computational Domain
The program allows for the termination of the
computational domain on either side by a perfect
electric conductor (PEC) or a first-order Mur absorbing boundary condition [4]. The PEC boundary
condition is enforced in the total field formulation
by setting Ey 0 at the boundary for all values of
the time index n. In the scattered field formulation,
the PEC boundary condition is enforced by setting
Eys Eyinc at the boundary.
The first-order Mur absorbing boundary is based on
the use of a one-way wave equation at each boundary,
which permits only outward propagation from the
computational domain. The FDTD update equations
for the electric field at the boundary in the total field
formulation can then be written as
c
t
x
c
t
x
 Eyn1 ib  1 Eyn ib ;

Eyn1 ib Eyn ib  1

where ib represents the index of either the right or left

boundary node, and the boundary condition is enforced using the upper sign in (8) for the left boundary
and the lower sign for the right boundary. All of the
quantities on the right-hand side of (8) are known
since they represent either the electric field at the
previous time step n or the electric field that has
already been computed using (2) at an interior node
(i 6 ib ) and the current time step n 1. Eq. (8) also
applies for the scattered field formulation if the

To start the program, the user should enter the

command
gui_fdtd_1d

from the command window of Matlab. The main

simulation window shown in Figure 1 will appear.
Pressing the start button in the main window will
begin the simulation with the default parameters. The
speed of the simulation can be controlled using the
slider bar or by typing a number from 1 (slow) to 100
(fast) in the edit box above the slider control. The
speed control can be changed interactively at any time
during a simulation.
Once the animation has been started, it can be
stopped temporarily and continued from its current
state using the stop and continue buttons. During
the simulation, the value of the y-component of the
electric field at each time step is shown as a function
of x=
x, where
x is the spatial increment used in the
simulation. When the simulation is stopped, the time
step number n and the time index value t n
t are
shown at the top of the graph. Figure 2 shows the
incident Gaussian pulse after one time step. The
positions of the boundaries of a single-material slab
region specified by the default parameter set are
shown as vertical green lines.
When the animation is stopped, the user can open
the data entry window to change the simulation
parameters by pressing the Set Data button. The data
entry window is shown in Figure 3. When the data

1D MATLAB FDTD CODE FOR EDUCATION

Figure 1 The main simulation window.

Figure 2 The incident pulse after one time step and the material region boundaries in the main
simulation window using the default parameters.

139

140

GLISSON AND ELSHERBENI

Figure 3 The data entry window.

entry window is open, the continue button is disabled,

so the simulation must be restarted regardless of
whether any data was actually changed in the data
entry window. The data entry window can be left open
while the simulation window is active, if desired, for
quick changes to the simulation parameters.

SIMULATION DATA ENTRY

Slab Parameters
The constitutive parameters r , "r , e , and m , and the
start and end locations for each of the two homogeneous slabs are set in the Slab Parameters section of
the data entry window. If the parameters of the second
slab are those of free space, the boundary lines for the
second slab do not appear in the simulation window.
The start and end locations of the two slabs are specified in terms of the index location x=
x. Rudimentary
error checking in the code does not permit the two
slabs to overlap.

Model and Excitation

The Model and Excitation section of the data entry
window allows one to control the spatial model in

terms of the sampling increment
x and the location

of the right boundary of the model xmax =
x. The left
boundary of the model is always at x 0. One can
also control whether or not media averaging is performed at the slab interfaces, and can set the Courant
FriedrichsLewy (CFL) number for the simulation
here. A CFL number greater than 1 is not allowed.
Note that, since
x is independently specified, setting
the CFL number effectively specifies the time step
t.
The time width and peak location of the incident
Gaussian pulse are also specified in this section. The
pulse width here refers to the distance between the
time points where the Gaussian pulse is down to 0.1%
of its peak value. In this code, the incident pulse must
appear completely within the computational domain
at time t 0, but it must reside within a free-space
portion of the computational window as suggested by
Figure 2, and not within the material slab. If any
portion of the incident pulse appears outside the
computational window or within the slab at time
t 0, the user is warned to change the data, and the
simulation will not run.
The peak plot range value for the electric field can
also be set in this section. The incident Gaussian pulse
is assumed to have a value of 1, so the plot range value
is also initially set to 1. Depending on the parameters
of the material slabs, however, it may be desirable to

1D MATLAB FDTD CODE FOR EDUCATION

change the plot range to avoid clipping the timedomain response in the graph window. Conversely,
reducing the plot range value also allows one to view
lower amplitude details in the plot.

141

electric field and the total electric field are equal, the
curve appears in green in the simulation window.

EXAMPLE SIMULATIONS
Computation Boundaries
The properties of both the left and right boundaries of
the computation domain can be chosen in this section.
The first-order Mur absorbing boundary condition is
used when absorbing is chosen. Since the incident
pulse starts within the computation domain, it is
possible to choose both the left and right boundaries
as PEC if desired.

Formulation Type
In this section the user can choose to use a total field
formulation, a scattered field formulation in which the
time derivative of the incident field is computed
exactly, or a scattered field formulation in which the
time derivative of the incident field is computed
numerically. When a scattered field formulation is
chosen, the simulation window shows plots of both the
scattered electric field (as a blue curve) and the total
electric field (as a red curve). When the scattered

Scattered Field Formulation With Exact

and Numerical Differentiation of the
Excitation Term
The first example considered compares the use of
exact and numerical differentiation of the excitation
term in a scattered field formulation. The default slab
geometry and constitutive parameters are used, so
the slab extends from x=
x 200 to 275, and the
constitutive parameters of the slab are defined
by "r 4, r 1, and e m 0. The default
model and excitation parameters are also used:

x 0:0015 m, c
t=
x 1, pulse width 400 ps,
initial pulse peak position x=
x 100, xmax =

x 400, plot range max value 1, and media
averaging is on. Absorbing boundaries are used at
both the left-hand right-hand sides of the computational domain.
Figure 4 shows the simulation result at time step
n 199 for the case in which exact differentiation of

Figure 4 Simulation using scattered field formulation with the default parameters, and with exact
differentiation of the excitation term.

142

GLISSON AND ELSHERBENI

the excitation term is performed. For this particular

choice of the time value, the peak of the incident
pulse, which always propagates in free space, has
passed beyond the location of the right boundary of
the slab. The total field, however, which represents the
true electromagnetic field, has not yet had time to exit
the slab, so should be zero to the right of the slab.
In Figure 4, one observes that the pulse reflected
from the left-hand side of the slab has a peak at about
x=
x 100 and that it has undergone a sign reversal.
Since the incident field has propagated away from this
region and is not affected by the presence of the slab,
the total field here is equal to the scattered field and
this portion of the curve in the simulation window is
green in color. The peak of the transmitted pulse in the
slab occurs at about x=
x 250. The transmitted
pulse is narrower in width and its speed of propagation
within the dielectric slab is slower. Since the incident
wave has also passed this region, the total field is
again equal to the scattered field. To the right of
the slab two curves are apparent. The blue curve is
the computed scattered field, while the red curve is the
computed total field. Since the total field, which is the
sum of the incident field and the scattered field, should
be zero to the right of the slab, the scattered field
should be equal and opposite to the incident wave.
The numerically computed scattered field, however,

contains some error and does not precisely cancel the

incident field. Thus, the computed total field shown in
red is not precisely zero.
For comparison, Figure 5 shows the same simulation result at time step n 199 for the case in which
numerical differentiation of the excitation term is
performed. One observes in this case that the total
field (red curve) to the right of the slab is essentially
zero, as it should be. It may at first seem surprising
that numerical differentiation of the known excitation
produces a more accurate result than exact differentiation. It is fairly easy to demonstrate for a
marching-in-time solution of even very simple differential equations how this can occur, however. We
note that if the time step size is reduced, which
improves the finite-difference approximation of
the time derivative, the result obtained using exact
differentiation of the excitation improves. The same
simulation as that of Figure 4, but with c
t=
x 0:5
is shown in Figure 6, where it is noted that the error in
the total field to the right of the slab is reduced.

Numerical Dispersion
The next example illustrates the problem of numerical
dispersion. For this example, the total field formulation is used and the default simulation parameters are

Figure 5 Simulation using scattered field formulation with the default parameters, and with
numerical differentiation of the excitation term.

1D MATLAB FDTD CODE FOR EDUCATION

143

Figure 6 Simulation using scattered field formulation with c
t=
x 0:5, and with exact
differentiation of the excitation term.

used except that the dielectric constant of the slab

has been set to "r 80. With this higher dielectric
constant, the pulse width within the slab is much
narrower and the propagation speed of the pulse in the
slab is much slower. Figure 7 shows the simulation
result after the pulse has had sufficient time to
propagate about three-fourths of the distance through
the slab.
Note that the time step n is much larger than in the
previous example (n 597) because of the slower
speed of pulse propagation in the dielectric. Thus, the
reflected wave has already been absorbed at the left
computational boundary. The ripples to the left of the
slab and the highly oscillatory tail of the pulse within
the slab are the result of numerical dispersion caused
by inadequate resolution of the pulse within the slab.
The much narrower width of the pulse within the slab
implies that an insufficient number of spatial samples
occur over the width of the pulse. (Compare the width
of the first peak within the slab with the width of the
incident pulse in Figure 2.) It should be noted also,
however, that the scale of Figure 7 has been increased
to better visualize the dispersion effect by setting plot
range maximum value 0.2.
The dispersion effect can be reduced by increasing
the pulse width in the simulation. Figure 8 shows the
simulation result for the same time step value when

the incident pulse has a width of 1000 ps instead of the

400 ps width used to obtain Figure 7.

Media Averaging
The final example illustrates the effect of changes in
the manner in which the medium parameters are
modeled at an interface. For this example, the total
field formulation is used and the default simulation
parameters are used except that the dielectric constant
of the slab has been set to "r 4, the relative permeability has been set to r 4, and the pulse width
has been set to 600 ps to reduce numerical dispersion.
With these parameters, the impedance of the slab is
the same as that of free space, so there should be no
reflection of the incident pulse at either slab interface.
For the case in which media averaging has been set
to off, Figure 9 shows the simulation result just after
the pulse has completely penetrated the left boundary
of the slab. When media averaging is off and a
medium parameter at the interface is required by an
FDTD update equation, the program simply uses the
value of the medium parameter immediately to the
right of the interface. This introduces an additional
error into one of the update equations and the result is
a significant reflection from the interface as shown in
Figure 9.

144

GLISSON AND ELSHERBENI

Figure 7 Simulation using total field formulation with the default parameters, except that the
dielectric constant of the slab is "r 80.

Figure 8 Simulation using total field formulation with an incident pulse of width 1000 ps for the
slab with dielectric constant "r 80.

1D MATLAB FDTD CODE FOR EDUCATION

Figure 9 Simulation showing computed reflection from the interface of a slab with parameters
"r 4 and r 4 when media averaging is off.

Figure 10 Simulation showing computed reflection from the interface of a slab with parameters
"r 4 and r 4 when media averaging is on.

145

146

GLISSON AND ELSHERBENI

For comparison, Figure 10 shows the simulation

result at the same time step for the case in which media
averaging has been set to on. When media averaging
is on and a medium parameter at the interface is required by an FDTD update equation, the program uses
the average value of the medium parameters immediately to the left and to the right of the interface.
Although a small reflection is clearly present in the
result in Figure 10, the reflection is significantly
smaller in amplitude than is the reflection shown in
Figure 9 for the case in which media averaging is off.

simulation that provides the user with the ability to

quickly change parameters and see the effects of the
new parameters, formulations, and models on the
simulation results. The program presents the results
of the FDTD simulation as an animation of the timedomain reflection and transmission of the electromagnetic wave. These program files are available
from the Applied Computational Electromagnetics
Society (ACES) web site, http://aces.ee.olemiss.edu.

REFERENCES
CONCLUSION
In this paper, we have described a simple 1D Matlab
FDTD code that has been developed primarily for
educational use and have provided examples of its
use. The program models the time-domain reflection
and transmission of a Gaussian plane wave through
one or two homogeneous material slabs. The program
allows the user to choose between several different
models for the same situation, such as using either a
total field formulation or a scattered field formulation.
It has a GUI for setting the parameters of the FDTD

[1] A. Taflove, Computational Electrodynamics: The

Finite-Difference Time-Domain Method. Artech House,
Norwood, MA, 1995.
[2] MATLAB Technical Computing Environment, The
MathWorks, Inc., Natick, MA.
[3] A. Z. Elsherbeni, and A. W. Glisson, The Basics of the
Finite-Difference Time-Domain Technique for Electromagnetic Applications, ACES Short Course Notes,
Monterey, CA, March 2000.
[4] G. Mur, Absorbing boundary conditions for the finitedifference approximation of the time-domain electromagnetic-field equations, IEEE Trans Electromagn
Compat, EMC-23 (1981), 377382.

BIOGRAPHIES
Allen W. Glisson received the BS, MS,
and PhD degrees in electrical engineering
from the University of Mississippi, in
1973, 1975, and 1978, respectively. In
1978, he joined the faculty of the
University of Mississippi, where he is
currently a professor of electrical engineering. He was selected as the Outstanding Engineering Faculty Member in
1986 and again in 1996. He received a Ralph R. Teetor Educational
Award in 1989. His current research interests include the
development and application of numerical techniques for treating
electromagnetic radiation and scattering problems, and modeling of
dielectric resonators and dielectric resonator antennas. Dr. Glisson
has been actively involved in the areas of numerical modeling of
arbitrarily shaped bodies and bodies of revolution with surface
integral equation formulations. He has also served as a consultant to
several different industrial organizations in the area of numerical
modeling in electromagnetics.

Dr. Glisson is a member of the Sigma Xi Research Society and

the honor societies Tau Beta Pi, Phi Kappa Phi, and Eta Kappa Nu.
He is a senior member of the IEEE and is a member of several
professional societies within the IEEE, a member of Commission B
of the International Union of Radio Science, and a member of the
Applied Computational Electromagnetics Society. He was a U.S.
delegate to the 22nd, 23rd, and 24th General Assemblies of URSI.
Dr. Glisson has received a best paper award from the SUMMA
Foundation and twice received a citation for excellence in
refereeing from the American Geophysical Union. Since 1984, he
has served as the associate editor for book reviews and abstracts for
the IEEE Antennas and Propagation Society Magazine. He is
currently a member of the IEEE Antennas and Propagation Society
Administrative Committee and IEEE Press Liaison Committee. He
is also currently coeditor of the Applied Computational Electromagnetics Society Journal and serves on the board of directors of the
Applied Computational Electromagnetics Society. Dr. Glisson has
also served as an associate editor for Radio Science and as the secretary of Commission B of the U.S. National Committee of URSI.

1D MATLAB FDTD CODE FOR EDUCATION

Atef Z. Elsherbeni received a BSc degree
(with honors) in electronics and communications, as BSc degree (with honors) in
applied physics, and a MEng degree in
electrical engineering, all from Cairo
University, Cairo, Egypt, in 1976, 1979,
and 1982, respectively, and a PhD degree
in electrical engineering from Manitoba
University, Winnipeg, Manitoba, Canada,
in 1987. He was a research assistant with the Faculty of Engineering
at Cairo University from 1976 to 1982, and from 1983 to 1986 at the
Department of Electrical Engineering, Manitoba University. He was
a part-time software and system design engineer from March 1980
to December 1982 at the Automated Data System Center, Cairo,
Egypt. From January to August 1987, he was a postdoctoral fellow
at Manitoba University. Dr. Elsherbeni joined the faculty at the
University of Mississippi in August 1987 as an assistant professor of
electrical engineering. He advanced to the rank of associate
professor in July 1991 and to the rank of professor in July 1997.
He spent his first sabbatical term in 1996 at the Department of

147

Electrical Engineering, University of California at Los Angeles

(UCLA).
His professional interests include scattering and diffraction of
electromagnetic waves, numerical techniques, antennas, remote
sensing, and computer applications for electromagnetic education.
He has published 55 technical journal articles and 10 book chapters
on applied electromagnetics, antenna design, and microwave
subjects, and presented over 160 papers at professional conferences.
Dr. Elsherbeni is a senior member of the Institute of Electrical and
Electronics Engineers (IEEE). He is the electronic publishing
managing editor of the Applied Computational Electromagnetic
Society (ACES). His honorary memberships include the Electromagnetics Academy and the Scientific Xi Society. He serves on the
editorial board of the Book Series on Progress in Electromagnetic
Research and the journals IEEE Microwave Theory and Techniques,
Electromagnetic Waves and Applications, and Computer Applications in Engineering Education. He is the chairman of the
Educational Activity Committee for the IEEE Region 3 Section.
Dr. Elsherbenis home page can be found at http://www.olemiss.
edu/~atef,

Copyright of Computer Applications in Engineering Education is the property of John Wiley & Sons, Inc. and
its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.