You are on page 1of 10

Aliakbar Dastranj

Optimization of a
Printed UWB Antenna
Application of the invasive weed optimization algorithm in antenna design.

image licensed by ingram publishing

T
his article presents a general approach to design and show good agreement with the simulated ones that validate the
optimize printed ultrawideband (UWB) antennas by proposed optimization approach. The optimized antenna has a
using invasive weed optimization (IWO), a well-known compact size of 50 mm 50 mm. The operational bandwidth
global optimization algorithm. To achieve the required of the antenna for S11 < 10 dB, from both measurement and
radiation parameters over a wide bandwidth, a frequency- simulation, is 150%, based on the center frequency of 6.4 GHz
related cost function with optimal weighting coefficients is (1.611.2 GHz), which also covers the UWB (3.110.6 GHz)
suggested. Two prototypes of the optimized antenna have been applications. The time-domain response of the antenna was also
manufactured and examined. The experimental outcomes investigated by measurement of the group delay. Finally, the
efficiency of the optimized antenna in terms of bandwidth, size,
Digital Object Identifier 10.1109/MAP.2016.2630025
and gain are compared with a number of previously proposed
Date of publication: 6 January 2017 designs. Comparison results show that the optimized antenna

48 1045-9243/172017IEEE february 2017 IEEE Antennas & Propagation Magazine


outperforms other designs cited. combination and feed gap width
The experimental outcomes in fre- on the impedance matching and
quency as well as the time domain UWB-printed slot radiation characteristics of the
show that the antenna is suitable for antennas are superior slot antennas were investigated.
use in UWB or other communica- In [14], double shorting walls
tion systems. candidates for wide- were used to enhance the 10-dB
band communication bandwidth of a compact fold-
BACKGROUND ed patch feed antenna. It has a
UWB-printed slot antennas are applications due to square ground plane with a side
superior candidates for wide-band their many advantages. length of 80 mm, and its imped-
communication applications due to ance bandwidth varied from 3.90
their many advantages, for exam- to 8.26 GHz (71.7%). Two CPW-
ple, they are lightweight, compact in size, low cost, easy to fed elliptical and circular slot antennas with approximately
fabricate, and join simply with active devices [1]. Compared omnidirectional radiation patterns for wide-band applications
with traditional microstrip antennas, the benefits of printed were presented in [15]. Also, an empirical formula was intro-
slot antennas are an omnidirectional radiation pattern and a duced to approximately determine the lower edge of the 10-dB
wide impedance bandwidth. In addition, the coplanar wave- operating bandwidth. Recently, various UWB CPW-fed print-
guide (CPW) feeding scenario has numerous preferences over ed antennas have been presented [16][18]. In [16], a mono-
microstrip feed lines, for example, low radiation spillage, low pole antenna printed on an acrylonitrile butadiene styrene
scattering, and easy control of characteristic impedance [2], [3]. polycarbonate substrate was proposed. The antenna can
Many wireless communication systems require miniaturized provide an impedance bandwidth from 3 to 13 GHz. A multi-
wide-bandwidth antennas. Moreover, several demands for increas- functional UWB antenna using a pair of split-ring resonators
ing the transfer speed of information in modern communication and metallic shunt strips on the CPW feed line was proposed
applications require operational bandwidth enhancement [4]. in [17]. The antenna also can provide both frequency-notched
Although many UWB-printed antennas were designed for and narrowband properties. In [18], two open-circuit stubs
broadband communication systems [5][8], the aforementioned were proposed to design a wide-band-printed antenna with an
requirements provided a number of challenging electromag- operating frequency range from 3.7 to 10.1 GHz.
netic parameters that necessitated new approaches for antenna IWO is a high-performance and effective numerical stochas-
design. In [9], by employing a fork-shaped stub, the impedance tic optimization algorithm that is extracted from the natural
bandwidth of a planar slot antenna was enhanced. An imped- colonization behavior of weeds. As was shown in [19], IWO
ance bandwidth of 55% centered at 2 GHz along with a maxi- is an efficient and robust algorithm to find global minima for
mum gain of 5 dBi was obtained. The electromagnetic coupling linear and nonlinear multiobjective optimization problems.
between the printed slot and feed line was controlled by tuning Furthermore, as was shown in [20], in contrast to the traditional
the geometrical parameters of the fork-like stub. In [10], a planar evolutionary algorithms such as the genetic algorithm (GA) and
antenna with a size of 50 mm 68 mm, including a round cor- particle-swarm optimization (PSO), IWO provides lower final
ner rectangular wide slot in the ground plane and an H-shaped error and higher convergence rate. Also, several advantages and
radiating patch, gave an impedance bandwidth of about various applications of the IWO algorithm for radiating struc-
6.2 GHz centered at 5.17 GHz. Also, in [11], a slot antenna tures were presented in [21][24].
printed on a substrate with a size of 72 mm 72 mm achieved The main purpose of the present research is to propose a
an impedance bandwidth of 60% with respect to the center novel optimization approach for a printed UWB antenna. To
frequency at 2.22 GHz (S11 < 10 dB). This broad bandwidth validate the proposed method, a compact CPW-fed printed
was obtained by using a broadened stub. The coupling between UWB antenna was designed and optimized by using the IWO
the square slot and the CPW feed line and, consequently, the algorithm. Some of the antenna geometrical dimensions were
impedance matching of the antenna was controlled by selecting chosen as the optimization parameters, and then the anten-
a suitable size and position of the tuning stub. na impedance bandwidth and gain were optimized toward
In [12], a square printed-slot antenna fed by a CPW the necessary specifications. To take the frequency variation
line, utilizing a U-shaped adjusting stub, provided a 10-dB effects into account, a frequency-related cost function hav-
impedance bandwidth of 110% with a maximum gain of ing optimal weighting coefficients was defined. The compact
5.1 dBi. The half power beamwidth and cross-polar gain size was achieved by exerting constraints on the optimization
in the E-plane at the upper end of the band were 100 parameters. Simulated results are in good agreement with the
and 13 dB, respectively. Two square planar-slot antennas measured data. The optimized antenna with a small size of
fed by microstrip lines were proposed in [13]. By etching 50mm50mm gives a broad bandwidth from 1.6 to 11.2 GHz
arc- and triangular-shaped slots into square flame resistant (150% for S11 < 10 dB and using the center frequency). As will
4 (FR4) substrates that had side lengths all measuring 110 be seen later in the Results and Discussion section, compared
mm, wide 10-dB impedance bandwidths of 120% and 110% to other designs reported in [11][17], the optimized antenna
were achieved, respectively. Also, the effects of the feed-slot provides a wider bandwidth as well as a smaller size.

IEEE Antennas & Propagation Magazine february 2017 49


Basic Antenna Design E-shaped tuning stub, hexagonal patch, and small slots) is
The geometry of the designed CPW-fed printed antenna with from [12]. The initial design of the proposed antenna is per-
the coordinate system is illustrated in Figure 1. As depicted formed through numerous simulations via a high-frequency
in this figure, both the feed line and ground plane have been structural simulator (HFSS). In the next stage, the IWO
printed on the top side of the dielectric substrate. The rect- algorithm is employed to optimize the proposed structure to
angular radiating slot with length ls and width ws is engraved obtain the optimal radiation parameters along with a broad
on a FR4 substrate with a size of 50 mm 50 mm, a relative impedance bandwidth.
electric permittivity f r = 4.4, and thickness h = 1.6mm. The
exact value of the characteristic impedance of the CPW IWO Procedure
line with dimensions of g = 1.9 mm and S = 0.5 mm is 49.9 . IWO is a robust and efficient numerical stochastic optimiza-
The antenna radiating patch that is embedded within the tion technique that is extracted from the natural behavior of
rectangular slot consists of two sections: an E-shaped tuning weeds such as colonization, distribution, and reproduction. As
stub and a hexagonal patch. As shown in Figure 1, the radiat- shown in [20], compared with the two traditional evolutionary
ing patch is connected to the end of the CPW feed line and is algorithms, GA and PSO, it can provide a lower final error and a
symmetric with respect to the y axis. It should be pointed out higher convergence rate. By catching the attributes of the obtru-
that, by joining the E-shape stub to the hexagonal patch, the sive weeds, IWO is exceptionally strong and versatile in captur-
impedance matching significantly improves. In addition, two ing in any environment, which prompts a high performance and
very small slots with equal lengths d and equal widths c are effective optimization algorithm. Due to the aforementioned
etched on a hexagonal patch to further increase the antenna significant features, it is an excellent optimizer for solving vari-
operational bandwidth. The initial length of the small slots is ous electromagnetic issues, for example, antenna design and
determined almost according to the subsequent formula [18]: optimization [21][24]. The algorithm can be expressed in the
following four stages [23].
y
d= , (1)
fr + 1
4f Produce the first population
2
A finite number of seeds composing the initial population are
where f, v, and f r are the midfrequency of the desired generated randomly over the multidimensional search space of
frequency range, the speed of light in free space, and elec- the problem.
tric permittivity of the substrate, respectively. Notice that
the general idea of the baseline design (without any of the Reproduction
Every seed that has grown to a new flowering plant is admit-
ted to create new seeds based on its fitness value. In the simple
Wg case, the number of seeds that are produced by each flowering
plant increases linearly from the minimum possible number
Y of seeds, corresponding to minimum fitness, to the maximum
possible number of seeds, corresponding to maximum fitness in
X the population.

Ws Spatial dispersal
The generated seeds in the former stage will be distributed sto-
Ig chastically in the problem space. They have a standard Gaussian
Wp
distribution with a median equal to zero and a variance param-
c eter decreasing over time. By adjusting the median of the dis-
Is Ip tribution equal to zero, the produced seeds will be distributed
a
stochastically around the parent plant. Due to the decreasing
d
behavior of the variance over the iterations, the best plants with
b p
higher fitness values are accumulated around better solutions,
and inappropriate plants with lower fitness values are eliminated
g over the iterations. The standard deviation (SD) that is the root
Ground Plane CPW Feed square of the variance of this distribution is calculated in every
S
step as follows:
z
^ itermax - iter hn
h v iter = n (v initial - v final) + v final , (2)
x
iter max
Dielectric Substrate
where v initial and v final represent the first and last values of
Figure 1. The configuration of the designed antenna and its SD for the standard Gaussian distribution, respectively. The
geometrical dimensions. SD at the present step is represented by v iter, the greatest

50 february 2017 IEEE Antennas & Propagation Magazine


number of repetitions before stop- variables does not decrease the
ping the algorithm is represented conclusive value of error, and,
by itermax, and the nonlinear mod- To achieve good radiation instead, it might increase the
ulation index is represented by n. characteristics along convergence time of the optimi-
As proposed in [21], the best value zation procedure. The least and
for n is three. with broad impedance greatest quantities of seeds are
bandwidth, various vital parameters in IWO. Results
Competitive exclusion showed that when the least num-
After some iteration, the maxi- geometrical parameters ber of seeds is set to zero or one,
mum number of plants (p ma x) of the antenna can excellent execution is accom-
will be produced through a rapid plished for some issues.
reproduction process. However, be selected as Also, choosing the greatest
it is anticipated that the plants optimization variables. number of seeds between three
with higher fitness values will be and six prompts a proper execu-
reported to be of better quality tion based on the convergence
than the inappropriate ones. Thus, the final step is to elimi- time and final error of the optimization procedure [24]. As
nate the inappropriate and weaker plants in a competitive mentioned previously, the nonlinear modulation index con-
manner to limit the maximum number of plants in the siderably affects the execution of the IWO algorithm. The
colony. The procedure continues until the greatest number recommended best value for this parameter is three [21].
of iterations is done, and the plant with the most superior Based on the problem dimension (the number of variables
fitness is selected as the optimum solution. that need to be optimized) and numbers of the optimiza-
To clarify the advantages of using the IWO, a few features tion goals and sampling frequencies, the maximum number
of this new algorithm are compared to two regular optimi- of iterations is selected. To obtain the desired radiation
zation algorithms, GA and PSO. Given that the size of the parameters, several values for the maximum number of itera-
population and quantity of repetitions are considered regular tions have been applied. The best value for this parameter is
prerequisites for every developmental algorithm, the least between 160 and 200.
and greatest number of seeds, nonlinear modulation index, To achieve good radiation characteristics along with
and the first and last SD are the key factors of the IWO algo- broad impedance bandwidth, various geometrical parame-
rithm that should be adjusted. As stated in [21], the maximum ters of the antenna can be selected as optimization variables.
speed, social and cognitive rates, inertial weight, and the In the optimization procedure, the size of the substrate is
boundary conditions are the most important parameters that kept fixed (50 mm 50 mm), and the other geometrical
affect the convergence of the PSO algorithm. Also, mutation parameters of the antenna (a, b, c, d, p, lp, wp, ls, ws) are
and crossover rates are the fundamental parameters of the chosen as the optimization variables. Thus, the optimiza-
GA that ought to be tuned to accomplish the convergence tion process includes a nine-dimensional solution space. The
[25]. Adjusting the basic parameters of the IWO algorithm, antenna gain and impedance matching over the 211-GHz
the starting and last SD, leads to a greater level of control in frequency range are optimization goals. Behind choosing
the final error and convergence of the algorithm; however, the
effects of the adjusting parameters on the convergence of the
PSO and GA are hard to distinguish. Furthermore, the IWO
can provide a superior steadiness with various boundary con- Table 1. TUNING PARAMETERS OF THE
ditions [19]. The above comparison elucidates a few points of IWO FOR ANTENNA OPTIMIZATION.
interest of the IWO algorithm. Parameter Description Value
Dim Problem dimension 9
Antenna Optimization by Using IWO
itermax Maximum number of iterations 180
In this section, the impressiveness and capability of the
IWO algorithm for the printed UWB antenna are inspected m Number of sampling frequencies 37
by an illustrative example. Table 1 shows the adjusting N Number of initial population 40
parameters of the IWO algorithm and their values for the n Nonlinear modulation index 3
antenna optimization. The beginning SD should be chosen
pmax Maximum number of plant 40
approximately 15% of the dynamic scope of every vari-
population
able. The last SD ought to be chosen deliberately to per-
mit the optimization program to discover the best answer Smax Maximum number of seeds 6
with the lowest final error as precise as could reasonably Smin Minimum number of seeds 1
be expected. A better response can be accomplished by SDinital Initial value of SD 10
diminishing this parameter. However, setting the last SD SDfinal Final value of SD 0.01
much less than the accuracy standard of the optimization

IEEE Antennas & Propagation Magazine february 2017 51


replicated antenna configurations (reproduced seeds and
Set the Initial Values of the Optimization parent weeds) are simulated by using an HFSS package
Parameters and Variables in MATLAB and then sorted based on their fitness. The greatest number
of flowering plants in the province is achieved as antenna
Initialize N Randomly Generated Weeds configurations (weeds) with better performance survive, and
(Trial Antenna Structures) in the Search Space inappropriate configurations with lower fitness values are
eliminated. This procedure proceeds until all prerequisites
Evaluate the Fitness of Each Trial are fulfilled, and the best antenna configuration with excel-
Antenna Structure lent fitness is selected as the optimum solution.
The essential stage of the optimization procedure is char-
Eliminate Plants (Antenna Structures) with acterizing the cost function. To achieve the optimal values
Lower Fitness to Reach the Maximum for the geometrical dimensions, at each repetition of optimi-
Number of Plants zation, the present antenna structure must be simulated by
using an HFSS, and the obtained radiation characteristics
Reproduce Based on Fitness Values: Create the Seeds must be contrasted with the desired characteristics. As a
Population (New Antenna Structures) from Each Plant result, the cost function of a produced antenna structure is
(Initial Antenna Structure) Depending on Its Fitness Value the difference between the obtained radiation parameters
and the desired performances. Notice that, in the repetitive
Distribute the New Generated Seeds optimization procedure, the standard is the minimization of
(Antenna Structures) over the Solution Space the cost function. The optimization program is implemented
in MATLAB. Ansoft HFSS and MATLAB are simultaneously
applied, and Visual Basic script (VBS) is used as an interface
Implement HFSS Using VBS Interface
between the two packages.
At each repetition of the optimization procedure, the anten-
na structures are simulated by an HFSS. After that, to calculate
Analyze Each Antenna Structure in HFSS and Obtain
Radiation Characteristics at the Sampling Frequencies the error (cost function), VBS is used to transfer the data of
radiation characteristics from the HFSS to MATLAB. In the
next stage, the cost functions of the structures are calculated in
Transfer Data of Radiation MATLAB, and then the antenna structures are sorted based
Characteristics from HFSS to MATLAB
on their fitness. Afterward, the best structure with lower cost
function is drawn and run in an HFSS by the command code
Evaluate the Fitness of MATLAB. The optimization procedure proceeds until either
of Each Structure the greatest number of repetitions is done or the radiation char-
in MATLAB
acteristics that are closest to the desired radiation parameters
are achieved. The procedure of the optimization program is
Desired Radiation
Characteristics shown in Figure 2. To take the frequency variation effects into
account, the antenna geometrical dimensions are optimized by
using a frequency-related cost function with optimal weighting
Is a Termination No
coefficients as follows:
Criterion Met?
m
Yes F= 1 / W j [E 1 ( f j) + E 2 ( f j)],(3)
m j=1

Keep the Best Fit


Sj +Sj
Result E 1 ( f j) = W a c m, (4)
2

Figure 2. A flowchart of the optimization program. S j = [S 11 ( f j) - (S 11 desired)], (5)

E 2 ( f j) = W b ^G desired ( f j) - G ( f j) h, (6)
optimization variables and setting optimization parameters,
initial antenna configurations (beginning seeds) are fortu- where E1 is the relative error of the desired S11, E2 is dif-
itously chosen in the search space (plant area). Afterward, ference between the obtained and the desired gains, m is
according to the least and greatest number of seeds and the number of sampling frequencies, fj represents the jth
ability of the developed plant, every initial configuration sampling frequency, and Wj is the weighting coefficient at
(weeds) produces new trial configurations. These recently the jth sampling frequency. In the optimization procedure,
produced configurations have a standard Gaussian dis- the desired S11 is selected as 10 dB so that, referring to
tribution in the neighborhood of the parent plants and (4) and (5), for S11 lower than 10 dB, the related error is
the SD defined by (2). In the next stage, both initial and not considered. The significance of the input impedance

52 february 2017 IEEE Antennas & Propagation Magazine


matching against the gain deter- after 180 iterations, after which
mines the (Wa /Wb) ratio. In the the optimum solution is achieved
cost function, the weighting val- At each repetition (the HFSS runs 180 times). The
ues Wa and Wb are chosen as 1 of theoptimization overall optimization framework
and 0.8, respectively. takes about 34.4 h on a personal
To reach the best antenna con- procedure, the computer platform of a 3.3-GHz
figuration with the desired perfor- antenna structures are six-core processor and 16-GB core
mance over the whole 211-GHz memory. The optimal values of the
frequency range, various weight- simulated by an HFSS. optimization variables are demon-
ing coefficients for the sampling strated in Table 3. The size reduc-
frequencies are utilized. Figure 3 tion can be achieved by limiting
shows the desired gain and the final optimum weighting coef- the size of the square FR4 substrate in the optimization
ficient for the sampling frequencies versus frequency. In this procedure. Figure 5 exhibits the fabricated prototype of the
figure, the values of the weighting coefficients are obtained optimized CPW-fed UWB antenna, and both the measured
by using an HFSS. Also, the central values of the search areas and simulated return loss curves of the optimized antenna are
for optimization variables are obtained through numerous illustrated in Figure 6. The impedance bandwidth of the opti-
simulations via an HFSS. Then, to obtain the best antenna con- mized antenna for S11 < 10 dB, from both the measurement
figuration with optimal geometrical dimensions, a wide range and simulation, is 1.6 11.2 GHz (150%), which also covers
is considered around each central value. The search areas for the bandwidth requirement for UWB (3.110.6 GHz) applica-
optimization variables are presented in Table 2, where m 0 is the tions. The measured and simulated copolar and cross-polar
wavelength of the free space at 6.5 GHz (the center frequency far-field patterns of the optimized antenna at the different typ-
of the desired bandwidth). ical frequencies are compared in Figure 7, where the E- and
The optimization process includes a nine-dimensional H-plane patterns represent y-z and x-z plane patterns of the
solution space, and the antenna gain and impedance match- antenna, respectively. A reasonable accordance between the
ing over the 211-GHz frequency range are optimization
goals. Therefore, to achieve the desired results over a wide
Table 2. SEARCH AREAS
bandwidth, a powerful optimization method linked with
FOR OPTIMIZATION VARIABLES.
an electromagnetic simulator package is required, and an
HFSS or other common full-wave analysis embedded opti- Parameter Search Area
mization program would not give the desired results.
a 0.030 < a < 0.080
Results and Discussion b 0.030 < b < 0.150
To justify the simulation results, two prototypes of the opti- c 0.0040 < c < 0.020
mized antenna were manufactured and examined in an anecho- d 0.030 < d < 0.20
ic chamber. The experimental results of the fabricated antenna
p 0.030 < p < 0.10
were compared to the simulation data. As mentioned previously,
the optimized antenna was simulated by using an HFSS pack- lp 0.10 < lp < 0.350
age that is based on the finite-element method. wp 0.20 < wp < 0.60
The convergence diagram of the cost (object) function ver- ls 0.50 < ls < 0.750
sus the number of repetitions is presented in Figure 4. It can ws 0.750 < ws < 1.080
be seen that minimum value of the cost function is achieved

5 5 20
Object Function (dB)
Weighting Coefficient

Desired Gain (dBi)

4 4 15
3 3
10
Weighting Coefficient
2 2
Desired Gain
5
1 1
1
0 0 0
2 3 4 5 6 7 8 9 10 11 0 20 40 60 80 100 120 140 160 180
Frequency (GHz) Number of Iterations

Figure 3. A desired gain and final optimum weighting Figure 4. The convergence curve of the object function
coefficient for the sampling frequencies. versus iteration.

IEEE Antennas & Propagation Magazine february 2017 53


simulated results and measured power gain of the E-plane occurs
patterns is obtained. As illustrated at the other direction, which is
in this figure, the H-plane pat- The measured outcomes far from the boresight direction.
terns are quasi-omnidirectionnal of group delay show In fact, the main beam is tilted
at low and midfrequencies, and away from the boresight direction
they are bidirectional at 11 GHz. that the optimized in the E-plane. Therefore, there
Also, all E-plane patterns exhibit antenna provides a are some differences between
almost bidirectional properties. the normalized E- and H-plane
However, the slight deformation time-domain response patterns at i = 0c. The desired,
of the E-plane patterns at higher without considerable measured, and simulated gains of
frequencies is due to the propaga- the optimized antenna against fre-
tion of the higher order modes. distortion. quency are compared in Figure 8.
Over the whole frequency range, The peak gain of the fabricated
the cross-polar component of antenna is about 4.65 dBi, which
the E-plane pattern is less than 20 dB. Furthermore, the occurs at higher band-edge frequency, and the gain variation
H-plane radiation patterns present a comparatively high cross- over the whole frequency band is less than nearly 2.6 dBi.
polarization level. One of the significant characteristics in a broadband com-
This deportment is due significantly to the radiation of munication system design is the group delay between the
the small slots, which causes a considerable increase of the transmitting and receiving antennas, which can cause a pulse
cross-polarization level. However, the cross-polarization level distortion at the receiving antenna. To prevent the pulse dis-
within the half power beamwidth is acceptable for H-plane tortion, the group delay must be constant. To verify the time-
patterns. Notice that, in Figure 7, E- and H-plane patterns domain response of the optimized antenna, the group-delay
are normalized to their maximum power gains. As shown in performance for face-to-face and side-by-side arrangements
this figure, the maximum power gain of the H-plane occurs was studied. Figure 9 depicts the simulated and measured
at i = 0c (the boresight direction), while the maximum results of the group delay. The distance between two antennas
is 300 mm, and the difference between the lowest and high-
est values of the group delay for the two arrangements is less
Table 3. OPTIMUM VALUES than 0.5 ns over the whole working band, except at 8.5 GHz
OF THE OPTIMIZATION VARIABLES. where (a) and (b) show a discontinuity. The group delay sudden
jump near 8.5 GHz is due partly to the scattering effect of the
Parameter a b c d p receiving antenna. Also, it can be due to the resonance at this
Value (mm) 2.3 2.7 0.3 3.5 3.1 frequency. However, any deviations from linear phase over
the frequency range will cause group-delay variations. The
Parameter lp wp ls ws
measured outcomes of group delay show that the optimized
Value (mm) 9.4 13.9 27.1 40.1 antenna provides a time-domain response without consider-
able distortion.
To perform a statistical error analysis, the bar graphs for
errors of S11 and group delay versus frequency are presented in
Figure 10. Notice that the error value is the deviation between
the simulated and measured outcomes. The average values

10
S11 (dB)

20
Simulation
Measurement
30
S11 Desired

1 2 3 4 5 6 7 8 9 10 11 12
Frequency (GHz)

Figure 5. The fabricated prototype of the optimized Figure 6. The desired, measured, and simulated return loss
antenna. (Photo courtesy of Yasouj University.) curves of the antenna.

54 february 2017 IEEE Antennas & Propagation Magazine


z z z

0 0 0
30 30 30 30 30 30

10 dB 10 dB 10 dB
60 60 60 20 dB 60 60 20 dB 60
20 dB

90 90 90 90 90 90

120 120 120 120 120 120

150 150 150 150 150 150


180 180 180 Z
(a) (b) (c)

X Y
z z z

0 0 0
30 30 30 30 30 30
10 dB
10 dB
10 dB 20 dB
60 60 60 60 60 20 dB 60
20 dB

90 90 90 90 90 90

120 120 120 120 120 120

150 150 150 150 150 150


180 180 180
Z
(d) (e) (f)

Y X

Copolar Simulation Copolar Measurement Cross-Polar Simulation Cross-Polar Measurement

Figure 7. The radiation patterns of the optimized CPW-fed antenna: (a)(c), E-plane and (d)(f) H-plane patterns at 4, 6.5, and
11 GHz, respectively. For (a)(c), the data is displayed using a theta plane angle on an X axis; and for (d)(f), the data is displayed
using a theta plane angle on a Y axis.

of the relative errors of S11 and the group delay are 1.277 dB
and 0.0335 ns, respectively, which imply that the measured
5
outcomes are in acceptable concordance with the simulated
data. Based on the obtained outcomes, the designed antenna 4
can be used in recent wideband communication systems such
Gain (dBi)

as high-speed wireless communication, ground penetrating 3


radar, phased array radars, electronic warfare, jammer systems,
electromagnetic compatibility testing systems, and biomedical 2 Simulation
microwave imaging systems. Specifically, the optimized antenna Measurement
Desired
can be fixed on the outer surface of communication satellites 1
2 3 4 5 6 7 8 9 10 11
and military airplanes. Also, it can be simply integrated with
Frequency (GHz)
mobile communication systems.
Finally, the efficiency of the optimized antenna in terms of
size, gain, and bandwidth are compared with previous designs Figure 8. The desired, simulated, and measured gains of the
reported in [11][17]. The comparison results are presented in antenna.

IEEE Antennas & Propagation Magazine february 2017 55


56
Error of Group Delay (ns) Error of S11 (dB) Group Delay (ns) Group Delay (ns)

0
0.5
1
1.5
0
0.5
1
1.5

0
1
2
3
4
5
1
1

0.025
0.075

0.05
0.1

0
1
2
2

2
3
3

3
face configurations.

3
4
4

4
30 cm 30 cm

and (b) error of group delay.


5
Simulation

6
(b)
(a)

(b)
7

(a)

7
8

8
Frequency (GHz)
Frequency (GHz)

Frequency (GHz)
Frequency (GHz)
9
5 6 7 8 9
5 6 7 8 9

9
Measurement

Figure 9. The numerical and experimental group-delay

10 11 12
10 11 12
10 11 12
10 11 12

Figure 10. The error values versus frequency: (a) error of S11
outcomes of the antenna for (a) side-by-side and (b) face-to-
Table 4. PERFORMANCE OF THE OPTIMIZED ANTENNA

february 2017
COMPARED WITH A NUMBER OF PREVIOUS PROPOSED DESIGNS.

[13] [15]
Proposed
Antenna [11] [12] Antenna 1 Antenna 2 [14] Antenna 1 Antenna 2 [16] [17]
2
Antenna size (mm ) 50 50 72 72 100 100 110 110 110 110 80 80 40 38 44 44 48 42 50 50
Frequency range (GHz) 1.611.2 1.562.88 2.799.84 1.827.23 2.428.48 3.908.26 3.110.6 3.7510.3 313 2.811
Impedance 150 60 110 120 110 71.7 109.5 93.2 125 118.8
bandwidth (%)
Antenna gain (dBi) 2.054.65 3.75 4.88 3.5 (average- 36.5 36.5 02.5 2.06.0 2.06.0 3.24.2 Not presented
simulation) (for 37 GHz)

IEEE Antennas & Propagation Magazine


Table 4, where it is demonstrated that the optimized antenna [5] J.-Y. Chiou, J.-Y. Sze, and K.-L. Wong, A broad-band CPW-fed striploaded
square slot antenna, IEEE Trans. Antennas Propag., vol. 51, no. 4, pp. 719721,
features a smaller size as well as a much broader working band Apr. 2003.
compared with those presented in [11][14]. Although the sizes [6] Y. C. Lin and K. J. Hung, Compact ultrawide-band rectangular aperture
of those antennas published in [15] and [16] are partly smaller antenna and band-notched designs, IEEE Trans. Antennas Propag., vol. 54,
than the proposed one, the impedance bandwidths are sig- no. 11, pp. 30753081, Nov. 2006.
nificantly less than the bandwidth of the optimized antenna. [7] J. Y. Sze, C. I. G. Hsu, and S. C. Hsu, Design of a compact dual-band
annular-ring slot antenna, IEEE Antennas Wireless Propag. Lett., vol. 6,
Also, the lower band-edge frequencies of these antennas are pp. 423426, Sept. 2007.
3.1, 3.75, and 3 GHz, which are at least 1.4 GHz higher than [8] D. D. Krishna, M. Gopikrishna, C. K. Anandan, P. Mohanan, and K. Vasude-
the lower band-edge frequency of the designed antenna. More- van, CPW-fed Koch fractal slot antenna for WLAN/WiMAX applications,
over, the antenna demonstrated in [17] with the same size of IEEE Antennas Wireless Propag. Lett., vol. 7, pp. 389392, May 2008.
50mm50mm, has a narrower impedance bandwidth com- [9] J. Y. Sze and K.-L. Wong, Bandwidth enhancement of a microstrip line-
fed printed wide-slot antenna, IEEE Trans. Antennas Propag, vol. 49, no. 7,
pared with the optimized structure presented in this work. As pp. 10201024, July 2001.
shown in this table, compared with other designs, the optimized [10] H. L. Lee, H. J. Lee, J. G. Yook, and H. K. Park, Broadband planar
antenna has a satisfactory gain. This comparison verifies the antenna having round corner rectangular wide slot, in Proc. IEEE Antennas
performance of the IWO method. and Propagation Society Int. Symp., 2002, pp. 460463.
[11] H. D. Chen, Broadband CPW-fed square slot antennas with a widened
tuning stub, IEEE Trans. Antennas Propag., vol. 51, no. 4, pp. 19821986, Aug.
Conclusions 2003.
The IWO algorithm has been prosperously employed for the [12] R. Chair, A. A. Kishk, and K. F. Lee, Ultrawideband coplanar waveguide-
optimization of a compact CPW-fed printed antenna over the fed rectangular slot antenna, IEEE Antennas Wireless Propag. Lett., vol. 3,
1.611.2-GHz frequency range. Some of the antenna geometrical pp. 227229, Dec. 2004.
dimensions were chosen as the optimization parameters, and [13] Y. F. Liu, K. L. Lau, Q. Xue, and C. H. Chan, Experimental studies of
printed wide-slot antenna for wide-band applications, IEEE Antennas Wireless
then the antenna impedance bandwidth and gain were optimized Propag. Lett., vol. 3, pp. 273275, Dec. 2004.
toward the necessary specifications. To obtain the desirable elec- [14] C. Y. Chiu, C. H. Chan, and K. M. Luk, Study of a small wide-band patch
tromagnetic characteristics over the whole frequency range, the antenna with double shorting walls, IEEE Antennas Wireless Propag. Lett.,
antenna geometrical dimensions were optimized by using a fre- vol. 3, pp. 230231, Dec. 2004.
quency-related cost function with optimal weighting coefficients. [15] P. Li, J. Liang, and X. Chen, Study of printed elliptical/circular slot anten-
nas for ultrawideband applications, IEEE Trans. Antennas Propag., vol. 54, no.
Moreover, the small size of 50 mm 50mm was achieved by 6, pp. 16701675, June 2006.
exerting constraints on the optimization parameters. The meas- [16] D. Unnikrishnan, D. Kaddour, S. Tedjini, E. Bihar, and M. Saadaoui,
ured outcomes were in good agreement with the full-wave simu- CPW-fed inkjet printed UWB antenna on ABS-PC for integration in molded
lations, which proves the validity of the presented optimization interconnect devices technology, IEEE Antennas Wireless Propag. Lett., vol.
14, pp. 11241128, May 2015.
technique. Also, the measured outcomes of group delay showed
[17] J. Y. Siddiqui, C. Saha, and Y. M. M. Antar, A novel ultrawideband (UWB)
that the optimized antenna provides a time-domain response printed antenna with a dual complementary characteristic, IEEE Antennas
without significant distortion, which assures that it is suitable for Wireless Propag. Lett., vol. 14, pp. 974977, Apr. 2015.
use in UWB or other communication systems. The antenna was [18] K. Xu, Z. Zhu, H. Li, J. Huangfu, C. Li, and L. Ran, A printed single-layer
compared with a number of previous proposed designs, and com- UWB monopole antenna with extended ground plane stubs, IEEE Antennas
Wireless Propag. Lett., vol. 12, pp. 237240, Mar. 2013.
parison results showed that the optimized antenna outperforms
[19] A. R. Mehrabian and C. Lucas, A novel numerical optimization algorithm
other designs. inspired from weed colonization, Ecol. Inform., vol. 1, no. 3, pp. 355366,
2006.
AUTHOR INFORMATION [20] K. Suresh, D. Kundu, S. Ghosh, S. Das, and A. Abraham, IWO with
Aliakbar Dastranj (dastranj@yu.ac.ir) is with Yasouj University, increased deviation and stochastic selection (IWO-ID-SS) for global optimiza-
tion of noisy fitness functions, in Proc. World Congr. Nature and Biologically
Yasouj, Iran. His research interests include novel designs of Inspired Computing, Coimbatore, India, Dec. 2009, pp. 215220.
modern antennas for advanced applications, design and model- [21] S. Karimkashi and A. A. Kishk, Invasive weed optimization and its
ing of microwave structures, evolutionary algorithms for electro- features in electromagnetics, IEEE Trans. Antennas Propag., vol. 58, no. 4,
magnetic applications, and electromagnetic theory. pp. 12691278, Apr. 2010.
[22] S. H. Sedighy, A. R. Mallahzadeh, M. Soleimani, and J. Rashed-Mohassel,
Optimization of printed Yagi antenna using invasive weed optimization (IWO),
References IEEE Antennas Wireless Propag. Lett., vol. 9, pp. 12751278, Dec. 2010.
[1] C. C. Yu and X. C. Lin, A wideband single chip inductor-loaded CPW- [23] A. A. Dastranj, H. Abiri, and A. R. Mallahzadeh, Design of a broadband
fed inductive slot antenna, IEEE Trans. Antennas Propag., vol. 56, no. 5, cosecant squared pattern reflector antenna using IWO algorithm, IEEE Trans.
pp. 14981501, May 2008. Antennas Propag., vol. 61, no. 7, pp. 38953900, Jul. 2013.
[2] X. C. Lin and C. C. Yu, A dual-band CPW-fed inductive slot-monopole hybrid [24] A. A. Dastranj, H. Abiri, and A. R. Mallahzadeh, Two-dimensional synthe-
antenna, IEEE Trans. Antennas Propag., vol. 56, no. 1, pp. 282285, Jan. 2008. sis and optimization of a broadband shaped beam reflector antenna using IWO
[3] A. U. Bhobe, C. L. Holloway, M. Piket-May, and R. Hall, Wide-band slot and PSO algorithms, Int. J. RF Microwave CAE, vol. 25, no. 2, pp. 129140,
antennas with CPW feed lines: Hybrid and log-periodic designs, IEEE Trans. Mar. 2015.
Antennas Propag., vol. 52, no. 10, pp. 25452554, Oct. 2004. [25] Y. Ramat-Samii and E. Michielssen, Eds., Electromagnetic Optimization
[4] S. I. Latif, L. Shafai, and S. K. Sharma, Bandwidth enhancement and size byGenetic Algorithms. New York: Wiley, 1999.
reduction of microstrip slot antennas, IEEE Trans. Antennas Propag., vol. 53,
no. 3, pp. 9941003, Mar. 2003. 

IEEE Antennas & Propagation Magazine february 2017 57

You might also like