Progress In Electromagnetics Research B, Vol.

54, 265–284, 2013

CIRCULAR ANTENNA ARRAY DESIGN BY USING

EVOLUTIONARY SEARCH ALGORITHMS

Pinar Civicioglu*
Department of Aircraft Electrics and Electronics, College of Aviation,
Erciyes University, Kayseri 38039, Turkey

Abstract—Evolutionary Search Algorithms (EA) have been inten-

sively used in solving numerical optimization problems. Since design
of antenna arrays is a numerical optimization problem, EAs have been
intensively used in solving antenna arrays design problems. Although
EAs are widely used in antenna array design problems, a performance
comparison study of the intensively used EAs for circular antenna ar-
ray design problem has been scarcely studied. In this paper, 3 different
circular antenna array design problems have been solved by using 15
different evolutionary search algorithms (i.e., ABC, ACS, BSA, CK,
CLPSO, CMAES, DE, E2-DSA, EPSDE, GSA, JADE, JDE, PSO,
SADE, S-DSA). The objective function designed for solution of the
relevant circular antenna array design problems ensures minimization
of side lobe levels, acquisition of maximum directivity, and null control
of the non-uniform, planar circular antenna array. Obtained statistical
analysis results show that S-DSA solves the relevant circular antenna
array design problems statistically better than the other evolutionary
algorithms used in this paper.

1. INTRODUCTION

Since antenna arrays are intensively used in mobile and wireless

communication systems, optimum design of array patterns is of
vital importance for increasing channel capacities of these systems,
broadening their coverage areas, and ensuring the efficient spectrum
utilization. When a single element based antenna is used, it
gets difficult to meet the gain or highly directive radiation pattern
conditions required in the long distance communication. Antenna
arrays use many individual antennas by means of a geometric and
Received 1 May 2013, Accepted 7 September 2013, Scheduled 8 September 2013
* Corresponding author: Pinar Civicioglu ([email protected]).
266 Civicioglu

electrical configuration, a variety of features of which have been

optimized previously. The antenna arrays are widely used in many
applications such as radar [1], GPS [2], sonar [3], radios [4], and third
generation wireless communication systems [5].
The objective function to be used for obtaining the optimum
parameters pertaining to antenna array geometry aims at determining
the positions of array elements [6–12]. Optimization of the positions
of array elements is very important for producing a radiation pattern
that is as similar as possible to the radiation pattern required to be
produced via the relevant antenna array [13–20]. Many researchers
studying in the field of electromagnetic optimization problems have
conducted many studies on the subject of non-uniformly spaced linear
antenna arrays [21–24].
Most of the classical optimization algorithms need very good
initial-solution values pertaining to the relevant problem in order
to solve an antenna array design problem. In addition, classical
optimization algorithms usually need the derivative of the relevant
problem. Since antenna array design problems are generally in a
multimodal form, it is a high possibility that most of the classical
optimization algorithms are caught up in a local-solution. Evolutionary
search algorithms can sub-optimally solve a multimodal numerical-
optimization problem defined by using an objective function without
needing the derivative of the relevant problem. Thus, the use of
evolutionary search algorithms has become widespread in solution of
antenna array design problems [10, 13–16, 18–22, 25]. Evolutionary
algorithms have been commonly used for optimizing Side Lobe Level
(SLL) [24–26] values of antenna arrays, and ensuring an optimized
null control from the designed arrays. Many studies were carried
out in order to develop antenna arrays having different geometric
characteristics [10–20]. Circular shaped antenna arrays [7, 9, 11, 12]
are intensively used in applications such as sonars, radars, mobile and
commercial satellite communication systems [27–32]. Evolutionary
algorithms, which are widely used for solution of the circular
array design problem, are Particle Swarm Optimization (PSO) [14–
16, 18, 20, 21], Differential Evolution Algorithm (DE) [19, 33–35],
Biogeography-Based Optimization algorithm [10], Invasive Weed
Optimization algorithm [13, 22, 26], Genetic Algorithm [16] and the
derivatives of these algorithms [36–41].
This paper includes 15 different evolutionary search algorithms
(Artificial Bee Colony Algorithm (ABC) [42], Artificial Cooperative
Search Algorithm (ACS) [43], Backtracking Search Optimization Al-
gorithm (BSA) [44], Cuckoo Search Algorithm (CS) [45], Compre-
hensive Learning Particle Swarm Optimizer (CLPSO) [46], Covari-
Progress In Electromagnetics Research B, Vol. 54, 2013 267

ance Matrix Adaptation Evolution Strategy (CMAES) [47], Differ-

ential Evolution Algorithm (DERND; DE/rand/1/bin) [48], Differen-
tial Evolution Algorithm with Ensemble of Parameters (EPSDE) [49],
Gravitational Search Algorithm (GSA) [50], Adaptive Differential Evo-
lution Algorithm (JADE) [51], Self-Adaptive Differential Evolution
Algorithm (JDE) [52, 53], Particle Swarm Optimization Algorithm
(PSO) [45, 54], Strategy Adaptation Based Differential Evolution Algo-
rithm (SADE) [55] and Differential Search Algorithm (DSA; E2-DSA
and S-DSA) [53, 56]) in order to solve 3 different circular antenna array
design problems.
This paper is organized as follows: Sections 2 and 3 include
the Evolutionary Search Algorithms and Problem Formulation,
respectively. Section 4 describes the Experiments and Section 5
presents the Conclusions.

2. EVOLUTIONARY SEARCH ALGORITHMS

The general structures and basic evolutionary computing philosophy

foundation of the evolutionary search algorithms used in this paper are
briefly explained below.
ABC is a population-based evolutionary search algorithm that
analogically simulates nectar source search behavior of honey-bees.
ABC has an elitist, and two-phased search strategy. Because ABC’s
search strategy is elitist, its problem solving success of multimodal
problems is limited. ABC has only two control parameters and its
mutation strategy is rather similar to DE. ABC has no crossover
operator.
ACS is a swarm intelligence algorithm developed for solving
real valued numerical optimization problems. ACS is a bijective
search algorithm, that means each random solution evolves towards
a random solution of the related problem. The non-elitist structure of
ACS gives it ability to solve multimodal problems more successfully.
The swarm intelligence philosophy behind ACS is based on the
migration of two artificial superorganisms as they biologically interact
to achieve the global minimum value pertaining to the problem. In
ACS, a superorganism consisting of random solutions of the related
problem corresponds to an artificial superorganism migrating to more
productive feeding areas.
BSA has a simple structure that is effective, fast and capable
of solving multimodal problems and that enables it to easily adapt
to numerical optimization problems. BSA can be considered as a
modernized PSO. BSA has new strategies for crossover and mutation
operations and it has quite powerful local and global search abilities.
268 Civicioglu

In particular, BSA possesses a short-range memory in which it stores

a population from a randomly chosen previous generation for use in
generating the search-direction matrix. Thus, BSAs memory allows
it to take advantage of experiences gained from previous generations
when it generates a new trial preparation.
CS is a population based, elitist stochastic search algorithm. CK
has a tendency to evolve each random solution towards to the best
solution obtained beforehand. CK is structurally similar to DE and
ABC. However, it has a superior problem solving success in comparison
to ABC, DE and some DE variants. CK has only two control
parameters.
CLPSO is an advanced PSO version. Differently from the
standard PSO, CLPSO uses the historical best information of all
the other particles simultaneously to update the speed of a particle.
Instead of the pbest and gbest values used in the standard PSO,
CLPSO uses the pbest values of all random solutions in the swarm.
In contrary to the other numerous PSO versions, CLPSO can also
solve the multimodal functions.
CMAES is an evolution-strategy based metaheuristic algorithm.
CMAES is based on the update of a covariance matrix used to model
the relations among the parameters of the optimization problem.
CMAES uses the ranks of the random solutions in the population to
update the elements of relevant covariance matrix. In consequence
of the recombination, mutation, selection, and adaptation stages it
structurally has, CMAES tries to evolve the population iteratively.
CMAES needs the dimension of the problem, the initial mean and the
standard deviation values of the multivariate normal distribution to
initialization process.
DE is a non-gradient-based, evolutionary raw-genetic algorithm.
Its mathematical structure is very simple and it can be adapted to
many problem types easily. For such reasons, DE is perhaps the mostly
used optimization algorithm in the literature. The basic difference
of DE from the genetic algorithm is its multi-structured mutation
operator. The most frequently used mutation operator in the literature
of DE is DERND (i.e., DE/rand/1/bin), which is used as DE in this
paper.
EPSDE uses the mutation strategies existing in the DE algorithm.
In EPSDE, a mutation strategy existing in DE algorithm is assigned to
each element of the population. The value of the mutation coefficient
to be used for the mutation strategy selected in EPSDE is randomly
selected from a pool in such a way that it will remain between 0.4
and 0.9 in each iteration. Despite of its low-speed, EPSDE is a quite
successful algorithm in problem solving.
Progress In Electromagnetics Research B, Vol. 54, 2013 269

GSA is an evolutionary search algorithm and it has been inspired

from the universal gravitational laws. Random solution of the
respective problem desired to be solved in GSA has been modeled
as artificial-bodies that apply newtonian gravitational force to each
other. Mass of an artificial-body is related to the quality of the
solution that artificial-body provides for the respective problem. The
higher the quality of the solution, the slower the speed that artificial-
body abandons that position due to the gravitation force applied to it
by other artificial-bodies. Speed of the artificial-bodies with inferior
quality of solution is higher in the search-space. This phenomenon
allows GSA to search the search space very efficiently to find a solution
for a problem.
JADE has a new mutation strategy (i.e., DE/current-to-pbest)
that has been developed to be used together with DE. The mutation
operator used in JADE evolves a solution that is randomly selected
from the population, towards a random top-best solution that provides
the best solution at the moment in the population. JADE algorithm
can solve numerical optimization problems with much greater success
than DE/current-to-best/1 and DE/best/1 strategies that are used in
standard DE algorithms.
JDE uses a developed form of DE/rand/1/bin of DE. The initial
values of the mutation coefficient and crossover coefficient that are the
basic control parameters of DE algorithm are generally determined on
the basis of the structure of the problem to be solved. JDE similar
to DE operates on the basis of a fixed population size. JDE uses a
different mutation and crossover coefficients for each element of each
random solution making up the population. Although the structure of
JDE is very simple its success in problem solving is much higher than
DE/rand/1/bin algorithm.
PSO is a metaheuristic algorithm inspired by the joint movements
of the members of superorganisms. The artificial particles creating a
population in PSO and corresponding to the chromosomes used in the
genetic algorithm benefit from their initiative and social experiences
while searching the solution of a problem. The initiative and social
experience notions used in PSO correspond philosophically to the
local search and global search notions in the global search algorithms.
PSO has been used to solve numerous numeric optimization problems.
As a result of the ongoing researches, numerous relatively new PSO
structures that are generally more successful than the standard PSO
have been developed (e.g., CLPSO, PSO2007). In this paper, the
PSO2011, which is an advanced version of PSO2007 that includes a
considerable part of the improvements acquired through the long years
of researches carried out on the PSO algorithm has been used.
270 Civicioglu

SADE is an advanced DE version capable of using the mutation

strategies used in the standard DE adaptively. In SADE, the
mutation strategy to be used in any area is determined using the
probability values calculated on the basis of the successes the relevant
mutation strategies have had in the past iteration steps. Global search
capabilities of the SADE are rather advanced.
DSA is a multi-strategy based, advanced evolutionary swarm-
algorithm. DSA analogically simulates a superorganism that migrates
between two stopovers. DSA has only unique mutation and crossover
operators. The structure of mutation operator of DSA contains just
one direction pattern apart from the target pattern. The structure
of crossover operator of DSA is very different from the structures of
crossover operators used in advanced DE algorithms. DSA has only
two control parameters. These parameters pertaining to the crossover
process of DSA are used for controlling the degree to which the trial

Table 1. Control parameters of the related algorithms.

Algorithm Initial Values of Control Parameters
limit = N · D
ABC
Size of EmployedBee = (Size of Colony)/2
ACS p = 0.10
BSA mixrate = 1
CK p = 0.25, β = 1.5
[c1 , c2 ] = [1.49445, 1.49445], m=0
et −et(1)
CLPSO pc = 0.5 · et(p
³ s ) −et(1) ´
where t = 0.5 ≤ 0 : ps1−1 :1
N = 4 + b3 · log(Dimension of problem)c
CMAES σ = 0.25
j k
µ = 4+b2·log(N
2
)c

DERND F = 0.50, CR = 0.90

E2-DSA p1 = p2 = 0.30 · κ|κ ∼ U [0, 1]
EPSDE F=[0.4 0.5 0.6 0.7 0.8 0.9] CR=[0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9]
GSA Rnorm = 2, Rpower = 1 α = 20, G0 = 100
JADE c = 0.10, p = 0.05 CRm = 0.50, Fm = 0.50, Af actor = 1
JDE Finit = 0.50, CRinit = 0.90
PSO C1 = 1.80, C2 = 1.80 ω = 0.5 + (1 − rand)|rand ∼ U [0, 1]
SADE F ∼ N (0.5, 0.3), CR ∼ N (CRm , 0.10) c = 0.10, p = 0.05
S-DSA p1 = p2 = 0.30 · κ|κ ∼ U [0, 1]
Progress In Electromagnetics Research B, Vol. 54, 2013 271

pattern will mutate in comparison to the target pattern. Each trial

pattern uses the corresponding target pattern for evolving towards
stopovers that provide a better fitness value. Standard DSA has 4
different options for obtaining direction matrix. In Bijective DSA; B-
DSA, population evolves in each cycle into the randomly permuted
form of current population. In Surjective DSA; S-DSA, population
evolves into artificial organisms in which relatively better solutions are
found. In Elitist DSA; E1-DSA, population evolves into the randomly
selected top-best solutions of the original population. In Elitist DSA;
E2-DSA, population evolves into the better solution of the original
population. In this paper, S-DSA and E2-DSA have been employed.
Table 1 gives the initial values of the relevant control parameters
for the evolutionary algorithms used in this paper.

3. PROBLEM FORMULATION

The antenna elements constitute a circular antenna array, as it is

illustrated in the Figure 1 [17]. Array factor value (AF) belonging
to a circular antenna array can be calculated by use of Eq. (1) when
the wavefront of incident plane wave is perpendicular to the x-y plane.
XN
£ ¡ ¡ ¢ ¡ ¢¢ ¤
AF (φ)= In · exp j ·k·r· cos φ0 −φnang −cos φ0 −φnang +βn (1)
n=1
Here,
• φnang = 2 · π · n−1
N : Angular position of the nth element on the x-y
plane,

Figure 1. Geometry of circular antenna array.

272 Civicioglu

• k: Wave-number, where k ·r = N ·d and d denotes angular spacing

between elements,
• r: Radius of the circle defined by the circular array,
• ϕ0 : Direction of maximum radiation,
• ϕ: Angle of incidence of the plane wave,
• In : Current excitation,
• βn : Phase excitation of the nth element.
Constraints between elements of the designed circular antenna
array were defined through Eq. (2).
I n2 +1 ∠β n2 +1 = conj (I1 ∠β1 ) ,
I n2 +2 ∠β n2 +2 = conj (I2 ∠β2 ) , . . . , (2)
³ ´
In ∠βn = conj I n2 ∠β n2

In this paper, the objective function defined in Eq. (3) has been
used for solution of the circular antenna array design problem [17].
¯ ³ ´¯.¯ ³ ´¯
¯ ~ ϕ0 ¯¯ ¯¯AR ϕmax , I,
~ β, ~ ϕ0 ¯¯
~ β,
F = w1 · ¯AR ϕsll , I,
1
+w2 · ³ ´ + w3 · |ϕ0 − ϕdes |
~ β
DIR ϕ0 , I, ~
num
X ¯ ³ ´¯
¯ ~ ϕ0 ¯¯
~ β,
+w4 · ¯AR ϕk , I, (3)
k=1
where w1 = w2 = w3 = w4 = 1 have been used in this paper.
The purpose of the objective function defined in Eq. (3) is to
minimize side-lobe levels, maximize directivity, and obtain null control.
To accomplish the purposes of the objective function, phase and
amplitude values exciting the antenna elements were investigated. The
range of normalized amplitude excitations is [0 1], and the range of
phase excitations is [−180 180].
In Eq. (3), the term of |AR(φsll , I, ~ φ0 )|/|AR(φmax , I,
~ β, ~ φ0 )|
~ β,
enables minimizing the side lobes. ϕsll shows the angle where the
maximum side lobe level is obtained.
1
The term of ~ used in Eq. (3) maximizes the value of
~ β)
DIR(φ0 ,I,
the directivity of the array pattern. The directivity is very useful
for meeting different array patterns. The term of |φ0 − φdes | strives to
drive the maxima of the array pattern close to the desired maxima ϕdes .
P
num
~ β,~ φ0 )| penalizes the objective function if
The term of |AR(φk , I,
k=1
Progress In Electromagnetics Research B, Vol. 54, 2013 273

sufficient null control is not achieved. num is the number of null control
directions and ϕk specifies the kth null control direction.
Table 2 presents the descriptive properties pertaining to the
circular antenna design problem used in this paper. The relevant
circular antenna array design problems were solved by the relevant
algorithms, and current excitation and phase of the antenna elements
values enabling to obtain the desired pattern were obtained.

Table 2. The circular antenna array design problems used in tests.

null
Test # low up dim φdes
control
[0.2, 0.2, 0.2, [1, 1, 1,
0.2, 0.2, 0.2, 1, 1, 1,
Test 1 12 180◦ -
−180, −180, −180, 180, 180, 180,
−180, −180, −180] 180, 180, 180]
[0.2, 0.2, 0.2, [1, 1, 1,
0.2, 0.2, 0.2, 1, 1, 1,
Test 2 12 180◦ 120◦
−180, −180, −180, 180, 180, 180,
−180, −180, −180] 180, 180, 180]
[0.2, 0.2, 0.2, [1, 1, 1,
0.2, 0.2, 0.2, 1, 1, 1,
Test 3 24 180◦ -
−180, −180, −180, 180, 180, 180,
−180, −180, −180] 180, 180, 180]

4. EXPERIMENTS

This section examines in detail the statistical parameters used in

analysis of the test results, algorithmic precision used in the statistical
analysis, common control parameters of related algorithms, stopping
conditions used by the relevant algorithms, and statistical results.
In the tests conducted, the relevant circular antenna array design
problems were solved 30 times, each time using a different initial
pattern-matrix [45]. In each test conducted, the relevant evolutionary
computing algorithms use the same initial population. Various values
acquired in the tests conducted (i.e., global minimum and global
minimizer ) are kept for detailed statistical analysis. All tests and
statistical analysis performed herein have been carried out in the
Matlab platform.
274 Civicioglu

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

(j) (k) (l)

(m) (n) (o)

Figure 2. The best array patterns for the Test-2, which is obtained
by (a) ACS, (b) JADE, (c) PSO, (d) BSA, (e) SADE, (f) JDE, (g) CK,
(h) DERND, (i) GSA, (j) EPSDE, (k) E2-DSA, (l) ABC, (m) S-DSA,
(n) CMAES, (o) CLPSO.
Progress In Electromagnetics Research B, Vol. 54, 2013 275

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

(j) (k) (l)

(m) (n) (o)

Figure 3. The best array patterns for the Test-2, which is obtained
by (a) BSA, (b) CLPSO, (c) PSO, (d) E2-DSA, (e) DERND, (f) JDE,
(g) SADE, (h) JADE, (i) GSA, (j) CK, (k) ACS, (l) ABC, (m) S-DSA,
(n) EPSDE, (o) CMAES.
276 Civicioglu

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

(j) (k) (l)

(m) (n) (o)

Figure 4. The best array patterns for the Test-2, which is obtained
by (a) EPSDE, (b) JADE, (c) JDE, (d) BSA, (e) SADE, (f) PSO,
(g) ACS, (h) CK, (i) CMAES, (j) E2-DSA, (k) DERND, (l) ABC,
(m) S-DSA, (n) GSA, (o) CLPSO.
Progress In Electromagnetics Research B, Vol. 54, 2013 277

4.1. Algorithmic Precision

The algorithmic precision of numerous modern software development
tools is at the level of 10−16 in the double-precision mode. If the
arithmetic precision value is selected higher than necessary, it becomes
difficult to compare the local search abilities of the algorithms. In the
statistical analysis conducted in this paper, the arithmetic precision
value has been determined as 10−16 so that it can cover the precision
level needed in many practical applications.

4.2. Common Control Parameters of Algorithms

The common control parameters of the relevant evolutionary search
algorithms are as follows.
• The maximum generation number value (i.e., maxcycle) is
2,000,000.
• Size of the pattern matrix (i.e., size of population) = 30.

Table 3. Statistical values of global minimum values obtained from

30-trials in Test 1; Mean: mean value of global minimum values, Std:
standard deviation of the mean global minimum value, Best: the best
value of global minimum values and Median: median value of global
minimum values.
Statistics
Algorithm
Mean Std Best Median
S-DSA −21.3959 0.955676 −21.9024 −21.7067
EPSDE −21.581 0.377924 −21.9023 −21.6606
CK −21.7273 0.150403 −21.9023 −21.7076
BSA −21.6781 0.198937 −21.9002 −21.7067
ACS −21.6273 0.174802 −21.899 −21.6389
CMAES −2.98977 27.6972 −21.8979 −12.1883
E2-DSA −21.5388 0.449843 −21.8976 −21.6158
DERND −20.947 1.759608 −21.8975 −21.4957
SADE −21.3964 0.621627 −21.8967 −21.5322
JADE −21.353 0.239282 −21.86 −21.3393
CLPSO −19.5979 1.560356 −21.0628 −20.1316
ABC −17.534 1.351429 −20.5099 −17.2709
GSA −16.5437 0.42498 −17.5645 −16.4553
JDE −9.72518 1.931982 −16.6028 −9.28692
PSO −13.4075 0.505283 −14.3721 −13.3552
278 Civicioglu

4.3. Stopping Conditions

The predetermined criteria to stop the search processes of the relevant
algorithms used in this paper are as follows.
• If absolute value of objective function is less than 10−16 , stop.
• If the algorithm has failed to find a better solution than the
existing solution even at the end of the last 200,000 function
evaluations, stop.
• When the number of function evaluations reaches 2,000,000, stop.

4.4. Statistical Analysis

Tables 3–5 gives basic statistical values (i.e., M ean, Std, Best, M edian)
of global optimum values obtained in 30 trials by the relevant algo-
rithms for the circular antenna array design problems. Array patterns
obtained by the relevant algorithms are showed in the Figures 2–4.

Table 4. Statistical values of global minimum values obtained from

30-trials in Test 2; Mean: mean value of global minimum values, Std:
standard deviation of the mean global minimum value, Best: the best
value of global minimum values and Median: median value of global
minimum values.
Statistics
Algorithm
Mean Std Best Median
S-DSA −21.4399 0.614433 −21.9017 −21.6274
CK −21.65 0.178681 −21.9017 −21.6677
SADE −21.277 0.701287 −21.9002 −21.4768
E2-DSA −21.4905 0.288214 −21.8987 −21.4796
BSA −21.6608 0.165671 −21.8981 −21.6674
EPSDE −21.6397 0.183391 −21.8915 −21.6602
ACS −21.6033 0.173024 −21.8824 −21.506
JADE −21.4089 0.271234 −21.8393 −21.3857
DERND −21.3465 0.426849 −21.8368 −21.4212
CLPSO −20.2716 0.933406 −21.2502 −20.5683
CMAES 18.59502 37.19945 −20.565 4.370296
ABC −18.0684 1.167902 −20.3474 −18.0741
GSA −16.6642 0.421773 −17.6169 −16.6748
JDE −9.61435 2.39578 −17.6116 −8.94924
PSO −13.2438 0.644399 −14.6641 −13.2706
Progress In Electromagnetics Research B, Vol. 54, 2013 279

Table 5. Statistical values of global minimum values obtained from

30-trials in Test 3; Mean: mean value of global minimum values, Std:
standard deviation of the mean global minimum value, Best: the best
value of global minimum values and Median: median value of global
minimum values.
Statistics
Algorithm
Mean Std Best Median
S-DSA −21.5036 6.7538 −33.4334 −19.3358
E2-DSA −24.6933 4.2943 −32.6903 −25.8875
ACS −22.2482 4.5013 −32.3133 −21.3762
BSA −23.1325 4.2970 −32.1408 −24.4438
EPSDE −19.4702 5.9493 −31.8883 −16.2445
GSA −22.8905 1.8213 −24.8898 −23.5310
DERND −17.7545 2.6231 −24.3769 −16.9840
CK −14.6113 1.7005 −22.1947 −14.4060
SADE −16.2070 1.2932 −18.7705 −16.4255
JADE −15.7014 1.0930 −18.4688 −15.5797
CLPSO −14.3385 0.6390 −16.6294 −14.2788
ABC −12.7516 0.7458 −15.1482 −12.7759
CMAES 15.3954 21.8090 −14.3321 12.1527
PSO −12.7375 0.5976 −13.9626 −12.7660
JDE −8.0792 1.4484 −12.1239 −7.5831

Examining the Figures 2–4, it is seen that S-DSA, in comparison to

the other algorithms, achieved better minimized side lobes and the
greatest directivity.

5. CONCLUSIONS

This paper includes solutions to 3 different circular antenna array

design problems using 15 different evolutionary search algorithms. 30
trials were solved for each problem by using different initial populations
in order to avoid negativities of the stochastic natures of evolutionary
search algorithms. Results were analyzed by using the well known
statistical parameters of Mean, Standard Deviation of Mean, Best
Solution and Median Value of the Solutions. Simulation results
show that results found by S-DSA are statistically better than those
produced by using ABC, ACS, BSA, CK, CLPSO, CMAES, DE, E2-
DSA, EPSDE, GSA, JADE, JDE, PSO and SADE.
280 Civicioglu

ACKNOWLEDGMENT

The studies in this paper have been supported within the scope of the
scientific research project of 110Y309 supported by TUBITAK.

REFERENCES

1. Akcakaya, M. and A. Nehorai, “Adaptive MIMO radar design

and detection in compound-gaussian clutter,” IEEE Transactions
on Aerospace and Electronic Systems, Vol. 47, No. 3, 2200–2207,
2011.
2. Heidari, A. A., M. Heyrani, and M. Nakhkash, “A dual-band
circularly polarized stub loaded microstrip patch antenna for GPS
applications,” Progress In Electromagnetics Research, Vol. 92,
195–208, 2009.
3. Bellettini, A. and M. A. Pinto, “Theoretical accuracy of synthetic
aperture sonar micronavigation using a displaced phase-center
antenna,” IEEE J. Oceanic Eng., Vol. 27, No. 4, 780–789, 2002.
4. Hasan, S. M. S. and S. W. Ellingson, “Integration of simple
antennas to multiband receivers using a novel multiplexer design
methodology,” IEEE Transactions on Antennas and Propagation,
Vol. 60, No. 3, 1550–1556, 2012.
5. Zaker, R., C. H. Ghobadi, and J. Nourinia, “A modified
microstrip-fed two-step tapered monopole antenna for UWB and
WLAN applications,” Progress In Electromagnetics Research,
Vol. 77, 137–148, 2007.
6. Weng, W. C., F. Yang, and A. Z. Elsherbeni, “Linear antenna
array synthesis using Taguchi’s method: A novel optimization
technique in electromagnetics,” IEEE Transactions on Antennas
and Propagation, Vol. 55, No. 3, 723–730, 2007.
7. Lee, W. S., K. S. Oh, and J. W. Yu, “A wideband planar
monopole antenna array with circular polarized and band-notched
characteristics,” Progress In Electromagnetics Research, Vol. 128,
381–398, 2012.
8. Gazzah, H. and S. Marcos, “Cramer-Rao bounds for antenna array
design,” IEEE Transactions on Signal Processing, Vol. 54, No. 1,
336–345, 2006.
9. Exposito-Dominguez, G., J. M. Fernandez-Gonzalez,
P. Padilla de la Torre, and M. Sierra-Castaner, “Dual cir-
cular polarized steering antenna for satellite communications in
X band,” Progress In Electromagnetics Research, Vol. 122, 61–76,
2012.
Progress In Electromagnetics Research B, Vol. 54, 2013 281

10. Li, X. and M. Yin, “Hybrid differential evolution with

biogeography-based optimization for design of a reconfigurable
antenna array with discrete phase shifters,” Int. J. Antennas
Prop., Article ID 685629, 685–629, 2011.
11. Krairiksh, M., P. Keowsawat, C. Phongcharoenpanich, and
S. Kosulvit, “Two-probe excited circular ring antenna for MIMO
application,” Progress In Electromagnetics Research, Vol. 97, 417–
431, 2009.
12. Gao, G. P., M. Li, S.-F. Niu, X.-J. Li, B.-N. Li, and J.-S. Zhang,
“Study of a novel wideband circular slot antenna having frequency
band-notched function,” Progress In Electromagnetics Research,
Vol. 96, 141–154, 2009.
13. Jabbar, A. N., A. S. Hasooni, and M. K. Ibrahim, “Maximizing
main-lobe to the side-lobe level for the non-uniform chaos game
planar antenna array using an IWO algorithm,” Int. J. Micro.
Wire. Tech., Vol. 4, No. 6, 635–646, 2012.
14. Khodier, M. M. and C. G. Christodoulou, “Linear array geometry
synthesis with minimum sidelobe level and null control using
particle swarm optimization,” IEEE Transactions on Antennas
and Propagation, Vol. 53, No. 8, 2674–2679, 2005.
15. Gies, D. and Y. Rahmat-Samii, “Particle swarm optimization for
reconfigurable phase-differentiated array design,” Microw. Opt.
Techn. Let., Vol. 38, 168–175, 2003.
16. Li, W. T., X. W. Shi, L. Xu, and Y.-Q. Hei, “Improved GA and
PSO culled hybrid algorithm for antenna array pattern synthesis,”
Progress In Electromagnetics Research, Vol. 80, 461–476, 2008.
17. Mandal, A., H. Zafar, S. Das, and A. V. Vasilakos, “Efficient
circular array synthesis with a memetic differential evolution
algorithm,” Progress In Electromagnetics Research B, Vol. 38,
367–385, 2012.
18. Lu, Z. B., A. Zhang, and X. Y. Hou, “Pattern synthesis
of cylindrical conformal array by the modified particle swarm
optimization algorithm,” Progress In Electromagnetics Research,
Vol. 79, 415–426, 2008.
19. Li, R., L. Xu, X. W. Shi, N. Zhang, and Z.-Q. Lv, “Improved
differential evolution strategy for antenna array pattern synthesis
problems,” Progress In Electromagnetics Research, Vol. 113, 429–
441, 2011.
20. Demarcke, P., H. Rogier, R. Goossens, et al., “Beamforming in the
presence of mutual coupling based on constrained particle swarm
optimization,” IEEE Transactions on Antennas and Propagation,
Vol. 57, No. 6, 1655–1666, 2009.
282 Civicioglu

21. Mohammed, H. and J. I. Ababneh, “Synthesis of aperiodic

linear phased antenna arrays using particle swarm optimization
Bataineh,” Electromagnetics, Vol. 26, 531–541, 2006.
22. Pal, S., A. Basak, and S. Das, “Linear antenna array synthesis
with modified invasive weed optimisation algorithm,” Int. J. Bio-
Insp. Comput., Vol. 3, No. 4, 238–251, 2011.
23. Chan, C. Y. and P. M. Goggans, “Using Bayesian inference for
linear antenna array design,” IEEE Transactions on Antennas and
Propagation, Vol. 59, No. 9, 3211–3217, 2011.
24. Guenes, F. and F. Tokan, “Pareto optimal synthesis of the linear
array geometry for minimum side lobe level and null control during
beam scanning,” International Journal of RF and Microwave
Computer-Aided Engineering, Vol. 20, No. 5, 557–566, 2010.
25. Abdul, R., N. Khairul, M. Abd, F. Mohd, and N. Siew-Chin,
“Nature-inspired cuckoo search algorithm for side lobe suppression
in a symmetric linear antenna array,” Radioengineering, Vol. 21,
865–874, 2012.
26. Zaharis, Z. D., C. Skeberis, and T. D. Xenos, “Improved antenna
array adaptive beamforming with low side lobe level using a
novel adaptive invasive weed optimization method,” Progress In
Electromagnetics Research, Vol. 124, 137–150, 2012.
27. Fan, Z. G., S. Qiao, J. T. Huangfu, and L.-X. Ran, “Signal
descriptions and formulations for long range UHF RFID readers,”
Progress In Electromagnetics Research, Vol. 71, 109–127, 2007.
28. Tarokh, V., H. Jafarkhani, and A. R. Calderbank, “Space-time
block codes from orthogonal designs,” IEEE Transactions on
Information Theory, Vol. 45, No. 5, 1456–1467, 1999.
29. Padros, N., J. I. Ortigosa, J. Baker, et al., “Comparative
study of high-performance GPS receiving antenna designs,” IEEE
Transactions on Antennas and Propagation, Vol. 45, No. 4, 698–
706, 1997.
30. Fan, Z. G., S. Qiao, J. T. Huangfu, and L.-X. Ran, “A
miniaturized printed dipole antenna with V-shaped ground for
2.45 GHz RFID readers,” Progress In Electromagnetics Research,
Vol. 71, 149–158, 2007.
31. Alamouti, S. M., “A simple transmit diversity technique for
wireless communications,” IEEE J. Sel. Area Comm., Vol. 16,
No. 8, 1451–1458, 1998.
32. Tarokh, V., N. Seshadri, and A. R. Calderbank, “Space-time codes
for high data rate wireless communication: Performance criterion
and code construction,” IEEE Transactions on Information
Progress In Electromagnetics Research B, Vol. 54, 2013 283

Theory, Vol. 44, No. 2, 744–765, 1998.

33. Guo, J. L. and J. Y. Li, “Pattern synthesis of conformal array
antenna in the presence of platform using differential evolution
algorithm,” IEEE Transactions on Antennas and Propagation,
Vol. 57, No. 9, 2615–2621, 2009.
34. Li, X. and M. Yin, “Optimal synthesis of linear antenna array
with composite differential evolution algorithm,” Scientia Iranica,
Vol. 19, 1780–1787, 2012.
35. Ho, M. H., C. C. Chiu, and S. H. Liao, “Bit error rate reduction
for circular ultra wideband antenna by dynamic differential evo-
lution,” International Journal of RF and Microwave Computer-
Aided Engineering, Vol. 22, No. 2, 260–271, 2012.
36. Rocca, P., G. Oliveri, and A. Massa, “Differential evolution
as applied to electromagnetics,” IEEE Antennas Propag. Mag.,
Vol. 53, No. 1, 38–49, 2011.
37. Lommi, A., A. Massa, E. Storti, and A. Trucco, “Side lobe
reduction in sparse linear arrays by genetic algorithms,” Microw.
Opt. Techn. Let., Vol. 32, 194–196, 2002.
38. Caorsi, S., A. Lommi, A. Massa, and M. Pastorino, “Peak
sidelobe level reduction with a hybrid approach based on a
Gas and difference sets,” IEEE Transactions on Antennas and
Propagation, Vol. 52, No. 4, 1116–1121, 2004.
39. Benedetti, M., R. Azaro, D. Franceschini, and A. Massa, “PSO-
based real-time control of planar uniform circular arrays,” IEEE
Antennas Wire. Propag. Lett., Vol. 5, 545–548, 2006.
40. Oliveri, G. and A. Massa, “GA-enhanced ADS-based approach for
array thinning,” IET Micro. Ant. Prop., Vol. 5, 305–315, 2011.
41. Oliveri, G., F. Caramanica, and A. Massa, “Hybrid ADS-based
techniques for radio astronomy array design,” IEEE Transactions
on Antennas and Propagation, Vol. 59, No. 6, 1817–1827, 2011.
42. Karaboga, D. and B. Akay, “A comparative study of artificial bee
colony algorithm,” Appl. Math Comput. Vol. 214, 108–132, 2009.
43. Civicioglu, P., “Artificial cooperative search algorithm for
numerical optimization problems,” Inform. Sci., Vol. 229, 58–76,
2013.
44. Civicioglu, P., “Backtracking Search Optimization Algorithm for
numerical optimization problems,” Appl. Math Comput., Vol. 219,
8121–8144, 2013.
45. Civicioglu, P. and E. Besdok, “A conceptual comparison of the
cuckoo search, particle swarm optimization, differential evolution
and artificial bee colony algorithms,” Artif. Intell. Rev., Vol. 39,
284 Civicioglu

315–346, 2013.
46. Liang, J. J., A. K. Qin, P. N. Suganthan, and S. Baskar,
“Comprehensive learning particle swarm optimizer for global
optimization of multimodal functions,” IEEE Transactions on
Evolutionary Computation, Vol. 10, No. 3, 281–295, 2006.
47. Hansen, N., S. P. N. Niederberger, L. Guzzella, and P. Koumout-
sakos, “A method for handling uncertainty in evolutionary opti-
mization with an application to feedback control of combustion,”
IEEE Transactions on Evolutionary Computation, Vol. 13, No. 1,
180–197, 2009.
48. Storn, R. and K. Price, “Differential evolution a simple and
efficient heuristic for global optimization over continuous spaces,”
J. Global Optim., Vol. 11, 341–359, 1997.
49. Mallipeddi, R., P. N. Suganthan, Q. K. Pan, et al., “Differential
evolution algorithm with ensemble of parameters and mutation
strategies,” Appl. Soft Comput., Vol. 11, 1679–1696, 2011.
50. Rashedi, E., H. Nezamabadi-pour, and S. Saryazdi, “GSA: A
gravitational search algorithm, information science,” Inform. Sci.,
Vol. 13, 2232–2248, 2009.
51. Zhang, J. and A. C. Sanderson, “JADE: Adaptive differential
evolution with optional external archive,” IEEE Transactions on
Evolutionary Computation, Vol. 13, No. 5, 945–958, 2009.
52. Brest, J., B. Bos̆kovic, S. Greiner, V. Zumer, and M. S. Maucec,
“Performance comparison of self-adaptive and adaptive differen-
tial evolution algorithms,” Soft Comput., Vol. 11, 617–629, 2007.
53. Civicioglu, P., “Transforming geocentric cartesian coordinates
to geodetic coordinates by using differential search algorithm,”
Comput. Geosci., Vol. 46, 229–247, 2012.
54. Bratton, D. and J. Kennedy, “Defining a standard for particle
swarm optimization,” IEEE Swarm Intelligence Symposium, 120–
127, Honolulu, 2007.
55. Qin, A. K. and P. N. Suganthan, “Self-adaptive differential
evolution algorithm for numerical optimization,” The 2005 IEEE
Congress on Evolutionary Computation, Vol. 2, 1785–1791, 2005.
56. Besdok, E., “A georeferenced-speech recognition based data
collection tool for GIS applications,” Technical Report, 110Y309,
Tubitak, 2013.