technologies

Article
The Optimization Design of a Novel Slotted
Microstrip Patch Antenna with Multi-Bands Using
Adaptive Network-Based Fuzzy Inference System
Mahmood Abbasi Layegh * ID
, Changiz Ghobadi and Javad Nourinia
Faculty of Electrical Engineering, Urmia University, Urmia 57591-57131, Iran; Ch.Ghobadi@urmia.ac.ir (C.G.);
J.Nourinia@urmia.ac.ir (J.N.)
* Correspondence: m.abbasilayegh@urmia.ac.ir; Tel.: +98-914-141-5082

Received: 29 September 2017; Accepted: 22 November 2017; Published: 27 November 2017

Abstract: This paper attempts to apply an adaptive network-based fuzzy inference system (ANFIS)
for analysis of the resonant frequency of a single-layer single-patch microstrip rectangular patch
antenna with two equal size slots which are placed on the patch in the form of parallel to resonance
edges. The resonant frequency is calculated as the position of the slots is shifted from the right
endpoint to the left endpoint on the patch between −4.2 mm ≤ Xslot ≤ 4.2 mm with the steps of
0.1 mm. The designed antenna is proposed for downlink of X band satellite, broadcasting satellite
service, fixed-satellite service uplink, satellite (Earth-to-space), radio navigation, mobile-satellite
(Earth-to-space), and KU band which can be achieved at the resonant frequencies of 7.2 GHz, 12.2 GHz,
14.6 GHz, 17.5 GHz and 19.3 GHz. Next, High Frequency Electromagnetic Field Simulation software
(ANSYS HFSS) results for the prototype microstrip antenna are compared with the values obtained
through ANFIS system. It can be concluded that the adaptive network-based fuzzy inference system
in such designs can be conveniently used due to fuzzy system’s high approximation capability and
much faster convergence rate. The best results for our ANFIS system can be obtained if Gaussian
membership is used which leads to the mean absolute error of 1.4653.

Keywords: microstrip antenna; slots parallel to resonance edges; adaptive network-based fuzzy
Inference system; resonant frequency; artificial neural networks

1. Introduction
Microstrip antennas have gained much attention for telecommunication engineers and researchers
especially in recent years because they have very simple installation. Furthermore, they have many
advantages, such as being cheap, light in terms of weight and easy to work with. Additionally,
when they have slotted structures of various forms, they can enhance the bandwidth to such a degree
that multi bands can be covered simultaneously. In addition, radiation loss in these antennas has
been reported to be much lower [1–5]. Obviously, it is much preferable to take advantage of only one
single multiband antenna instead of several antennas which is able to operate in different frequency
bands. Reportedly, when a slot is cut appropriately on the microstrip patch with the right dimensions,
a multiband slot antenna can be fabricated more efficiently [1,6]. Hereupon, it is attempted to evaluate
the effect of changing the position of the two slots which are located parallel to resonance edges on the
antenna characteristics especially the parameter of resonant frequency. As well as, the feed line of the
antenna is fixed in order to make the analysis of the antenna much simpler. In previous works Artificial
Neural Network (ANNs) have successfully been introduced to compute different parameters of the
rectangular antennas [7–10]. In [11], the frequency in which a rectangular microstrip antenna resonates
can be computed by using connectionist systems namely artificial neural networks. Nevertheless,
almost all the previously published papers focus on estimation of a single resonant frequency [7–11].

Technologies 2017, 5, 75; doi:10.3390/technologies5040075 www.mdpi.com/journal/technologies

Technologies 2017, 5, 75 2 of 12
Technologies 2017, 5, 75 2 of 12
estimation of a single resonant frequency [7–11]. ANNs will perform better than other techniques if
large data sets have to be analyzed. However, in this work, since the number of data points is limited
ANNs
to will perform
84, ANFIS betterapplied
has been than other techniquesmultiple
to simulate if large data sets have
resonant to be analyzed.
frequencies at the However,
same timeinandthis
work, since the number of data points is
fortunately the results are quite satisfactory. limited to 84, ANFIS has been applied to simulate multiple
resonant frequencies at the same time and fortunately the results are quite satisfactory.
2. Resonant Frequency of Rectangular Microstrip Antenna with Slots Parallel to Resonance Edges
2. Resonant Frequency of Rectangular Microstrip Antenna with Slots Parallel to Resonance Edges
A rectangular microstrip antenna is shown in Figure 1. This microstrip antenna has a rectangular
shapeA of the dimensions
rectangular microstripof antenna
L and isWshown
representing
in Figurethe length
1. This and theantenna
microstrip width hasof athe antenna
rectangular
respectively. There is a of
shape of the dimensions conducting surface acting
L and W representing as the and
the length ground planeofwhich
the width is located
the antenna over a
respectively.
substrate
There is awith the thickness
conducting surfaceof acting
h and relative
as the groundpermittivity
plane of
which . Equation
is located(1) indicates
over the frequency
a substrate with the
of the antenna
thickness of h andin which permittivity
relative the antenna ofresonates and it(1)
ε r . Equation can be computed
indicates as:
the frequency f mn of the antenna
mn
in which the antenna resonates and it can be computed as:
c m n 1/2
f mn = c h [(m
2)2 + ( n
)22i]1/2 (1)
f mn = 2√ ε e Le + We (1)
2 ε e Le We
where is the effective relative permittivity for the patch, c is the velocity of electromagnetic waves
where
in r is themeffective
free εspace, relative and
and n integers, permittivity
Le and Wfor thethe
e are patch, c is the
effective velocity of Reportedly,
dimensions. electromagnetic waves of
the width in
freepatch
the space,has m and n integers,
direct effect onand Le and
some We are the
important effective of
parameters dimensions.
the antennaReportedly, the width
such as radiation of the
pattern
patchthe
and hasfrequency
direct effect
in on somethe
which important
antennaparameters
probably of the antenna
resonates. Thesuch as radiation
resonant pattern and
frequency the
of the
mn
frequency in which the antenna probably resonates. The resonant frequency f mn of the antenna at its
antenna at its fundamental TM10 mode can be calculated from Equation (1):
fundamental TM10 mode can be calculated from Equation (1):
c
f10 = c
f 10 = 2 Le √ε e
2Le ε e
Le can be defined as follows:
Le can be defined as follows: Le = L + 2∆L
Le = L + 2∆L
is referred to as the effective relative permittivity and Le is the effective length showing the
field εfringing
e is at the
referred to end of effective
as the the patch. They are
relative indicatingand
permittivity theLeffects
e is the of the non-uniform
effective medium
length showing and
the field
the fringing fields at each end of the patch. Two formulas for Equations (2) and (3)
fringing at the end of the patch. They are indicating the effects of the non-uniform medium and the presented by
Schneider (1969)
fringing fields and Hammerstad
at each (1975)
end of the patch. Twocan be usedfor
formulas to Equations
calculate (2)(W)andand ∆L.
(3) presented by Schneider
(1969) and Hammerstad (1975) can be used to calculate ε e (W) and ∆L.
εr +1 ε r −1
ε e (W ) = + (2)
ε e (W ) =
εr 2
+ 1 2 1 +ε10
+ p
r −h
1 /W( ) (2)
2 2 1 + 10(h/W )
ε e (W ) + 0.300  (W / h ) + 0.264 
= 0.412
ΔL ∆L h  [ε e (W ) + 0.300][(W/h) + 0.264]  (3)
ε[eε e((W
W)) − (W /) h+)0.813
= 0.412h (3)
0.258][(W/h
−0.258 ] 
+ 0.813

Figure 1.
Figure Geometry of
1. Geometry of rectangular
rectangular MSA.
Technologies 2017, 5, 75 3 of 12
Technologies 2017, 5, 75 3 of 12

Figure22illustrates
Figure illustratesinset-fed
inset-fedmicrostrip
microstripantenna
antennawith
withtwo
twoslots
slotswhich
whichare
areplaced
placedparallel
paralleltotothe
the
resonanceedges
resonance edgeson onthe
thepatch.
patch.The
Theantenna
antennaisisfabricated
fabricatedon
onFR-4
FR-4substrate,
substrate,with
withaathickness
thicknessof
of1.6
1.6mm
mm
andpermittivity
and permittivityofofε r = 4.2.
= 4.2.Table
Table 1 gives
1 gives the
the values
values ofof the
the proposed
proposed microstrip
microstrip antenna.
antenna.

Figure2.2.Full
Figure Fullground
groundplane
planemicrostrip
microstrip antenna
antenna without
without slots
slots on on
thethe
leftleft
sideside
andand
withwith
two two
slotsslots on
on the
the right
right side side
beingbeing parallel
parallel to resonance
to resonance edges
edges fabricated
fabricated on FR-4
on FR-4 substrate
substrate withwith a thickness
a thickness of of
1.61.6
mm.mm.

Table1.1.The
Table Thevalues
valuesof
ofthe
theproposed
proposedmicrostrip
microstripantenna
antennawith
withits
itsparameters
parameterswithout
withoutslots.
slots.

Parameter mm Parameter mm
Parameter mm Parameter mm
L1 = L feed 3 LG 15.528
L1 = L feed
L4 39 LG
WG 15.528
18.44
L4 L5 95 WG
W4 18.44
1
L5 L6 55 W4
W1 1
3.2
L6 5 W1 3.2
L7 15.2 -
L7 15.2 -

The first thing to do is to estimate the feeding line. As can be seen from Figure 3, the best feed
line The first thingwhen
is calculated to doL1 is to
= 3estimate
mm. the feeding line. As can be seen from Figure 3, the best feed line
is calculated when L1 = 3 mm.
After estimating the feed line, the length and width of the radiating patch must be examined. L5
and After
L6 areestimating
consideredthe to feed
be equalline,inthe length
this andIt width
design. meansof thehave
they radiating
the samepatch must
value beasexamined.
but the value
L5
of and
L5 isL6changed,
are considered
the valueto beofequal
L6 isinvaried
this design.
as wellIt means
to obtain they have
the the same
optimum valueand
length but width
as the value
of the
of L5 is changed, the value of L6 is varied as well to obtain the optimum
patch. Figure 4 shows S11 curves in dB versus frequency to evaluate and obtain the best results from length and width of the patch.
Figure
the length4 shows S11 curves
and width of theinproposed
dB versus frequency
patch antenna. to evaluate and obtain the best results from the
lengthAs and width ofbefore,
explained the proposed
the two patchslots areantenna.
placed parallel to resonance edges on the patch. In the case
As explained
of having two slots before,
beingthe two slots
parallel are placededges,
to resonance parallel to resonance
they are exactlyedgesplacedoninthethe
patch.
middleIn the case
point of
of
L5having
and L6. two slotsthey
Then, beingareparallel
shifted to fromresonance edges, point
the beginning they are exactly
of L5 and L6 placed
to theinendpoint
the middle point
of the of
patch
L5
in and
order L6.toThen,
find thethey arepoint
best shifted and fromthe the
mostbeginning
appropriatepoint of L5 and
location. TheL6results
to the reveal
endpoint thatofasthe
thepatch
slots
in order to find the best point and the most appropriate location. The
get closer to the right and left sides of the patch (the endpoints), a better response can be expected.results reveal that as the slots
get closer
Figure to thethe
5 shows right and of
values left sides
Xslot of the
when −2 patch
mm ≤ (the
Xslotendpoints),
≤ 2 mm from a which
better response can be is
the best position expected.
selected
Figure 5 shows the values of Xslot when − 2 mm ≤ Xslot
as Xslot = 1 mm. i.e., a distance of 1 mm from the right and left sides of the patch. ≤ 2 mm from which the best position is
selectedAs as canXslot = 1 mm.
be seen fromi.e., a distance
Figure 6, with of changing
1 mm from thelength
the right and leftpatch,
of the sides of
thethe patch. frequency
resonant
As imperceptibly
varies can be seen from butFigure 6, with changing
the bandwidth the length matching
and impedance of the patch, varythetangibly.
resonant Accordingly,
frequency varies the
imperceptibly but the bandwidth and impedance matching vary tangibly.
length of the feed line is selected longer owing to the existence of more resonant frequencies, high Accordingly, the length of
the feed linematching
impedance is selected(little
longer owing
return to the
loss) andexistence of more resonant
more bandwidth. Both L4 frequencies,
= 9 mm and high 10 mm, impedance
and L5 =
matching (little return loss) and more bandwidth. Both L4 =
5 mm and 6 mm are appropriate for the length and width of the patch because they have good 9 mm and 10 mm, and L5 = 5 mm and
6bandwidths
mm are appropriate for the resonant
and acceptable length and width of the
frequencies. In patch because
this work, they have
we selected L5good
to bebandwidths
4 mm and L4 and to
acceptable
be 9 mm. Inresonantthe 1 mm frequencies.
distance leftIntothis the work,
ends ofwe theselected
patch the L5twoto be 4 mm
slots and L4and
are placed to be 9 mm.
after In the
estimation
1of
mmthedistance
length ofleft thetoslots,
the ends of the
the gain ofpatch the twoisslots
the antenna are placed
calculated and afterlengths
for different estimation
and of the length
widths of the
slots. As the width of the slots increases, the bandwidth and impedance matching decrease. Figure 6
shows the gain of the proposed antenna versus frequency based on changes in the width and the
Technologies 2017, 5, 75 4 of 12

of the slots, the gain of the antenna is calculated for different lengths and widths of the slots. As the
width of the slots increases, the bandwidth and impedance matching decrease. Figure 6 shows the gain
Technologies 2017, 5, 75 4 of 12
of theTechnologies
proposed antenna
2017, 5, 75 versus frequency based on changes in the width and the length of4 the of 12 slots.

L8 and W8 of
length represent
slots. the
the slots. length
L8 and
and and the width
W8 represent
represent of the
the length
length andslots which
the width
width of are placed
the slots
slots on the
which patch on
are placed
placed in the
parallel
length of the L8 W8 the and the of the which are on the
form patch
to resonance
in paralleledges
parallel form on
form to the patch
to resonance and
resonance edges the
edges on two
on the slots
the patch are
patch and shifted
and the
the two to the
two slots right
slots are and
are shifted left
shifted to
to theside during
the right and the
right and
patch in
wholeleft
procedure.
left side during
side during the the whole
whole procedure.
procedure.

0
0

-5
-5

-10
-10

-15
-15

-20 L1=3mm
-20 L1=3mm
L1=4mm
L1=4mm
L1=5mm
L1=5mm
L1=6mm
-25 L1=6mm
-25

-30
-30 0 2 4 6 8 10 12 14 16 18 20
0 2 4 6 8 10 12 14 16 18 20
Frequency (GHz)
Frequency (GHz)

Figure 3. Estimation
Figure 3. Estimationofofthe
the feed line for
feed line forthe
theproposed
proposed antenna.
antenna.
Figure 3. Estimation of the feed line for the proposed antenna.

Figure 4. S11 curves for calculation of the best dimensions of the proposed patch antenna.
Figure 4. S11 curves for calculation of the best dimensions of the proposed patch antenna.
Figure 4. S11 curves for calculation of the best dimensions of the proposed patch antenna.
Technologies 2017, 5, 75 5 of 12
Technologies 2017, 5, 75 5 of 12

Technologies 2017, 5, 75 5 of 12

Figure 5. Finding the best coordination for Xslot on the patch with two slots being parallel to resonance edges.
Figure 5. Finding the best coordination for Xslot on the patch with two slots being parallel to
Figure 5. Finding
resonance the best coordination for Xslot on the patch with two slots being parallel to resonance edges.
edges.
Gain(dB)
Gain(dB)

Figure 6. The gain of the proposed slotted-patch antenna as the length and the width of the slots are changed.
Figure
Figure 6. The
6. The gaingain ofproposed
of the the proposed slotted-patch
slotted-patch antennaantenna as theand
as the length length and the
the width width
of the slotsof
arethe slots
changed.
are changed.
Technologies 2017, 5, 75 6 of 12

Technologies 2017, 5, 75 6 of 12

Table 2 gives the values of resonant frequency, bandwidth and return loss for the proposed
Table 2 gives the values of resonant frequency, bandwidth and return loss for the proposed
antenna with parallel slots to resonance edges when L8 = 7.5 mm, W8 = 0.25 mm. Resonant frequency
antenna with parallel slots to resonance edges when L8 = 7.5 mm, W8 = 0.25 mm. Resonant frequency
and bandwidth are measured in GHz and return loss and gain are measured in dB. Figures 7 and 8
and bandwidth are measured in GHz and return loss and gain are measured in dB. Figures 7 and 8
show S11 and gain of the proposed antenna without slots and in the presence of slots respectively.
show S11 and gain of the proposed antenna without slots and in the presence of slots respectively.
Table 2. Different parameters of the proposed antenna.
Table 2. Different parameters of the proposed antenna.

Microstrip Patch Antenna with Slots

Slots (L8
(L8 == 7.5 mm, W8 = 0.25 mm)
Resonant
Resonant Frequency
Frequency (GHz)(GHz) 7.4 7.4 12.2
12.2 14.6
14.6 17.517.5 19.3 19.3
Bandwidth
Bandwidth 7–7.8 11.9–13.2
7–7.8 11.9–13.2 && 13.2–15.5
13.2–15.5 16.9–18
16.9–18 18–18.9
18–18.918.9–20
18.9–20
Return
Return loss (dB)
loss (dB) −26−26 −−33
33 &&− −20
20 −28
−28 −18−18 −28 −28
GainGain
(dB) (dB) 4 4 2.8
2.8 &
& 5.3
5.3 5 5 7.757.75 10.8 10.8

In
In Figures
Figures 77 and
and 8,
8, S11
S11 curves
curves and
and the
the gain
gain ofof the
the proposed
proposed antenna
antenna are
are illustrated
illustrated in
in the
the presence
presence
of slots and without them respectively.
of slots and without them respectively.
A comparisonbetween
A comparison between resonant
resonant frequency,
frequency, maximummaximum gain
gain and and dimensions
dimensions between between the
the proposed
proposed
antenna and antenna
some and
othersome other
recent recent
works worksin
is shown is Table
shown 3. in Table 3.
Figure 9 shows the radiation patterns of the proposed
Figure 9 shows the radiation patterns of the proposed microstrip microstrip patch
patch antenna
antenna in
in the presence
the presence
of slots at resonant frequencies of 7.3 GHz, 12.5 GHz, 17.4 GHz and 19.2 GHz
of slots at resonant frequencies of 7.3 GHz, 12.5 GHz, 17.4 GHz and 19.2 GHz for E- and H-plane for E- and H-plane
including
including both co-polarization and
both co-polarization and cross-polarization.
cross-polarization. ItIt can can be
be seen
seen that
that the radiation patterns
the radiation patterns are
are
nearly omnidirectional in most of the frequencies mentioned
nearly omnidirectional in most of the frequencies mentioned above. above.

Figure 7. S11 parameters for the patch antenna without and with slots.
Technologies 2017,
Technologies 5, 75
2017, 7 of
12 12
Technologies 2017, 5,5,75
75 77ofof12

Figure 8.8.The
Figure Thegain
gainof
ofthe patch
thepatch antenna
patchantenna without
antennawithout andwith
withoutand
and withslots.
slots.
Figure 8. The gain of the with slots.

(a)
(a) (b)
(b)

(c)
(c) (d)
(d)
Figure9.9.Radiation
Figure Radiationpatterns
patternsof
ofthe
theproposed
proposedmicrostrip
microstripantenna
antennawith
withslots
slotsparallel
paralleltotoresonance
resonanceedges
edges
Figure 9. Radiation patterns of the proposed microstrip antenna with slots parallel to resonance edges
atat(a)
(a)7.4GHz
7.4GHz(b)
(b)12.5GHz
12.5GHz(c)
(c)17.4GHz
17.4GHz(d)
(d)19.2GHz.
19.2GHz.
at (a) 7.4 GHz (b) 12.5 GHz (c) 17.4 GHz (d) 19.2 GHz.
Technologies 2017, 5, 75 8 of 12
Technologies 2017, 5, 75 8 of 12

Table3.3.Comparison
Table Comparisonbetween
betweenthe
theproposed
proposedantenna
antennaand
andother
othersimilar
similarmicrostrip
microstripantennas.
antennas.

Reference Resonant Freq Max Gain dBi Size (mm3)

Reference Resonant Freq Max Gain dBi Size (mm3 )
[12] 3.39/4.29/5.46/5.77 8 25 × 25 × 1.6
[12] [13] 3.39/4.29/5.46/5.77
2.6/5.45/0.08 85 25××1625× ×
29 1.61.6
[13] [14] 2.6/5.45/0.08
1.6/1.9/3.8 5
6.58 29××7116× ×
52 1.61.6
[14] [15] 1.6/1.9/3.8
3.35/3.70/5.20/5.80 6.58
9 52 ××50
50 71××5 1.6
[15]
The Proposed Antenna 3.35/3.70/5.20/5.80
7.4/12.2/14.6/17.5/19.3 9
10.8 50 × 50 ×× 51.6
15.528 × 18.44
The Proposed Antenna 7.4/12.2/14.6/17.5/19.3 10.8 15.528 × 18.44 × 1.6

3. Adaptive Neuro-Fuzzy Inference System (ANFIS)

3. Adaptive Neuro-Fuzzy Inference System (ANFIS)
Basically a fuzzy inference system is composed of five functional blocks (see Figure10). The main
Basically aoffuzzy
contribution inferencecan
this system system is composed
be attributed of five functional
to computation blocksdata
of some (seeboth
Figurein10). The main
granular and
contribution of this system can be attributed to computation of some
imprecise form. Here, membership functions are employed to compute numerically small and big data both in granular and
imprecise form. inference
datasets. Fuzzy Here, membership
system (FIS)functions
originatesare employed
form to compute
the principles numerically
pertaining to fuzzysmall and
sets, fuzzy
big datasets. Fuzzy inference system (FIS) originates form the principles
if-then rules and fuzzy reasoning. Fuzzy inference system can be strongly used to classify data ifpertaining to fuzzy sets,
fuzzy if-then
necessary. rules andthe
Whenever fuzzy reasoning.
inputs Fuzzyofinference
and outputs system are
a fuzzy system can determined
be strongly by used to classify
variables, the
data
following steps ought to be done successively. The first step, which is called fuzzification,variables,
if necessary. Whenever the inputs and outputs of a fuzzy system are determined by indicates
the followingvariables
expressing steps ought to be
in the done
form successively.
of fuzzy and then Thecalculating
first step, which is called fuzzification,
their dependence on the fuzzyindicates
set. As
expressing variables in the form of fuzzy and then calculating their dependence
the membership function can take numerous shapes, the ones in a smooth shape can be more efficient. on the fuzzy set. As the
membership function can take numerous shapes, the ones in a smooth shape
After that, the level of statement is appraised and some algebraic operators are used to approximate can be more efficient.
After that, the task.
classification level Activation
of statement is appraised
is the next step and sometoalgebraic
in order calculateoperators are used
the activations to approximate
of the rules applied
classification
to the system. Finally all the output activations are connected to each other and thisrules
task. Activation is the next step in order to calculate the activations of the step applied
is called
toaccumulation
the system. [16–19].
Finally all the output activations are connected to each other and this step is called
accumulation [16–19].

Figure 10.AAfuzzy
Figure10. fuzzyinference
inferencesystem.
system.

4.4.How
Howto
toTake
TakeAdvantage
Advantageofofthe
theANFIS
ANFISininCalculating
Calculatingthe
theResonant
ResonantFrequency
FrequencyofofRectangular
Rectangular
Antennas
AntennasOperating
OperatingatatMulti
MultiBands
Bands
Figure
Figure 11 belowillustrates
11 below illustratesS11S11 curves
curves for for the whole
the whole possible
possible situations.
situations. Since theSince
antennathe resonates
antenna
resonates
in more thanin more
onethan one frequency,
frequency,it it is necessary
is necessary that an appropriate
that an appropriate algorithmalgorithm
to detecttothe detect
resonantthe
resonant
frequencies be selected. Experimentally, artificial neural networks are unable to diagnose the
frequencies be selected. Experimentally, artificial neural networks are unable to diagnose the
frequencies
frequenciesbecause
becausein inmany
many cases
cases the
the resonant
resonant frequencies are very
frequencies are very similar
similar or
or close
closeto toeach
eachother.
other.
Hence,
Hence,thetheANFIS
ANFISwas wasutilized
utilizedin inorder
orderto toanalyze
analyzemultiple
multipleresonant
resonantfrequencies
frequenciesof ofthe
theproposed
proposed
rectangular
rectangular patch antenna having slots being parallel to resonance edges on it. For thethe
patch antenna having slots being parallel to resonance edges on it. For ANFIS,
ANFIS, the
the inputs were the position of slots and the output was the measured
inputs were the position of slots and the output was the measured resonant frequency resonant frequency f
. The.
ME
The network was trained with 84 samples. The database was divided fortuitously
network was trained with 84 samples. The database was divided fortuitously into two separate into two separate
groups
groupsfor fortraining
trainingandandtesting
testingasaswellwellasasvalidation.
validation. TheThe random
random divisions
divisions were
wereperformed
performed by by
dividing the whole data into two equal datasets. But in this work, the final division
dividing the whole data into two equal datasets. But in this work, the final division for training, for training, testing
and validation
testing was selected
and validation was as 80:10:10.
selected The training
as 80:10:10. Thewas carried
training was out in 220out
carried epochs selected
in 220 epochsbased
selectedon
MATLB default. To implement and test the proposed architecture MATLAB
based on MATLB default. To implement and test the proposed architecture MATLAB Fuzzy Logic Fuzzy Logic Toolbox (FLT)
Toolbox (FLT) from MathWorks was selected. The ANFIS is able to do simulation and analyses of
existing relationship between the input and output data through a hybrid learning. The next step is
Technologies 2017, 5, 75 9 of 12

from MathWorks
Technologies 2017, 5, 75was selected. The ANFIS is able to do simulation and analyses of existing relationship 9 of 12
between the input and output data through a hybrid learning. The next step is to identify optimal
to identify optimal
parameters parameters
of the presumed FIS. of the presumed
Commonly, FIS. Commonly,
measurement measurement
and simulation and simulation
are considered as two waysare
considered
of generating asdata
twowhich
ways can
of generating
be used in data whichHere,
antennas. can the
be used
datasetin was
antennas.
obtainedHere, the dataset
through was
simulation
obtained
results through
using HFSSsimulation
because itresults using HFSS
is practically because
impossible it is practically
to design impossible
and fabricate the wholeto design and
antennas
fabricate the whole antennas mentioned earlier. Table 4 provides the results of
mentioned earlier. Table 4 provides the results of the ANFIS tested with 8 different membership the ANFIS tested with
8 different(trimf,
functions membership functionspimf,
trapmf, psigmf, (trimf, trapmf,
dsigmf, psigmf,
gaussmf, pimf, dsigmf,
gauss2mf, gaussmf,Ingauss2mf,
and gbellmf). our proposedand
gbellmf). In our proposed architecture the best result belonged to Gaussmf with
architecture the best result belonged to Gaussmf with mean absolute error of 1.4653. Some data weremean absolute error
of 1.4653.
left Some data
for validation. were
Here, the left for validation.
validated data wereHere, the validated
not included data were
in the training andnot included
testing phasesin and
the
training
the lowestand testing phases
computational andwas
time thealmost
lowest106.578
computational
s. Also, time
Tablewas almost
4 gives 106.578
detailed s. Also, Table
information about4
gives detailed
different information
membership about different
functions membership
applied here as well asfunctions applied
quantitative here of
values as the
wellmargin
as quantitative
of error
values ofthe
between theestimated
margin of errorand
values between the estimated
the experimental values
data (Rootandmean thesquare
experimental
error, meandata (Root mean
absolute error
square
and error, mean
coefficient absolute error and coefficient of determination).
of determination).

Figure 11.
Figure 11. S11
S11 obtained
obtained from
from the
the whole
whole possible
possible cases
cases simulated
simulated by
by High
High Frequency
Frequency Electromagnetic
Electromagnetic
Field Simulation software (HFSS).
Field Simulation software (HFSS).

Table 4.
Table 4. Results
Results of
of the
the adaptive
adaptive network-based
network-based fuzzy
fuzzy inference
inference system
system (ANFIS)
(ANFIS) tested
tested with
with different
different
membership functions.
membership functions.
MFs
Stage Trapmf Trimf Dsigmf MFs Psigmf Pimf Gbellmf Gaussmf Gauss2mf
Stage MAE Trapmf
1.5722 Trimf
1.536 Dsigmf
1.5386 Psigmf 1.5317
1.5385 Pimf 1.483
Gbellmf 1.4653
Gaussmf 1.5012
Gauss2mf
training MAE
RMSE 1.5722
2.6385 2.6797
1.536 2.6823
1.5386 2.616
1.5385 2.6066
1.5317 2.3713
1.483 2.515
1.4653 2.341
1.5012
training CC
RMSE 0.7982
2.6385 0.8213
2.6797 0.8511
2.6823 0.8552
2.616 0.8695
2.6066 0.8997
2.3713 0.9187
2.515 0.9234
2.341
CC:CC 0.7982
correlation 0.8213RMSE:
coefficient; 0.8511 0.8552
root mean 0.8695MAE:0.8997
square error; 0.9187
mean absolute error. 0.9234
CC: correlation coefficient; RMSE: root mean square error; MAE: mean absolute error.
As can be seen from Table 4, this algorithm can determine the resonant frequencies with high
precision.
As canFigure 12 shows
be seen the flow
from Table chart
4, this of ANFIS
algorithm training
can process
determine the step by step.
resonant frequencies with high
precision. Figure 12 shows the flow chart of ANFIS training process step by step. Fuzzy output are
In Figure 13, the results of the approximation of both training data and
depicted.
In Figure 13, the results of the approximation of both training data and Fuzzy output are depicted.
Likewise, Figure
Likewise, Figure 14 illustrates the
14 illustrates same outputs
the same outputs approximated
approximated by by line
line graphs
graphs and the error
and the error
function obtained through curve fitting indicating the difference between original and
function obtained through curve fitting indicating the difference between original and ANFIS outputs. ANFIS
outputs.
Technologies 2017, 5, 75 10 of 12
Technologies 2017,
Technologies 2017, 5,
5, 75
75 10 of
10 of 12
12
Technologies 2017, 5, 75 10 of 12

Figure 12.
Figure 12. ANFIS
ANFIS training
training process
process flowchart.
flowchart.
Figure
Figure ANFIStraining
12.ANFIS
12. trainingprocess
processflowchart.
flowchart.

Figure 13.
Figure 13. Approximation
Approximation of
of training
training data
data and
and fuzzy
fuzzy output
fuzzy output (O
output (O and * are
and *
(O and are symbols
are symbols representing
symbols representing the
representing the
Figure 13. Approximation
trained data
trained data and
and FIS of
FIS output training data
output respectively).
respectively). and fuzzy output (O and * are symbols representing the
trained data and FIS output respectively).
trained data and FIS output respectively).

Figure 14.
Figure 14. Original and
14. Original
Original and fuzzy
fuzzy output with
output with error
error function
with error function obtained
obtained from
from curve
curve fitting
fitting method.
method.
Figure and fuzzy output function obtained from curve fitting method.
Figure 14. Original and fuzzy output with error function obtained from curve fitting method.
5. Conclusions
5. Conclusions
5. Conclusions
Microstrip patch
Microstrip patch antennas
antennas are
are versatile
versatile structures
structures which
which can
can be
be modified
modified by
by adding
adding simple
simple slots
slots
Microstrip
either “parallelpatch
to antennasedges”
radiation are versatile
or structures
“parallel to which can be
resonance modified
edges” in by adding
the design simple slots
structure to
either “parallel to radiation edges” or “parallel to resonance edges” in the design structure to
either “parallel to radiation edges” or “parallel to resonance edges” in the design structure to
Technologies 2017, 5, 75 11 of 12

5. Conclusions
Microstrip patch antennas are versatile structures which can be modified by adding simple
slots either “parallel to radiation edges” or “parallel to resonance edges” in the design structure to
overcome selected limitations of conventional patch antennas. The antenna can provide improved
bandwidth enhancement, under certain conditions, while maintaining many of the desirable features
of conventional patches. However, it is difficult to determine or predict resonant frequencies and
bandwidths especially when the changes in terms of position of slots and the widths of slots are
relatively small. As a result, machine algorithms can be applied to interpret the relationship between
inputs and outputs of the system. Depending on the problem, the appropriate algorithms such as
ANNs can be selected to estimate the outputs. This paper concludes that the results obtained using
adaptive network-based fuzzy inference system (ANFIS) technique are quite satisfactory and far
outweigh some algorithms such as ANNs because the data set is small and its complexity can be
analyzed by ANFIS more successfully.

Author Contributions: All authors contributed equally in the design and performance of the experiments,
the analysis of the data and the writing and revision of the manuscript.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Garg, R.; Bhartia, P.; Bahl, I.; Ittipiboon, A. Microstrip Antenna Design Handbook; Artech House: Norwood,
MA, USA, 2001.
2. Wong, K.L. Compact and Broadband Microstrip Antennas; J. Wiley and Sons: New York, NY, USA, 2002.
3. Chen, W.S. A novel broadband design of a printed rectangular slot antenna for wireless applications. Microw. J.
2006, 49, 122. [CrossRef]
4. Azadegan, R.; Sarabandi, K. Bandwidth Enhancement of Miniaturized Slot Antennas Using Folded,
Complementary, and Self-Complementary Realizations. IEEE Trans. Antennas Propag. 2007, 55, 2435–2444.
[CrossRef]
5. Zhang, L.; Jiao, Y.C.; Zhao, G.; Song, Y.; Wang, X.M.; Zhang, F.-S. A novel CPW-FED monopole antenna for
multiband operation. J. Electromagn. Waves Appl. 2008, 22, 741–747. [CrossRef]
6. Sze, J.Y.; Wong, K.L. Bandwidth enhancement of a microstrip line-fed printed wide-slot antenna. IEEE Trans.
Antennas Propag. 2001, 49, 1020–1024. [CrossRef]
7. Guney, K.; Gultekin, S.S. Artificial Neural Networks for Resonant Frequency Calculation of Rectangular
Microstrip Antennas with Thin and Thick Substrates. Int. J. Infrared Millim. Waves 2004, 25, 1383. [CrossRef]
8. Gultekin, S.S.; Guney, K.; Sagiroglu, S. Neural networks for the calculation of bandwidth of rectangular
microstrip antennas. Appl. Comput. Electromagn. Soc. (ACES) J. 2003, 18, 46–56.
9. Karaboga, D.; Guney, K.; Sagiroglu, S.; Erler, M. Neural computation of resonant frequency of electrically
thin and thick rectangular microstrip antennas. IEE Proc. Microw. Antennas Propagat. 1999, 146, 155–159.
[CrossRef]
10. Guney, K.; Erler, M.; Sagiroglu, S. Artificial neural networks for the resonant resistance calculation of
electrically thin and thick rectangular microstrip antennas. Electromagnetics 2000, 20, 387–400. [CrossRef]
11. Guney, K.; Sagiroglu, S.; Erler, M. Comparison of neural networks for resonant frequency computation of
electrically thin and thick rectangular microstrip antennas. J. Electromagn. Waves Appl. 2012, 15, 1121–1145.
[CrossRef]
12. Amanpreet, K.; Gurmohan, S.; Manjit, K. Miniaturized Multiband Slotted Microstrip Antenna for Wireless
Applications. Wirel. Pers. Commun. 2017. [CrossRef]
13. Jiachen, Y.; Huanling, W.; Zhihan, L.; Huihui, W. Design of Miniaturized Dual-Band Microstrip Antenna for
WLAN Application. Sensors 2016, 16, 983. [CrossRef]
14. Naresh, K.; Darimireddy, R.; RamanReddy, A.; Mallikarjuna, P. Design of triple-layer double U-slot patch
antenna for wireless applications. J. Appl. Res. Technol. 2015, 13, 526–534.
15. Liu, S.; Qi, S.S.; Wu, W.; Fang, D.G. Single-Layer Single-Patch Four-Band Asymmetrical U-Slot Patch Antenna.
IEEE Trans. Antennas Propag. 2014, 62, 4895–4899. [CrossRef]
Technologies 2017, 5, 75 12 of 12

16. Jang, J.S.R. Anfis: Adaptive-network-based fuzzy inference system. IEEE Trans. Syst. Man Cyber. 1993, 23,
665–685. [CrossRef]
17. Jang, J.S.R.; Sun, C.T.; Mizutani, E. Neuro-Fuzzy and Soft Computing: A Computational Approach to Learning and
Machine Intelligence; Prentice-Hall: Upper Saddle River, NJ, USA, 1997.
18. Bassam, A.; May Tzuc, O.; Escalante Soberanis, M. Temperature Estimation for Photovoltaic Array Using an
Adaptive Neuro Fuzzy Inference System. Sustainability 2017, 9, 1399. [CrossRef]
19. Mashaly, A.F.; Alazba, A.A. Application of adaptive neuro-fuzzy inference system (ANFIS) for modeling
solar still productivity. J. Water Supply Res. Technol. AQUA 2017, 66, 367–380. [CrossRef]

© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).