CHAPTER 3

DESIGN AND ANALYSIS OF FRACTAL ANTENNAS

3.1 FRACTAL GEOMETRY AND FRACTAL ANTENNAS

The term „fractal‟ is used to represent a class of geometry with very unique properties.
This term was originally employed by Mandelbrot to describe recursively generated
self-similar geometric shapes (Werner et al., 1999). The shape of fractal structures is
similar at different scales i.e. the subparts of the overall geometry are similar to the
overall global shape. Due to this property, the fractal shapes are called self-similar
shapes (Puente et al., 1996). Another important property of the fractal geometries is
the space filling property that means the possibility to enclose an infinitely long curve
in a finite area (Baliarda et al., 2000) e.g. a surface with very large perimeter can be
enclosed in fractal loops.

Another important property of fractal shapes is the fractional dimensions i.e. the
dimensions of the fractal geometry are not whole numbers but fractional numbers
(Werner et al., 1999). The self-similarity property of fractals state that the fractional
dimension (Dim) of a fractal geometry is calculated by the equation given below
(Rahim et al., 2005):

…(3.1)

where M represents the number of similar copies of fractal geometry and s is the
inverse of the scaling down ratio of fractal geometry which is calculated as 1/sd
(Huang et al., 2005; Vinoy et al., 2003).

Many fractal shapes are generated by applying repetitive procedures such as multiple
reduction copy machine algorithm (Puente-Baliarda et al., 1998) or by using an IFS
(Werner and Ganguly, 2003). In these repetitive algorithms, a starting base shape

67
named as ‘Initiator’ is selected and another shape named as ‘Generator’ is copied a
number of times at various locations, scaling ratios and orientations, to achieve the
end geometry (Puente-Baliarda et al., 1998). Ideally fractal geometries are designed
by iterating an infinite number of times, however, practically, few starting geometries
(also called as pre-fractals) are considered (Gianvittorio and Rahmat-Samii, 2002).
The repetitive designing technique is shown in Figs. 3.1 and 3.2 for two different
geometries. Fig. 3.1 shows the repetitive procedure for the implementation of the
Sierpinski gasket geometry. The initiator triangle is replicated at different scales and
shifted to form the various iterations. Each triangle is replaced with the three small
triangles arranged as shown in Fig. 3.1. The repetitive-designing procedure for
Minkowski island fractal shown in Fig. 3.2 involves the replacement of each straight
line of the structure by the generator shape.

Zeroth Iteration First Iteration Second Iteration Third Iteration

Fig. 3.1 Development of Sierpinski Gasket Geometry (Puente-Baliarda et al.,
1998)

Fig. 3.2 Development of Minkowski Island Fractal Geometry (Gianvittorio and

Rahmat-Samii, 2002)

68
Fractal geometries have been used in antenna structures to extend antenna design
concept beyond Euclidean geometry (Vinoy et al., 2003). The fractal antennas are
antennas in which the fractal geometry concepts are used to design the radiating
shapes (Panoutsopoulos, 2003). The properties of fractal geometries result in number
of advantages in antenna design. The space filling property of fractal shapes leads to
fitting of very long electrical lengths into compact physical spaces. This results in
miniaturization of antennas (Gianvittorio and Rahmat-Samii, 2002). The self-similar
property of fractals means that various segments of the structure are like the other
parts but at different scales. This leads to multiband behaviour of antennas (Tiehong
and Zheng, 2003). The fractional dimensions of the fractals are considered as an
important mathematical property (Yusof et al., 2006). In self similar shapes, the
variation of the scale ratio results in changed fractal dimensions of the geometry
(Huang et al., 2005). There exist a direct relationship between antenna characteristic
and variation of dimensions (Vinoy et al., 2003). This property is used for tuning of
fractal antennas for desired frequencies. The other advantages of fractal antennas
include enhanced bandwidth, improved gain and directivity, and better efficiency
(Song et al., 2006; Azaro et al., 2006).

3.2 IMPORTANT FRACTAL ANTENNAS AND THEIR

FEATURES

Some of the important fractal antennas are introduced in this section. The
electromagnetic behaviour and important features are also described to highlight the
potential/advantages of the fractal antennas.

3.2.1 Sierpinski Gasket Monopole Fractal (SGMF) Antenna

The most popular fractal antenna is an antenna based on a fractal shape named as
Sierpinski triangle. This fractal shape, also called the Sierpinski gasket, was proposed
in 1915 by a Polish mathematician named Waclaw Sierpinski. This fractal shape is
used to design a monopole antenna by Puente-Baliarda et al. (1998) and the fractal

69
antenna is named as the SGMF antenna. The first four iterations of Sierpinski gasket
are shown in Fig. 3.1. The starting shape i.e. the zeroth iteration is an equilateral
triangle from which the first iteration gasket is constructed by subtracting central
inverted triangle. After the subtraction, three equal sized triangles remain on the
structure each one being half of the size of the original. So the first iteration shape is a
self similar structure and each one of its three main parts has exactly same shape as
that of whole object but reduced by a factor of two. Each triangle of the first iteration
geometry is replaced by three small size rectangles to obtain the second iteration and
similarly this iterative procedure is carried out further to obtain the next iterations
(Puente-Baliarda et al., 1998).

The SGMF antenna, as shown in Fig. 3.3, is constructed by printing the Sierpinski
gasket on a substrate and then mounted over a conducting surface perpendicular to the
plane of printed substrate which acts as ground plane. This arrangement is similar to
a monopole feeding structure. The geometry is excited by coaxial signal source from
the reverse side of the ground plane.

Fig. 3.3 SGMF Antenna (Puente-Baliarda et al., 1998)

70
The SGMF antenna is a self similar structure and this property is also observed in its
S11 results and the radiation patterns (Puente-Baliarda et al., 1998). This antenna has a
multiband performance and the number of bands depends on the number of iterations
n. The zeroth iteration has single fundamental resonant frequency. The first iteration
has two resonant frequencies; second iteration has three resonant frequencies and so
on. Also the spacing between the bands is equal to the scale factor existing among
similar structures on the fractal shape i.e. a log-periodic factor of two.

3.2.2 Sierpinski Carpet Fractal Antenna

The fractal antennas are also designed using the rectangular Sierpinski shape, also
known as the Sierpinski carpet. Fig. 3.4 shows the construction of Sierpinski carpet
geometry. The zeroth iteration i.e. the base shape is a square. The base shape is
divided into nine equal squares from which the central square is removed to obtain the
first iteration. In first iteration geometry, eight squares are left and every square is
again partitioned into nine equal squares and the central square is erased to design the
second iteration. This procedure is repeated to obtain next iterations.

Zeroth Iteration First Iteration Second Iteration

Fig. 3.4 Development of Sierpinski Carpet Fractal Antenna (Rahim et al., 2005)

The Sierpinski carpet geometry is printed on grounded substrate and fed with a
transmission line feeding to implement a fractal antenna by Wong et al. (2001). The

71
simulation and experimental results depict that this antenna has a multiband
performance. The monopole configuration of this geometry results in a multiband and
wideband performance (Rahim et al., 2005).

3.2.3 Koch Curve Antenna

The Koch monopole antenna is also an effective example to demonstrate that fractals
can improve some features of common Euclidean shapes. The Koch curve is
constructed as shown in Fig. 3.5. The zeroth iteration i.e. the first element is a straight
segment and the first iteration is obtained by applying the four similarity
transformations known as Affine transformation to zeroth iteration (Baliarda et al.,
2000). The next iterations are obtained by applying the same Affine transformation
iteratively. The length of the curve increases by a factor of 4/3 in each iteration, but
maintains exactly the same height (Puente et al., 1998).

Zeroth Iteration First Iteration Second Iteration Third Iteration

Fig. 3.5 Development of Koch Curve Fractal Geometry (Baliarda et al., 2000)

72
A Koch curve monopole antenna of height 6 cm is analyzed by Puente et al. (1998)
and Baliarda et al. (2000). It is found that the Q of this antenna reaches the
fundamental limit for small antennas when the number of iterations is increased.
However, measurement of input return loss of Koch curve monopole over a wide
frequency range suggests that this fractal antenna has harmonic behavior rather than
multiband behavior. This antenna is very useful in applications where the reduction of
the antenna size is an ultimate goal.

3.2.4 Hilbert Curve based Fractal Antennas

Hilbert curve based fractal antennas are examples of space-filling fractal geometries.
The generation of the Hilbert curve is shown in Fig. 3.6 which depicts that first
iteration is implemented by arranging the four replicas of the zeroth iteration having
different directions and joining these replicas using additional line segments. This is
repeated to obtain the second iteration and so on. For higher iterations, these
additional line segments has very little contribution as compared to the complete
length of the structure so these additional line segments can be ignored and it results
in self-similar structure. Also, the comparison of the starting iterations of this fractal
shape, shown in Fig. 3.6, depicts the space-filling property (Vinoy et al., 2001).

Zeroth Iteration Second Iteration Third Iteration

Fig. 3.6 Development of Hilbert Curve Antenna (Vinoy et al., 2001)

73
A Hilbert curve based fractal antenna analyzed by Vinoy et al. (2001) shows a
multiband behaviour due to its self-similar shape. For the same overall dimensions of
the antenna, an increase in the iteration number results in reduction of the lowest
resonant frequency. This leads to miniaturization in size and it is due to the space-
filling shape of the antenna. Vinoy et al. (2001) described a Hilbert curve based
antenna of size less than λ/10 and has reported that the behavior of their antenna is
comparable to a dipole of resonant length nearly equal to λ/2. A size reduction of the
order of 78% is proposed for Hilbert curve fractal antenna designed for RFID
application by (Murad et al., 2006).

3.2.5 Hexagonal Fractal Antenna

Another example of the fractal geometry is a hexagonal fractal antenna as shown in

Fig. 3.7. The base shape is hexagon and the first iteration shape is designed by
grouping six hexagons of size one third of original size. Similarly, the further
iterations are obtained by applying this procedure. A dipole antenna, as shown in Fig.
3.7, is implemented using these hexagonal fractal shapes and the corner feeding is
used for excitation (Tang and Wahid, 2004).

Zeroth Iteration First Iteration Second Iteration

Fig. 3.7 Development of Hexagonal Fractal Antenna (Tang and Wahid, 2004)

74
The reflection coefficients of the hexagonal fractal antenna show that it resonates in a
number of bands like SGMF antenna. However, due to self-similarity factor of 1/3,
the resonant frequencies repeat by a factor of three. The hexagonal fractal antenna is
useful in wireless applications where a broader frequency separation is desired.

3.2.6 Crown Square Fractal Antenna

This fractal antenna named as the crown square fractal antenna is another example of
self-similar shape based on square shape. The implementation of this antenna is
shown in Fig. 3.8. To obtain the kth iteration geometry, the size of the (k – 1)th
iteration geometry is reduced by 1/2k and then this is merged to the (k – 1)th iteration
geometry.

Zeroth Iteration First Iteration Second Iteration

Fig. 3.8 Development of Crown Square Fractal Antenna (Dehkhoda and
Tavakoli, 2004)

The results of the crown square fractal antenna show that it has a circular polarization
and multiband performance (Dehkhoda and Tavakoli, 2004). When compared to a
simple square antenna, a size reduction of 16.84% is observed in this crown square
fractal antenna.

3.2.7 Rectangular Sierpinski Carpet Fractal Antenna

A small size edge-fed rectangular Sierpinski carpet microstrip patch antenna is

proposed by Chen and Wang (2008). A size reduction of 33.9% is achieved by the

75
second iteration geometry shown in Fig. 3.9 for resonant operating frequency of 1.8
GHz. The experimental verification of the results is also reported.

Zeroth Iteration First Iteration Second Iteration

Fig. 3.9 Development of Edge-Fed Rectangular Sierpinski Carpet Fractal
Antenna (Chen and Wang, 2008)

3.2.8 Sierpinski-Koch Hybrid Fractal Antenna

A miniaturized fractal patch antenna is designed by Chen et al. (2008) by combining

the two fractal geometries: the Koch curve and the Sierpinski carpet. The Koch curves
are used to etch the edges of the patch and Sierpinski carpet shape is used for the
inner patch as shown in Fig. 3.10. Using this hybrid geometry, a size miniaturization
of 77.1% is achieved along with an improvement in operating frequency bandwidth.
The microstrip line feed in the form of impedance transformer is used to excite the
antenna.

76
 Zeroth Iteration First Iteration Second Iteration
Fig. 3.10 Development of Sierpinski-Koch Hybrid Fractal Antenna (Chen et al.,
2008)

3.2.9 H-Fractal Antenna

A new fractal antenna based on H shape is proposed by Weng and Hung (2014). The
first five stages of their antenna are shown in Fig. 3.11. The antenna has multiband
performance and it is optimized for 2.45 GHz and 5.5 GHz WLAN applications using
the PSO algorithm. The optimized antenna is fabricated on an FR4 substrate and
tested experimentally.

Stage 1 Stage 2 Stage 3

77
 Stage 4 Stage 5

Fig. 3.11 Five Stages of H Shape Fractal Antenna (Weng and Hung, 2014)

3.3 FRACTAL ANTENNAS DEVELOPED IN PRESENT

RESEARCH WORK

A number of fractal antennas have been proposed in the last decade. The fractal
antennas have been designed for improved performances such as multiband operation,
enhanced gain, miniaturization etc. The main motive of the presented research work is
to develop fractal antennas with simple structures having size reduction capabilities.
This section introduces the fractal antennas developed in this research work. The
detailed analysis and design is discussed in subsequent chapters.

3.3.1 Miniaturized Crown Rectangular Fractal (CRF) Antenna

The first fractal antenna designed in this proposed research work is based on the
rectangular shape and is named as CRF antenna because of its similarity to the
antenna proposed by Dehkhoda and Tavakoli (2004). The realization of this fractal
antenna is shown in Fig. 3.12.
78
The base geometry i.e. zeroth iteration is a rectangle as shown in Fig. 3.12(a). The
first iteration geometry, shown in Fig. 3.12(b), is obtained by cutting an ellipse from
the base shape and then inserting a rectangle such that the corners of the inserted
rectangle touch the boundary of elliptical slot. The same procedure is repeated for the
inner rectangle of first iteration geometry to obtain the second iteration geometry
which is shown in Fig. 3.12(c). Similarly the further iterations can be obtained.

L L

L L1 L2
L1

W2
W W1 W1
W W

(a) Base Geometry (b) First Iteration (c) Second Iteration

(Zeroth iteration)
Fig. 3.12 Proposed CRF Antenna

The base geometry shown in Fig. 3.12(a) has two design variables: the length L and
the width W of the antenna; Fig. 3.12(b) depicts that the first iteration shape has four
variables: the lengths L & L1 and the widths W & W1; and the second iteration
geometry has six variables: the lengths L, L1 & L2 and the widths W, W1 & W2 as
shown in Fig. 3.12(c). In addition, the resonant frequency also depends on the
substrate parameters i.e. the height of substrate h, dielectric constant εr of the
substrate, and also on copper patch thickness t. Therefore, the resonant frequency of
second iteration of proposed fractal antenna depends on eight parameters assuming t
<< λ (the wavelength). Consequently, its design for a user defined frequency requires
the optimal values of all these parameters. So, design of the proposed antenna is a
multivariable problem.

79
However, to reduce the number of design variables and hence, to simplify the design
procedure, size of the rectangle inserted in first iteration is selected as 50% of the size
of base rectangle and size of the rectangle inserted in second iteration is 50% of the
size of the first iteration rectangle. These assumptions make the number of design
variables reduce to four i.e. the L and W of the base rectangle and the two substrate
parameters. Further, in the presented design, the RT-Duroid substrate with h = 3.175
mm and εr = 2.2 is used. Therefore, the resonant frequency finally depends only on
two parameters: the L and W of the base rectangle. The various dimensions of the first
iteration and second iteration geometry are calculated from the base rectangle
dimensions using the above assumptions. The feed location of the antenna is found by
the trial and error approach; however, it can also be taken as another variable during
antenna design.

In presented work, the dimensions of the starting shape i.e. the zeroth iteration are
taken as L = 30 mm and W = 37 mm respectively. The dimensions of the first and
second iteration geometries are calculated using the above assumptions and are given
as follows:

Base geometry (zeroth iteration) dimensions:-

- Length of base rectangle (L) = 30 mm

- Width of base rectangle (W) = 37 mm

First iteration dimensions:-

- Length of rectangle to be inserted (L1) = 15 mm

- Width of rectangle to be inserted (W1) = 18.5 mm
- Primary axis radius of ellipse to be cut = 12 mm
- Secondary axis radius of ellipse to be cut = 11.8495 mm

Second iteration dimensions

- Length of rectangle to be inserted (L2) = 7.5 mm

- Width of rectangle to be inserted (W2) = 9.25 mm

80
 - Primary axis radius of ellipse to be cut = 6 mm
- Secondary axis radius of ellipse to be cut = 5.9247 mm

After a number of trials, the feeding point of base geometry is selected as (8, 2) and
that of first iteration and second iteration geometries is selected as (5, 7) with respect
to (w.r.t.) center of all geometries at (0, 0).

The fractal antennas described in Fig. 3.12 are simulated using the IE3D software.
The S11 plots shown in Fig. 3.13 depict that the resonance frequency of the base
geometry is 3.12 GHz and that of first and second iterations is 2.35 GHz and 2.32
GHz respectively. So the iterated shapes result in shifting of resonant frequency
values towards the lower end of the frequency scale.

Fig. 3.13 S11 Results of CRF Antenna

The simulated elevation and azimuthal radiation patterns of the antennas shown in
Fig. 3.12 are presented in Fig. 3.14. These patterns depict that the values of peak gain
for base, first and second iteration geometries are 6.97 dBi, 5.59 dBi, and 5.54 dBi
respectively, which are sufficient for low power application in this frequency range.
The patterns also show that the CRF antenna has omni-directional radiation patterns.
Among the proposed three geometries, the base geometry has more metallic area
followed by first iteration geometry and then the second iteration geometry. The
values of gain also follow this order as the gain directly depends upon the metallic
area of the antenna.

81
(a) Elevation Radiation Pattern of (b) Azimuthal Radiation Pattern of
Zeroth Iteration Geometry Zeroth Iteration Geometry

(c) Elevation Radiation Pattern of (d) Azimuthal Radiation Pattern of

First Iteration Geometry First Iteration Geometry

82
 (e) Elevation Radiation Pattern of (f) Azimuthal Radiation Pattern of
Second Iteration Geometry Second Iteration Geometry

Fig. 3.14 Simulated Radiation Patterns of CRF Antenna

The L and W of simple rectangular microstrip antenna resonating at a frequency fr are

calculated using the standard equations (Balanis, 2005) described below.

…(3.2)

…(3.3)

…(3.4)

…(3.5)

where c is velocity of light in free space and is effective dielectric constant.

83
The expressions given in equations 3.2 to 3.5 are used to calculate the L and W of a
simple rectangular antenna for a resonant frequency of 2.32 GHz considering
substrate parameters as h = 3.175 mm & εr = 2.2 and these values come out to be L =
41.79 mm and W = 51.11 mm resulting in an area of 2135.88 mm2. The second
iteration geometry of CRF antenna described above has resonant frequency of 2.32
GHz and has the dimensions as of 30 mm x 37 mm resulting in an area of 1110 mm 2.
Therefore, the second iteration CRF antenna has an area which is only 51.97% of
2135.88 mm2. So, the CRF antenna results in a size reduction of 48.03%.

The above analysis shows that the designed CRF antenna has frequency reduction
property which can be used for the design of miniaturized antennas. The drift in
frequency is due to the fact that this geometry has multiple slots which result in larger
current path leading to the lowering of resonant frequencies.

3.3.2 Tapered CRF Antenna

The scale ratio of base and inserted rectangle in a specific iteration is 50% in the CRF
antenna of previous section. The fractal antenna of Fig. 3.12 is also simulated for
different scale ratios and the frequency reduction characteristic has been observed in
all simulations. However, the main constraint of this fractal antenna geometry is that
it has very small bandwidth. Various approaches for the bandwidth enhancement of
microstrip antennas have been proposed in recent years e.g. slotting (Chen et al.,
2009; Jan and Su, 2005), tapering (Ghatak et al., 2012), and coplanar waveguides
(Jahromi et al., 2011) etc.

In the presented work, diagonally opposite corners of the antenna are tapered to
enhance the bandwidth. The fractal geometry of the Fig. 3.12 (c) with scale ratio of
60% for the inner rectangles is selected for bandwidth enhancement. The taper
dimensions are finalized as 4 mm x 4 mm using trial and error method. Fig. 3.15
shows the values of various dimensions of the selected antenna.

84
 4 4

48.4 29.04 17.42

14.1
5
23.58
39.3

Fig. 3.15 Tapered CRF Antenna (All Dimensions in mm)

The S11 results of the tapered and non-tapered second iteration CRF antenna given in
Fig. 3.16 show that the tapering results in a significant improvement in the bandwidth.
The bandwidth of the non-tapered antenna is 38 MHz and that of tapered is 92.8 MHz
resulting in an increase of about 144%. However, the tapering results in shifting of
resonant frequency from 1.856 GHz to 1.95 GHz i.e. a slight upwards shift in resonant
frequency.

Fig. 3.16 S11 Results of Tapered and Non-Tapered CRF Antenna

85
The radiation patterns of the non-tapered and tapered antennas shown in Fig. 3.17
depict that the pattern shape is not disturbed due to tapering. However, the gain of
tapered antenna has slightly reduced to 5.82 dBi from 5.97 dBi in case of non-tapered.

(a) Elevation Radiation Pattern of (b) Azimuthal Radiation Pattern of

Non-Tapered Antenna Non-Tapered Antenna

(c) Elevation Radiation Pattern of (d) Azimuthal Radiation Pattern of

Tapered Antenna Tapered Antenna

Fig. 3.17 Radiation Patterns of the Non-Tapered and Tapered Antennas

86
The tapered CRF described above has resonant frequency of 1.95 GHz and has the
outer dimensions of 39.3 mm x 48.4 mm resulting in an area of 1902.12 mm2. The
expressions given in equations 3.2 to 3.5 are used to calculate the L and W of a simple
rectangular antenna for the same resonant frequency i.e. 1.95 GHz considering
substrate parameters as h = 3.175 mm & εr = 2.2 and these values comes out to be L =
50.13 mm and W = 60.81 mm resulting in an area of 3048.41 mm2. So, the tapered
CRF antenna has an area of only 62.4% of area of simple rectangular antenna for
same frequency so it results in a size reduction of 37.6%.

3.3.3 Miniaturized Crown Circular Fractal (CCF) Antenna

The CCF antenna geometry proposed here is based on circular shape and is inspired
from the fractal antenna of Ding et al. (2006). The development of the CCF antenna is
shown in Fig. 3.18. The base geometry is a circle as shown in Fig. 3.18(a). The zeroth
iteration geometry, shown in Fig. 3.18(b), is obtained by cutting an ellipse from the
base circular shape. To obtain the first iteration geometry shown in Fig. 3.18(c), the
reduced size copy of the zeroth iteration geometry is merged with the zeroth iteration
such that the circumference of the inserted reduced shape touches the boundary of
elliptical slot. The same procedure can be repeated for the inner circular shape of first
iteration geometry to obtain the second iteration geometry and similarly further
iterations can be designed.

R2
S1
S2
R R R1

(a) Base Circular (b) Zeroth Iteration (c) First Iteration

Shape
Fig. 3.18 CCF Antenna Geometries

87
Fig. 3.18(a) shows that the base geometry has one design variable i.e. the radius R of
the circular antenna, Fig. 3.18(b) depicts that the zeroth iteration shape has three
variables: the radius R of the base circular antenna, the primary axis radius R1 and
the secondary axis radius S1 of the elliptical slot, whereas the first iteration geometry
has total five variables i.e. all the three variables (R, R1, and S1) of zeroth iteration and
two more variables: the primary radius R2 and the secondary radius S2 of the inner
elliptical slot as shown in Fig. 3.18(c). In addition, the substrate parameters h & εr and
the copper patch thickness t also affect the resonant frequency. Therefore, as shown in
Fig. 3.19(a), the resonant frequency of first iteration of proposed CCF antenna
depends on seven parameters assuming t << λ. Consequently, its design for a desired
frequency requires the optimal values of all these parameters. So, the design of the
proposed CCF antenna is a multivariable problem.

However, to reduce the number of design variables and hence, to simplify the design
procedure, the value of R1 is taken as 80% of R and that of S1 is taken as 60% of R.
The radius of the inner circle of the first iteration shape is taken as equal to S1 so that
the circumference of the inserted reduced shape touch the boundary of elliptical slot
and the value of R2 is taken as 80% of S1 and that of S2 is taken as 60% of S1. These
assumptions make the number of design variables reduce to three i.e. the radius R of
base circular shape, and the two substrate parameters h & εr. Further, in the presented
design, the RT-Duroid substrate with h = 0.787 mm and εr = 2.2 is used. Therefore, as
shown in Fig. 3.19(b), the resonant frequency finally depends only on one parameter:
the radius R of base circular shape. The other dimensions of the zeroth and first
iteration geometries are calculated from the base radius R value using the above
assumptions. The feed location of the proposed antenna is found by the trial and error
approach; however, it can also be taken as another variable during antenna design.

88
 Antenna
R Analysis
Algorithm
R1
Antenna R1 = 0.8R
S1
Analysis S1 = 0.6R
R2 Algorithm fr R R2 = 0.8S1 fr
= 0.48R
S2 S2 = 0.6S1
h = 0.36R

εr h = 0.787 mm
εr = 2.2

(a) Full Model (b) Reduced Model

Fig. 3.19 Proposed CCF Antenna Analysis Model

In presented work, the radius R of the base geometry of Fig. 3.18(a) is taken as 10
mm. The zeroth iteration geometry of Fig. 3.18(b) is achieved by cutting an ellipse
from the base circular shape. As per the assumptions, the primary axis radius R1 of the
ellipse is 80% of R i.e. 8 mm and secondary axis radius S1 of the ellipse is 60% of R
i.e. 6 mm. To obtain the dimensions of first iteration geometry shown in Fig. 3.18(c),
the zeroth iteration geometry is reduced by 60% i.e. outer radius of the reduced
geometry is made equal to the secondary axis radius S1 and then this reduced
geometry is inserted in zeroth iteration geometry. As per the assumptions, the primary
axis radius R2 of the ellipse of first iteration geometry is 80% of S1 i.e. 4.8 mm and
secondary axis radius S2 of the ellipse of first iteration geometry is 60% of S1 i.e. 3.6
mm. The feeding points of base circular geometry, zeroth iteration, and first iteration
are finalized by the trial and error method as (1.9, 1), (1.5, 6.2), and (1.7, 6.4)
respectively with (0, 0) co-ordinate at the centre of the geometry.

The proposed CCF antenna geometries are simulated using IE3D software and the
simulated S11 plots of the zeroth and first iterations along with that of simple circular
antenna are shown in Fig. 3.20. The simulated S11 plots illustrate that the resonance
89
frequency of the circular geometry is 5.68 GHz and that of zeroth and first iterations
is 4.59 GHz and 4.57 GHz respectively. These graphs show that the resonant
frequency of the zeroth and first iteration antennas shift towards the lower end of the
frequency scale as compared to the resonant frequency of circular antenna. This
characteristic means that this geometry has frequency lowering properties which can
be exploited for designing miniaturized antennas. Also it has been found using
simulation results that second and further iterations do not provide much further
frequency reduction.

The radiation patterns of the CCF antenna are shown in Fig. 3.21. This antenna also
has the omni-directional radiation pattern. The base geometry has a peak gain of 6.53
dBi. The peak gain of first iteration geometry is equal to 6.03 dBi whereas that of
zeroth iteration geometry is 5.99 dBi.

Fig. 3.20 S11 Results of CCF Antenna

90
(a) Elevation Radiation Pattern of (b) Azimuthal Radiation Pattern of
Base Circle Base Circle

(c) Elevation Radiation Pattern of (d) Azimuthal Radiation Pattern of

Zeroth Iteration Geometry Zeroth Iteration Geometry

91
 (e) Elevation Radiation Pattern of (f) Azimuthal Radiation Pattern of
First Iteration Geometry First Iteration Geometry

Fig. 3.21 Simulated Radiation Patterns of CCF Antenna

The radius a of a simple circular antenna resonating at a frequency fr is calculated

using the expressions (Balanis, 2005) given below as equations 3.6 and 3.7.

…(3.6)

…(3.7)

The radius of a simple circular patch antenna resonating at 4.57 GHz frequency (i.e.
resonant frequency of first iteration CCF antenna discussed above) is calculated using
the equations 3.6 & 3.7 and it comes equal to 12.43 mm resulting in an area of 485.39
mm2. However, the area of first iteration CCF antenna resonating at this frequency is
314.16 mm2 which is 64.72% of area of simple circular antenna. Therefore, the
proposed CCF antenna has a size reduction of 35.28% at this frequency.

92
3.3.4 Low Cost Miniaturized CCF Antenna

The performance of CCF antenna geometry described in previous section is analyzed

on FR4 substrate of h = 1.47 mm and εr = 4.3 to design a low cost miniaturized fractal
antenna. The first and second iteration CCF antenna geometries of outer radius 10 mm
are simulated and the S11 results as shown in Fig. 3.22 are obtained which depict that
the resonant of zeroth iteration antenna is 3.45 GHz and that of first iteration antenna
is 3.39 GHz. The resonant frequency of simple circular antenna of 10 mm radius is
also shown in Fig. 3.22 and it is 4.08 GHz. The comparison of resonant frequencies
show that the resonant frequency of zeroth and first iteration antennas shift towards
the lower end of frequency scale as compared to the resonant frequency of circular
antenna. So, the CCF antenna has frequency lowering characteristic which can be
used to design the miniaturized antennas for different wireless applications.

The elevation and azimuthal radiation patterns of first iteration of this FR4 CCF
antenna are shown in Fig. 3.23. These plots show that this antenna also has the omni-
directional radiation pattern and has a peak gain of 1.15 dBi. The peak gain of this
antenna is less as compared to CCF on RT-Duroid because the FR4 substrate is a

Fig. 3.22 S11 Results of Low Cost CCF Antenna

93
 (a) Elevation Radiation Pattern (b) Azimuthal Radiation Pattern

Fig. 3.23 Radiation Pattern Plots of First Iteration Low Cost CCF Antenna

lossy substrate. However, as the FR4 substrate is a low cost substrate and it is easily
available so it is helpful in designing the low cost CCF antenna for applications where
moderate value of gain is acceptable.

The radius of a simple circular patch antenna resonating at 3.39 GHz frequency (i.e.
resonant frequency of first iteration CCF antenna discussed above) is calculated using
the equations 3.6 & 3.7 and it comes equal to 12.06 mm resulting in an area of 456.92
mm2. However, the area of first iteration CCF antenna resonating at this frequency is
314.16 mm2 which is 68.75% of area of simple circular antenna. Therefore, the
proposed low cost CCF antenna results in a size reduction of 31.25% at this
frequency.

3.3.5 Multiband Miniaturized Split Ring Fractal (SRF) Antenna

The starting shape for SRF antenna is a square ring of side-length 10 mm with ring
width of 0.5 mm as shown in Fig. 3.24(a). A slit of size 0.5 mm x 0.5 mm is inserted
in the top side of the ring as shown in Fig. 3.24(b) to obtain a split ring.
94
 0.5mm
0.5mm

10mm

(a) Square Ring (b) Split Ring

Fig. 3.24 Geometries of Square Ring and Split Ring Patch Antennas

The split ring shape of Fig. 3.24(b) is converted into fractal shape by using the
Minkowski curve fractal geometry. The first two stages i.e. the initiator and the
generator of the Minkowski curve are shown in Fig. 3.25(a). All straight sides of the
split ring of Fig. 3.24(b) are replaced by the generator of Fig. 3.25(a) and the resulting
shape of the antenna is shown in Fig. 3.25(b). As the antenna is implemented with a
split ring with Minkowski curve sides so it is named as the SRF Antenna. The coaxial
probe is used to feed the antenna structures shown in Fig. 3.24 and 3.25. The feed
points are (0.35, 4.75) in Fig. 3.24(a) and (b), and (0.35, 2.975) in Fig. 3.25(b). The
feed point locations are defined by taking the center of the structures at (0, 0). The
substrate used for implementation is RT-Duroid substrate having εr = 2.2, h = 3.175
mm, and tan δ = 0.0009.

95
 Initiator

10mm

Generator
3 mm 4 mm 3 mm

1.8 mm

(a) First Two Stages of Minkowski (b) SRF Antenna

Curve

Fig. 3.25 Geometry of Proposed SRF Antenna

The antenna structures shown in Figs. 3.24 and 3.25 are simulated using IE3D
software for 5–11 GHz frequency range. The S11 results shown in Fig. 3.26 depict that
the ring structure of Fig. 3.24(a) resonates at 8.418 GHz. Theoretically, the square
patch antenna of side-length 10 mm i.e. the side-length equal to that of the proposed
ring structure will resonate at 8.364 GHz, so the ring structure has performance
comparable to the square patch of same dimensions. The split ring structure of Fig.
3.24(b) has dual band performance with the resonant frequencies of 7.616 GHz and
10.279 GHz due to the presence of the slit in the upper side of the ring. As revealed
by Fig. 3.26, the SRF antenna also has dual band of resonance with resonant
frequencies 6.587 GHz and 8.658 GHz. The replacement of the sides of the split ring
structure by Minkowski curve resulted in fitting of more electrical length in same
area. So, it causes the drift of the resonant frequencies toward lower end of frequency.
In reference to the lower resonance frequency, the proposed SRF antenna has resulted
in a size reduction of 57.77% compared to simple rectangular microstrip antenna for
the same frequency.

96
 Fig. 3.26 S11 Results of Ring, Split Ring, and SRF Antennas

The radiation pattern plots of the SRF antenna are shown in Fig. 3.27. The azimuthal
pattern shown in Fig. 3.27(a) illustrates that the peak values of gain are 4.85 dBi and
1.88 dBi in first and second band respectively, which are within 3 dBi range. The
elevation pattern graph is shown in Fig. 3.27.

(a) Azimuthal Radiation Pattern (b) Elevation Radiation Pattern

Fig. 3.27 Radiation Pattern Plots of Proposed SRF Antenna
97
The prototype of SRF antenna discussed above is fabricated and it is shown in Fig.
3.28. The prototype antenna is tested using Anritsu‟ vector network analyzer (Model
No. MS46322A) for frequency range of 5 GHz to 15 GHz. The measured results are
shown in Fig. 3.29 which depict that the antenna has dual-band performance.

Fig. 3.28 Prototype of SRF Antenna

Fig. 3.29 Measured S11 Results of SRF Antenna

98
The comparison of measured and simulated results is shown in Table 3.1 which
illustrates that the there is reasonable matching of simulated and experimental
resonant frequency values.

Table 3.1 Comparison of Simulation and Experimental Results of SRF Antenna

S. Simulation Results Experimental Results

No. Resonant S11 Resonant S11

Frequency Frequency
(dB) (dB)
(GHz) (GHz)

1. 6.587 –25.9 7.00 –10.75

2. 8.658 –14.85 9.20 –17.8

The main reason of the minor deviation of measured results from simulated results is
improper co-axial connector used in prototype antenna. The split ring of the antenna
has a width of 0.5 mm. But the diameter of the central conductor of the co-axial
connector used is 1.2 mm. The large value of the diameter of connector resulted in
small shift in resonant frequency.

3.3.6 Metamaterial Loaded SRF Antenna

The SRF antenna described in previous section is further loaded with a metamaterial
split ring resonator (SRR) structure of dimensions as shown in Fig. 3.30(a). This SRR
structure has metamaterial characteristics at frequency 5 GHz (Joshi et al., 2010). The
SRR structure is placed at the center of the SRF antenna as shown in Fig. 3.30(b). The
simulation results of this antenna, named as metamaterial loaded SRF antenna, shown
in Fig. 3.31 depicts that this antenna resonates at three frequencies as compared to two
in SRF antenna.

99
 0.2

0.2
5

0.2

(a) SRR Structure (b) SRR Loaded SRF Antenna

Fig. 3.30 Structure of Metamaterial Loaded SRF Antenna (All Dimensions in

mm)

Fig. 3.31 S11 Results of Metamaterial Loaded SRF Antenna

The radiation plots shown in Fig. 3.32 illustrate that the fundamental resonant band of
metamaterial loaded SRF antenna has omni-directional radiation pattern and the shape
of the pattern in other resonant bands is distorted to some extent as compared to omni-
directional.

100
 (a) Elevation Radiation Pattern (b) Azimuthal Radiation Pattern
Fig. 3.32 Radiation Pattern Plots of Metamaterial Loaded SRF Antenna

The S11, bandwidth, and gain results of SRF antenna with and without metamaterial
loading are summarized in Table 3.2. The comparison of these parameters with those
of simple SRF antenna shows that the first two resonant frequencies of the proposed
metamaterial loaded SRF antenna have drifted toward the lower end of frequency
scale and a third resonant band centered at 10.594 GHz is observed. The bandwidth
and gain values observed in all bands are sufficient for low power applications.

Table 3.2 Results of SRF Antenna with and without Metamaterial Loading
S SRF Antenna Metamaterial Loaded SRF Antenna
No. Resonant S11 Band- Gain Resonant S11 Band- Gain
Frequencies (dB) width (dBi) Frequencies (dB) width (dBi)
(GHz) (MHz) (GHz) (MHz)
1 6.587 -26.26 213 4.85 6.407 -12.16 225 3.87
2 8.658 -15.22 332 1.88 8.495 -11.82 259 2.45
3 10.836 -6.13 -- 1.34 10.594 -18.68 294 3.54

101
In reference to the lowest resonance frequency, the proposed metamaterial loaded
SRF antenna has resulted in a size reduction of 60.26% compared to simple
rectangular microstrip antenna for the same frequency.

3.3.7 Dual Band Crown Elliptical Fractal (CEF) Antenna

The proposed CEF antenna is based on elliptical base shape with primary axis radius
of 5 mm and secondary axis radius of 2 mm. The first iteration shape is achieved by
first cutting a rectangle of length 7 mm and width 2.8 mm and then adding an ellipse
with primary axis diameter equal to length of rectangular slot and secondary axis
diameter equal to width of rectangular slot. These dimensions are selected so that the
size of the inner ellipse comes out to 70% of the base ellipse. Similarly, the second
iteration geometry is designed by cutting a rectangle of length 4.9 mm and width 1.96
mm from inner ellipse and then adding an ellipse of size equal to 70% of inner ellipse
of first iteration geometry. The h and εr of substrate used is 2.2 and 2 mm
respectively. The development of the antennas is shown in Fig. 3.33.

The CEF antenna geometries of Fig. 3.33 are excited by using a coaxial feed at point
(4, 0) w.r.t. centre of the geometries at the (0, 0). The S11 results are shown in Fig.
3.34 which depict that the base elliptical shape resonates at 13.85 GHz and has a
bandwidth of 2.4 GHz. The first iteration has dual band with resonating frequencies at
13.52 GHz and 20.12 GHz with bandwidth 1.9 GHz and 2.1 GHz respectively. The
second iteration shape also has dual band resonating characteristics with resonating
frequencies equal to that of first iteration shape. However, the bandwidth of this is
enhanced to 2.2 GHz and 3.6 GHz in first and second band respectively.

(a) Base Shape (b) First Iteration (c) Second Iteration

Fig. 3.33 CEF Antenna Geometries

102
 Fig. 3.34 S11 Results of CEF Antenna

The simulated radiation patterns of the CEF antennas shown in Fig. 3.35 demonstrate
that the peak gain of base shape is 2.56 dBi, gain of first iteration antenna in first and
second band is 1.95 dBi and 2.09 dBi respectively, and that of second iteration
antenna in first and second band is 2.3 dBi and 1.5 dBi respectively.

(a) Elevation Radiation Pattern of (b) Azimuthal Radiation Pattern of

Base Geometry Base Geometry

103
(c) Elevation Radiation Pattern of (d) Azimuthal Radiation Pattern of
First Iteration Geometry First Iteration Geometry

(e) Elevation Radiation Pattern of (f) Azimuthal Radiation Pattern of

Second Iteration Geometry Second Iteration Geometry

Fig. 3.35 Radiation Pattern Plots of CEF Antenna

104
The second iteration CEF geometry has dual band behavior with wide bandwidth in
both bands and also has sufficient gain for indoor applications so this geometry can be
used to design dual band antennas for such applications.

3.3.8 Dual Band Wearable Fractal Antenna

The wearable antennas are used for various types of body-centric communication
applications such as medical healthcare, GPS, and personal safety (Hertleer et al.,
2009; Lui et al., 2013). Different types of wearable antennas have been reported in the
literature in recent years. Hertleer et al. (2009) have presented a 2.4 GHz ISM band
textile planar antenna on flexible protective foam which is suitable for firefighter
garments. A body-worn circularly polarized textile antenna resonating in 2.4 GHz
ISM band is proposed by Lui et al. (2013) and is used for the wirelessly-powered
temperature sensor system on a human arm to demonstrate the successful power
transmission over a distance of 1.7 m. The use of metamatrial loading for designing
wearable antennas has been explained by Joshi et al. (2012) and Joshi et al. (2013).
The wearable fractal antennas have also been explored in recent years. A design of a
triple band textile dipole antenna based on Koch fractal geometry for wearable
applications is described by Jalil et al. (2013). A dual band wearable fractal monopole
antenna integrated with an electromagnetic band-gap structure is presented by Velan
et al. (2015).

A new dual band wearable fractal antenna is discussed in this section. The antenna
geometry is based on circular shape and its first and second iterations are shown in
Fig. 3.36. First iteration geometry is designed by cutting a square and then adding a
circle of diameter equal to the side-length of square slot. To reduce the number of
variable, the side-length of the square slot is taken as 60% of radius of base circular
shape. Similarly, the second iteration shape is designed by repeating the same
procedure on the inner circle of first iteration i.e. the circle inserted in first iteration. A
square slot of side-length equal to 60% of diameter of inner circle is cut and then a
circle of radius equal to side-length of square slot is inserted to implement the second

105
iteration antenna. The antenna is designed on polypropylene (PR 30), a geo-textile
material substrate of h = 1.9 mm and εr = 2.2 which is a tested and easily available
substrate for wearable antennas (Joshi et al., 2013). The outer radius of the wearable
antenna discussed in this section is taken as 10 mm. Other dimensions are calculated
using the above described relations and first two iterations are simulated using the
IE3D software. The feed-points of base, first, and second iteration geometries are
finalized as (4, 0), (1.5, 5), and (4, 3.8) respectively by following the trial and error
approach. The feed point co-ordinates are specified by considering the centre of
antenna geometries at (0, 0).

The S11 results of the wearable antenna geometries are shown in Fig. 3.37. These
results show that while the base circular geometry is resonating at a single frequency,
the first and second iteration geometries are having two resonant frequencies. The
resonant frequency of base geometry is 5.51 GHz, the two resonant frequencies of
first iteration geometry are 5.35 GHz and 9.10 GHz, and the second iteration
geometry is having the resonant frequencies of 5.28 GHz and 8.95 GHz.

(a) Base Circle (b) First Iteration (c) Second Iteration

(Zeroth Iteration)

Fig. 3.36 Wearable Fractal Antenna Geometry

106
 Fig. 3.37 S11 Results of Wearable Fractal Antenna

So, in first iteration geometry, there is a small drift of frequency value towards lower
end of the frequency scale for first resonant frequency and a new resonant band is
introduced. Similarly, the second iteration geometry resulted in further frequency drift
in both the bands.

The simulated radiation patterns of this wearable fractal antenna geometries have
shown that the base iteration geometry has a peak gain of 6.85 dBi, the first iteration
geometry has peak gain values of 6.02 dBi and 3.1 dBi in first and second resonant
band respectively, and for second iteration geometry, the values of peak gain are 6.23
dBi and 3.55 dBi in first and second resonant band respectively. So the antenna has
very good value of peak gain for first resonant band while having a sufficient value of
gain in second band. Therefore, the antenna can be used for body centric and wearable

107
wireless applications in these bands. However, it can be designed to operate at desired
frequencies by tuning the outer radius and the outer to inner circle scale ratio.

3.4 SUMMARY

The chapter starts with an introduction to fractal geometry and its design procedure.
Then the fractal antennas are defined and the importance of self similarity and space
filling properties in fractal antennas is described. The fractal antennas based on
certain fractal shapes like Sierpinski gasket, Sierpinski carpet, Koch curve, Hilbert
curve, hexagonal, crown square, and rectangular Sierpinski carpet are discussed. The
features of the antennas based on above fractal geometries are highlighted and it is
seen that the fractal antennas have improved performances as compared to standard
shapes of equivalent dimensions.

All the fractal antennas developed in presented research work are also described in
this chapter. The miniaturized CRF antenna is designed on the rectangular base
geometry and it is seen that this antenna has size-reduction capability. A
miniaturization of 48.03% is achieved for 2.32 GHz frequency of operation. The
bandwidth enhancement of this CRF antenna by tapering the opposite corner is
discussed and it is seen that tapering of corners resulted in increase in bandwidth from
38 MHz to 92.8 MHz, an improvement of about 144%. However, the slight upward
shift in resonant frequency is also observed due to tapering. The CCF antenna
designed in presented thesis work is based on base circular shape. This fractal shape is
developed on RT-Duroid and FR4 substrates and it is seen that it has miniaturization
features for both the cases. However, the gain of CCF antenna developed on FR4
substrate is relatively less. Various parameters of this CCF geometry are interlinked to
reduce the number of design variables. The next antenna developed in presented work
is SRF antenna which is designed using a fractal ring of width 0.5 mm and having a
gap of 0.5 mm x 0.5 mm in top side of the geometry. This fractal geometry has
multiband performance and a size reduction of about 58% is achieved for lowest
resonant frequency of 6.599 GHz. This antenna has moderate gain values for all the
108
resonant frequencies. The measured results of SRF antenna have shown that there is
reasonable matching of simulated and measured resonant frequencies. The non-
availability of proper co-axial connecter for fabrication is main reason of minor
deviations.

The performance of the SRF antenna is enhanced by loading it with the metamaterial
SRR structure. The metamaterial loaded SRF antenna has more resonant bands and
improved bandwidth as compared to unloaded case.

The CEF antenna is another geometry introduced in this work. This geometry has dual
band performance and it has been observed that an increase in iteration number results
in an improvement in bandwidth of resonant bands.

The wearable fractal antenna introduced in the last section is another important
contribution of this chapter. The antenna is designed using a flexible geo-textile
material and is based on circular geometry. The antenna has dual band performance
with very good values of gain.

ANN models for parameter estimation of the proposed antennas are developed in the
next chapter.

109