Fractal Antenna Engineering: The Theory and
Design of Fractal Antenna Arrays
Douglas H. Werner’, Randy L. Haupt,
‘and Pingjuan L. Werner’
"Communications and Space Sciences Laboratory
The Pennsylvania State Univesity
Department of Blecrical Engineering
211A Electrical Engineering Eat
University Park, PA 16802
E-mail: dhw@psu.edu
Department of Electrical Enginaéxing
Utah State University
Logan, UT 84322-4120
Tel: (435) 797-2840
ans (435) 797-2054
E-mail randy haupt@ece-usu.edu or haupi@ieee org
The Pennsylvania State University
College of Engineering
‘DuBois, PA 15801
E-xil: piv7@paueds
Keywords: Fractals antenna arays; antenna theory; antenna
radiation pattems frequeney-independent antennas log-periodic
sntennas owsidelebe antennas; aay thinning, tay signal
processing
1 Abstract
A fractal isa recursively generated object having a fractional
dimension. Many objects, including antennas, can be designed
using the recursive nature of fatal. In this ace, we provide a
comprehensive overview of recent developments in the field of
fractal antenna enginoerng, with particular emphasis placed on the
theory and design of fractal amays. We introduce some important
properties of fractal anays, including the frequeney-independent
‘mult-band characteristics, schemes. for realizing lowsidelobe
designs, systematic approaches to thinning, and the ability to
1). In many cases,
the generating subarray has elements that are tumed on and off in &
‘cerain pattem. A set formula for copying, scaling, and translation
‘ofthe generating subarray is then followed in order to produce the
fractal aay. Hence, fractal arrays that are created inthis manner
‘will be composed ofa Soquence of sesimilarsubarrays, In other
‘words, they may be conveniently thought of as arays of aay (6
‘The aray factor for a frctal aray of this type may be
‘expressed inthe general form [23-25]
:
AFp(e)=T]64( 0°"
@
where GA(y) represents the aay factor associated wth the gen
crating suhaay. The parameter 5 js sae or expansion factor
{hat govems how large the say grows with each recursive appli-
cation of the generating subaray. The expression forthe facta
fury factor given in Euuatin (1) i simply the product of sealed
‘anions ofa generating subaray factor. Therefore, wo may reat
Equation (1) as representing formal semen ofthe pater
‘alpiction theore for fatal arrays. Applications of hie pe
Gialid pattern-mliptication theorem tothe analysis an dssign
Of ica aswell s planar cll ays willbe considered in he
following sections.
ed}
Figure 1. The geometry fora linear array of uniformly spaced
Isotropic sources.
3.1 Cantor linear arrays
A linear aray of isotropic elements, uniformly spaced a dis-
tance apart along the # axis, s shown in Figure 1. The arayfic-
tor coresponding to this linear array may be expressed inthe form
(27,28)
{in+23%rpcosfuy}. foe N41 elements
arty ®
[PS iaeeelle-ti2)e]} fer Nemes
vere
y=Hened conte} °
sot
peeia o
‘These amays become ffacal-ike when approprinte clements are
tumed off o removed, suc that
o
A, ifelement mis tumed on
0, element mi trned off
Hence, fractal arays produced by following this procedure belong
toa special category of thinned uray
(One of the simplest schemes for constructing 2 fractal linear
ray follows the eve forthe Cantor set (29). Cantor linear arrays
were frst proposed and studied in (21] for thee potential use in the
design of low-sidelobe arrays. Some other aspects of Cantor arays
have been investigated more recently in (23-25]
‘The basic triadic Cantor array may be eeated by tating with
1 tires-element generating subaray, and then applying it repeat-
edly over P scales of growth. The generating subarray inthis ase
has three uniformly spaced elements, with the center clement
tured off or removed, i. 101. The triadic Cantor array i gener-
ated recursively by replacing 1 by 101 and 0 by 000 at each stage
(of the construction. For example, atthe second stage of coastruc-
tion (= 2) the array pattern would look Hike
IEEE Antennas and Propagation Magazine, Vol 41, No.5, October 1998101000101,
and atthe third stage (P=3), we would have oy
101000101 000000000101 000101 od
The aray factor ofthe theee-lement generating subaray with the 84F~
representation 101 8 . {|
GAly)=200s(v). o 4 LI
which may be derive fom Equation (2) by setting N'=1. Jy =0, —azfme~
and 4, =1.Substtuing Equation (6 into Egpation (1) and choos:
ing an expansion factor ofthe (ie, 8 3) rls a expres
sion forthe Cantor aay factor given by
) Figure 2s. plot of the tiadie Cantor
third stage of growth, P=3.
Jeos(3¥)e0s(9y).
Figure 2a, A plot ofthe tr
the first stage of growth, ?
ie Cantor fractal array factor for
1. The array factor is co5(¥).
[i
: ] \
a
igure 2, A plot ofthe triadic Cantor fractal array factor for
the second stage of growth, P=2. The array factor i
cos(yeos(3)-
IEEE Antonnas and Propagation Magazine, Vol. 41, No, 8, Otober 1988
Figure 24. 4 plot of the triadie Cantor fractal array factor for
fhe fourth stage of growth, Px. The array factor is
as(y oot }eos oy )eos(27)-
‘where the hat notation indicates thatthe quantities have been nor
‘alized, Figure 2 contains plots of Equation (7) forthe it four
sages inthe growth ofa Cantor aay.
‘Suppose thatthe spacing between array elements is a quater-
wavelength (ie, d= 4/4), and that 0) =90°. Thea, an expression
forthe directivity of a Cantor ary of isotropic point sources may
be derived from
®
where y =u, with a= cord and