You are on page 1of 5

2019 1st Global Power, Energy and Communication Conference (IEEE GPECOM2019), June 12-15, 2019, Cappadocia, Turkey

Linear Antenna-Array with Log-increasing


Inter-element Spacing and Non Uniform Weights
Waseem Khan Sarah Saeed Atiq-ur-Rehman
Dept. of Electrical Engineering Dept. of Electrical Engineering Dept. of Electrical Engineering
Air University Air University Air University
Islamabad, Pakistan Islamabad, Pakistan Islamabad, Pakistan
waseem.khan@mail.au.edu.pk sarahsaeed@mail.au.edu.pk atiq@mail.au.edu.pk

Abstract—In this paper, linear antenna-array with logarithmi- optimization algorithms are generally computationally expen-
cally increasing inter-element spacing and non-uniform excitation sive.
amplitudes or weights has been investigated for beamwidth (BW), In this paper, we have proposed a NULA with non-uniform
side lobe level (SLL) and directivity (D). This non-uniform linear
array has been compared with uniform and non-uniform linear weights. For inter-element spacing, logarithmic function has
arrays, employing uniform and non-uniform weights, proposed been employed and the weights have been derived from a
in literature. Conventionally, non-uniform weighting techniques standard window function. In literature, a number of window
such as binomial, Dolph-Chebhychev, etc. have been applied to functions are available, but for the proof of concept, only
suppress side lobes at the cost of increased beamwidth. In this Bartlett-Hanning window function has been used here. The
paper, we have shown that applying non-uniform weights to the
proposed non-uniform linear array, SLL can be suppressed with- proposed array has been simulated for 9, 11, 15 and 19
out compromising beamwidth, resulting in increased directivity. elements and compared with various ULAs and NULAs in
The proposed array has been compared with other linear array terms of HPBW, PSLL and directivity. The rest of the paper is
geometries and weighting schemes in terms of beamwidth, peak organized as follows. In section II, the proposed system model
side lobe level (PSLL) and directivity. is described. In section III, simulation results are presented and
Index Terms—non uniform array, beampattern, directivity
the performance metrics of the proposed and other arrays have
been compared. Finally, section IV concludes the paper.
I. I NTRODUCTION II. S YSTEM M ODEL
In wireless communications and radars, antenna arrays Consider a broadside linear array comprising of an odd
have been used for synthesizing beampattern having BW number, L, of isotropic antenna elements placed along x-axis
and PSLL as small as possible. Thus, half-power beamwidth centered at origin, as shown in Fig. 1. Let the position of
(HPBW) and PSLL are the key parameters while comparing mth element is xm ; m = −M/2, −M/2 + 1, ..., M/2, where
the performance of different types of arrays. Directivity is M = L − 1. Array factor (AF ) of non-uniform linear array is
also an important performance parameter which is a meausure given by[9]
M/2
of ratio of energy transmitted in the desired direction to X
the total energy transmitted. Conventionally, linear arrays AF = am ejψm (1)
m=−M/2
with uniform inter-element spacing have been widely used[1].
Uniform linear arrays (ULA) with uniform excitation-current 2
and the beampattern is calculated as |AF | . Here am is the
amplitudes or weights yield quite small BW but high PSLL. weight applied to the signal transmitted by mth element, ψm =
Employing non-uniform weights, such as triangular, binomial 2πf0 xm cos θ/c, f0 is the operating frequency of the antenna
or Chebyshev[2], PSLL can be suppressed but the beamwidth array, θ is the angle of propagation of electromagnetic waves
increases drastically. Arrays with non-uniform spacing, termed
as non-uniform linear array (NULA) in this paper, have also
been investigated for the desired beampattern[3],[4]. Heuristic
optimization techniques have been applied to find the optimum
weights and inter-element distances. In particular, Particle
Swarm Optimization (PSO)[5], [6], Genetic Algorithm (GA)
[3], [7], Artificial Bee Colony (ABC) algorithm [8] and Firefly
algorithm (FA) [4] have been successfully used to design
ULAs and NULAs. However, in most of the reported works,
e.g. [3], [4], [5], [8], reduced PSLL is accompanied with
increased bandwidth (BW) and/or decreased directivity, when
compared with ULA. Further, it may be noted that heuristic Fig. 1: System Model

978-1-5386-8086-5/19/$31.00 © 2019 IEEE


5
2019 1st Global Power, Energy and Communication Conference (IEEE GPECOM2019), June 12-15, 2019, Cappadocia, Turkey

w.r.t. x-axis and c is the speed of light. Now consider that


locations of the elements are given by
(
0, m=0
xm = λ
sign(m) 2 (|m| + logM |m|), m = ±1, ±2, ..., ±M/2
(2)
Here logM (.) represents log base M and λ denotes the
wavelength. In this scenario, the positions of antenna
elements are

x0 = 0
λ
x±1 = ±
2
λ
x±2 = ± {2 + logM (2)} (3)
2
..
.
λ (a) Bartlett-Hanning Window and array weights combined
x±M/2 = ± {M + logM (M/2)}
2
Note that minimum inter-element distance is λ/2 and in-
creases by logM (m) times λ/2 as we go away from the center
of the array. It may be further noted that removing log term
in (2) and (3) leads to the case of uniform linear array with
λ/2 spacing.
The proposed geometry, in this work, has been tested
with non-uniform weights that have been derived from a
standard window function namely Bartlett-Hanning window.
The window function is given by[10]:
n h n i
wn = 0.62 − 0.48 − 0.5 + 0.38cos 2π − 0.5 ;

N N
n = 0, ..., N − 1 (4)
and the array weights am are given by
am = wm+3M/2 ; m = −M/2, ..., M/2 (5)
(b) Weights derived from Bartlett-Hanning window
Here the window length is taken as N = 3M + 1. Thus,
the window length is nearly three times the number of array Fig. 2: Bartlett-Hanning Window wm (N = 3M + 1 = 31)
elements, however, only central portion of the window is and the array weights am
employed for weighting, as explained pictorially in Fig. 2.
III. S IMULATION
Fig. 3 shows the beampattern of the proposed array, for
The proposed array geometry has been simulated in Matlab
L = 9, 11, 15, 19, with weights derived from Bartlett-
and the normalized beampattern is plotted and compared with
Hanning window function. This figure shows that HPBW
that of ULAs and NULAs given in literature. In all ULAs,
decreases and side-lobes go down when the array length
inter-element spacing of λ/2 is assumed. PSLL, HPBW and
increases. Fig. 4, Fig. 5 and Fig. 6 show a comparison of the
directivity of these arrays have been tabulated for comparison.
beampattern of the proposed array with other types of arrays
Directivity has been calculated using the relation[11]
!2 namely (i)ULA with uniform weights (ii)ULA with binomial
M/2
P weights (iii)ULA with Dolph-Chebychev weights (iv)ULA
am with weights optimized by GA[7] (v)NULA with inter-element
m=−M/2
D= spacings optimized by FA(NULA-FA)[4] (vi)NULA designed
M/2 M/2
sin(k(xm − xp )) using parabolic relation[3] and (vii)ULA with weights opti-
am ap ej(αm −αp )
P P
m=−M/2 p=−M/2 k(xm − xp ) mized by ABC(ULA-ABC)[8]. As obvious from Fig. 4b and
(6) Fig. 5b, HPBW offered by the proposed array is the smallest
where k = 2πf0 /c is the wavenumber and αm are the phases among all the compared here. Performance metrics of all these
of array weights. In our case αm = 0, as we are considering arrays are given in Tables I, II and III for L = 9, 11, 19
broadside arrays. respectively. The tabulated results show that HPBW of the

6
2019 1st Global Power, Energy and Communication Conference (IEEE GPECOM2019), June 12-15, 2019, Cappadocia, Turkey

(a) Beampattern(whole view)


Fig. 3: Beampattern (partial view) of the proposed array with
number of elements L = 9, 11, 15, 19

proposed array is the smallest with an exception of Dolph-


Chebychev array with L=19 having HPBW slightly smaller
than that of the proposed one. However, in terms of directivity,
the proposed array outperforms all the arrays, including Dolph-
Chebychev, with considerable margin. Although PSLL of the
proposed array is not the lowest but due to the smallest HPBW
and considerably low PSLL, directivity of the proposed array
is the largest.
In [4], [6], arrays have been optimized, using FA and
PSO respectively, with an objective of keeping PSLL less
than a threshold. The resultant arrays have PSLL less than
the threshold but all the side lobes have the same level, in
contrast to most of the arrays, where the SLL decays as the
angular distance from the main lobe increases. Moreover, this (b) Beampattern(partial view)
reduction in PSLL is accompanied by widening of main lobe,
Fig. 4: Beampattern of the proposed array compared with ULA
which causes loss of directivity, despite quite low PSLL. In
(with uniform, binomial and Dolph-Chebychev weights) and
[7], [3], GA has been employed with the constraint of no
the array designed using GA[7], L=9 for all, except ULA(GA)
side lobes. This approach yielded quite low PSLL in [3], but
where L=10
due to increased beamwidth, directivity is compromised. It
may also be noted in Tables I and II that ULA with binomial
weights offer extra-ordinary low side lobes but simultaneously
very wide main lobe, due to which its directivity is the lowest TABLE I: Performance Parameters of the Proposed and Other
among all the arrays compared. Arrays with 9 antenna elements
IV. C ONCLUSION
HPBW PSLL
Name of Array Directivity
In this paper a novel geometry of non-uniformly spaced (◦ ) (dB)
antenna array has been proposed and it is shown that apply- ULA with uniform weights 11.09 -12.9 8.99
ULA with binomial weights 21.37 -335 5.08
ing non-uniform excitation amplitudes to the proposed array ULA with Dolph-Chebychev
generates a beampattern with quite small HPBW and the 11.09 -15 8.83
weights
highest directivity, among the arrays compared here. It has ULA, with 10 elements,
designed using GA (example 10.89 -16.5 9.71
been shown that the proposed array outperforms arrays in 3 in [7])
which heuristic optimization algorithms have been used to Proposed array 10.27 -16.9 10.03
find optimum amplitudes or optimum inter-element spacings.
Although some of the existing arrays offer PSLL lower than

7
2019 1st Global Power, Energy and Communication Conference (IEEE GPECOM2019), June 12-15, 2019, Cappadocia, Turkey

(a) Beampattern(whole view) (a) Beampattern(whole view)

(b) Beampattern(partial view) (b) Beampattern(partial view)

Fig. 5: Beampattern of proposed array compared with ULA Fig. 6: Beampattern of proposed array with Bartlett-Hanning
(with uniform, binomial and Dolph-Chebychev weights) and weights compared with ULA (with uniform and Dolph-
NULA designed using parabolic relation[3], L = 11 for all Chebychev weights), NULA-FA[4] and ULA-ABC[8], L=19
for all, except NULA-FA and ULA-ABC where L=20

TABLE II: Performance Parameters of the Proposed and Other TABLE III: Performance Parameters of the Proposed and
Arrays with 11 antenna elements Other Arrays with 19 Antenna Elements
HPBW PSLL HPBW PSLL
Name of Array Directivity
(◦ ) (dB) Name of Array
(◦ ) (dB)
Directivity
ULA with uniform weights 9.04 -13 10.99 ULA with uniform weights 5.138 -13.2 18.98
ULA with binomial weights 18.9 -325 5.67 ULA with Dolph-Chebychev
ULA with Dolph-Chebychev 4.93 -15 16.77
9.04 -15 10.62 weights
weights ULA-ABC[8], L=20 6.99 -15.32 15.32
NULA designed using NULA-FA (First design
11.5 -24.5 9.53 5.75 -23 18.48
parabolic relation[3] example in [4]), L=20
Proposed array 8.22 -15.8 12.15 Proposed array 5.138 -16.5 19.85

8
2019 1st Global Power, Energy and Communication Conference (IEEE GPECOM2019), June 12-15, 2019, Cappadocia, Turkey

the proposed one, but due to its small HPBW, the proposed
array achieves the highest directivity. In this paper, only
Bartlett-Hanning window has been considered to find the
weights of the array, but other standard window functions,
such as Blackman, Gaussian, Kaiser, etc., can also be applied
and investigated.
R EFERENCES
[1] C. A. Balanis, Antenna Theory and Design. John Wiley&Sons, 1997.
[2] H. L. V. Trees, Optimum Array Processing. Wiley, 2002.
[3] F. Enache, D. Deparateanu, and F. Popescu, “Optimization of non-
uniform linear antenna array with linear and parabolic parameters vari-
ations,” in 2017 International Conference on Optimization of Electrical
and Electronic Equipment (OPTIM) & 2017 Intl Aegean Conference on
Electrical Machines and Power Electronics (ACEMP), IEEE, may 2017.
[4] M. A. Zaman and M. A. Matin, “Nonuniformly spaced linear antenna
array design using firefly algorithm,” International Journal of Microwave
Science and Technology, vol. 2012, pp. 1–8, 2012.
[5] T. Chen, Y. Chen, Y. Jiao, and F. Zhang, “Synthesis of antenna array
using particle swarm optimization,” in 2005 Asia-Pacific Microwave
Conference Proceedings, IEEE.
[6] M. A. Zaman, S. A. Mamun, M. Gaffar, S. M. Choudhury, M. M. Alam,
and M. A. Matin, “Phased array synthesis using modified particle swarm
optimization,” Journal of Engineering Science and Technology Review,
vol. 4, pp. 68–73, apr 2011.
[7] M. Ridwan, M. Abdo, and E. Jorswieck, “Design of non-uniform
antenna arrays using genetic algorithm,” Journal of Wireless Networking
and Communications, vol. 2, pp. 7–10, aug 2012.
[8] A. Zaman, M. Gaffar, M. M. Alam, S. A. Mamun, and M. A. Matin,
“Synthesis of antenna arrays using artificial bee colony optimization
algorithm,” International Journal of Microwave and Optical Technology,
vol. 6, pp. 234–241, 07 2011.
[9] H. Unz, “Linear arrays with arbitrarily distributed elements,” IRE
Transactions on Antennas and Propagation, vol. 8, pp. 222–223, Mar.
1960.
[10] Y. H. Ha and J. A. Pearce, “A new window and comparison to
standard windows,” IEEE Transactions on Acoustics, Speech, and Signal
Processing, vol. Vol. 37, pp. 298–301, Feb. 1989.
[11] W. L. Stutzman and G. A. Thiele, Antenna Theory and Design. John
Wiley Sons, 1998.

You might also like