3038
16, 2017
Interference Suppression of ULA Antennas by
Phase-Only Control Using Bat Algorithm
Tong Van Luyen and Truong Vu Bang Giang
Abstract—This letter proposes an adaptive Bat algorithm (BA)-
based beamformer for pattern nulling of half-wave dipole uni-
formly spaced linear array (DULA), of which the imposed nulls
have been placed at directions of interferences. This pattern nulling
is obtained by controlling only the phase of each array element. In
order to verify the proposal, several scenarios of DULA pattern
with the preset nulls have been performed and compared with
those of genetic algorithm (GA) and accelerated particle swarm
optimization (APSO) considering mutual coupling. The proposed
beamformer has shown the ability to suppress sidelobes, to main-
tain predefined beamwidth, to place precisely single, multiple, and
broad nulls at an arbitrary direction of interferences. Further-
more, the beamformer is much faster and more effective in terms
of null-steering and sidelobe suppression in pattern synthesis than
GA and APSO based ones.
Index Terms—Array pattern synthesis, Bat algorithm (BA),
beamformer, interference suppression, null-steering, uniformly
spaced linear array (ULA).
I. INTRODUCTION
DAPTIVE beamformers are widely applied in radar,
A sonar, and communication systems to enhance the perfor-
mance by increasing the efficiency of radio spectrum utilization,
interference suppression, and power saving. Beamformers are
able to produce appropriate weights for array antennas to obtain
the required pattern [1].
Today, the prevalence of wireless devices causing serious pol-
lution in the electromagnetic propagation environment, smart
antennas with null-steering capabilities emerge as a promising
solution for suppressing interferences in wireless communica-
tions, and radar applications.
Some nulling methods, such as controlling the amplitude-
only, the phase-only, the position-only, the complex weights
(both the amplitude and the phase), and array thinning, which
have their own advantages and limitations, have been widely
studied and implemented [1], [2].
Among those, array thinning does not require digital beam-
forming. It takes advantages of adaptive algorithm to make ar-
ray element active or inactive. It is able to apply well for large
Manuscript received July 27, 2017; accepted September 28, 2017. Date of
publication October 5, 2017; date of current version November 8, 2017. This
work was supported in part by Vietnam National University, Hanoi, Vietnam,
under Project QG. 16.27. (Corresponding author: Truong Vu Bang Giang.)
T. Vu Bang Giang is with the Vietnam National University, Hanoi 10000,
Vietnam (e-mail: giangtvb@vnu.edu.vn).
T. Van Luyen is with the Faculty of Electronic Engineering, Hanoi University
of Industry, Hanoi 100000, Vietnam (e-mail: luyen.tv@ haui.edu.vn).
Color versions of one or more of the figures in this letter are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/LAWP.2017.2759318
1536-1225 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL.
arrays but not for small arrays [3]. The complex weights method
has been considered the most flexible and efficient, but the most
complicated and expensive, since each array element must have
controllers, phase shifters and attenuators [4]. The position-only
method [5] requires a mechanical driving system such as ser-
vomotors to adjust the array element position, which makes the
system more complicated, and accuracy control more difficult.
The amplitude-only control [6] is a simple method as it only
changes the amplitude excited at each element. Specifically, in
the case of even number of elements, symmetrical at the center of
the array, the number of attenuators and the computational time
will be reduced by a half. Applying this control method, in our
previous work [7], BA-based beamformer for adaptive steering
nulls of ULA antennas pattern has been proposed and imple-
mented successfully. However, [6], [7] have a disadvantage that
the nulling points have been simultaneously and symmetrically
placed at both sides of the main beam. Phase-only pattern nulling
is less complicated, and more attractive for phased arrays, since
the required controls are available at no extra cost. Besides, it is
easy to control the main beam direction by adjusting the phase
weights [8], [9].
Recently, optimization techniques have been commonly ap-
plied in beamforming for antenna array pattern synthesis includ-
ing null-steering. The classical optimization techniques used for
the array pattern synthesis are likely to be stuck in local minima
if the initial guesses are not reasonably close to the final solu-
tion. Most of the classical optimization techniques and analyti-
cal approaches suffer from a lack of flexible solutions to a given
antenna pattern synthesis problem. To overcome this, various
evolutionary optimization algorithms based on computational
intelligence approaches have been developed. For instance, ge-
netic algorithm (GA), ant colony optimization, particle swarm
optimization (PSO), differential evolution, and clonal selection
have been proved to be better and more flexible than the classi-
cal techniques [4]–[8], [10]–[12]. These techniques have been
proposed and implemented with their own benefits and limita-
tions in the array pattern synthesis. For array pattern synthesis,
GA and PSO have been widely implemented [2], [10]–[12].
BA is an evolutionary computation technique based on the
Bat behavior of using echolocation to detect prey, avoid obsta-
cles, and locate their roosting crevices in the dark. It has been
successfully used to solve various kinds of engineering prob-
lems [13], [14]. BA has been proved to be better than PSO and
GA optimization in terms of convergence, robustness, and pre-
cision [13]. This algorithm has been applied for the first time for
beamforming in [15]. Yao and Lu [15] showed that the BA is a
promising optimization tool for adaptive beamforming regard-
ing computation time. Still, this letter was just in preliminary
phase, and thus, it lacked adequate analysis on the application
of BA in beamforming.
VAN LUYEN AND VU BANG GIANG: INTERFERENCE SUPPRESSION OF ULA ANTENNAS BY PHASE-ONLY CONTROL
In this letter, a beamformer based on BA is proposed for
uniformly spaced linear array (DULA) to adaptively suppress
interferences with or without mutual coupling. In our proposal,
the phase of excitation for each array element is the only control-
ling parameter, and the main aim is to synthesize array patterns
with nulls imposed on directions of interferences. The proposed
beamformer has been verified in five scenarios and compared
with accelerated particle swarm optimization (APSO) and GA.
The results show that the beamformer operates well to suppress
sidelobes, steer the nulls to interference directions, with more
efficiency than those of APSO and GA.
II. PROBLEM FORMULATION
In our study, the DULA antenna of 2N half-wave-dipole
elements is considered. The array is positioned symmetrically.
In the ideal case (without mutual coupling) the array factor can
be expressed as

N
jndksin(θ )
AF (θ) = ωn e (1)
n =−N
where ωn = an ej δ n , {n = (−N, . . . , −2, −1, 1, 2, . . . , N }, is
the complex weight of nth array element; k = (2π)/λ is the
wave number; λ is wave length; d is the distance between adja-
cent elements.
In order to gain faster convergence, minimum weight pertur-
bation phase-only null synthesis requires odd phase shift [16]
(δ−n = −δn ). Therefore, an antisymmetrical pattern through
the main beam direction (θ = 0) is obtained. When a−n = an
and δ−n = −δn , the array factor in (1) can be rewritten as

N rate ri; and loudness Ai. Each Bat is corresponding to a
AF (θ) = 2 an cos (ndksin (θ) + δn ) . (2)
n =1
Additionally, inspired by the objective function implemented
in [6] and [18], a new objective function F , which meets the
requirements in (1), has been developed as follows:
   
N (θ) M i=1 |AFo (θi )| ,
2
for θ = θi (3.1)
F = 90 ◦  
θ =−90 ◦ |AFo (θ)−AFd (θ)| , elsewhere
2
(3.2)
where N (θ) is a parameter, which is experimentally defined
during the investigation of the proposal. In this letter, N (θ) has
been chosen as 10 000; (3.1) is for placing the null points, in
which M is the maximum number of interferences; (3.2) is used
to reduce the sidelobe level (SLL) and to keep the beamwidth
of the main beam within a maximum allowable change; AF0
and AFd are the optimized array factor obtained by using an
optimization algorithm, which will be BA in this letter, and the
desired array factor, respectively; θi are the angles of null points.
In fact, when array elements are close enough to each other,
array elements transfer electromagnetic energy from one to an-
other, causing mutual coupling effects. It has been shown in
[19] that mutual coupling between array elements plays an im-
portant role in an applied adaptive array as it directly affects
the efficiency and the performance of an adaptive array such
as sidelobes and directions of nulls. This is very important to
null-steering beamforming.
The mutual impedance matrix of the investigated DULA
has been computed by the induced electromotive force method
3039
presented in [19]

73.1291 + 42.5446j, if m = n
Zm n = 30[2Ci (u0 ) − Ci (u1 ) − Ci (u2 )]
− 30j[2Si (u0 ) − Si (u1 ) − Si (u2 )],if m = n
(3)
where dh is thedistance between the dipole elements; u 0 =
2πdh ; u1 = 2π d2h + 0.25 + π u2 = 2π d2h + 0.25 − π. Ci
and Si are the cosine and sine integrals given by
cos (x) sin (x)
Ci (u) = ∫∞
u
dx; Si (u) = ∫0u dx.
x x
If the mutual coupling is taken into account, the input current
I can be calculated from the excitation voltages V and will be
ZI = V (4)
where Z is defined by (4).
III. PROPOSAL OF THE BEAMFORMER
Based on studies in [20] and [21], a null-steering beamformer
using phase-only control for interference suppressing is being
proposed and its flowchart is being presented in Fig. 1. Operation
of the beamformer is described as follows.
Initializations (I)
1) Setting the input data such as: number of array elements
(N ), direction of arrival (DOA) of interferences; number
of iteration (i); maximum number of iterations (Max I);
and the termination criterion (Threshold).
2) Initializing Bat population in which parameters of each
Bat are: location xi; velocity vi; pulse frequency f i; pulse
potential solution.
Finding the Best Solution (F)
The beamformer consecutively calculates and searches for
the current best solution based on BA. The operation is
finished when the termination criterion or maximum num-
ber of iterations is satisfied. Then, the final best solution is
obtained.
Building Array Element Weights (B)
From the best solution, the beamformer calculates the cor-
responding weight excited at each element of the DULA
antenna. These parameters will be used for null-steering.
IV. NUMERICAL RESULTS
To demonstrate the capability and the flexibility of the pro-
posal for pattern nulling, five scenarios will be investigated. It
is well known that the Chebyshev array weights distribution
produces the optimum pattern in terms of a tradeoff between
the sidelobe level and the first-null beamwidth of main beam
for equally spaced arrays [22]. Therefore, in this letter, an array
factor of the Chebyshev array has been chosen as a desired one
to control SLL and beamwidth of the main beam. A Chebyshev
array pattern with −30 dB SLL for λ/2 inter-element spacing
20 half-wave-dipole elements has been utilized as the initial
pattern. The step size of the theta angle in (3) is 1°.
Initial parameters for BA have been chosen for all investi-
gation scenarios as: the population size (pop) is 1000 and the
3040
Fig. 1. Flowchart of the proposed beamformer.
number of iterations (ite) is 5 (except for scenario 1); step size
of a random walk is 0.01; boundary frequency values: fm in = 0
and fm ax = 1.
Parameters for GA: pop: 1000; ite: 5; selection rate: 50%;
mutation rate: 20%; crossover type: roulette wheel and crossover
rate: 90%; elitism: 1.
Due to the simplicity and improved convergence speed, APSO
by Yang [14] has been considered a good choice, and the pa-
rameters have been chosen as α = 0.2 and β = 0.5 [14], [23].
To show the ability of BA utilization in our proposed beam-
former for interference suppression, five scenarios have been
built. Scenario 1 named Convergence Rate is the first step to
evaluate the operation of the proposal beamformer by compar-
ing the convergence rate of the objective functions based on
BA, APSO, and GA (see Section IV-A). Scenarios 2–4 are for
investigating and comparing the capability of null-steering of
the proposal (see Sections IV-B–IV-D) with APSO and GA-
based beamformers in the case of no mutual coupling (ideal).
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 16, 2017
Fig. 2. Objective function comparisons of BA, PSO, and GA.
Fig. 3. Optimized pattern with a single null at 14°.
Fig. 4. Optimized pattern with three nulls at −48°, 20°, and 40°.
Furthermore, the proposal has been investigated in the presence
of mutual coupling in scenario 5. As a result, simulation results
of all scenarios have been presented in Figs. 2–6 and Table I, in
which the results are average values of Monte Carlo simulations
with 1000 times for scenario 1, and 100 times for the others.
A. Convergence Rate
In scenario 1, first, convergence ability of our proposed beam-
former has been investigated and compared with APSO- and
GA-based ones in the case of obtaining the desired optimization
pattern as the Chebyshev array pattern with −30 dB SLL. To
do that, the initial population has been randomly generated; pop
has been chosen as 1000; ite is 50; and search value xi is in the
range of −π to π. The simulation results in Fig. 2 show that the
BA-based beamformer converges much faster than the APSO
and GA ones.
VAN LUYEN AND VU BANG GIANG: INTERFERENCE SUPPRESSION OF ULA ANTENNAS BY PHASE-ONLY CONTROL 3041

−24.48 dB. The null depth level (NDL) at 14° is −87.15 dB.
Fig. 3 indicates that the single null-pattern optimized by the BA
is better than that of the APSO and GA in terms of NDL at the
desired null point.

C. Multiple Nulls
In scenario 3, the proposal will be used to separately set
multiple nulls at −48°, 20°, and 40°, corresponding to the peaks
of three sidelobes next to the main beam of the Chebyshev array
pattern. As shown in Fig. 4, the patterns with multiple nulls at
predefined locations have been obtained. All NDLs are deeper
Fig. 5. Optimized pattern with a broad null from 30° to 40°.
than −73 dB, all SLLs are lower than −24 dB, and HPBW
roughly equals to that of the Chebysev pattern. The BA pattern
shows advantages over the APSO and GA ones in terms of NDL.

D. Broad Null
In interference suppression applications, if the DOAs of in-
terferences vary slightly with time or cannot be known exactly,
or a null is continuously steered to obtain an appropriate signal-
to-noise ratio, a broad null is required. So to show the capability
of broad interference suppression, in scenario 4, the pattern with
an imposed broad null at the target sector of [30°, 40°] has been
obtained and illustrated in Fig. 5. A broad null (minimum NDL
Fig. 6. Optimized pattern (nulls: −48°, 20°, 40°) with mutual coupling. < −52 dB) on the BA patterns at that target sector has been ob-
tained. The beamwidth is nearly unchanged and the maximum
SLL is −20.69 dB. The results show that the BA pattern is better
TABLE I than APSO and GA ones in terms of NDL.
NDL AND MAXIMUM SLL OF THE PATTERNS SHOWN IN FIGS. 3–6

Fig. Parameters GA (dB) PSO (dB) BA (dB) E. Patterns with Mutual Coupling
Ideal Ideal Ideal MC (IV.E) Scenario 5 has been carried out to investigate the ability of the
3 NDL at 14° −66.43 −71.55 −87.15 −66.00
proposal in the presence of mutual coupling. This mutual cou-
Maximum SLL −24.93 −24.53 −24.48 −24.52 pling effect has been modeled based on the mutual impedance
4 NDL at: −48° −56.59 −63.26 −73.24 −49.73 matrix in (4). To do that, the proposed beamformer has been
20° −56.07 −62.29 −73.48 −54.17 applied for pattern nulling with single null, multiple nulls, and
40° −56.60 −60.68 −74.68 −51.63 broad null similarly to as was done in Sections IV-B, IV-C, and
Maximum SLL −24.23 −24.04 −24.35 −25.51
5 Maximum NDL −56.20 −62.51 −69.06 −51.55 IV-D. For detailed demonstration, the simulation results in the
Minimum NDL −48.40 −47.56 −52.00 −40.01 case of multiple nulls have been presented in Fig. 6, while the
Maximum SLL −23.48 −24.00 −20.69 −20.64 others have been shown in Table I. It is seen that null points
have been exactly placed at predefined locations but with lower
NDLs.
The NDL and maximum SLL of the patterns of all scenarios
B. Single Null have been summarized in Table I for more detailed reference of
In scenario 2, the optimized pattern with a single null has been data in Figs. 3–6.
demonstrated. This null can be arbitrarily set at any angle, which
is chosen at the peak of the second sidelobe (14°) in this test case.
The population has been initialized as Chebyshev array weights V. CONCLUSION
with −30 dB SLL. In [2], Haupt found that the phase had more In this letter, a BA-based beamformer has been proposed
effect on the main beam than the amplitude; and adaptive nulling for DULA antennas pattern nulling to suppress interferences.
with little perturbation of phase would not result in significant The pattern nulling capability of the proposal has been veri-
degradation to the main beam. Moreover, the smaller range of fied by five scenarios including operation speed, pattern nulling
variable phase weight is, the faster convergence can be obtained. with single, multiple, and broad nulls. The results show that the
Thus, in the simulation process for all investigated scenarios of above-mentioned nulls can be precisely imposed to arbitrary in-
pattern nulling, variable phase weights have been limited in the terference directions using our proposed beamformer while the
range of −0.5–0.5 rad. Fig. 3 presents optimized patterns with pattern maintains the HPBW and low SLL in both cases as with
single null obtained by BA, APSO, and GA. The BA pattern or without mutual coupling. Furthermore, compared with APSO
preserves almost all characteristics of the initial Chebyshev pat- and GA, our proposal is more efficient in terms of computation
tern such as half-power beamwidth (HPBW = 7.64◦ ) and SLL time, adaptive null-steering, and sidelobe suppression in array
(−30 dB) except for a few sidelobes with maximum SLL of pattern nulling synthesis.
3042
REFERENCES
[1] H. L. Van Trees, Optimum Array Processing: Part IV of Detection, Es-
timation, and Modulation Theory. New York, NY, USA: Wiley, 2002,
pp. 1–12.
[2] R. L. Haupt, Antenna Arrays: A Computational Approach. Hoboken, NJ,
USA: Wiley, 2010, pp. 156–176; 484–515.
[3] R. L. Haupt, “Adaptively thinned arrays,” IEEE Trans. Antennas Propag.,
vol. 63, no. 4, pp. 1626–1632, Apr. 2015.
[4] D. Karaboga, K. Guney, and A. Akdagli, “Antenna array pattern nulling
by controlling both the amplitude and the phase using modified touring
ant colony optimisation algorithm,” Int. J. Electron., vol. 91, pp. 241–251,
2004.
[5] K. Guney and M. Onay, “Bees algorithm for null synthesizing of linear
antenna arrays by controlling only the element positions,” Neural Netw.
World, vol. 16, pp. 153–169, 2007.
[6] K. Guney and M. Onay, “Amplitude-only pattern nulling of linear antenna
arrays with the use of bees algorithm,” Prog. Electromagn. Res., vol. 70,
pp. 21–36, 2007.
[7] T. V. Luyen and T. V. B. Giang, “Null-steering beamformer using BAT
algorithm,” J. Appl. Comput. Electromagn. Soc., submitted for publication.
[8] A. Akdagli and K. Guney, “Null steering of linear antenna arrays by phase
perturbations using modified tabu search algorithm,” J. Commun. Technol.
Electron., vol. 49, pp. 37–42, 2004.
[9] M. Mouhamadou, P. Vaudon, and M. Rammal, “Smart antenna array
patterns synthesis: Null steering and multi-user beamforming by phase
control,” Prog. Electromagn. Res., vol. 60, pp. 95–106, 2006.
[10] D. W. Boeringer and D. H. Werner, “Particle swarm optimization versus
genetic algorithms for phased array synthesis,” IEEE Trans. Antennas
Propag., vol. 52, no. 3, pp. 771–779, Mar. 2004.
[11] K. C. Lee and J. Y. Jhang, “Application of particle swarm algorithm to the
optimization of unequally spaced antenna arrays,” J. Electromagn. Waves
Appl., vol. 20, pp. 2001–2012, 2006.
[12] M. R. Khan and V. Tuzlukov, “Null steering beamforming for wireless
communication system using genetic algorithm,” in Proc. IEEE Int. Conf.
Microw. Technol. Comput. Electromagn., Beijing, 2011, pp. 289–292.
IEEE ANTENNAS AND WIRELESS PROPAGATION LETTERS, VOL. 16, 2017
[13] X. S. Yang, “A new metaheuristic bat-inspired algorithm,” in Nature In-
spired Cooperative Strategies for Optimization (NISCO 2010) (Studies in
Computational Intelligence, vol. 284), J. R. González et al., Ed., Berlin,
Germany: Springer-Verlag, 2010, pp. 65–74.
[14] X. S. Yang, Nature-Inspired Optimization Algorithms. London, U.K.:
Elsevier, 2014, pp. 141–154.
[15] Q. Yao and Y. Lu, “Efficient beamforming using bat algorithm,” in Proc.
IEEE MTT-S Int. Conf. Numer. Electromagn. Multiphys. Model. Optim.,
Beijing, 2016, pp. 1–2.
[16] R. A. Shore, “A proof of the odd-symmetry of the phases for minimum
weight perturbation phase-only null synthesis,” IEEE Trans. Antennas
Propag., vol. AP-32, no. 5. pp. 528–530, May 1984.
[17] H. Steyskal, R. A. Shore, and R. L. Haupt, “Methods for null control
and their effects on the radiation pattern,” IEEE Trans. Antennas Propag.,
vol. AP-34, no. 3, pp. 404–409, Mar. 1986.
[18] S. U. Khan, I. M. Qureshi, F. Zaman, and A. Naveed, “Null placement and
side lobe suppression in failed array using symmetrical element failure
technique and hybrid heuristic computation,” Prog. Electromagn. Res. B,
vol. 52, pp. 65–184, 2013.
[19] C. A. Balanis, Antena Theory: Analysis and Design, 4th ed. Hoboken, NJ,
USA: Wiley, 2016, pp. 431–480.
[20] J. Litva and T. K.-Y. Lo, Digital Beamforming in Wireless Communica-
tions. London, U.K.: Artech House, 1996, pp. 13–55.
[21] T. V. Luyen and T. V. B. Giang, “Design of LMS based adaptive beam-
former for ULA antennas,” VNU J. Sci., Comput. Sci. Commun. Eng.,
vol. 32, no. 3, pp. 72–79, 2016.
[22] C. L. Dolph, “A current distribution for broadside arrays which optimizes
the relationship between beam width and side-lobe level,” Proc. IRE,
vol. 34, no. 6, pp. 335–348, Jun. 1946.
[23] P. V. Florence and G. S. N. Raju, “Synthesis of linear antenna arrays using
accelerated particle swarm optimization algorithm,” Int. J. Comput. Appl.,
vol. 103, no. 3, pp. 43–49, 2014.