IETE Journal of Research

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/tijr20

Multi-objective Naked Mole-Rat Algorithm for UWB

Antenna Design

Gurdeep Singh & Urvinder Singh

To cite this article: Gurdeep Singh & Urvinder Singh (2021): Multi-objective Naked
Mole-Rat Algorithm for UWB Antenna Design, IETE Journal of Research, DOI:
10.1080/03772063.2021.1912657

To link to this article: https://doi.org/10.1080/03772063.2021.1912657

Published online: 22 Apr 2021.

Submit your article to this journal

Article views: 5

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tijr20
IETE JOURNAL OF RESEARCH
https://doi.org/10.1080/03772063.2021.1912657

Multi-objective Naked Mole-Rat Algorithm for UWB Antenna Design

Gurdeep Singh and Urvinder Singh
Department of Electronics and Communication Engineering, Thapar Institute of Engineering and Technology, Deemed University, Patiala
147004, India

ABSTRACT KEYWORDS
This paper presents a novel multi-objective naked mole-rat (MONMR) algorithm to optimize two CST-MATLAB integration
printed monopole patch antenna structures for ultrawideband (UWB) applications. The design (CMI); Electromagnetics;
includes basic-ultrawideband and triple band-notch characteristic antennas. The performance of Metaheuristic;
MONMR is evaluated for two objectives: minimize pass-band signal reflection and maximize the gain Multi-objective optimization;
Naked mole-rat algorithm;
of the antenna. In resultant, MONMR provided multiple optimum antenna designs instead of a sin- Triple band-notched; UWB
gle antenna design by mapping solutions to the Pareto-front (PF) boundary in the objective space. antenna
The simulated results matched with the experimental results and hence MONMR has proved to be
an effective approach in the field of electromagnetics. Moreover, the obtained PF designs are better
than the reported UWB antennas in terms of the gain and reflection coefficient.

1. INTRODUCTION as multi-objective salp swarm algorithm [12–14], multi-

The design of an antenna in the wireless communication objective cuckoo search (MOCS) [15–17], and multi-
system is a challenging optimization problem because objective evolutionary algorithm (MOEA) [18]. These
it consists of non-linear/smooth and non-differentiable methods are capable enough to handle multiple con-
multiple conflicting constraints. For this, classical opti- straints simultaneously and to generate the entire PF
mization methods are not a good choice as their per- boundary in a single run but these have high compu-
formance heavily relies on the initial guess and if it tational cost (thousands of objective evaluations). The
is not available there may occur local minima stagna- MOO methods have several advantages associated with
tion problem. So, these techniques have failed to reach them like the ability to handle multiple constraints, fast
a global optimal solution in complex structure design solution approaching strategy, improved exploration and
problems. Therefore, many nature-inspired optimization exploitation search capability. Hence, these have been
algorithms have been introduced to solve the electromag- used to solve different domain problems such as cog-
netic (EM) antenna problems using single-objective opti- nitive radio [19], wireless sensor network [20], Inter-
mization (SOO) [1–2], hybrid [3], and multi-objective net of things (IoT) [21], cloud computing [22], fea-
optimization (MOO) [4–6] algorithms. Again, SOO algo- ture selection [12,23], automobile [16,24], electrical [25],
rithms are not the appropriate technique for best antenna military [26–27], and weather forecasting [28] applica-
optimization because antenna design is inherently a tions. In recent years, the MOO algorithms have been
multi-objective problem as it usually consists of several widely used to investigate ultrawideband (UWB) antenna
contrary objectives, which are required to be handled at [29–31], Multiple-In Multiple-Out (MIMO) [32–33],
the same time. These objectives include size reduction fragment-type antenna (FTA) [34], and linear antenna
[7], gain maximization [8], improve electrical perfor- array (LAA) [17] in terms to improve the data rate
mance in terms of reflection-coefficient (S11 ) and voltage (channel capacity), reduce signal reflection, increase gain
standing wave ratio (VSWR) [9], and maneuver the radi- with a small foot-print, and reduce the cost of wireless
ation pattern [10]. Hence, to identify the best trade-off communications systems. Since 2002, the UWB antenna
solutions, generally called Pareto-front (PF) set, MOO designing approach is very popular because of its high-
algorithms are required [11]. speed data rate (approximately 1 Gb/s), huge bandwidth
(approximately 7.5 GHz band), low power consumption,
In the last decade, the most popular algorithms for find- and short-range communication. Hence, Federal Com-
ing optimal PF boundary of various engineering MOO munication Commission (FCC) has recommended a
problems include population-based metaheuristics such 3.1–10.6 GHz as an unlicensed frequency band for UWB

© 2021 IETE
2 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN

applications. Some specific UWB systems suffer from 2.1.1 Initialization

strong narrowband EM-interference of unlicensed ISM- It is the first step almost in every algorithm, generates
band wireless communication devices such as Wi-Max a uniformly distributed random population of n NMRs,
(3.3–3.6 GHz), WLAN (5.15–5.85 GHz), X-band of satel- each NMR ranges from [1, 2 . . . n] is a D dimensional vec-
lite communication (7.25–8.4 GHz) [35]. Therefore, for tor. D is the number of variables or parameters of a given
UWB antennas in-built band-notch (BN) characteristics problem.
are required [36–39].
NMRi,j = NMRmin,j + U(0, 1)
This work proposes a novel multi-objective naked mole- × (NMRmax,j − NMRmin,j ) (1)
rat (MONMR) optimization algorithm to solve EM ultra-
wideband antenna design problems, which are classified where i varies from [1, 2, 3 . . . n], j from [1, 2, 3 . . . D],
as basic UWB (B-UWB) and triple band-notched UWB NMRi,j is ith solution for the jth dimension, Smin,j and
(TBN-UWB) antenna designs. The number of opti- Smax,j are lower and upper bounds of the problem under
mization variables for B-UWB and TBN-UWB anten- consideration and U(0,1) is a uniform random number
nas is 9 and 13, respectively. In order to reduce the in the range of [0,1].
computational cost, the proposed antennas have been
designed with low fidelity models, which provide only 2.1.2 Worker Phase
0.8–2 min/simulation. After the simulation, the TBN- It is the second stage for which each worker’s fitness is cal-
UWB antenna is validated through experimental testing, culated. It is also called an exploration phase, each worker
which shows good agreement between simulated and is highly explored to maximize their performance with a
experimental results. tendency to become a breeder. The new worker solution
is calculated based on Equation (2).
The next section presents the basic NMR, MOO and
the proposed MONMR algorithm. Section 3 describes Wit+1 = Wit + λ(Wjt − Wkt ) (2)
two UWB antenna geometries along with their fitness
functions, PF representations, and convergence curves. Here, Wit signifies ith worker in the jth iteration, Wit+1
In Section 4, the experimental performance of the pro- is the new worker solution, WJt and Wkt are two random
totype is examined. The paper has been concluded in solutions, λ is the Levy flight based scaling factor ranges
Section 5. from [0,1].

2.1.3 Breeder Phase

2. PROPOSED ALGORITHM It is the third stage for which all the breeders also updated
themselves to reach the final stage of the queen’s mat-
2.1 Basic Naked Mole-Rat (NMR) Algorithm ing. The breeder NMRs are updated based upon breed-
The NMR is a newly introduced swarm intelligent meta- ing probability (bp), which helps to push back the low-
heuristic algorithm [40], which has been designed using performing breeders.
worker and breeders capacity to balance both exploration
and exploitation abilities for achieving better optimiza- bt+1
i = (1 − λ)bti + λ(d − bti ) (3)
tion of any problem under test. It is based on some key
where bti corresponds to the ith breeder in tth iteration,
factors as:
bt+1
i is the new breeder solution, λ handles the breeder’s
mating frequency, and bp kept 0.5 as the initial value.
• The algorithm mimics the mating patterns in NMRs
of mostly living in a group of 295 members with an
average of 70–80 members in each group. 2.2 Multi-objective Optimization (MOO)
• Each group leads by a female, which divides the pop-
MOO problem defines several conflicting objectives in
ulation into workers and breeders.
terms of min-min, min-max, and max-max strategies
• The best worker from the workers pool always
with some optional constraints to be satisfied. The gen-
wants to become a breeder and breeder with poor-
eralized expression of the MOO problem is defined as:
performance has pushed back to the worker’s pool.
• The high-performing breeder among all breeder class Minimize Fj (X), j = [1, 2 . . . M] (4)
finally mates with the queen.
Subject to hk (X) ≤ 0, k = [1, 2 . . . N] (5)
The algorithm steps are explained as follows: XLb ≤ X ≤ XUb ; X = [x1 , x2 . . . xD ]
 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN 3

where X is the design variables vector [1, D], D is the problem under consideration.
dimension size, M and N are the numbers of objectives
and constraints. XLb and XUb are the lower and upper Si,j = Smin,j + rand × (Smax,j − Smin,j ) (6)
bound limits.
Here, i and j are the [1:n] population and [1:D] dimen-
sion vectors, respectively. Si,j is an ith solution for the jth
Generally, an SOO has only one optimal solution whereas
dimension, Smin,j and Smax,j are lower and upper bounds
a MOO problem has a set of optimal solutions known
as the Pareto optimal set. The solutions which are not of the problem to be tested. rand is a uniform random
dominated by others, located at the outer edge of the number that lies between [0,1]. Here, every solution pro-
objective space, called the Pareto-front. All the solutions vides M objective fitness values.
are optimal for one/more objective/s, but none of them
is optimal for all the objectives. A PF boundary selection 2.3.2 Worker Phase
is based on the problem designer’s decision, expressed in The worker phase of MONMR is the same as that of
Equation (4). the basic NMR algorithm and has been discussed in the
previous section.
The MOO algorithm aims to converge maximum solu-
tions over the entire PF and extract most of them towards 2.3.3 Breeder Phase
the true PF with maximum constraints satisfaction. The breeder phase of the MONMR is the same as that of
the basic NMR algorithm and only Equation 8 has been
introduced in the proposed MONMR algorithm which
2.3 Proposed Multi-objective NMR (MONMR) is based on the SA-IW strategy. After investigation, bp is
Algorithm taken as 0.05 for getting better optimization results.
Most of the engineering design problems are typically
Bt+1
i = (1 − λ)Bti + λ(best − Bti ) (7)
multi-objective including non-smooth/non-linear con-
straints. Though it needed MOO algorithms, which are where Bti corresponds to the ith breeder in tth iteration,
different from SOO. The number of function evalua- Bt+1 is the new breeder solution, λ controls the breeder
i
tions and computation effort may increase significantly mating frequency, which is now based on the SA-IW
in MOO. strategy.

For MONMR with M objectives, there are three modifi- λ = αmin + [(αmax − αmin ) × pk−1 ] (8)
cations in NMR rules can be described as:
where αmin and αmax constant values are taken as 0.5 and
• Each NMR group, a combination of worker and breed- 0.9, respectively. p is 0.95 and k is any random number
ers, puts n number of solutions. n corresponds to the [0,1].
solution of the Mth objective.
• The best worker or breeder is selected among M objec- 2.3.4 Sorting Process for Pareto Optimal Set
tives instead of single-objective in NMR. A solution vector s = (s1 , . . . , sn )T ∈ F is said to domi-
• Some poor-performing breeders have been eliminated nate another vector r = (s1 , . . . , sn )T if and only if si ≤ ri
with a probability bp. The new breeder solution is for ∀i ∈ [1 . . . M] (in particular, all solutions of s are less
made according to similarities/difference to old solu- than or equal to r), ∃i ∈ [1 . . . M] : si < ri indicates at
tions. For this, a simulated annealing-inertia weight least one component is smaller. Therefore, a solution x∗ ∈
(SA-IW) based mating factor has been introduced F is selected as a non-dominated solution if no solution
[41]. A small random mixing can be used to increase is better than it or dominates it.
diversity if needed.
The PF boundary of a MOO problem can be defined as a
Every multi-objective algorithm has a target to achieve PF set of non-dominated solutions as described below:
optimal set, which is defined by the “domination” concept
as described in the following subsections. PF = {s ∈ S|s ∈ S : s ≺ s} (9)

or in terms of the PF set in the objective space,

2.3.1 Initialization
In this stage, all the n NMRs firstly initialize with a uni- PF∗ = {x ∈ F | x ∈ F : f (x ) ≺ f (x)} (10)
formly distributed random population, each NMR is a D
dimensional vector. D is the number of variables of any where f = (f1 , . . . , fM )T , T is the current iteration.
4 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN

This is for the minimization problem. If we want to find end for

the best PF set for the maximization MOO problem, then
notation ≤ is to be changed with ≥ symbol. To achieve combine old and new solutions with population 2∗n
the best representation of solutions on PF, strong explo-
ration search is required. There are different techniques Sort the 2n population. Keep best n non-dominated
to obtain a diverse range of solutions. In this case, Levy sets
flights assure the good diversity of solutions. For local
search, SA-IW based exploitation mating factor is used. update the overall best d

All the above-discussed modifications are added inde- update current Pareto optimal solutions
pendently without compromising the basic properties of
NMR, to formulate the new algorithm as described in the end until
Pseudocode shown in Algorithm 1.
Post processing results and visualization, save best (d,n)
Algorithm 1: Pseudocode of the proposed MONMR
Algorithm End

Begin:
3. ANTENNA DESIGN AND SIMULATION
Inputs: Initialize objective function with Eq. (4) RESULTS
To check the effectiveness and accuracy of the proposed
NMRs population: n, maxiter, d, Lb, Ub, bp,
MONMR, it is applied to the design of EM antenna
problems. In the present case, two different antenna
Breeders: B = n/5, Workers: W = n-B
structures are considered for optimization: basic UWB
with 9 optimization parameters and triple band-notched
Generate an initial population of n NMRs with Eq. (6)
UWB antenna with 13 variables. After that, the antenna
was fabricated, tested, and compared to some recently
Sort the initialized population
reported BN techniques. All the simulations have been
performed on window 7 PC with a 64-bit operating
while (t ≤ maxiter)
system, 32 GB RAM and Intel(R) Core(TM) i7 CPU
3.20 GHz machine including the software versions of
for i = 1: W
MATLAB R2016a and CST studio suite 2016.
perform worker phase: using updated Eq. (2)
3.1 Basic-Ultrawideband (B-UWB) Antenna Design
evaluate
In the case of the first antenna design, MONMR is intro-
update new solution and current best duced to obtain UWB characteristics (3.1–10.6 GHz)
including two objectives: minimize reflection to ensure
end for S11 ≤ −10 dB and maximize the gain. The correspond-
ing fitness functions are formulated in Equations (11)
for i = 1:B and (12). The antenna initial structure dimensions
are taken as W × L of 20 mm × 28 mm, consists of
if (rand > bp) ground structure on bottom plane 20 mm × 11 mm,
microstrip patch 20 mm × 15 mm, microstrip feed line
perform breeder phase: using updated Eq. (7) 3.316 mm × 13 mm, as shown in Figure 1. Both the top
and bottom metal layers (35 μm) are separated by a flame
evaluate retardant-4 (FR-4) dielectric substrate of the relative per-
mittivity (ε r ) 4.4 and a height of 1.6 mm. The parameters
update new solution and current best for MONMR optimization are selected as the popula-
tion size of 30, iterations 100, bp 0.05, dimension size of
end if 9 antenna parameters with the lower and upper bound
limits, as listed in Table 1.
 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN 5

Table 1: Optimization parameters of the B-UWB antenna

Optimization parameters Bound limits Optimized Ant-1 Optimized Ant-2 Optimized Ant-3
L1 05–12 8.76 8.72 9.06
L3 00–02 0.53 0.48 0.30
L4 10–15 10 10 10
W1 1.5–2.5 1.50 1.50 1.55
W2 0.8–1.3 0.80 0.80 0.80
W3 2.5–4.5 4.49 4.50 4.50
Lg 10–17 10.61 10.59 10.41
d1 02–06 2.67 2.73 3.00
d2 02–06 2.04 2.04 2.04
L2 (adaptive) L − (L1 + L4 + 2 × L3 ) 8.18 8.32 8.34
F 1 (S11 ) 0 0 0.41
F 2 (avg. gain in dBi) 2.91 2.94 2.95

Figure 1: The optimized proposed B-UWB antenna design: front,

back, and side view

The antenna design optimization has been performed by

creating an optimization-interface (OI) of commercially Figure 2: B-UWB antenna PF based on the MONMR, MOCS
available finite integration technique (FIT) based EM algorithms
simulator (CST) and the MATLAB. In the CMI interface,
all the MONMR updated antenna solutions are simulated
in CST, which are directly controlled by MATLAB com- illustrated in Table 1. Figure 3 gives the fitness functions
mand window with the help of visual basic script (VBS) F 1 and F 2 convergence curves of the MONMR com-
macro codes. After that, it analyzes the electrical output pared to the MOCS algorithm. It can be seen that the
performance to calculate M objective functions, which MONMR has faster convergence among MOCS. Finally,
should be further utilized in MONMR to update and the S11 characteristics of the best antenna design (Ant-2),
generate new solutions. is shown in Figure 4.
 10.6GHz 
1  Ant-2 provided a minimum reflection of zero with an
F1 = min Sx (f ) (11)
Z average gain of 2.94 dBi means that it worked effi-
3.1GHz
ciently (|S11 | ≤ −10 dB) on every frequency component
 of UWB among the other B-UWB antennas. There-
0 ; S11 ≤ −10dB
Sx (f ) = fore, none of the frequency components exists where
S11 + 10 ; S11 > −10dB
|S11 | > −10 dB. The peak gain of Ant-2 is denoted as
  3.82 dBi.
1 
10.6GHz
F2 = max G(f ) (12)
Z
3.1GHz
3.2 Triple Band-Notch (TBN-UWB) Antenna Design
where z is the number of frequency samples for UWB,
126 out of the total 201 frequency samples (2–14 GHz), The band-notch characteristics are required to avoid
Sx (f ) and G(f ) are the |S11 | and gain values, min/max the undesirable narrowband signal interference from the
signifies minimization/maximization problem. unlicensed ISM band wireless communication devices.
Here, in this case, MONMR is applied to optimize BN
After completing all the MONMR iterations with an exe- elements which are included on the top surface of B-
cution time of ≈ 56 h, the non-dominated solutions on UWB antenna to obtain the TBN characteristics. These
PF representation, are shown in Figure 2. Here, we can see BN elements consist: Inverted-U1 (∩1 ) and ∩2 slots
three different B-UWB antenna solutions on PF bound- are etched on a top patch-plane to reject the Wi-Max
ary for analysis and show the MONMR diversity, also (3.5 GHz) and ITU-8 (International telecommunication
6 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN

Optimal goals of this design are the same as that of

the previous design. The first objective was to minimize
reflection and secondly maximize the gain of TBN-UWB
antenna. The fitness functions are formulated in Equa-
tions (14)–(17), as below:

1  1 
3.2GHz 5GHz
1
F1A = Sy (f ) + Sy (f )
4 P Q
2.7GHz 3.71GHz

1  1 
7.79GHz 10.6GHz
+ Sy (f ) + Sy (f ) (13)
R S
6.1GHz 8.81GHz
 3.7GHz
1 1  1 
6GHz
F1B = Sz (f ) + Sz (f )
3 L M
3.21GHz 5.1GHz

1 
8.8GHz
+ Sz (f ) (14)
N
7.8GHz

F1 = min(F1A − F1B ) (15)

Figure 3: Convergence curves of B-UWB antenna F 1 and F 2 using 
MONMR and MOCS −10 ; S11 ≤ −10dB
Sy (f ) =
S11 ; S11 > −10dB

S11 ; S11 ≤ −10dB
Sz (f ) =
−10 ; S11 > −10dB
  3.2GHz
1 1  1 
5GHz
F2 = max G(f ) + G(f )
4 P Q
2.7GHz 3.7GHz

1  1 
7.79GHz 10.6GHz
+ G(f ) + G(f ) (16)
R S
6.1GHz 8.81GHz

Figure 4: Simulated |S11 | of the three optimal B-UWB antennas

Here, P is 9, Q 22, R 30, S 31 are the number of fre-
quency samples of antenna pass-bands (|S11 | ≤ −10 dB)
whereas L 8, M 16 and N 17 are the frequency samples of
antenna stop bands (|S11 | > −10 dB). F 1A varies from {0,
−1 . . . −10} and F 1B {∞, . . . −11, −10}. Hence, the F 1
value should be required as zero. Sy (f ) and Sz (f ) define
the S11 values of antenna pass and stop bands, respec-
tively. F 2 represents the average realized gain among
Figure 5: Front view of the optimized TBN-UWB antenna UWB except for the notched bands. The simulation
structure was done with 201 frequency samples in the range of
2–14 GHz.

union band of 8 GHz) X-band and C-strip pair on both After completing the OI iterative process with the exe-
sides of microstrip-fed line is used to reject WLAN band cution time of ≈ 95 h, the non-dominated solutions on
(5.2/5.8 GHz) (Figure 5). PF boundary, are shown in Figure 6. Five diverse solu-
tions are marked as Ant-4 through Ant-8 on PF for
The MONMR parameters were selected as the population analyzing the MONMR diversity-the detail is listed in
size of 30, iterations 150, bp 0.05, and 13-dimensional Table 2. Ant-5 is selected as the best TBN-UWB antenna
variables of BN elements. All the variables are listed in in respective to “min-max” optimization problem among
Table 2 with their lower and upper bound limits. the other TBN-UWB antennas. Figure 7 illustrates F 1
 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN 7

Table 2: Optimization parameters of the TBN-UWB antenna

Opt. variable Bound limits Optimized Ant-4 Opt. Ant-5 Opt. Ant-6 Opt. Ant-7 Opt. Ant-8
x1 14–19 17.35 15.78 14.64 15.65 16.98
x2 07–14 9.00 8.59 7.00 7.51 7.70
x3 02–04 3.10 3.69 2.00 2.00 4.00
x4 0.3–1.0 0.51 0.69 0.69 0.58 0.75
y1 03–08 6.00 6.44 7.53 7.19 7.61
y2 01–04 1.56 2.09 2.63 2.50 1.00
y3 06–10 7.75 8.39 8.10 8.09 8.21
y4 02–04 2.90 2.00 3.99 3.99 3.99
m1 0.3–1.0 0.80 0.83 0.81 0.86 0.83
m2 0.3–1.0 0.48 0.72 0.58 0.73 0.79
gap1 02–04 2.27 2.99 2.20 2.00 2.45
gap2 06–10 8.34 8.16 8.80 8.17 8.74
gap3 0.2–1.0 0.50 0.56 0.70 1.00 0.99
F 1 (S11 ) 0.12 0.30 1.00 2.12 7.96
F 2 (avg. gain in dBi) 2.27 2.45 2.77 3.01 3.41

Figure 6: TBN-UWB PF based on the MONMR, MOCS algorithms

Figure 8: Simulated |S11 | of the three optimal TBN-UWB

antennas

for more than 95% frequency components, the required

threshold levels of |S11 | have been achieved. Moreover,
the average and peak gain of the Ant-5 are denoted as
2.45 and 3.89 dBi, respectively, over the entire UWB
bandwidth except for notched-bands.

4. FABRICATION, MEASUREMENT AND

ANALYSIS
Every simulated antenna performance is validated if
its physical oriented structure-electrical output perfor-
mance is the same, not concern to minor variations. For
this, the proposed TBN-UWB Ant-5 structure was firstly
formulated through the fabrication process and then the
prototype was considered for testing on E5063A vec-
tor network analyzer (VNA). The experimental testbed
Figure 7: Convergence curves of TBN-UWB antenna using is placed in the antenna design Lab. Figure 9 shows
MONMR and MOCS
Ant-5 top view, bottom view, and the VNA measure-
ment setup. VNA is used to evaluate frequency domain
and F 2 convergence curves of the MONMR compared analysis of the proposed antenna in terms of S11 and
to MOCS performance and is found that MONMR has VSWR. The simulated and measured S11 results are com-
faster convergence among MOCS. pared in Figure 10, which shows a reasonably good
agreement. A minor variation occurs at high frequen-
The S11 characteristics of Ant-5 are shown in Figure 8. cies due to the tolerance in fabrication, SMA connector
It provided a minimum reflection of 0.3 means that losses, feed line mismatching, improper soldering, loss
8 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN

Table 3: Comparison of the proposed Ant-5 with some recently reported BN antenna techniques
Dimensions
Reference (mm2 ) Area (mm2 ) Gain (dBi) Optimization Rejection bands (GHz) BN technique used
[36], 2018 60 × 60 3600 <6 NA 3.0–3.75 and 5.2–5.6 Spiral defected microstrip
structure and SRR
[42], 2015 30 × 30 2080 < 5.8 SOO 3.55, 5.5 and 7.2 C and arc-shape slots and
annular strips
[43], 2020 34.2 × 40.7 1392 < 5.4 SOO 3.2–3.9 and 5.1–6.05 FTA structure
[37], 2019 26 × 35 910 < 3.5 NA 3.3–3.57 and 5.17–5.25 C-slot on patch and horizontal
stripes on ground
[44], 2015 30 × 30 900 < 3.8 NA 3.29–3.61, 4.65–5.49 and 7.3–8.41 Spiral slot loading
[45], 2013 28 × 32 896 <2 SOO 5.06–5.9 V-shaped slot
[34], 2019 24 × 30 720 < 3.7 MOO 3.3–3.6 and 5.15–5.84 FTA structure
[35], 2020 24 × 28 672 < 4.08 NA 5.02–5.97 and 7.23–7.72 S-slot on feed line and C-strip
on ground
[38], 2019 22 × 28 616 < 3.35 NA 3.15–4 and 5.18–5.95 V-shaped slot and C-strips
[3], 2018 20 × 29 580 < 3.3 SOO 3.4–3.7 and 5.1–5.9 Inverted U-slit and C-strips
Proposed Ant-5 20 × 28 560 < 3.89 MOO 3.5, 5.2/5.8 and 8.4 Inverted U-slots and C-strips

dBi, sharp TBN characteristics, efficient PF solution sets,

and best fitness [0.3, 2.45] based Ant-5 structure.

5. CONCLUSION
This paper proposes a novel MONMR algorithm to
Figure 9: Photograph of the fabricated TBN-UWB monopole Ant- optimize two antenna design problems in the field of
5: front, back-view, and VNA measurement setup electromagnetics. These designs included B-UWB and
TBN-UWB antennas. In MONMR, the exploration and
exploitation search has been improved by using Levy-
flight based scaling factor and SA-IW based mating fac-
tor, respectively. Here, two objective functions are consid-
ered for optimization: minimize reflection and maximize
the antenna gain over the entire UWB range. As a result,
Ant-5 gives the best TBN-UWB antenna structure, the
size is reduced to 20 mm × 28 mm along with sharp TBN
characteristics. Further, the prototype measurement has
also been reasonably good in agreement with the simu-
lated results. The measured working range is observed as
Figure 10: Simulated and measured S11 of proposed TBN-UWB 2.95–13.3 (FBW of 127%) including three notched-bands
Ant-5 of Wi-Max (3.15–3.95 GHz), WLAN (5.05–6.05 GHz)
and ITU-8 X-band (7.65–8.67 GHz). The peak gain is
denoted as 3.89 dBi with an excellent average gain of
tangent of FR-4 substrate, and environmental effects etc.
2.45 dBi throughout the UWB except for band-notched
Also, it is observed that the Ant-5 provided a sharp
regions. Hence, the proposed antenna is a good candidate
TBN characteristic at Wi-Max (3.15–3.95 GHz), WLAN
for UWB applications.
(5.05–6.05 GHz) and ITU-8 (7.65–8.67 GHz). The mea-
sured S11(min) is noted as −17.1, −35.6, −31 and −39 dB
at 3.05, 4.75, 6.98, 11.82 GHz frequencies, respectively. ACKNOWLEDGEMENTS
This work was supported by Council of Scientific and Industrial
In the last section, the performance of the proposed Research (CSIR) HRD Group, Govt. of India (GoI) under SRF
trapezoidal-shaped TBN-UWB Ant-5 monopole struc- research grant No. 09/677(0042)/2019-EMR-I.
ture is compared to some recently reported BN tech-
niques in terms of patch size, covered area, peak gain,
rejected bands and their corresponding band-notch tech- FUNDING
niques, as shown in Table 3. From this table, it can be con- This work was supported by Council of Scientific and Industrial
cluded that the MONMR optimized antenna provided Research (CSIR) HRD Group, Govt. of India (GoI) under SRF
good size reduction of 20 mm × 28 mm, peak gain of 3.89 research grant No. 09/677(0042)/2019-EMR-I.
 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN 9

ORCID 12. I. Aljarah, M. Habib, H. Faris, N. Al-Madi, A. A. Heidari,

M. Mafarja, M. A. Elaziz, and S. Mirjalili, “A dynamic
Gurdeep Singh http://orcid.org/0000-0001-8471-1077
locality multi-objective salp swarm algorithm for feature
Urvinder Singh http://orcid.org/0000-0002-0811-2037 selection,” Comput. Ind. Eng., Vol. 147, p. 106628, 2020.

13. R. Salgotra, U. Singh, S. Singh, G. Singh, and N. Mit-

tal, “Self-adaptive salp swarm algorithm for engineering
REFERENCES optimization problems,” Appl. Math. Model., Vol. 89, pp.
188–207, 2021.
1. A. Darvish and A. Ebrahimzadeh, “Improved fruit-fly opti-
mization algorithm and its applications in antenna arrays
14. R. Salgotra, U. Singh, G. Singh, S. Singh, and A. H. Gan-
synthesis,” IEEE Trans. Ant. Propag., Vol. 66, no. 4, pp.
domi, “Application of mutation operators to salp swarm
1756–66, Feb. 2018.
algorithm,” Expert. Syst. Appl., Vol. 169, p. 114368, 2021.
2. I. M. Danjuma, M. O. Akinsolu, C. H. See, R. Abd-
15. S. Etedali, K. Hasankhoie, and M. R. Sohrabi, “Optimal
Alhameed, and B. Liu, “Design and optimization of a slot-
design of pure-friction isolators with and without restoring
ted monopole antenna for ultra-wide band body centric
device: A multi-objective cuckoo search-based approach
imaging applications,” IEEE J. Electromag. RF Microw. Med.
for seismic-excited structures,” Structures, Vol. 25, pp.
Biol., Vol. 4, no. 2, pp. 140–7, 2020.
708–19, Jun. 2020.
3. D. Ustun and A. Akdagli, “Design of band-notched UWB
16. K. Chandrasekaran, S. P. Simon, and N. P. Padhy, “Cuckoo
antenna using a hybrid optimization based on ABC and DE
search algorithm for emission reliable economic multi-
algorithms,” AEU-Int. J. Elect. Comm., Vol. 87, pp. 10–21,
objective dispatch problem,” IETE J. Res., Vol. 60, no. 2, pp.
2018.
128–38, 2014.
4. S. Koziel and A. Bekasiewicz, “Multi-objective optimiza-
17. K. N. A. Rani, M. Abdulmalek, H. A. Rahim, N. S. Chin,
tion of expensive electromagnetic simulation models,”
and A. Abd Wahab, “Hybridization of strength Pareto
Appl. Soft Comp, Vol. 47, pp. 332–42, 2016.
multiobjective optimization with modified cuckoo search
algorithm for rectangular array,” Sci. Rep., Vol. 7, p. 46521,
5. S. Chamaani, S. A. Mirtaheri, M. Teshnehlab, and M. A.
2017.
Shooredeli, “Modified multi-objective particle swarm opti-
mization for electromagnetic absorber design,” in Asia-
18. D. Lu, L. Wang, E. Yang, and G. Wang, “Design of high-
Pacific Conf. on Applied Electromag. IEEE, Dec. 2007, pp.
isolation wideband dual-polarized compact MIMO anten-
1–5.
nas with multiobjective optimization,” IEEE Trans. Ant.
Propag., Vol. 66, no. 3, pp. 1522–7, 2017.
6. P. Di Barba, M. E. Mognaschi, D. A. Lowther, and J.
K. Sykulski, “A benchmark TEAM problem for multi-
19. A. Kaur and K. Kumar, “A reinforcement learning based
objective Pareto optimization of electromagnetic devices,”
evolutionary multi-objective optimization algorithm for
IEEE Trans. Magnetics, Vol. 54, no. 3, pp. 1–4,
spectrum allocation in cognitive radio networks,” Phys.
2017.
Comm., Vol. 43, p. 101196, 2020.
7. S. Koziel, A. Bekasiewicz, and W. Zieniutycz, “Expedited
20. A. Prasanth, and S. Jayachitra, “A novel multi-objective
EM-driven multiobjective antenna design in highly dimen-
optimization strategy for enhancing quality of service in
sional parameter spaces,” IEEE Ant. Wireless Propag. Lett.,
IoT-enabled WSN applications,” Peer-to-Peer Netw. Appl.,
Vol. 13, pp. 631–4, 2014.
Vol. 13, no. 6, pp. 1905–20, 2020.
8. Y. Kuwahara, “Multiobjective optimization design of Yagi-
21. H. R. Chi and A. Radwan, “Multi-objective optimization
Uda antenna,” IEEE Trans. Ant. Propag., Vol. 53, no. 6, pp.
of green small cell allocation for IoT applications in smart
1984–92, 2005.
city,” IEEE Access, Vol. 8, pp. 101903–14, 2020.
9. S. Chamaani, S. A. Mirtaheri, and M. S. Abrishamian,
22. K. Sreenu and S. Malempati, “MFGMTS: Epsilon constra
“Improvement of time and frequency domain performance
int-based modified fractional grey wolf optimizer for
of antipodal Vivaldi antenna using multi-objective particle
multi-objective task scheduling in cloud computing,” IETE
swarm optimization,” IEEE Trans. Ant. Propag., Vol. 59, no.
J. Res., Vol. 65, no. 2, pp. 201–15, 2019.
5, pp. 1738–42, 2011.
23. Q. Al-Tashi, S. J. Abdulkadir, H. M. Rais, S. Mirjalili,
10. S. Koziel, S. Ogurtsov, W. Zieniutycz, and L. Sorokosz,
H. Alhussian, M. G. Ragab, and A. Alqushaibi, “Binary
“Simulation-driven design of microstrip antenna subar-
multi-objective grey wolf optimizer for feature selection in
rays,” IEEE Trans. Ant. Propag., Vol. 62, no. 7, pp. 3584–91,
classification,” IEEE Access, Vol. 8, pp. 106247–63, 2020.
2014.
24. W. Yin, D. Mavaluru, M. Ahmed, M. Abbas, and A. Darvis-
11. E. G. Talbi, Metaheuristics: From Design to Implementation,
han, “Application of new multi-objective optimization
Vol. 74. John Wiley & Sons, 2009.
10 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN

algorithm for EV scheduling in smart grid through the asymmetrical parasitic stub,” IETE J. Res., pp. 1–11, Sep.
uncertainties,” J. Ambient Intell. Humaniz. Comp., Vol. 11, 2020.
no. 5, pp. 2071–103, 2020.
36. U. A. Dash and S. Sahu, “UWB Dual-Band notched conical
25. B. Esenboga and T. Demirdelen, “Efficiency and cost based dielectric resonator antenna with improved gain,” IETE J.
multi-optimization and thermal/electromagnetic analyses Res., Vol. 66, no. 5, pp. 643–53, Sep. 2020.
of 3-phase dry-type transformer,” IETE J. Res., pp. 1–13,
2020. 37. A. Bhattacharya, B. Roy, S. K. Chowdhury, and A. K.
Bhattacharjee, “An isolation enhanced, printed, low-profile
26. P. R. Jenkins, B. J. Lunday, and M. J. Robbins, “Robust, UWB-MIMO antenna with unique dual band-notching
multi-objective optimization for the military medical evac- features for WLAN and WIMAX,” IETE J. Res., pp. 1–8,
uation location-allocation problem,” Omega (Westport), May 2019.
Vol. 97, p. 102088, Dec. 2020.
38. G. Singh and U. Singh, “Dual band rejected low profile
27. A. Hebbal, L. Brevault, M. Balesdent, E. G. Talbi, and planar monopole antenna for UWB application,” in Int.
N. Melab, “Multi-objective optimization using deep Gaus- Conf. on Automa., Computation and Tech. Management
sian processes: Application to aerospace vehicle design,” in (ICACTM). IEEE, Apr. 2019, pp. 534–538.
AIAA Scitech 2019 Forum, 2019, p. 1973.
39. G. Singh and U. Singh, “Analysis of triple band rejected
28. C. Wu, J. Wang, X. Chen, P. Du, and W. Yang, “A novel compact planar octagon shape monopole antenna for
hybrid system based on multi-objective optimization for UWB applications,” in Int. Symp. on Networks, Computers
wind speed forecasting,” Renew. Energy, Vol. 146, pp. and Comm. (ISNCC). IEEE, 2020, pp. 1–6.
149–65, 2020.
40. R. Salgotra and U. Singh, “The naked mole-rat algorithm,”
29. W. Chien, C. C. Chiu, Y. T. Cheng, S. H. Liao, and H. S. Neural Comput. Appl., Vol. 31, no. 12, pp. 8837–57, 2019.
Yen, “Multi-objective optimization for UWB antenna array
by APSO algorithm,” Telecommun. Syst., Vol. 64, no. 4, pp. 41. J. C. Bansal, P. K. Singh, M. Saraswat, A. Verma, S. S.
649–60, Apr. 2017. Jadon, and A. Abraham, “Inertia weight strategies in par-
ticle swarm optimization,” in Third World Congress on
30. P. Palanisamy and M. Subramani, “Closely mounted UWB Nature and Biologically Inspired Computing. IEEE, Oct.
MIMO antenna with notch characteristics for short-range 2011, pp. 633–640.
wireless video transmission application,” IETE J. Res., pp.
1–13, May 2020. 42. H. J. Mohammed, A. S. Abdullah, R. S. Ali, R. A. Abd-
Alhameed, Y. I. Abdulraheem, and J. M. Noras, “Design
31. G. Singh, U. Singh, and R. Salgotra, “Effect of paramet- of a uniplanar printed triple band-rejected ultra-wideband
ric enhancements on naked mole-rat algorithm for global antenna using particle swarm optimisation and the firefly
optimization,” Eng. Comput., Vol. 37, pp. 1–29, 2021. algorithm,” IET Microwav. Ant. Propag., Vol. 10, no. 1, pp.
31–7, 2016.
32. S. Koziel and A. Bekasiewicz, “Low-cost multi-objective
optimization and experimental validation of UWB MIMO 43. M. C. Derbal, A. Zeghdoud, and M. Nedil, “A dual band
antenna,” Eng. Comput. (Swansea), Vol. 33, no. 4, pp. notched UWB antenna with optimized DGS using genetic
1246–58, 2016. algorithm,” Prog. Electromagn. Res., Vol. 88, pp. 89–95,
2020.
33. P. Palanisamy and M. Subramani, “Design and experimen-
tal analysis of miniaturized octa-port UWB/SWB-MIMO 44. S. Das, D. Mitra, and S. R. B. Chaudhuri, “Design
antenna with triple-band rejection characteristics,” IETE J. of UWB planar monopole antennas with etched spiral
Res., pp. 1–15, 2020. slot on the patch for multiple band-notched character-
istics,” Int. J. Microwav. Sci. Tech., Vol. 2015, p. 303215,
34. Y. Du, X. Wu, J. Sidén, and G. Wang, “Design of ultra- 2015.
wideband antenna with high-selectivity band notches
using fragment-type etch pattern,” Microw. Opt. Technol. 45. Y. L. Li, W. Shao, L. You, and B. Z. Wang, “An improved
Lett., Vol. 62, no. 2, pp. 912–8, Feb. 2020. PSO algorithm and its application to UWB antenna
design,” IEEE Ant. Wirel. Propag. Lett., Vol. 12, pp. 1236–9,
35. V. K. R. Devana and A. M. Rao, “Design and analysis of 2013.
dual band-notched UWB antenna using a slot in feed and
 G. SINGH AND U. SINGH: MULTI-OBJECTIVE NAKED MOLE-RAT ALGORITHM FOR UWB ANTENNA DESIGN 11

AUTHORS Urvinder Singh was born in India and

received his BTech in electronics and com-
Gurdeep Singh was born in India and munication engineering (ECE) from Guru
received his BTech in electronics and com- Nanak Dev University, Punjab, India, in
munication (ECE) from Guru Nanak Dev 1998. He received his MTech and PhD
Engineering College, Ludhiana, Punjab, in ECE from Punjab Technical University,
India, in 2009 and ME (ECE) from Pan- Jalandhar, India, in 2007 and 2013, respec-
jab University, Chandigarh, India, in 2012. tively. He is now working with Thapar
He has teaching experience of 5 years. His University, Patiala, India. His research interests include anten-
research interest includes microstrip patch nas, computational electromagnetics, filter design and wireless
antenna design and simulation, UWB, soft computing, nature- sensor networks. He has published research papers in journals
inspired algorithms, electromagnetic designs optimization, and of repute like IEEE, IET, Springer, and PIER.
wireless communication. He has published research papers in
reputed journals of IEEE and Elsevier.
Email: [email protected]
Corresponding author. Email: [email protected]