You are on page 1of 6

International Journal of Speech Technology

https://doi.org/10.1007/s10772-020-09701-2

Synthesis of phased array antenna for side lobe level reduction using
the differential evolution algorithm
Ravi Tej1 · K. Ch. Sri Kavya1 · Sarat K. Kotamraju1

Received: 2 December 2019 / Accepted: 14 March 2020


© Springer Science+Business Media, LLC, part of Springer Nature 2020

Abstract
In this paper, we synthesis an array consisting of uniformly excited unequally spaced elements for reducing SLLs. A glob-
ally optimal solution for the array elements phase and position is determined by using a differential evolutionary algorithm
for a given radiation pattern. Reduced side-lobe levels are observed with position-phase synthesis. The obtained results are
compared with the possible two cases i.e., uniform amplitude equal phase (variable position-control) and uniform amplitude
unequal phase (variable phase-control). The obtained results justify that the position-phase synthesis technique is not only
reduces the side-lobe levels to (− 25.28) dB, but also preserves the advantages of both variable position-controlled synthesis
and variable phase-controlled synthesis with a minimum number of array elements.

Keywords Phased array · Uniform excitation · Side lobe level · Equal phase · Unequal spacing

1 Introduction can be enlarged to use multiple fitness functions. There are


three evolutionary algorithms; they are PSO (Particle Swarm
Different mathematical methods for the pattern synthesis of Optimization) an algorithm, GA (Genetic Algorithm), DE
antenna arrays have evolved from the last few years. There are (Differential Evolution) algorithm. For discrete optimiza-
several optimization weighting methods for the linear array tion mainly we use GA (genetic algorithm). For continuous
synthesis. They are Chebyshev and Taylor methods and also optimization, we use PSO (particle swarm optimization) and
there are some statistical-based weighting methods. These DE (differential evolution) algorithms. These two algorithms
have some different types of algorithms based on various fac- are more natural compared to GA (Genetic algorithm). In this
tors. Some execution comparison on various types of arrays paper, we apply the DE algorithm which has given priority or
is using Least Mean Square (LMS) and Root Mean Squared the equally spaced antenna array e.g., by changing the phase
(RMS) algorithms are applicable to a smart antenna (Ali et al. and the amplitudes of the elements, the minimum side-lobe
2012). And some direct search methods involve random deci- level are achieved. In the non-uniform amplitude array model
sion making are Particle Swarm Optimization (PSO), Genetic there are few functional obstacles but the amplitude, in particu-
Algorithm (GA), Differential Evolution (DE) algorithms. In lar, decreased the antennas of the micro-strip array which are
Artificial Intelligence the Evolutionary algorithms utilize the well known among the antenna models due to their minimum
mechanisms essentially and take care of issues through proce- size (Zubir et al. 2010). For example, look at the micro-strip
dures that imitate the practices of the living life forms. These antenna model which exhibits feed on the micro-strip board.
If we use the substrate in the feed line of the micro-strip with a
high dielectric constant for high amplitude ratios the width can
* Ravi Tej
ravitej025@gmail.com; ravi.tej489@gmail.com be very small causing problems with suppliers. However to
achieve the enormous amplitude ratios used by the microstrip
K. Ch. Sri Kavya
kavya@kluniversity.in line on a substrate with small dielectric constant substrate the
width grows excessively greater. The consequence radiation
Sarat K. Kotamraju
kksarat@kluniversity.in pattern has equal ripple side-lobes like a Chebyshev design
(Miller and Goodman 1983). Chebyshev channels are utilized
1
Department of Electronics and Communication Engineering, to isolate one band of frequencies from another. The essen-
Koneru Lakshmaiah Education Foundation, Greenfields, tial characteristic of Chebyshev’s design is its speed. This is
Vaddeswaram, Guntur, AP 522502, India

13
Vol.:(0123456789)
International Journal of Speech Technology

because they are completed by recursion instead of convolu- EM problems (Modi et al. 2019). GA is indicated as one of
tion. In the Taylor arrangement extension strategy, a produc- the dynamic approaches that are able to handle complicated
tive calculation for playing out the famous Jacobi arrangement problems among several self-sufficient variables in scheming
is built up for the development of radiation which is necessary arrays with many elements. For optimizing large huge arrays
for portraying the activity of reflector and other roundabout classical optimization techniques are not strongly suitable
gap radio wires. This methodology brings about a Bessel (Ares-Pena et al. 1999). For attaining good control over the
arrangement extension for the radiation vital. The strategy is antenna patterns, high gain than a separate antenna element
then applied for effortlessness to various instances of source and electronic beam steering a phased array antenna is more
work with rotational balance. advantageous than other antennas. In various classical models,
The main drawback of the variable-phase synthesis which the phase and amplitude excitation are obtained for elements
uses the same element size is the need for a large number among equivalent separation. Also, the far-field pattern of the
of components contrast to the amplitude decreased arrays array is controlled by the amplitudes and phases of the excita-
for accomplishing minimum side lobe rates (Tian and Qian tion of currents (Varahram and Rashed-Mohassel 2002). In
2005). The measure of the decrease in the sidelobe level is several new models, the position of the elements is controlled
achievable is very large particularly for huge arrays. The with an assumed current distribution. Evolutionary algorithms
performance of the non-uniform element spacing dispers- have no restrictions on the selection of variables in optimiza-
ing practically is regularly easier than the generally utilized. tion. As a result, the amplitudes and phases of the excitation
Likewise, the powerful abilities of an array having all the can be used as separate variables in the optimization proce-
elements are superior to at the point when the amplitude dure. The necessity of optimization of phased array antennas
is decreased (Rocha-Alicano et al. 2006) In this paper, is compressing side lobe level or null position control of the
although it appears as design problem may be bypassed by patterns, half-power bandwidth.
the uniform amplitude uniformly spaced arrays with equal
levels, the power-productive antenna arrays in the case of 3 Differential evolution algorithm
cellular communications allow us to operate the power
amplifiers with less non-linearity. Therefore increasing the It is a stochastic population-based optimization algorithm for
radiated energy with fewer elements is important for some solving nonlinear optimization problems following the systems
applications as achieving decreasing rates of side lobes. The a framework of the DE calculation is as per the following.
antenna elements whose spacing is unequal are connected to Like GAS DE additionally works on a populace with people
the uniform spacing of the time-space signals. The synthesis and every individual is an emblematic portrayal of the vector
of pole-zero patterns differs from most others in that it gives comprising of the streamlining parameters (Naga Jyothi and
origin areas only the example and the number of sources that Sriadibhatla 2017, 2019; Jyothi and Sridevi 2018, 2019). The
needs to be determined. The DE algorithm which uses the mutation happens first and the freak vector can be produced
best of irregular differential transformation applies to the as shown in Eq. (1).
synthesis of antenna arrays that are unequally spaced. The
best differential shift is used as the differential mutation by V Mj = vnopi + 𝛽(vnp1 + vnp2 ), i ≠ P1 andi ≠ P2 (1)
the best person among the arbitrarily selected ones (Lin et al. where i, p1 and p2 are haphazardly selected individual indi-
2010). The differential evolution with N = 32 components cates in the parent populace and the superscript opi alludes
will reduce non-uniformity. Compared to other algorithms to the ideal individual in the populace. The genuine consist-
the algorithm is strong (Kumari and Sridevi 2016). Trouble ent p is the transformation factor. Once in a while, a few
in analyzing and modeling unequally spaced arrays stem qualities of the freak vector surpass their pursuit ranges and
from the array factor’s non-linear and non-convex phases these qualities need to be adjusted. After mutation, the vec-
and locations of components. The elements of the positions tor with corresponding father vector to give out child vector
and angles for given radiation are trouble as a synthesis of is given in equation.2.
equal amplitude and unequal phases of uniformly spaced
sets. (V cj )j = {(vMj )j , 𝛾 < PCross , (v(n)j )j , otherwise (2)

where gamma is random number in range of (0,1) and real


2 Literature survey constant Pcross is probability of crossover.

Low side-lobe antennas are getting a progressively more


essential element of highly realized electronic setup, mainly
those working in intense clutter and jamming situations. The
purpose of evolutionary algorithms is now frequent in complex

13
International Journal of Speech Technology

4 Synthesis of arrays to the GA (genetic algorithm) the amount of interest in an


evolution process may be more than two. For the synthesis
The synthesis of antenna arrays utilizes the differential of the series we also set K=2 to adjust the vector according
evolution algorithm which gives detail in Fig. 1. Initially, to the age of the two trail individuals represented as by Eqs.
the parent population P̄i , i 𝜀[1, Np ] is assigned uniformly. (3) and (4).
At the point when the best parent, as far as its objective
meets the optimization criteria or increases the repetitions t1 = Pb + F(Pi − Pr ) (3)
itn achieves the highest indicated repetitions max the opti-
mization is ended. For each repetition itn, given Pi , which t2 = pr + F(Pi − Pr ) (4)
are utilizing K dissimilar procedures, produce parent Pi ,
become child ci . However, it produces a uniformly assign a where F is the true and permanent variable governing dif-
random number in [0, 1], the likelihood of producing trail ferent varieties like (P̄i − P̄r ) and (P̄i − P̄r ). Individuals and
members is pc , and the likelihood of the parent Pi straight parent population sub-set satisfy the condition that the files
forwardly it turning in to child ci is (1 − Pc ). If at that point are not exactly the same as the running index from Fig. 1.
the trail members are made and the trail portion is created More on the diverse DE algorithm techniques and the list
by the positive procedure contends with the parent and the age system for r and s can be found. It is to be noticed that
champ is mutated with a low likelihood for ending up with not every one of the element of the vectors t1 and t2 experi-
the child. That DE algorithm is essentially a straight mix of ence the changes are appointed earlier the elements of the
vector contrasts between a subset of the parent population vector. We are shared with the parent population when all
together with parent Pi or Pb. The quantity of the individuals the children are named as above. The DE algorithm planned
in the subset is lower than Np, the complete number of indi- Pm =0 and K = 1. By having K = 2 and for a few cases with
viduals in the parent population. Subsequently, as opposed 0 ≤ Pm ≤ 0.2, improved attributes have been accomplished.
The ranges of other control parameters are 0.5 ≤ Pc ≤ 1 and
0.4 ≤ F ≤ 1.

4.1 Problem statement for the synthesis

For a linear array the array factor AF (𝜃) with arbitrary posi-
tions is given by the Eq. (5):
Ne
∑ j2𝜋xi
AF (𝜃) = Wi e 𝜆 sin 𝜃 (5)
i=1

where wi indicates the current which is situated at xi and 𝜆


is the wavelength. We consider the components with steady
sufficiency random stage and random positions which array
factor becomes as shown in Eq. (6).
Ne
∑ j2𝜋xi
AF (𝜃) = e 𝜆 sin 𝜃 + 𝜙i (6)
i=1

4.2 Aim to minimize the side‑lobe level

The problem of optimization (or) minimization of the side


lobe level can be given by Eq. (7) for f s (P)
̄ as:

AF (𝜃)
̄ = Max𝜃𝜖s
f s (P) ̄ (7)
AFP (𝜃0 )

Here s is the space spread over by the point eliminate the


main lobe which comprising of elements positions and the
Fig. 1  Flow-diagram of the algorithm of DE used for array synthesis phases which limits all the side lobe levels and high the

13
International Journal of Speech Technology

power in the main lobe is situated at the 𝜃 = 𝜃0 Hence it can


be given by the Eq. (8) as
p̄ = xi , 𝜙i (8)
where 1 ≤ i ≤ Pi
We utilized an explore procedure to decide as on the basis
of the above equation. In general, it varies starting with one
repetition than on to the next. Since both the positions and
phases appear in the array factors exponential term, there is
no reason to advance both the positions and phases given
in the meantime by the above formula. Because we use the
DE algorithm in the above section three, we demonstrate
that there is a tradeoff between the most extreme distance
between the positions and the phase of a particular set.

Fig. 2  This graph reflects the synthesis of both phase and the position
5 Results and discussion of the side lobe level suppression

In this section uniformly excited unequally spaced phased


array antennas are considered with different cases: almost equal i.e − 55.42 dB. The synthesis results are com-
pared and tabulated for variable-position, variable-phase and
5.1 Comparison of position‑phase synthesis position-phase when dmin=0.5𝜆 and dmax= 0.7𝜆 in Table 1.
and variable‑position synthesis Figure 4 represents the different mutations used in the dif-
ferential evolution algorithm.
For the variable-position and position-phase instances the
synthesis of an arbitrarily distributed array is performed 5.2 Comparision of variable‑phase synthesis
separately for a different range of boundaries between the and the position‑phase synthesis
elements and the corresponding array patterns are shown
in Figs. 2 and 3. The number of elements required for the We have seen in the earlier chapter that the unequally spaced
DEA-based synthesis is 32 so the optimization parameters array consequential from the synthesis of the position-phase
the number is 16 for variable-position synthesis and 32 for has diminished side lobe levels as distinguished to those of
phase- position synthesis. dmin = 0.5𝜆 where dmin is the the unequally spaced arrays derived from the synthesis of
minimum distance between two adjacent elements is the
poor restriction in the synthesis of the positions of the ele-
ments for both the cases. The upper limit of the length of the
elements and dmax is diverse from 0.5 to 1𝜆. The sidelobe
level that we have achieved in variable-position was drasti-
cally reduced to − 55.45 dB whereas in position phase was

Table 1  Comparison of different synthesis techniques


Number of Variable phase Variable Position phase
iterations position

I X∕𝜆 X∕𝜆 X∕𝜆 𝜋


1 1.5 1.5 6.7 2.1
2 2.5 2.5 4.6 2.5
3 2.89 3.5 3.8 0.3
4 4.09 4.5 2.5 6.9
5 5.18 5.5 − 2.75 3.7
6 5.97 6.5 − 5.78 2.4
7 6.4 6.99 − 7.90 3.6 Fig. 3  This graph reflects the variable-postion of the the elements
with the suppression of side lobe level

13
International Journal of Speech Technology

sidelobe level minimizing for a uniformly excited unequally


spaced array is derived, that influence for any scan direction,
beam width, and type of antenna element used. In this paper,
The DEA is used to develop a globally optimum and efficient
synthesis technique for the design of uniform amplitude
unequally spaced arrays. From the analysis of DEA based
variable-position and position-phase synthesis it is observed
that position-phase synthesis derives less side lobe level and
enhanced array gain by preserving all the advantages under
the same radiation conditions.

References

Fig. 4  Indicates the mutation used in the DE algorithm Ali, M., Siarry, P., & Pant, M. (2012). An efficient differential evo-
lution based algorithm for solving multi-objective optimization
problems. European Journal of Operational Research, 217(2),
404–416.
Ares-Pena, F. J., Rodriguez-Gonzalez, J. A., Villanueva-Lopez, E., &
Rengarajan, S. R. (1999). Genetic algorithms in the design and
optimization of antenna array patterns. IEEE Transactions on
Antennas and Propagation, 47(3), 506–510.
Jyothi, G. N., & Sriadibhatla, S. (2018). Low power, low area adaptive
finite impulse response filter based on memory less distributed
arithmetic. Journal of Computational and Theoretical Nanosci-
ence, 15(6–7), 2003–2008.
Jyothi, G. N., & Sriadibhatla, S. (2019). Asic implementation of low
power, area efficient adaptive fir filter using pipelined DA. In S.
C. Satapathy, N. B. Rao, S. S. Kumar, C. D. Raj, V. M. Rao, & G.
V. Sarma (Eds.), Microelectronics, electromagnetics and telecom-
munications (pp. 385–394). Singapore: Springer.
Kumari, K. K., & Sridevi, D. P. (2016). Pattern synthesis of non uni-
form amplitude equally spaced microstrip array antenna using
GA, PSO and DE Algorithms. International Journal Advanced
Research in Engineering and Technology, 7, 132–147.
Lin, C., Qing, A., & Feng, Q. (2010). Synthesis of unequally spaced
antenna arrays by using differential evolution. IEEE Transactions
Fig. 5  This represents suppression in the sidelobe level with variable
on Antennas and Propagation, 58(8), 2553–2561.
phases
Miller, E. K., & Goodman, D. M. (1983). A polezero modeling
approach to linear array synthesis: 1. The unconstrained solution.
Radio Science, 18(1), 57–69.
the variable-position and the corresponding array patterns Modi, J. K., Gangwar, R. K., Ashwin, P., Gangwar, V. S., Singh, A. K.,
are shown in Fig. 5. To design a pencil beam array in this & Singh, S. P. (2018). Investigation on novel synthesis approach
for thinned planar arrays employing modified GA optimizer: Pro-
section we first compare the DEA-based position-phase syn-
ceedings of the nternational Microwave and RF Conference 2018
thesis and variable-phase synthesis. In both methods of syn- IEEE MTT-S I (IMaRC) (pp. 1–4). IEEE.
thesis, the number of elements are 32. The previous limits Naga Jyothi, G. & Sriadibhatla, G. (2017). Distributed arithmetic
assumed in the least and highest distance connecting the ele- architectures for FIR filters-a comparative review: Proceedings
of the International Conference on Wireless Communications (p.
ments for the synthesis of the position-phase aredmin = 0.5𝜆
2017). IEEE: Signal Processing and Networking (WiSPNET).
and dmax = 0.7𝜆. The sidelobe level that we have achieved NagaJyothi, G., & Sriadibhatla, S. (2019). High speed and low area
in variable-position was drastically reduced to (− 25.28) dB decision feed-back equalizer with novel memory less distributed
where as in position-phase (− 15.42) dB. arithmetic filter. Multimedia Tools and Applications, 78, 1–15.
Rocha-Alicano, C., Covarrubias-Rosales, D., Brizuela-Rodrguez, C., &
Center, C. R. (2006). Performance evaluation of two array factor
synthesis techniques for steerable linear arrays: Proceedings of
6 Conclusion the IASTED Conference on Antennas, Radar, and Wave Propaga-
tion, (pp. 3–5), Banff, Alberta, Canada.
Tian, Y. B., & Qian, J. (2005). Improve the performance of a linear
As the first side lobe level which regulates maximum SLL
array by changing the spaces among array elements in terms of
is the major cause of electromagnetic interference, it must genetic algorithm. IEEE Transactions on Antennas and Propaga-
be reduced. A strategy of finding the optimum solution for tion, 53(7), 2226–2230.

13
International Journal of Speech Technology

Varahram, A. A., & Rashed-Mohassel, J. (2002). Sidelobe level opti- Publisher’s Note Springer Nature remains neutral with regard to
mization using modified genetic algorithm: Proceedings of the jurisdictional claims in published maps and institutional affiliations.
IEEE Antennas and Propagation Society International Symposium
(IEEE Cat. No. 02CH37313) (Vol. 1, pp. 742–745). IEEE.
Zubir, F., Abd Rahim, M. K., Ayop, O. B., & Majid, H. A. (2010).
Design and analysis of microstrip reflectarray antenna with
minkowski shape radiating element. Progress In Electromagnet-
ics Research, 24, 317–331.

13

You might also like