2014 3rd Asia-Pacific Conference on Antennas and Propagation

Antenna Pattern Synthesis of Shaped-Beam Using a

New Combined Algorithm
GAO ZhiGuo YANG Bing
Beijing Institute of Remote Sensing Equipment Beijing Institute of Remote Sensing Equipment
Beijing, China Beijing, China
[email protected] [email protected]

Abstract—This Paper presents a new combined algorithm for 1


N 1
shaped-beam antenna pattern synthesis. After main shaped-beam x ( n)  k 0
X (k )W N nk n=0,1,…,N-1 (2)
is obtained by DFT synthesis optimization, the side lobe is N
optimized through genetic algorithm (GA), which greatly
improves the optimization speed and effectively avoids the 2j / N
shortcoming of GA. Since only with GA, the optimization where, W N  e .
converges slowly and optimized shaped-beam fluctuated greatly.
A low side lobe square cosec beam antenna array is illustrated. III. APPLICATION OF UNIFORM SPACING LINEAR ARRAY
Synthesis results prove the feasibility of the algorithm.
Radiation pattern of uniform spacing linear array, as Figure
Keywords—shaped-beam, GA, DFT, synthesis 1, with element spacing d and weighted coefficient w(n), can
be expressed as:
I. INTRODUCTION j 2d N 1
F ( )  n 0 w(n) exp[
N 1
Generally in radar, communication and many other areas, (n  ) sin( )] (3)
 
specific shape of antenna beam (e.g. Csc square beam) is
required, called antenna shaped-beam. The method to find the
excitation, including amplitude and phase of the antenna array, The relations between radiation pattern F ( ) and
through the desired radiation pattern, is referred to as synthesis. coefficient w(n) can be represented with DFT[8] :
Shaped-beam synthesis is a quite difficult nonlinear
optimization problem. F ( )  N  exp( j 0 ) IDFT ( w(n)) (4)

GA is a self-organized, adaptive method, which analogs the w(n)  DFT ( N1 exp( j 0 ) F ( ) (5)
process of natural evolution, to generate solution to
optimization and search problems. GA predominates for
complicated nonlinear optimization problem, and has been
applied in many areas [1~6]. GA is random algorithm, therefore
it behaves great to jump out of local optimization. However,
while synthesis csc square beam, main beam fluctuates greatly
and convergences slowly appear commonly. In this paper, a
new optimization method which combines GA with discrete
Fourier transform (DFT) is proposed. Firstly, main beam is
synthesized by DFT. Afterwards, GA is applied to optimize the
side lobe. The new combined algorithm compared with only
GA, get the synthesis results more accurate and calculation
speed faster.
Fig. 1. Uniform spacing linear array
II. DFT TRANSFORMATION
Fouried transform is analysis method widely used in
various areas, and DFT is the basis of signal processing. As IV. PROCESSES OF SYNTHESIS WITH DFT
defined in [7], DFT of sequence {xi} (i=0,1,2...N-1) is : DFT is used to restrict main lobe region of radiation pattern.
Steps are as follows:
X (k )  n 0 x(n)WNnk k=0,1,…,N-1
N 1
(1)
Step 1: Set initial weighted coefficient w(n) to satisfy side
lobe request of Taylor distribution.
And discrete Fourier inverse transform (IDFT) is :

978-1-4799-4354-8/14/$31.00 ©2014 IEEE 291 Harbin, CHINA

 Step 2: Get the amplitude F ( ) and Phase  of radiation TABLE I. THE REQUIREMENT OF SHAPED-BEAM

pattern F ( ) , by IDFT. Angle（°）

Requirement of shaped-beam
(dB)
Step 3: Divide the region of F ( ) into main lobe and side 0 -2.0
2 0
lobe two parts. Use the desired pattern of main lobe instead of
6 -2.3
the former value, as F~ ( ) .
10 -6.7

Step 4: Compute new radiation pattern F~ ( ) with F~ ( ) .

14 -9.6
18 -11.7
~ (n) by F~ ( )
Step 5: Calculate new weighted coefficient w
20 -12.6
25 -14.5
with DFT method.
30 -15.9
Step 6: Repeat Step 1 to Step 5 until main beam of shaped- 5
beam pattern meets the requirement.
0

Radiation pattern(dB)
V. APPLICATION OF GA IN COMBINED ALGORITHM -5

GA requires 3 basis elements, as follows: -10

-15
1) Generation of initial values:
Use the result from DFT synthesis as initial value of GA. -20

-25
2) A fitness function to evaluate the solution domain:
In general, during evolution searching process, GA doesn't -30

need other extra information, except fitness function, which -35

evaluates the individual or solution, and then is used in the next
-40
iteration of the algorithm. On the base of meeting requirement -80 -60 -40 -20 0 20 40 60 80

of main lobe of shaped-beam, to optimize side lobe, fitness Angle(°)

function design must include three parts: main lobe deviation,
Fig. 2. Radiation pattern after DFT synthesis (The red dot is the requirement
main lobe and side lobe fluctuation. of shaped-beam)

F  F ( )  F0 ( ) (6) 5

where,  is the angle of main beam. 0

Fitness function is defined as: -5

Radiation pattern(dB)
Fitness  a1  sum(F )  a 2  max(F )  a3  SLL (7)
-10

-15

where a1, a2, a3 are the weights of three parts, SLL is the side -20

lobe. -25

3) Genetic operation: selection, crossover and mutation -30

The programs of selection, crossover and mutation are -35
already written in Matlab.
-40
-80 -60 -40 -20 0 20 40 60 80

VI. ENGINEERING INSTANCE Angle(°)

A 17 elements linear array, operating at X band, with Fig. 3. Radiation pattern after combined algorithm (The red dot is the
requirement of shaped-beam)
element spacing λ/2, is synthesized. The main lobe of the
shaped-beam requirements are shown in Table 1, and the side
lobe must less than -20dB. VII. CONCLUSION
Set -25dB Taylor distribution as the initial value of DFT The combined algorithm, which compounds DFT synthesis
synthesis. After 25 times iteration, excitation distribution that method with GA, accelerates the calculation speed of shaped-
satisfies main lobe of shaped-beam is obtained. Result of beam synthesis, effectively avoids shortcomings of show
radiation pattern is illustrated in Figure 2, and result of convergence and great fluctuation in main lobe by only GA.
weighted coefficients is shown in Table 2. The next step,
optimize side lobe with GA. After 5000 times iteration,
radiation pattern is illustrated in Figure 3, side lobe is -23 dB,
and the deviation of shaped-beam in main lobe is within 0.1 dB.

292
 [4] Liu Hao, Zheng Ming, Fan Desen, Lu Jiaguo, “Synthesis of antenna
REFERENCES arrays shaped-beam using genetic algorithm,” Chinese Journal of Radio
[1] Diogenes Mareano and Filinto Duran, “Synthesis of antenna arrays Science, 2000, 17(5), pp.539-542.
using genetic algorithms,” IEEE Antennas and Propagation Magazine, [5] Shang Fei, Cai Yaxing,Zhang Ying,”Synthesis of array antenna using
2000, 42(3), pp. 12-20. double population genetic algorithm,” Chinese Journal of Radio Science,
[2] Das, S.,Bhattacherjee, S., Mandal, D.,Bhattacharjee, A.K.,” Optimal 2007, 22(2), pp.224-228
sidelobe reduction of symmetric linear antenna array using Genetic [6] Fan Yu, Jin Ronghong. “Pattern synthesis of antennas based on a novel
Algorithm,” 2010 Annual IEEE India Conference (INDICON). genetic algorithm,” Chinese Journal of Radio Science, 2004, 19(2) ,
[3] Soltankarimi, F.; Nourinia, J.; Ghobadi, C.” Side lobe level optimization pp.182-186.
in phased array antennas using genetic algorithm,” 2004 IEEE Eighth [7] Cooley JW, Tukey JW. “Analgorithmn for machine calculation of
International Symposium on Spread Spectrum Techniques and complex Fourier series,” Mathematics of Computation. 1965,19:297-301.
Applications, pp. 389_394, 30 Aug.-2 Sept. 2004. [8] Hu Hongfei, Fu Demin, Yu Ding, Mao Naihong,”A fast algorithm for
planar transformation from near-field to far-field ,” Chinese Journal of
Radio Science, 2007, 22(2).

TABLE II. THE WEIGHTED COEFFICIENTS

Weighted Coefficients
Weighted Coefficients
After Combined
No. After DFT Synthesis
Algorithm
amplitude Phase(°） amplitude Phase(°）
1 0.385 165.90 0.391 164.83
2 0.434 138.95 0.431 139.31
3 0.487 113.38 0.483 114.94
4 0.599 108.51 0.603 109.81
5 0.900 82.37 0.902 83.15
6 1.000 50.96 1.000 51.58
7 0.808 19.29 0.809 20.07
8 0.444 14.90 0.455 15.43
9 0.401 33.81 0.412 32.28
10 0.481 38.69 0.481 36.27
11 0.439 21.58 0.437 18.88
12 0.390 5.21 0.389 2.08
13 0.363 14.86 0.350 10.53
14 0.149 40.50 0.117 34.70
15 0.209 35.11 0.175 35.59
16 0.147 20.79 0.118 21.62
17 0.123 22.54 0.087 27.43

293