Design of the Sector Array Antenna Based on Genetic Algorithm for Smart

Antenna System Front End

An-Tao Bu I, Xiao-Wei Shi I, Ying Liu2, Liang-Yong Xiao’

I. P. 0. 223, School ofElectronic Engineering of Xidian University, 710071,Xi’an, China
2. Institute ofAntennasand EM Scattering of Xidian University, 710071,Xi’anChina
3. Xi’an Haitian Antenna Technologies Co. Ltd., Xi’an, China
Email: [email protected]

Abstract: Genetic Algorithm (CA) is used to design a type of sector array antenna for
smart antenna system front end. The sector array antenna is mainly applied to the base
station of TD-SCDMA. Firstly, to meet the demand of smart antenna system sector
coverage mode (namely broadcast mode), the arrays configuration and element weight are
obtained by the global-search GA. Secondly, element weight is optimized to simulate the
tracking mode (namely working mode) by CA. The sector antenna array of practical smart
antenna system is analyzed, which is meaningful for the smart antenna technology.

1. Introduction
Smart antennas refer to a group of antenna technologies that increase the system
capacity by reducing the CO-channelinterference and increase the quality by reducing the
fading effects. Smart antenna array containing M identical elements can steer a directional
beam to maximize the signal from desired user while nullifying the signals from all other
directions. The character of antenna array is important for the smart antenna system.
The geometry of the antenna locations can vary widely, but the most common
configurations are to place antennas around a circle (circular array), along a line (linear
array) or in plane (planar array). Most analysis methods [2-31 assume ideal antenna
element and ignore mutual coupling which is far from practical antenna structure.
However in this paper the antenna array with reflector board is analyzed by the Method of
Moment (MM) in order to realize the sector beam for the TD-SCDMA. The analysis
process is by the exact MM with the mutual coupling considered and optimization process
the global-search C A [4]. In the sector coverage mode the pattem subject to a ripple
constraint covers the sector uniformly. Based on the optimized antenna array configuration
for sector coverage mode, the weight is optimized again for the tracking mode to simulate
the beam forming (BF). The numerical results show that the designed antenna array can
not only cover the sector uniformly, but also track desired user exactly and suppress the
co-channel interference efficiently.

2. Analysis Process
The antenna array is the planar array with reflector board as shown in Fig.1. The
reflector board can not only make the antenna meet the sector coverage requirement, but
also increase the array gain. The radiating element is the cylindrical dipole with diameter

0-7803-7846-6/03/%17.0002003 IEEE 686

Authorized licensed use limited to: ANNA UNIVERSITY. Downloaded on August 04,2023 at 09:05:52 UTC from IEEE Xplore. Restrictions apply.
 0. The elements of the same column along the 2 axis are excited with equal amplitude
and phase, whose number is decided by the gain request. A column is so-called one
element of smart antenna system and the analysis target here is a four-element array.
The antenna array is composed of linear and planar structure, which is simulated
with wire-grid model. The whole structure is analyzed with MM.

3. Optimization Process
In cellular mobile communication, a cell is divided into three or four sectors, to
increase the system capacity. The smart antenna system sends public controlling
information to all the users in sector and judges the amount and direction of users
according to the retumed information. So the antenna pattem should cover the whole
sector equably which is called the broadcast mode. Also the main beam needs to direct to
the expected user and null points are steered in the direction of interference, which is
called the working mode.
Genetic Algorithms are global search optimization algorithms. Because of their
versatility and ability to optimize in complex search process, they are on the rise in
electromagnetics as design tools and problem solvers. The detailed process can refer to
related reference. The problem about the fitness function will be discussed below.
According to the two modes, broadcast mode is first realized. The Half Power Beam
Width (HPBW), the ripple of main lobe and the first side lobe level (SLL) are targets, with
the element distance and element weight (phase only) being optimized. According to the
requirement, the cost function is designed as,

Fit = -ma&, (HP5W - HPBW, y, u2(Ripple - Rippleo)",u3(SLL - SLL, )"}(I)

where Fit is the cost function. a l a2, a3 are the respective constraint coefficients which
are used to control the contribution of three variables. HPBW. Ripple and SLL denote the
half power beam width, ripple of main lobe and the first side lobe level respectively. The
subscript 0 represents the desired values.
There has been much work on the algorithms for the working mode of smart
antenna system. The antenna array used as the front end of smart antenna system is
optimized by GA for the working mode after the broadcast mode is satisfied. The element

weight (amplitude, phase) is optimization variables. The expected user direction is @d

and the interference direction is # , (n=1,2,3) .The mainbeam with low sidelobe is

directed to the user direction and also the null is steered in the direction of interference.
The cost function is

687

Authorized licensed use limited to: ANNA UNIVERSITY. Downloaded on August 04,2023 at 09:05:52 UTC from IEEE Xplore. Restrictions apply.
 whereg(h”) ,(n=1,2,3) is the level of interference direction, go is the desired value.

4. Results
The working frequency of the antenna array shown in Fig.1 is 1880-1920MHz and
20102-2025MHz. The center frequency is chosen as the optimization frequency. Fig2
shows the broadcast mode pattern of the two frequency bands with detailed parameters
shown in Table 1. The HPBW changes with frequency in the two frequency bands are just
3.37%(1880-1920MHz) and 3.06% (2010-20251vWz), which are far smaller than the
industry standard *lo%. Fig.3 shows the optimized pattern with one expected user
direction and one, two and three interference directions. It can be seen that when the
number of interference directions is smaller than two, very deep nulls below 4 O d B are
obtained. For three interference directions, as the number of elements is limited, deep nulls
are not obvious but the level of null points are all below -33dB, which still can suppress
the interference efficiently.

5. Conclusion
GA is used to design the sector array antenna for smart antenna system. The antenna
array is analyzed with MM, which takes the mutual coupling into account. Two kinds of
modes are synthesized which show that the antenna array can not only cover the sector
uniformly, but also track desired user exactly and suppress the co-channel interference
efficiently.
Reference
Joseph C. Liberti, Jr. Bellcore, Theodore S. Rappaport. Smart Antennas for Wireless
Communications IS-95 and Third Generation CDMAApplications.
J.E.Evans, “Synthesis of equiripple sector antenna patterns”, IEEE Trans. Antenna Propagat.,
pp.347-353, May 1967.
ASafaai-Jazi, “A new formulation for the design of Chebyshev arrays”, IEEE Trans. Antenna
Propagat., Vo1.42,pp.439-443,Mar. 1994.
Daniel S. and Eric Michielssen. “Genetic Algorithm Optimization Applied to Electromagnetics: A
Review”. IEEE Trans. On AP. 199745(3): 343-353.
Table1 Parameters ofBroadcast Mode
f
HF’BW(”) Ripple(dB) Gain(dB)

2010 88.14 1.35 14.00

2018 187.69 I 1.18 113.96
I 2025 I 87.25 I 1.17 I 13.94 I
-x
Fig1 Sketch ofihe Amy
1920 89.94 13.39

688

Authorized licensed use limited to: ANNA UNIVERSITY. Downloaded on August 04,2023 at 09:05:52 UTC from IEEE Xplore. Restrictions apply.
 0- 0-

5- 5-
-10- -10-

-15- -15-

-20 - -20 -

.25
-30
-
-
-
0 -25-
c -M-
-25 - 8 -25-

-20 - -20 -
.15- -15-
-10- .*a7

5- 5-
0-
270

NULLI=-SI. 03d0(1137
NULLZ=-33.27dO (I25")

0 45 PO 135 I80 225 270 315 360

689

Authorized licensed use limited to: ANNA UNIVERSITY. Downloaded on August 04,2023 at 09:05:52 UTC from IEEE Xplore. Restrictions apply.