Chamber Array Antenna Layout for Compact OTA Measurements

Full text

Chamber Array Antenna Layout for
Compact OTA Measurements
Mohammad Poordaraee1and Andr´
es Alay´
on Glazunov1,2
1Department of Electrical Engineering, University of Twente, Enschede, The Netherlands.
2Department of Electrical Engineering, Chalmers University of Technology, Gothenburg, Sweden.
Abstract—An optimized irregular planar array antenna lay-
out with uniform excitation of antenna elements is presented
for Random-LOS OTA (Random Line-Of-Sight Over-The-Air)
characterization setups. A plane wave is synthesized within a
cylindrical 3D test zone at 2.7GHz, which can be scaled to
mmWave frequencies. The obtained thinned array achieves a 52%
reduction of the number of elements and a 45% aperture size
as compared to a uniform fully populated planar array with an
inter-element distance of 0.93λ, which is the optimum distance
through [λ/2, λ]based on the presented cost function at this
paper . The obtained maximum phase deviation and the maxi-
mum field amplitude deviation from the average field distribution
in the 3D test zone of the proposed optimized chamber array
antenna layout are approximately 6.4◦and 3.9dB, respectively.
The numerical computation of the radiation pattern of a 10 ×10
element uniform planar array antenna as AUT (Antenna Under
Test) placed within the test zone was performed too. Results show
a rather good agreement between the radiation pattern to that
obtained assuming far-field conditions.
I. INTRODUCTION
The Over-The-Air (OTA) performance evaluation of wire-
less devices and antenna systems is in high demand since
conducted measurements standards are becoming obsolete.
Massive multiple-input multiple-output (MIMO) antenna sys-
tems are in focus due to their large physical sizes, e.g., at
sub-6GHz bands. Compact OTA measurement setups are
desirable to keep costs down. For example, in a Random Line-
Of-Sight (Random LOS) OTA setup, far-field-like conditions
ought to be emulated within the test zone [1], [2]. The
objective is to synthesize a plane wave within the test zone,
e.g., of cylindrical shape, in the near-field of the chamber
array antenna. The emulated plane wave field can be used
to experimentally verify the performance of massive MIMO
systems or, in general, any antenna systems [3], [4], [5],
[6]. In [7] a numerical method was proposed to synthesize
a plane wave in the near field, while in [8], [9], the fully-
populated uniform linear and uniform planar array antennas
were considered. In [10], thinned planar array antenna layouts
were produced by means of the Genetic Algorithm (GA).
The optimized thinning minimized the fluctuations of the field
intensity, but not the phase fluctuations. In this paper, we
consider the joint minimization of the phase and the intensity
fluctuations of a vertically polarized wave field within the test
zone by means of the GA. We further compute the radiation
pattern of an AUT with 10×10 elements in the test zone of the
proposed optimized irregular planar array (OIPA) antenna and
Fig. 1. Planar array layouts for compact OTA measurement setup.
compare it with the numerical computation of the theoretical
radiation pattern of the considered AUT.
II. CO MPAC T OTA SETUP AND IRREGULAR PLANAR
AR RAY LAYO UT D ESIGN PROC ED URE
Fig. 1 shows a schematic representation of a Random-LOS
OTA setup emulating a plane wave field in a cylindrical test
zone of width 2Rand height Hin an anechoic chamber. The
AUT, e.g., a base station or an array antenna may be placed
on a turn table at a distance Dfrom the chamber antenna in a
shielded anechoic or semi-anechoic chamber. The performance
of the AUT could then be evaluated by rotating the platform
which axis coincides with the axis of the cylindrical test
zone. The center of the planar chamber array antenna and
the center of the imaginary cylinder are aligned as shown in
Fig. 1. The center of the chamber array is selected as center of
the Cartesian coordinate system to perform the optimization
procedure explained in the following.
Two Figures of Merit (FoM) are considered for character-
izing the uniformity of the plane wave. The first FoM is used

to evaluate the maximum field intensity ratio and is given by
EF I Rmax = max {|EF I R|} ,(1)
where max {.}yields the maximum and parameter EF I R is
the Electrical Field Intensity Ratio in dB scale and defined as
follows
EF I R = 20 log |Ez|
Eavg .(2)
where |Ez|is the magnitude of the z-component of the
electrical field at any point within the test zone and Eavg =
MEAN {|Ez|} is the average value of the magnitude of the
z-component of the electrical field. The average is taken over
the points lying in the central parallel plane within test zone,
as shown in Fig. 1. As mentioned above, the antenna ele-
ments of the chamber array are vertically polarized, thus only
the z-component of the electrical field is considered. Other
components are negligible compared to the z-component. The
second FoM evaluates the phase fluctuations as the maximum
absolute deviation from the mean value on the central parallel
plane to the chamber array defined as
∆φmax = max {| − k(D−x)−MEAN {φ},|} .(3)
where k= 2π/λ is the wave number, λis the free space
wavelength at the operation frequency, φ= arg {Ez}is the
phase of the complex amplitude of z-component of the E-field
at a point within the zone. The x-coordinate of each point
within test zone is used to compute −k(D−x)to compensate
the phase difference between the central parallel plane and
the other parallel planes within the test zone. MEAN {φ}is
the average phase on the central parallel plane. For an ideal
plane wave, the phase and magnitude fluctuations of the E-
Field shall be zero. Therefore, the optimization problem aims
at minimizing the linear combination of the EF I Rmax and
∆φmax within the cylindrical test zone.
min
d,nj
αEF I Rmax + ∆φmax (4a)
s.t. λ/2≤d < λ, (4b)
nj= 0,1, j = 1 . . . Nelem,(4c)
where one of the optimization parameters is the inter-element
distance d, which is set to be the same in both the vertical
and the horizontal dimensions of the chamber array (zand
yaxes in Fig. 1). The constraint λ/2≤d≤λis used to
prevent mutual coupling effects as well as the appearance
of grating lobes. The other optimization parameter is the
weighting coefficient njtaking binary values 1or 0depending
on whether an antenna element is present or not in the
array layout, respectively. The index jdenotes the number
of an antenna element and Nelem denotes the total number
of antenna elements of the fully populated array. An element
with zero weighting coefficient nj= 0 indicates that it can
be removed from the planar array which leads to reduction
of chamber antenna cost and complexity. The parameter αis
fixed coefficient in order to establish a balance between the
impacts of EF I Rmax and ∆φmax in the cost function and is
chosen empirically.
(a) (b)
Fig. 2. Phase and field intensity distribution over the central parallel plane
as shown in Fig. 1 at x= 3.5m a) φnorm[°], b) EF I R [dB].
(a) (b)
(c) (d)
Fig. 3. Phase and field intensity distribution over a square 5×5m2area on
the x-yplane at z= 0 m a) ∆φnorm , [°], b) EF I R, [dB]. Phase and field
intensity distribution over circular test zone area with R= 0.5m on the x-y
plane at z= 0 m as shown in Fig. 1, c) φnorm, [°] and d) E F IR, [dB] .
III. SIMULATION RES ULTS
The simulations are specialized to synthesize a plane wave
field within the 3D test zone, i.e., an imaginary cylinder with
dimensions R= 0.5m and H= 1 m. As a starting point, a
fully populated uniform planar array with Nelem = 23 ×23 =
529 isotropic antenna elements and inter-element distances
d=λat f= 2.7GHz is assumed [8].
In order to consider the phase fluctuations within the volume
of the test zone, ∆φmax and EF I Rmax are computed through-
out the test zone. Furthermore, to balance the contributions of
the phase and amplitude variations in (4), α= 1.4was chosen.
Indeed, α= 1.4means that 1dB increase in EF I Rmax
changes the value of the cost function to a similar amount as
a1.4◦increase in ∆φmax. Hence, the influence of phase and
intensity variation of the electrical field on the cost function
could be tuned by α. It should be stressed that αwill impact

TABLE I. Field uniformity and Aperture size (Ap. size) comparison.
FPPA Fully Populated Planar Array, OIPA Optimized Irregular Planar Array,
Ref.[10]
FoM ∆φmax EF I Rmax Nelem Ap. size d/λ
FPPA 23.5°3.9dB 529 86% ×23λ×23λ0.93
OIPA 6.38°3.9dB 255 48% ×23λ×23λ0.69
Ref.[10] 46.5°4.8dB 291 35% ×23λ×23λ0.59
(a)
(b)
Fig. 4. Plane wave uniformity evaluation on planes parallel to z-yas function
of x, (a) φmax and (b) σdB.
the optimized configuration.
The OIPA antenna configuration optimized by GA is shown
in Fig. 1. The optimized parameters for the optimized irregular
planar array are d= 0.0763 m = 0.69λwith Nelem = 255.
The colored circles show the positions populated by the
antenna elements, while the white circles where an antenna
element is absent as a result of the optimization. The op-
timization, produces an array with a 52% reduction of the
antenna elements and a 45% reduction of array aperture. The
phase and amplitude distributions of the E-field on the central
parallel (z-y) plane at x= 3.5m are plotted in Figs. 2 (a) and
(b). Fig. 3 shows the E-Field distribution on the perpendicular
plane to the chamber array plane at z= 0 (x-yplane). Planar
test zone areas are outlined in Figs. 3 (a) and (b) by circles.
For the sake of comparison we also considered two addi-
tional planar array layouts, i.e., the Fully Populated Planar
Array (FPPA) with identical inter-element distance of 0.93λ
in both directions and the irregular array in [10].
Figs. 4 (a) and (b) show the plane wave uniformity at
each z-yparallel plane within the test zone as function of
(a)
(b) (c)
Fig. 5. Computed radiation pattern of an uniform planar array with 10 ×10
elements and symmetric inter-element space of λ/2in both directions, within
the test zones of FPPA, OIPA and Ref. [10] arrays. a) full radiation pattern
b) magnified first null c) magnified first side lobe.
x[m], evaluated as ∆φmax and EF I Rmax on each parallel
plane, respectively. As can be seen from Fig. 4, Ref.[10]
has the largest phase fluctuations, since [10] is not optimized
for low variation of phase. Also the proposed array layout
has the lowest phase deviation. In terms of EF I Rmax on
parallel planes, FPPA and OIPA have better performance in
comparison with [10] to implement more uniform amplitude
although the OIPA provides a layout implementation that is
simpler and more cost-effective. A summary of the FoMs
of the wave field uniformity (plane wave) for all of three
layouts is given in Table I. An OTA setup serves the purpose
of characterizing a device’s performance. Therefore we also
compare the performance of the aforementioned array layouts
by calculating the radiation pattern of an AUT in the test zone.
For this purpose we consider an uniform planar array with
10 ×10 elements with inter-element distance of λ/2in both
dimensions. The comparison of the main beam level, the side
lobe levels and nulls is obtaining by computing the radiation
pattern of the AUT. The results obtained assuming different
OTA array layouts are compared with the computed radiation
pattern of the theoretical (ideal) reference case (Ref. pattern)
as shown in Fig. 5. To better highlight the differences, the
first null and side lobe level are shown separately in Figs. 5
(b) and (c), respectively. As can be seen from Fig. 5 (b) the
OIPA antenna “reproduces” the null more precisely than the
FPPA and the layout in Ref.[10]. FPPA, OIPA and Ref.[10]

have 21.2dB, 1.58 dB and 29.26 dB difference in comparison
with theoretical pattern in absolute value of the null shown in
Fig. 5 (b), respectively. For the absolute peak value of the first
side lobe level as shown in Fig. 5 (c) there are 2.47 dB, 0.7
dB and 0.7dB difference between FPPA, OIPA , Ref.[10] and
value of the theoretical pattern, respectively. Hence, the OIPA
antenna is an improvement as compared with the previously
presented result.
IV. CONCLUSIONS
In this paper we compare various layouts of planar arrays
used to synthesize a plane wave field within a 3D cylindrical
test zone for compact OTA applications. It is concluded that
among the considered layouts and optimization criteria the
irregular planar array with uniform excitation of elements
where minimization of phase and amplitude fluctuations has
been done over the 3D volume gives the better results as
compared to the other methods and the resulting array an-
tennas. The obtained array antenna has 52% less antenna than
the fully populated counterpart. The difference between the
absolute value of the first null of the radiation pattern obtained
from the optimized irregular planar array and the theoretical
pattern is 1.58 dB. On the other hand the same difference for
considered fully populated planar array antenna 21.2dB. The
corresponding differences between the absolute value of the
first side lobe levels are 0.7dB and 2.4dB, respectively. This
result shows that a Random-LOS OTA measurement setup
with a chamber array which layout has been optimized over
a 3D test volume can be used to measure an antenna’s (or a
device’s) radiation pattern with good accuracy.
V. ACKNOWL ED GMENT S
This work was supported by the European Union’s H2020:
ITN program for the “mmWave communications in the Built
Environments - WaveComBE” project under the grant no.
766231.
REFERENCES
[1] P.-S. Kildal et al., “Cost-effective measurement setups for testing wire-
less communication to vehicles in reverberation chambers and anechoic
chambers,” IEEE Conference on Antenna Measurements & Applications
(CAMA), Antibes Juan-les-Pins, 2014, pp. 1-4.
[2] A. A. Glazunov, “Test zone characterization in an automotive random-
LOS OTA test setup,” IEEE-APS Topical Conference on Antennas and
Propagation in Wireless Communications (APWC), Verona, 2017, pp.
170-173.
[3] P.-S. Kildal and J. Carlsson, “New approach to OTA testing: RIMP and
pure-LOS reference environments &amp; a hypothesis,” 7th European
Conference on Antennas and Propagation (EuCAP), Gothenburg, 2013.
[4] A. Razavi et al., “Characterizing Polarization-MIMO Antennas in
Random-LOS Propagation Channels,” in IEEE Access, vol. 4, pp.
10067-10075, 2016.
[5] M. S. Kildal et al., “Initial measured OTA throughput of 4G LTE
communication to cars with roof-mounted antennas in 2D random-LOS,”
International Symposium on Antennas and Propagation (ISAP), Hobart,
TAS, 2015, pp. 1-4.
[6] S. M. Moghaddam et al., “Improvement of an octave bandwidth bowtie
antenna design based on the analysis of a MIMO efficiency metric in
Random-Los,” Microw. Opt. Technol. Lett., 59: 1229-1233, 2017.
[7] D. A. Hill, “A Numerical Method for Near-Field Array Synthesis”, IEEE
Transactions on Electromagnetic Compatibility, vol. 27, no. 4, pp. 201-
211, 1985.
[8] M. S. Kildal, J. Carlsson and A. A. Glazunov, “Measurements and
Simulations for Validation of the Random-LOS Measurement Accuracy
for Vehicular OTA Applications,” IEEE Transactions on Antennas and
Propagation, vol. 66, no. 11, pp. 6291-6299, Nov. 2018.
[9] A. A. Glazunov, A. Razavi and P.-S. Kildal, “Simulations of a planar
array arrangement for automotive Random-LOS OTA testing,” 10th
European Conference on Antennas and Propagation (EuCAP), Davos,
2016.
[10] M. Poordaraee, A. A. Glazunov, “Plane Wave Synthesis with Irregular
Chamber Planar Antenna Arrays for Compact OTA Measurements,” 13th
European Conference on Antennas and Propagation (EuCAP), Krakow,
2019.