2654 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 59, NO.

7, JULY 2011

Spherical Horn Array for Wideband Propagation

Measurements
Ondřej Franek, Member, IEEE, and Gert Frølund Pedersen

Abstract—A spherical array of horn antennas designed to Horn antennas are well known for being used as primary re-
obtain directional channel information and characteristics is flector feeds, in electromagnetic compatibility measurements
introduced. A dual-polarized quad-ridged horn antenna with and for standard gain calibration purposes [4]. Their main ad-
open flared boundaries and coaxial feeding for the frequency band
600 MHz–6 GHz is used as the element of the array. Matching vantages are relatively wide bandwidth, high gain, and the pos-
properties and coupling between the elements are investigated sibility to accommodate both polarizations within one antenna.
via measurements and numerical simulations. Radiation patterns The bandwidth can be further enlarged by introducing ridges,
and sum beams of the array on selected frequencies throughout which, together with symmetrical feeding, substantially extend
the band are also presented. Based on the obtained results it is the waveguide single-mode operation [5]. As our measurement
concluded that the array is a good candidate for a wideband
multipath propagation studies. campaigns were expected to require wideband measurements
from 600 MHz to 6 GHz (i.e., LTE interfaces span from 700
Index Terms—Antenna arrays, antenna array mutual coupling, MHz up to 2.7 GHz) as well as both polarizations, we decided
horn antennas.
to use the ridged design with dual polarization, i.e., quadruple-
ridged horn antenna.
I. INTRODUCTION Double ridged horn antennas have been studied widely in
the literature, for examples we refer to [6]–[9]. Publications on
quad-ridged horns, on the other hand, are few. An early presen-
N order to measure the propagation channel characteristics
I without antenna influence and extensive postprocessing, a
spherical array of antennas is needed. The elements of the array
tation of quad-ridged design, although very brief, can be found
in [10]. A dual-polarized ridged horn with coaxial feeding and
bandwidth from 2 to 26.5 GHz was introduced in [11]. An-
should theoretically have medium gain, both polarizations, and other original design was presented in [12], where the quad-
cover wide frequency range. Also, in order to directly retrieve ridged horn intended for anechoic chamber operation in 2–18
the directional channel information and to obtain sufficiently GHz band was stripped off the flare boundaries while main-
isotropic coverage, we need moderately narrow radiation beam taining very good radiation and matching parameters. Recently,
covering the whole sphere. One possibility is to use a directional two novel designs of quad-ridged horn antennas, with full flare
antenna with the desired characteristics on a pedestal, scanning boundaries though, were presented in [13] and [14], with respec-
the space to obtain directional information [1], but this solu- tive bandwidths 8–18 GHz and 2–18 GHz and coaxial feeding.
tion cannot accommodate the velocities typically involved in However, none of the papers addressed the possibility of using
mobile experiments. Fast and simultaneous measurements are, such antenna as an array element.
however, feasible with spherical arrays. A spherical array com- In this paper, we introduce a spherical array of quad-ridged
posed of microstrip patch antennas was introduced in [2], but the dual-polarized horn antennas with frequency range 600 MHz-6
bandwidth offered by this type of antennas is generally poor. GHz. Focus is given on aspects relevant to using this array for
In our previous measurement campaign, we used a spherical obtaining channel characteristics and directional information. In
array of monopoles [3] with the view of increasing the band- particular, we present reflection coefficient , coupling
width. Although successful in some aspects, it did not allow and radiation patterns. The data were obtained from anechoic
a simple estimation of directional information because the chamber measurements and from simulation by the finite-dif-
elements of the array, monopoles, were not directional, instead ference time-domain (FDTD) method [15].
having maximum radiation around the circumference. This
made the determination of signal direction particularly difficult
in terms of postprocessing. For the next measurement campaign II. DESCRIPTION
we therefore decided to use horn antennas to compose the array.
A. The Antenna
Manuscript received August 18, 2010; revised November 08, 2010; accepted The element of the proposed array is a diagonal horn antenna
November 25, 2010. Date of publication May 10, 2011; date of current version with four ridges added for simultaneous use of both vertical
July 07, 2011. This work was supported by the Danish Center for Scientific
Computing.
and horizontal polarizations (see Figs. 1 and 2). The initial de-
The authors are with the Antennas, Propagation and Radio Networking Sec- sign of the horn antenna has been prepared using simulations
tion, Department of Electronic Systems, Aalborg University, DK-9220 Aalborg in our in-house FDTD code. Optimization has been performed
øst, Denmark (e-mail: [email protected]; [email protected]). by sweeping through many values for the feed position, the gap
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. between the ridges, the thickness of the ridges, the chamfering
Digital Object Identifier 10.1109/TAP.2011.2152349 angle near the feed, and the position of the shorting strip. The
0018-926X/$26.00 © 2011 IEEE
FRANEK AND PEDERSEN: SPHERICAL HORN ARRAY FOR WIDEBAND PROPAGATION MEASUREMENTS 2655

Fig. 1. The realization of the quad-ridged dual-polarized horn antenna.

Fig. 3. Detail of the feed with two ridges present: (a) side-view, (b) cross-sec-
tion. Arrangement of the perpendicular ridges is similar, only the feed is shifted
2 mm forward.

Fig. 2. Dimensions of the horn antenna.

optimization had strong influence on matching, while the radi-

ation pattern has been found to be largely insensitive to small Fig. 4. Simulations for various gaps between the ridges.
changes in the antenna geometry. Fine tuning of the parameters
has been done using measurements on a prototype.
The flare is designed with open boundary, but includes a perpendicularly to each other for excitation of both polariza-
shorting strip similar to [7] resulting from matching optimiza- tions. The probes are positioned 2 mm apart, counting for 14
tion; the strip is 20 mm wide at a distance of 80 mm from degrees of phase shift at the upper end of the frequency range.
the transition waveguide. The antenna is 348 mm long, with The gap between the ridges at the feedpoint is one of the most
an aperture size of 288 mm. The ridges are tapered approx- crucial dimensions of the antenna. While the tapering length was
imating an exponential transition between the characteristic kept constant, we eventually chose 1 mm as the optimum dis-
impedance of the feeding waveguide and free space. Note that tance between the ridges at the feedpoint with respect to the re-
the tapering of the ridges runs right from the feed in full length flection coefficient . We also found, however, that the ten-
of the antenna. Towards the feed the ridges are chamfered to dency of the reflection coefficient is generally not monotonic
accommodate the very narrow gap resulting from tapering and with the gap width (see Fig. 4 with FDTD simulations). This is
to achieve higher bandwidth and lower impedance [5]. in contrast to [16] and [17] which report that the matching im-
The antenna is fed by a coaxial probe—a coaxial cable passes proves with decreasing the gap between the ridges. Exact phys-
throughout the ridge with outer conductor connected to it and ical explanation remains unclear to us, but we hypothesize that
the inner conductor is attached to the opposite ridge (see Fig. 3). the size of the feed may be the cause, since the diameter of the
The passage is filled with dielectric material and keeps the char- inner conductor of the coaxial probe is 1 mm, comparable to the
acteristic impedance at 50 Ohms. There are two probes arranged gap itself.
2656 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 59, NO. 7, JULY 2011

Fig. 6. Distribution and numbering of the horn antennas on the spherical array;
(a) top view, (b) bottom view, (c) theta angles of the element rows.

Fig. 5. The quad-ridged horn antennas in the spherical array arrangement.

Thickness of the ridges has been determined by numerical

optimization as 8 mm. Reducing the overall thickness of the
ridges negatively influences the performance of the antenna and
also poses some practical difficulties, e.g., does not allow for
the coaxial feeding through the ridges. However, large portion
of the ridges has been milled out resulting in 2 mm thickness ev-
erywhere except a 10 mm wide frame at the edges—we did not
find any significant influence of this modification. The benefit
is reduced weight of the antenna, which is especially important
when it is used as an array element. Another component of the
antenna of which dimensions were numerically investigated was
the cavity behind the feed, but again without noticing any con-
siderable impact. Fig. 7. Reflection coefficient (s ) of the quad-ridged horn antenna. Solid line:
measurement, dashed line: simulation.

B. The Array
The antenna array presented in this paper has 16 previously this approach. All the mounting pods for the antennas were
described horn antennas arranged in a quasi-spherical pattern arranged in such a way that the antennas are positioned on
(Fig. 5), with emphasis on the coverage of the upper hemisphere. a sphere, i.e., with approximately equal distance from the
The elements in the array (schematic in Fig. 6) are arranged geometrical center of the array. The homogeneity of the final
into 4 groups (rows) with antennas sharing the same theta angle array turned out to be satisfactory, as shown below in results.
from the vertical axis in each group. The first group contains
one element pointing directly upwards, the second group has 4
III. RESULTS
elements with , the third group has 7 elements with
, pointing slightly above the horizon, and the fourth Fig. 7 shows the parameter of the horn antenna across the
group has 4 elements with , pointing slightly below intended operation frequency range, obtained by measurement
the horizon. and FDTD simulations. This plot is a result of extensive opti-
The described constellation has been chosen seeking the mization on the shape of the antenna, as we tried to find dimen-
highest possible homogeneity in obtaining the directional sions giving maximum possible bandwidth, favorable matching
channel information, but also with respect to practical matters and well-defined radiation beam, all at the same time. The re-
like mounting and mobility. In particular, the lower hemisphere sult is lower than dB in most of the frequency range of
is not covered entirely due to the presence of the mounting interest and lower than dB in few segments. However, there
rod and also because of generally lower probability of signal are notable differences between the measurement and the sim-
incoming from the downward direction. Generally, we were ulation which are caused by subtle mechanical changes on the
searching for the highest possible distance between any two final prototype (ridge gap, waveguide walls perforation) which
antenna elements in order to minimize the coupling. The pattern could not be reflected in the simulation. Differences appeared
homogeneity was, however, also expected to be improved by also when comparing radiation patterns, therefore we decided
FRANEK AND PEDERSEN: SPHERICAL HORN ARRAY FOR WIDEBAND PROPAGATION MEASUREMENTS 2657

Fig. 8. Measured mutual coupling (s parameter) between the two nearest Fig. 11. Measured radiation pattern for the antenna cluster in H- (solid line)
neighboring antennas (11–15) with both polarizations (solid: horizontal, dashed: and E-plane (dashed line), and for single antenna in H- (dash-dotted line) and
vertical). E-plane (dotted line), at 2.3 GHz.

Fig. 12. Measured radiation pattern for the antenna cluster in H- (solid line)
Fig. 9. Measured mutual coupling (s parameter) between the two ports of and E-plane (dashed line), and for single antenna in H- (dash-dotted line) and
the same antenna. E-plane (dotted line), at 4.5 GHz.

One of the most important parameters of any antenna array is

the amount of mutual coupling between the array elements. The
results for the worst coupling in the array (approx. 40 apart) are
shown in Fig. 8. The maximum coupling in the frequency band
of interest occurs at the lower end with magnitude at most
dB. Only the ports with the same orientation (H-H, V-V) are
significantly coupled as shown in Fig. 8, whereas mixed polar-
ization coupling is less than dB in all cases (not shown).
It is the H-plane coupling (V-V) which turned out to be the
strongest. The coupling between the two ports of the same an-
tenna is shown in Fig. 9.
The radiation pattern is quite insensitive to small changes in
the geometry of the antenna. However, its properties change
throughout the band as it is demonstrated for four selected
Fig. 10. Measured radiation pattern for the antenna cluster in H- (solid line)
frequencies: 776 MHz, 2.3 GHz, 4.5 GHz and 6 GHz (see
and E-plane (dashed line), and for single antenna in H- (dash-dotted line) and Figs. 10–13). In this case, the measurement setup consisted of
E-plane (dotted line), at 776 MHz. the measured element and four neighboring identical antennas
tilted by 40 from the main axis (“antenna cluster”), in order to
capture an image of the real-world performance of the antenna
to rely entirely on measurement in obtaining the remaining an- as a part of the array. Radiation patterns for the single antenna
tenna parameters. only are also included in Figs. 10–13 for comparison.
2658 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 59, NO. 7, JULY 2011

Fig. 13. Measured radiation pattern for the antenna cluster in H- (solid line)
and E-plane (dashed line), and for single antenna in H- (dash-dotted line) and
E-plane (dotted line), at 6 GHz. Fig. 15. Measured beamwidth (full-width half-maximum) in H- (solid line) and
E-plane (dashed line) versus frequency.

Fig. 16. Sum beam of array (relative power to maximum),  -polarization, 776
Fig. 14. Measured directivity versus frequency. MHz.

The directivity (directive gain) grows with frequency

(Fig. 14) and the single beam becomes narrower (Fig. 15),
while the front-to-back ratio improves. The beam is generally
wider in the H-plane than in the E-plane. At the upper frequency
limit (6 GHz, Fig. 13) the beam is still intact with a wide flat
top in H-plane and mild sidelobes in E-plane. Still, some of
the sidelobes occur throughout the entire band, but these have
generally lower level and they are also pointing around 40
degrees or more from the main beam, suggesting that they are
caused by the influence of the four neighboring antennas.
Figs. 16–21 show the sum beams in -and -polarization of
the whole spherical array composed of 16 horn antennas. The
sum beam is obtained by summing power contributions from Fig. 17. Sum beam of array (relative power to maximum), -polarization, 776
all of the antennas in the array and both polarization feeds. MHz.
It shows how homogeneous the array will be in retrieving the
directional channel information over the sphere. Each antenna
is represented by a full 3D radiation pattern in both polariza- pole”. The horizontal radiation, or “the equator”, is denoted by
tions, including the influence of its four neighbors. These pat- a dashed line at half distance from the center to the circumfer-
terns are then rotated into the appropriate direction and added to- ence.
gether. The plot center in Figs. 16–21 corresponds to the “north From the schematic of the array (Fig. 6) it follows that the
pole” of the spherical domain (radiation upwards) while the cir- coverage in the downwards direction will be minimal, and this
cumference corresponds to the downwards radiation, or “south is indeed the case. The homogeneity is generally better when
FRANEK AND PEDERSEN: SPHERICAL HORN ARRAY FOR WIDEBAND PROPAGATION MEASUREMENTS 2659

Fig. 18. Sum beam of array (relative power to maximum),  -polarization, 2.3 Fig. 21. Sum beam of array (relative power to maximum), -polarization, 6.0
GHz. GHz.

upper hemisphere (inner part of the circle), while at 2.3 GHz we

have only 4 dB and for 6 GHz even 6 dB span due to sporadic
dips. These numbers apply to a 16-element array. For 32 and 62
elements in the array (uniform distribution around the sphere,
not displayed) the homogeneity would be better than 6 dB and
4 dB, respectively, for both hemispheres at 6 GHz, but then the
array would have a larger diameter and corresponding weight.

IV. CONCLUSION
In this paper, a spherical array with dual-polarized
quad-ridged horn antenna as an element has been introduced.
The 16-element array has an isotropy of max. 6 dB over the
sum radiation pattern in the upper hemisphere within frequency
range from 600 MHz to 6 GHz. The reflection coefficient of
Fig. 19. Sum beam of array (relative power to maximum), -polarization, 2.3
GHz. the horn antenna is below dB across the frequency range
and the coupling between the elements is typically better than
dB. The radiation pattern of a single antenna in the array
shows a single well-defined beam with small sidelobes and
good front-to-back ratio. The directivity reaches 12 dB for the
most part of the band. Although there was a quite significant
proximity effect of the neighboring antennas on the radiation
pattern, this did not manifest in the coupling. We therefore
conclude that the presented array is suitable for wireless prop-
agation studies with potential for accurate estimation of signal
direction.

ACKNOWLEDGMENT
The authors would like to thank the TAP reviewers and Dr.
E. de Carvalho for their useful comments on the manuscript.

REFERENCES
Fig. 20. Sum beam of array (relative power to maximum),  -polarization, 6.0 [1] M. Knudsen and G. Pedersen, “Spherical outdoor to indoor power spec-
GHz. trum model at the mobile terminal,” IEEE J. Sel. Areas Commun., vol.
20, no. 6, pp. 1156–1169, Aug. 2002.
[2] K. Kalliola, H. Laitinen, L. Vaskelainen, and P. Vainikainen, “Real-
the antenna beam is wider, as at the lower frequency (776 MHz, time 3-D spatial-temporal dual-polarized measurement of wideband
radio channel at mobile station,” IEEE Trans. Instrum. Meas., vol. 49,
Figs. 16, 17), or when it has many secondary sidelobes, which no. 2, pp. 439–448, Apr. 2000.
can be observed at the upper end of the band (6 GHz, Figs. 20, [3] O. Franek, G. Pedersen, and J. Andersen, “Numerical modeling of a
21). However, at higher frequencies there is also a higher prob- spherical array of monopoles using FDTD method,” IEEE Trans. An-
tennas Propag., vol. 54, no. 7, pp. 1952–1963, July 2006.
ability of sharp dips in the sum radiation pattern. In particular, [4] C. Balanis, Modern Antenna Handbook. New York: Wiley-Inter-
the homogeneity of the sum beam at 776 MHz is within 3 dB for science, 2008.
2660 IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION, VOL. 59, NO. 7, JULY 2011

[5] W. Sun and C. Balanis, “Analysis and design of quadruple-ridged Ondřej Franek (S’02–M’05) was born in 1977. He
waveguides,” IEEE Trans. Microw. Theory Tech., vol. 42, no. 12, pp. received the M.Sc. (Ing., with honors) and Ph.D. de-
2201–2207, Dec. 1994. grees in electronics and communication from Brno
[6] J. Kerr, “Short axial length broad-band horns,” IEEE Trans. Antennas University of Technology, Czech Republic, in 2001
Propag., vol. 21, no. 5, pp. 710–714, Sep. 1973. and 2006, respectively.
[7] C. Bruns, P. Leuchtmann, and R. Vahldieck, “Analysis and simulation Currently, he is working in the Department of
of a 1–18 GHz broadband double-ridged horn antenna,” IEEE Trans. Electronic Systems, Aalborg University, Denmark,
Electromagn. Compat., vol. 45, no. 1, pp. 55–60, Feb. 2003. as a Postdoctoral Research Associate. His research
[8] V. Rodriguez, “New broadband EMC double-ridge guide horn an- interests include computational electromagnetics
tenna,” RF Design, vol. 27, no. 5, 2004. with focus on fast and efficient numerical methods,
[9] M. Botello-Perez, H. Jardon-Aguilar, and I. Ruiz, “Design and simula- especially the finite-difference time-domain method.
tion of a 1 to 14 GHz broadband electromagnetic compatibility DRGH He is also involved in research on biological effects of non-ionizing elec-
antenna,” in Proc. 2nd Int. Conf. on Electrical and Electronics Engi- tromagnetic radiation, indoor radiowave propagation, and electromagnetic
neering, Sep. 2005, pp. 118–121. compatibility.
[10] S. Soroka, “A physically compact quad ridge horn design,” in Proc. Dr. Franek was the recipient of the Seventh Annual SIEMENS Award for
Antennas and Propagation Society Int. Symp., Jun. 1986, vol. 24, pp. outstanding scientific publication.
903–906.
[11] Z. Shen and C. Feng, “A new dual-polarized broadband horn antenna,”
IEEE Antennas Wireless Propag. Lett., vol. 4, pp. 270–273, 2005.
[12] V. Rodriguez, “An open-boundary quad-ridged guide horn antenna for Gert Frølund Pedersen was born in 1965. He
use as a source in antenna pattern measurement anechoic chambers,” received the B.Sc. E.E. degree (with honour) in
IEEE Antennas Propag. Mag., vol. 48, no. 2, pp. 157–160, Apr. 2006. electrical engineering from College of Technology
[13] R. Dehdasht-Heydari, H. Hassani, and A. Mallahzadeh, “Quad ridged in Dublin, Ireland, and the M.Sc. E.E. and Ph.D.
horn antenna for UWB applications,” Progr. Electromagn. Res., vol. degrees from Aalborg University, Denmark, in 1993
79, pp. 23–38, 2008. and 2003, respectively.
[14] R. Dehdasht-Heydari, H. Hassani, and A. Mallahzadeh, “A new 2–18 He has been employed by Aalborg University
GHz quad-ridged horn antenna,” Progr. Electromagn. Res., vol. 81, pp. since 1993 where he is now Full Professor heading
183–195, 2008. the Antennas, Propagation and Radio Networking
[15] A. Taflove, Computational Electrodynamics: The Finite-Difference Group and is also the Head of the Doctoral School
Time-Domain Method, 3rd ed. Boston: Artech House, 2005. on Wireless which has close to 100 Ph.D. students
[16] M. Abbas-Azimi, F. Arazm, and J. Rashed-Mohassel, “Sensitivity enrolled. His research has focused on radio communication for mobile ter-
analysis of a 1 to 18 GHz broadband DRGH antenna,” in Proc. minals, and especially on small antennas, diversity systems, propagation and
IEEE Antennas and Propagation Society Int. Symp., Jul. 2006, pp. biological effects, and he has published more than 75 peer reviewed papers
3129–3132. and holds 20 patents. He has also worked as consultant for developments
[17] M. Kujalowicz, W. Zieniutycz, and M. Mazur, “Double-ridged of more than 100 antennas for mobile terminals including the first internal
horn antenna with sinusoidal ridge profile,” in Proc. Int. Conf. on antenna for mobile phones in 1994 with lowest SAR, first internal triple-band
Microwaves, Radar & Wireless Communications, May 2006, pp. antenna in 1998 with low SAR and high TRP and TIS, and lately various
759–762. multi antenna systems rated as the most efficient on the market. He has been
one of the pioneers in establishing the over-the-air measurement systems. The
measurement technique is now well established for mobile terminals with
single antennas and he is now chairing the COST2100 SWG2.2 group with
liaison to 3 GPP for over-the-air tests of MIMO terminals.