Real-time 3-D spatial-temporal dual-polarized measurement of wideband radio channel at mobile station

Abstract

This paper describes a measurement system enabling the complete
real-time characterization of the wideband radio channel. The system is
based on a wideband radio channel sounder and a spherical antenna array,
and it aims to describe the three-dimensional (3-D) spatial radio
channel seen by the mobile terminal, including polarization. This
information is highly valuable in designing antennas for mobile
terminals. The spatial properties of the measurement system are analyzed
through test measurements in an anechoic chamber. The system has a
40° spatial resolution and a 17 dB cross polarization
discrimination. The values are well above those of a small mobile
terminal antenna. The dynamic range in the spatial domain is 12 dB. The
measurement is very fast, which makes real-time channel acquisition
practical at normal mobile speeds

Full text

Page 1

IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 49, NO. 2, APRIL 2000 439
Real-Time 3-D Spatial-Temporal Dual-Polarized
Measurement of Wideband Radio Channel
at Mobile Station
Kimmo Kalliola, Heikki Laitinen, Leo I. Vaskelainen, and Pertti Vainikainen, Member, IEEE
Abstract—This paper describes a measurement system en-
abling the complete real-time characterization of the wideband
radio channel. The system is based on a wideband radio channel
sounder and a spherical antenna array, and it aims to describe
the three-dimensional (3-D) spatial radio channel seen by the
mobile terminal, including polarization. This informationis highly
valuable in designing antennas for mobile terminals. The spatial
properties of the measurement system are analyzed through test
measurements in an anechoic chamber. The system has a 40?
spatial resolution and a 17 dB cross polarization discrimination.
The values are well above those of a small mobile terminal
antenna. The dynamic range in the spatial domain is 12 dB.
The measurement is very fast, which makes real-time channel
acquisition practical at normal mobile speeds.
Index Terms—Conformal antennas, microstrip antennas, mi-
crowave measurements, mobile antennas, mobile communication,
multipath channels, polarization, radio propagation.
I. INTRODUCTION
T
cient use of the limited frequency spectrum requires thorough
understanding of thewideband radio channel, alsoin the spatial
domain. At the mobile terminal the spatial information can be
exploitedbymatching theantennaproperlytotheradioenviron-
ment, by correct design of both the powerand polarization radi-
ation patterns. The advantages are in lower transmission power
and less radiation toward the user, which lead to increased bat-
tery lifetimes and lower interference levels in the networks. In
addition, space and polarization diversity in reception are seen
asapplicableinmobileterminalsoperatingincomplexradioen-
vironments. The obtained diversity gain depends on the spatial
properties of the radio channel.
Traditionally, mobile antennas have been evaluated based on
their radiation properties, e.g., gain, measured in an anechoic
chamber. However, the actual performance of a mobile handset
HEdemandforhigherdataratessetshighrequirementsfor
the future wireless mobile communication systems. Effi-
Manuscript receivedMay 26, 1999; revised February 3, 2000. This work was
supported by the Wihuri Foundation, Tekniikan edistämissäätiö, Academy of
Finland,TechnologyDevelopmentCentreofFinland,Nokia,Sonera,andFinnet
Group.
K. Kalliola is with Helsinki University of Technology, Radio Laboratory,
FIN-02015 HUT, Finland, and with Nokia Research Center, FIN-00045 Nokia
Group, Finland (e-mail: kimmo.kalliola@hut.fi; kimmo.kalliola@nokia.com).
H. Laitinen and L. I. Vaskelainen are with VTT Information Technology,
Telecommunications, FIN-02044 VTT, Finland (e-mail: heikki.laitinen@vtt.fi;
leo.vaskelainen@vtt.fi.).
P. Vainikainen is with Helsinki University of Technology, Radio Laboratory,
FIN-02015 HUT, Finland (e-mail: pva@radio.hut.fi).
Publisher Item Identifier S 0018-9456(00)03584-1.
antennainarealisticoperationenvironmentdependsonitsmean
effective gain (MEG) [1], which is the ratio of the power re-
ceived by the antenna and the total mean incident power. In ad-
dition to the three-dimensional (3-D) radiation pattern of the
antenna, MEG depends on the 3-D angular density functions
of both the vertically polarized (VP) and horizontally polarized
(HP) components of the incident waves in the environment, and
on thecross polarization power ratio(XPR). These could beob-
tained through simulations, but the complexity of the operating
environments makes the modeling of the spatial radio channel
seenbythemobileveryobscure.Particularly,thespatialchannel
models have to be extended to three dimensions at the mobile
end, due to the scatterers close to the mobile e.g., inside build-
ings or in street canyons, that cause signal spreading in three
dimensions.
Deterministic indoor channel models are usually based on
ray-tracing or finite-difference time-domain (FDTD) methods
[2]. The problem with those is that they are computationally
complex, and also require a very detailed description of the par-
ticular environment, which makes the 3-D case even more diffi-
cult. Stochastic models for the spatial radio channel have also
been developed [3], but there is a need for measurements to
verify the models, and to adjust the parameters for different en-
vironments.
The 3-D angular density functions for the VP and HP signal
components at the mobile station could be obtained by per-
forming extensive measurement campaigns in realistic environ-
ments.Previously,thesyntheticaperturemeasurementapproach
has been applied for 3-D radio channel measurements [4]. Also
directional scanning with a narrow-beam antenna has been used
[1], [5]. Both methods are, however, very slow and thus infea-
sible for collecting statistically significant amounts of data for
directional signal distributions. In addition, they do not allow
the characterization of the time-variant behavior of the channel
(e.g., transients), and thus disable the verification of real dy-
namic channel models. Such would be needed for extensive
evaluationof mobile terminal antennas in realisticoperating en-
vironments.
A method for real-time spatial radio channel measurements
has been presented in [6]. The method is based on using a com-
plex wideband radio channel sounder together with a fast RF
switch and an antenna array. The directions-of-arrival (DOA)
of the incident waves can be estimated based on the relative
phases of the array elements. So far the reported real-time spa-
tial channel measurements have been limited to the azimuthdo-
main [6], [7]. Typically these aim to describe the spatial radio
0018–9456/00$10.00 © 2000 IEEE
Copyright 2000 IEEE. Reprinted, with permission, from IEEE Transactions on Instrumentation and Measurement, Volume 49, No. 2, April 2000, pp. 439-448.

Page 2

440IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 49, NO. 2, APRIL 2000
environment seen from a base station (BS) of a cellular system,
where normally the elevation angles of incident waves do not
vary significantly.
In this paper, we apply the same method for the real-time
spatial channel measurement in the 3-D case of the mobile
station. The measurement enables the complete real-time
spatio-temporal characterization of the wideband mobile radio
channel. The excess delay, azimuth angle, elevation angle,
polarization state, amplitude, and phase of the multipath
components of the transmitted signal can be measured. The
fast measurement allows also the calculation of the Doppler
spectrum of each wave. The 3-D spatial measurement is based
on a dual-polarized spherical antenna array.
The measurement system is described, and two different ap-
proaches to resolve the spatial information are presented and
compared in this paper. The directional properties of the mea-
surement system including angular resolution, spatial dynamic
range, cross polarization discrimination, accuracy of measured
incidence angle, and amplitude and phase ripple versus inci-
dence angle are defined with test measurements.
II. DESCRIPTION OF MEASUREMENT SYSTEM
The measurement system is based on a spherical array of 32
dual-polarized antenna elements and a complex wideband radio
channel sounder [8]. A wideband signal is transmitted using a
single antenna with linear polarization, and received separately
with each of the 32 elements and both ? and ? polarizations,
using a fast 64-channel RF switch. The carrier frequency of the
sounder is 2.154GHz,and thechipfrequency of themodulating
pseudo-random sequence in the transmitter can be selected to
be up to 30 MHz, resulting in delay resolution of 33 ns. In
the receiver, the demodulated signal is divided into ?- and ?-
branches and sampled with two 120 Msps A/D-converters. The
signal samples from each branch of the switch are then stored
for off-line processing to obtain the complex impulse response
(IR) of thechannel corresponding to each element and polariza-
tion.
A. Array Geometry
The ideal way to cover the whole ?? angle in space with di-
rection-independent angular resolution would be to use equally
spaced elements on the surface of a sphere. However, it is not
possibletodistributeanynumberofpointsonaspherewitheach
oneequidistant from itsneighbors.In other words,regular poly-
hedra exists only for certain numbers of vertices, and 32 is not
one of them.
The geometry of the built array is based on the dual of the
Archimedian solid, the truncated icosahedron, which is prob-
ably best known today as the geometry of the soccer ball. One
element is located in each vertex of the polyhedron, and two
differentdistances betweenadjacent elementsexist. The nearest
neighbors of an element are always at a distance of ?????? and
the next at ?????? from the element, where ? is the distance
from the center of the structure to each element. Every element
has either five or six such neighbors. It is shown in [9] that this
arrangement minimizes the Coulombic potential of a system of
Fig. 1.Geometry of the microstrip patch element used in the spherical array.
32 point charges distributed on the surface of a sphere, thus ap-
proximating an even distribution.
B. Array Element
The main requirements for the element of the array are two
separate feeds for orthogonal polarizations with high cross po-
larization discrimination (XPD), and sufficient bandwidth to
meet the requirement of the measurement system. The used el-
ement is a stacked microstrip patch antenna with separateprobe
feeds for ? and ? polarizations [10]. The geometry of the an-
tennaelementis presentedinFig.1.Fig.2presentstheradiation
pattern of the element, measured with all elements attached to
the spherical array.
The 6-dBbeamwidthof theelement is90?inthe?-planeand
100?in the ?-plane. The XPD is better than 18 dB within the
6-dB beamwidth. The measured gain of the element is 7.8 dBi.
The return loss is over 10 dB inside the frequency band of the
channel sounder (???? ? ?? MHz).
C. Array Realization
Theantennaelementsaremountedonasphericalsurfacecon-
sisting of two hollow aluminum hemispheres with outer diam-
eterof330mm (2.37?at2.154GHz). Theelementsare isolated
from the mount, and the distance from the center of the driven
patch (see Fig. 1) to the center of the sphere is 170 mm. The
elements point toward the normal of the sphere, and they are
oriented so that the polarization vectors corresponding to the
feeds (? and ?) are parallel to unit vectors ??and ??, respec-
tively. One of the elements is located in the zenith of thesphere,
pointing at the positive ?-axis direction. Fig. 3 presents thecon-
figuration of the spherical array.
To reduce cabling and losses, the 64-channel RF switching
unit is placed inside the sphere together with its control elec-
tronics. Only the RF signal cable, two coaxial control cables,
and the power supply wires are leading outside of the sphere,
which minimizes the disturbance to the near field of the array.
The diameterof thesphere(330mm) is determinedby thephys-
ical size of the switching unit, although a lower sidelobe level
(directional dynamic range) could be obtained by a smaller di-
ameter. The grating lobes (secondary maxima in radiation pat-
tern), however, are not a problem in the spherical array, as in
most array geometries when wide element spacing is used. As

Page 3

KALLIOLA et al.: REAL-TIME 3-D SPATIAL-TEMPORAL DUAL-POLARIZED MEASUREMENT441
Fig. 2.
attached to the spherical array. (a) ?-plane. (b) ?-plane.
Radiation pattern of antenna element measured with all elements
calculated in [11], the grating lobe level of a spherical array
is 27 dB below the main beam even with an element spacing
of 1.0 ?. The average distance between adjacent elements in
the built array is approximately 0.8 ?, and the directional ele-
ment pattern further suppresses the grating lobe. Moreover, the
mutual coupling between the elements is reduced with wider
spacing. Nomutual coupling compensation has been performed
for the array.
III. DIRECTIONAL MEASUREMENT WITH SPHERICAL ARRAY
The far field of an arbitrary group of ? dual-polarized an-
tenna elements at direction ????? can be expressed as
Fig. 3.Spherical array of 32 dual-polarized microstrip patch elements.
?????? ? ?
?
?
???
??????
????????????????
??????
? ????
???????
??? ?????? ?????? ??? ??
??????? ? ???????? ? ??? ???????? ? ??? ???? ? ??
(1)
where
?
??
?
array input signal;
complex weight of element ?;
carrier wavelength;
position vector of element ?;
elevation angle from positive ?-axis;
azimuth angle CCW from positive ?-axis (looking
from positive ?-axis direction);
unit vector pointing at direction ?????;
?-polarized component of the field pattern of ele-
ment ?;
?-polarized component of the field pattern of ele-
ment ?.
The element field pattern is defined by
??
?
?
??
??
??????
??
??????
??
??
?????? ? ???? ? ???? ? ???
?????? ? ???? ? ????? ???
(2)
where ??and ??are the elevation and azimuth pointing direc-
tionsof element?,respectively, and??and??arethemeasured
elementradiationpatternsforthetwopolarizations.Coefficients

Page 4

442IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 49, NO. 2, APRIL 2000
Fig. 4.
Vertical polarization. (b) Horizontal polarization.
Spatial response of the spherical array calculated with element phasing in the case of one vertically polarized signal source in direction (????, 0?). (a)
??and ??are applied to take into account the location-depen-
dent orientation of the element polarization vectors, and are de-
fined as
??? ? ? ?????????
? ? ? ??????????? ?????????
??? ? ? ?????????
? ? ? ??????????????
? ??????????? ??? ???? ? ???
(3)
where ? is the desired polarization vector, for which the field
is to be calculated. In dual-polarized spatial channel measure-
ments the two most practical values for vector ? are the HP
and VP components of the incident wave, namely, ??????? and
???????. In thiscase,theequationsfor??and??canbe written
as
?? ?
?
?
?
?
?
?
?
?
?
??? ??????? ? ?????????
? ?????????? ?????????
??? ??????? ??????????
? ???????????????
???????????????
?? ?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
??? ??????? ??????????
? ??????????????
??????????????
??? ??????? ? ?????????
? ???? ??????????????
???????????????????
????? ??????
(4)
Thecomplexweights??canbeusedtoformabeamtowarda
desired direction. The 3-D spatial response of the radio channel
can be measuredby forminga gridof beams covering the whole
?? solid angle. Two methods to determine the corresponding
weight vectors are presented in the following.
A. Element Phasing
The simplest method to create a beam toward a certain direc-
tion ??????? is to phase the element signals in that direction. In
this case the weights are written as
????????? ? ??????????????
?????????????
(5)
where ??is an amplitude tapering coefficient applied to sup-
press the array sidelobes.
If ??and ??are the signals measured from element ? at two
orthogonalpolarizations(?and?),thespatialresponsebecomes

Page 5

KALLIOLA et al.: REAL-TIME 3-D SPATIAL-TEMPORAL DUAL-POLARIZED MEASUREMENT 443
Fig. 5.
(a) Vertical polarization. (b) Horizontal polarization.
Spatial response of the spherical array calculated with element phasing in the case of two signal sources in directions (?????, 0?) and ????????????.
?????? ?
?
?
???
??????????? ?????????????????
?? ? ?????????
? ? ????????? ????
(6)
which still has to be normalized to equalize the response for all
directions. Let us assume a single plane wave with field vector
? arriving at the array from direction ???????. In this case sig-
nals ??and ??contain the direction-dependent phase informa-
tion corresponding to the location of the element, and they are
weighted by the element pattern at ???????
???
??
?
? ????
?????????????
???
??
???????? ? ?
???????? ? ?
?
?
(7)
Substituting this for (6) and using element phasing from (5),
gives
?????? ?
?
?
???
??????
????????????
??????????????????????????
???????? ? ? ? ????
???????? ? ???
(8)
It can be seen that the maximum of the response is produced
toward direction ?? ? ???? ? ???. The highest level is pro-
duced if the field vector ? is matched to the desired polariza-
tion vector. The remaining part of the response is not zeros, but
spurious responses whose level depends on the geometry of the
array, and the amplitude tapering term ??.
The tapering function that has been used for the spherical
array is the normalized power function written as
??????? ?
?? ? ??????????
??
??
(9)
where ? is the radius of the sphere. The weight of the ele-
ment pointing at thedesired direction is unity while theelement
pointing at the opposite direction is nulled.
B. Beam Synthesis
In addition to simple phasing, the weight vector corre-
sponding to a beam in a certain direction can be found through
numeric beam synthesis. In the directional channel measure-
ment approach this method can be applied to produce a weight
vector that suppresses the spurious responses in the directional