Progress In Electromagnetics Research, PIER 64, 257–278, 2006

EFFECT OF ANTENNA ARRAY GEOMETRY AND ULA

AZIMUTHAL ORIENTATION ON MIMO CHANNEL
PROPERTIES IN URBAN CITY STREET GRID

A. A. Abouda, H. M. El-Sallabi, and S. G. Häggman

Helsinki University of Technology
P.O. Box 3000, FIN-02015 HUT, Finland

Abstract—Propagation environment and antenna array configuration

have significant effect on spatial correlation properties of multiple-
input multiple-output (MIMO) wireless communications channels. In
this paper the effect of different antenna array geometries on MIMO
channel properties is investigated in urban city street grid propagation
environment. Four antenna array geometries with the same number
of antenna elements and fixed inter-element spacing are considered,
namely, uniform linear array (ULA), uniform circular array (UCA),
uniform rectangular array (URA) and uniform cubic array (UCuA).
The effect of ULA orientation in azimuthal plane on MIMO channel
ergodic capacity is also investigated. Varying orientation angle from 0
to π at the two communication ends is considered. The investigation is
carried out based on three dimensional (3D) spatial multi-ray realistic
propagation channel model covering different propagation types. It
is shown that the antenna array geometry have significant impact on
MIMO channel properties. Under different propagation scenarios the
ULA shows superiority to the other considered geometries in terms of
the ergodic channel capacity and number of spatial parallel channels.
However, this superiority depends largely on the array azimuthal
orientation.

1. INTRODUCTION

More services and higher data rates are the demands of future
wireless communication system end users. In order to accommodate
these expectations while maintaining robustness against wireless
impairments, contemporary technologies have to be developed.
Multiple-input multiple-output (MIMO) techniques stand as a strong
candidate to allow robustness against channel fading and interference
258 Abouda, El-Sallabi, and Häggman

as well as to enable high data rates [1, 2]. However, the performance of
future MIMO wireless communication systems strongly depends on the
propagation environment and the antenna array configuration [3][4].
Previous research results have focused largely on evaluating MIMO
system performance under the assumption of uniform linear array
(ULA) geometry at both ends with a specific array orientation. Despite
the implementation advantages of other array geometries, they have
not been extensively investigated. Recently, in [5] the impact of five
antenna array geometries on wireless MIMO system performance has
been studied using the clustered channel model [6] in indoor scenario.
It is shown in [5] that in low spatial correlation environment the ULA
geometry outperforms the other considered array geometries in terms
of channel capacity and bit error rate performance. In [7] a compact
MIMO antenna array was proposed by combining polarization diversity
and space diversity into one arrangement consisting of a cube. It
is shown that even for very small inter-element spacing considerable
capacity is obtained due to polarization diversity.
The assumption of a specific array orientation at both of the
transmitter and the receiver ends requires the arrays at the two
ends to be fixed to a specific direction. In reality, this is a valid
assumption for fixed wireless communications systems but in mobile
communications fixed array orientation at base station (BS) side is a
realistic assumption meanwhile it is not in the mobile side. The mobile
station (MS) is unlikely to be fixed to a specific direction. The effect
of azimuthal array orientation on MIMO system performance has been
studied in [8] and [9]. In [8] it is shown that the maximum channel
capacity is obtained when the ULA at the two communication ends are
’broadside’ orientated to each other. However, their investigation was
carried out based on a stochastic channel model [10]. In [9] based on
data measured in an office corridor it is shown that under line-of-sight
(LOS) conditions the channel capacity varies significantly depending
on the receiver array orientation. However, due to cost and complexity
of field measurements only a few array orientation angles at the receiver
side were considered.
The performance of MIMO wireless communication systems highly
depends on the propagation environment, antenna array geometry and
the antenna element properties. In previous work we have studied
the influence of environment physical parameters on the capacity of
outdoor MIMO channel [11]. Parameters such as street width, wall
relative permittivity and reflection order were considered. In this
paper we present detailed investigation on the effect of antenna array
geometry and ULA azimuthal orientation on outdoor MIMO channel
properties. Four antenna array geometries with eight antenna elements
Progress In Electromagnetics Research, PIER 64, 2006 259

and fixed inter-element spacing are considered, namely, uniform linear

array, uniform circular array (UCA), uniform rectangular array (URA)
and uniform cubic array (UCuA). The effect of antenna array geometry
on MIMO channel properties is studied by analyzing the eigenvalues
of the normalized channel correlation matrix. To study the effect of
ULA azimuthal orientation on MIMO channel properties, the obtained
ergodic channel capacity with different array orientation angles at
both ends is analyzed. The investigation is carried out in outdoor
microcellular environment using a three dimensional (3D) deterministic
propagation channel model based on electromagnetic theory. Results
for different propagation types are presented and analyzed.
The rest of this paper is organized as follows: The considered
antenna array geometries and the ULA azimuthal orientation are
presented in Section 2. The propagation channel model utilized in
this study is described in Section 3. The eigenvalues and the channel
capacity calculations are presented in Section 4. Numerical results and
intuitive discussions are given in Section 5. Our conclusions are drawn
in Section 6.

2. ARRAY GEOMETRIES AND ULA AZIMUTHAL

ORIENTATION

Fig. 1 depicts the considered antenna array geometries and shows

the reference azimuthal orientation angle of each geometry. All the
orientation angles are calculated relative to the x-axis (0◦ ) with anti-

X
X

X
UCA URA UCuA

X
ULA

Figure 1. Considered antenna array geometries and reference

azimuthal orientation angle φ.
260 Abouda, El-Sallabi, and Häggman

x1
C2 C3
BS y1 E y1
C D G H
y0 A
C1 C4
x1

Y
B

Figure 2. Urban street grid showing traveling routes under study.

clock wise rotation. Each geometry has eight antenna elements that
is distributed in an uniform shape. These geometries represent three
types of antenna arrays, one dimensional ULA, two dimensional UCA
and URA and three dimensional UCuA. In principle the antenna
elements could be of different field pattern, polarization and inter-
element spacing, however, in this study we consider omnidirectional
radiation patterns and vertical polarization with 0.5λ inter-element
spacing, where λ is the wavelength.
This study is conducted in one of the typical deployment scenarios
defined in standardization [12] and is shown in Fig. 2. Typical
traveling routes that represent different propagation scenarios in urban
microcellular environment are considered, namely, traveling route A-
B, C-D, E-F and G-H. The different antenna array geometries are
deployed at both ends and the eigenvalues of the normalized channel
correlation matrix are computed in each traveling route under fixed
array orientation. Later the effect of ULA orientation angle in
azimuthal plane on the ergodic channel capacity is studied on the
same traveling routes by computing the ergodic channel capacity under
different ULA azimuthal orientation angles at both ends. The effect
of the azimuthal array orientation on the performance of the other
array geometries is not considered in this work. However, it should be
noticed that the results obtained from the ULA azimuthal orientation
studies can be projected into the performance of the other geometries.
Progress In Electromagnetics Research, PIER 64, 2006 261

3. UTILIZED CHANNEL MODEL DESCRIPTION

Successful design and deployment of MIMO wireless communication

systems require detailed channel characterization. In order to carry
out this characterization, two approaches are widely common, field
measurements, e.g., [13], and model-based, e.g., [14]. The field
measurement is costly, time consuming, the results are site dependent
and it also requires skilled personnel. Due to the difficulties of field
measurement-based characterization, many researchers have turned to
model-based characterization approach. The advantage of model based
analysis is the flexibility of testing the influence of different parameters
that can not be controlled in field measurements in addition to the
possibility of interpreting the obtained results more accurately.
A three dimensional (3D) spatial variant multi-ray radio wave
propagation models for main street and perpendicular streets in an
urban street grid were developed in [15] and [16], respectively. The
propagation channel models are different from ray tracing models in
a sense that there is no searching for coupling paths between BS and
MS since all the ray characteristics, such as angle of arrival (AOA),
angle of departure (AOD) and path length, are given in closed form
mathematical expressions by using the set membership criteria. The
advantage of getting the ray characteristics in explicit mathematical
expressions is significant reduction in computation time which is the
main limitation of the available simulation tools. Performing this
study based on field measurement is costly and complex in terms of
design and implementation of different antenna array geometries and
orientation. Therefore, these propagation channel models are adopted
to carry out this study.
For MIMO channels, the complex channel gain between jth
transmit antenna element and ith receive antenna element is given
by [15, 16]:
RR RDR
gV,H (i, j) = gV,H (i, j) + gV,H (i, j) (1)
RR and g RDR are the total complex channel gain due to
where gV,H V,H
reflected-reflected (RR) rays group and reflected-diffracted-reflected
(RDR) rays group, respectively, with transmission in vertical
polarization (VP) and horizontal polarization (HP), respectively.
The complex channel gain of the RR rays and the RDR rays are
given by [16]:
λ  1
RR
gV,H (i, j) = fBS (ϑk , ϕk )(RkV,H (i, j))g
4π k≡(m,s,n,u,g)
r k (i, j)
−j2π
kn
(RH,V (i, j))n (RH,V
km
(i, j))m fM S (θk , φk )e( λ
rk (i,j))
(2)
262 Abouda, El-Sallabi, and Häggman

and
λ 
RDR
gV,H (i, j) = fBS (ϑk , ϕk )(RkV,H (i, j))g (RkH,kV
kn
(i, j))n
4π k≡(m,s,n,u,g,Ck )
−j2π
km (i, j))m D k
(RH,V ( (D1 (i,j)+D2 (i,j)))
H,V (i, j)fM S (θk , φk )e
λ

D1 (i, j)D2 (i, j)(D1 (i, j) + D2 (i, j))
(3)
where each ray k is represented by a set of five integers (m, s, n, u, g)
which characterize ray k, m and n are the wall reflection orders in
the main street and perpendicular streets, respectively, s = 1, 2 and
u = 1, 2 are for two sidewalls on the street for BS and MS, respectively,
g = 0, 1 is for the ground reflection, when m = 0, s = 0, u = 0 ray k is
characterized by g, if g = 0, ray k is the LOS path and if g = 1 ray k
is paired ground reflection path, rk is the path length, fBS (ϑ, ϕ) and
fM S (θ, φ) are the BS and MS antenna field patterns with polarization
information, respectively, (ϑk , ϕk ) and (θk , φk ) are the elevation and
azimuth AOD and AOA for the kth ray at BS and MS, respectively,
RkV,H , RH,V
km and Rkn are the well-known Fresnel reflection coefficients
H,V
for ground and wall reflections in main and perpendicular streets,
respectively, with transmission in VP and HP, respectively, DH,V k is
the diffraction coefficient at the vertical edge of the building corner
with transmission in VP and HP, respectively, D1,2 is the distance
from the BS and the MS to the diffraction point, respectively, C1,2,3,4
are four building corners.
For a MS located at (x,y) with antenna height hm , the path length
of the kth ray that couples the MS to a BS located at (0,0) with antenna
height hB is given by [16]:


 ((n + (−1)u Ξ(n))w2 + (−1)u (x − 2Ξ(n)x1 ))2

rk =  +((−1)s ((m + (−1)s Π(m))w1 − 2(−1)s Ξ(m)y1 ) − y)2 (4)
+(hB − (−1)g hm )2
where 
1, ∀ χ even
Ξ(χ) = (5)
0, ∀ χ odd
w1 and w2 are widths of the main and perpendicular streets, x1 and
y1 are the distances shown in Fig. 2. The azimuthal AOD and AOA
of the kth ray are given by [16]:

(−1)m sgn(y)((m−sgn(y)Ξ(m))w1+2sgn(y)Ξ(m)y1+|y|)
ϕk = arctan
(−1)u ((n+(−1)u Ξ(n))w2 +(−1)u (x − 2Ξ(n)x1 ))
(6)
Progress In Electromagnetics Research, PIER 64, 2006 263


(−1)s ((m + (−1)s Ξ(m))w1 − 2(−1)s Ξ(m)y1 ) − y
φk = arctan (7)
(−1)n+1 ((−1)u ((n + Ξ(n))w2 − 2Ξ(n)x1 ) + x)
where arctan(.) returns the angle in the corresponding quadrant. More
details can be found in [15] and [16].

4. EIGENVALUES AND CHANNEL CAPACITY

CALCULATIONS

The eigenvalue decomposition of the instantaneous channel correlation

matrix is a useful tool for MIMO system performance investigation.
The distribution of the eigenvalues of the channel correlation matrix
reveals valuable information about the MIMO channel properties and
consequently predicts the MIMO system performance. While the
number of the non zero eigenvalues reflects the spatial multiplexing
gain [17] that can be obtained over a given channel, the eigenvalues
represent the power gain [18] in each spatial channel. The eigenvalues
are obtained by applying the eigenvalue decomposition on the channel
correlation matrix as follows:

λi = EVD(HHH ), i = 1, 2, ..., R(HHH ) (8)

where H ∈ C Nr ,Nt is the normalized channel matrix with Nt transmit

antennas and Nr receive antennas, λi is the ith eigenvalue of the
channel correlation matrix, EVD(A) returns the eigenvalues of matrix
A and R(A) denotes the rank of matrix A.
The channel capacity under fixed average receive signal to noise
ratio (SNR) is usually used as a performance measure of MIMO
systems. It maps the eigenvalues of the channel correlation matrix to
a single number that can be used for easy benchmark. Under spatial
multiplexing scheme [19] and with equal power allocation strategy,
the channel capacity will be the sum of the capacities supported by
each spatial subchannel. Therefore, the total channel capacity at each
channel realization can be written as [18]:
R(HHH )
 
ρ
c= log 1 + λi b/s/Hz (9)
i=1
Nt

where ρ is the average receive SNR.

264 Abouda, El-Sallabi, and Häggman

5. RESULTS AND DISCUSSIONS

5.1. Simulation Environment and Parameters

The geometry of the urban microcellular environment under study is
depicted in Fig. 2 where the street grid pattern has 100 m × 50 m
blocks of building and 25 m street widths for the main street, where the
BS is located, and perpendicular streets. The building street surface
electrical parameters are relative permittivity r = 5 and conductivity
σ = 0.005 S/m. These are practical values for city street concrete
walls [20]. The environment multipath richness, from channel modeling
perspective, is a function of the maximum reflection orders in the main
and perpendicular streets. Since one reflection results in about 5 dB
loss, rays reflected more than 7 times are very weak [11]. Therefore, the
maximum reflection order is set to 7 in both main and perpendicular
streets. The BS is mounted below the rooftops with antenna height 13
m at distances of 75 m from corner C1 and 5 m from the wall, y1 = 5 m.
A MS with antenna height 1.8 m moves a distance of 100 m in each
traveling route with x1 = 13 m and y1 = 5 m.
In the following numerical results a carrier frequency of 2 GHz
is assumed. Throughout the traveling routes, channel realizations
are computed every 10 cm. The channel capacity calculations are
performed at 20 dB average receive SNR.

5.2. Traveling Routes Description

It is widely known that the performance of wireless MIMO
communication systems depends on the dominant propagation
mechanism [21] and the propagation environment geometry [22]. The
traveling routes under study represent different propagation types
and there are different dominant propagation mechanisms throughout
the routes. For instance, traveling route A-B is a non line of sight
(NLOS) scenario. The dominant propagation mechanism in this route
changes with the MS location. At the beginning of the route there
is a strong LOS component due to the presence of the direct path
and its ground reflected pair. The LOS component disappears when
the MS moves inside the perpendicular street. Then the environment
will be dominated by the reflection mechanism. As the MS goes
inside the perpendicular street the RR rays group will suffer from high
attenuation due to multiple reflections and the diffraction mechanism
will become more dominant. At the end of the route the RR rays
group will vanish and only RDR rays group will be available. Most
of the dominant RDR rays group will be due to the signal diffracted
from corner C3 , (see Fig. 2), which results in a scenario similar to
Progress In Electromagnetics Research, PIER 64, 2006 265

low k-facor situation though it is NLOS. The propagation scenario in

traveling route E-F is NLOS where both RR rays group and RDR rays
group are available but the RR rays group exist for short distances of
the route due to large separation distance between BS and MS. Most
of the dominant RDR rays group in this route will be due to the signal
diffracted from corner C4 in the second crossing perpendicular street.
On the other hand, traveling routes C-D and G-H are LOS scenarios.
Throughout the routes the direct ray and its ground reflected pair
and the two sidewalls reflected rays are the dominant propagation
components. In traveling route G-H the separation distance between
the BS and the MS is larger than that of route C-D. This results in a
LOS scenario with low angular spread (AS) compared to route C-D.
In this case the channel matrix will be dominated by the direct path
and its ground reflected pair.

5.3. Effect of Antenna Array Geometry on MIMO Channel

Eigenvalues
The different antenna array geometries are deployed at both ends and
the channel matrices are computed in each traveling route. Fixed array
orientation for the all considered geometries is assumed. Orientation
angle, φ = 0, at both ends for the UCA, URA and the UCuA is
assumed. For the ULA broadside array orientation is assumed for the
BS array, φBS =π/2. In the MS side two possible ULA orientations are
considered; 1)- Transversal to the main street for routes C-D and G-H,
φM S = π/2, 2)- Transversal to the perpendicular street for routes A-B
and E-F, φM S = 0.

5.3.1. Eigenvalues Analysis

Figs. 3 and 4 show the effect of different antenna array geometries
on the eigenvalues of MIMO channel at 10% outage probability in
LOS traveling routes, C-D and G-H, respectively. The power gain
distribution of single input single output (SISO) system arbitrarily
chosen from each of the obtained channel matrices is calculated and
denoted as γS . In addition, the power gain distribution in the case of
SISO system with Rayleigh distribution is also calculated and denoted
as γR . Both γS and γR serve as references in the following discussions.
It can be noticed that in both traveling routes γS > γR which is the case
of Ricean channel when there is a strong LOS component. In terms of
the number of parallel channels, using the ULA results in four parallel
channels with λi ≥ γS available 90% of the time in traveling route
C-D. These parallel channels are capable of carrying high data rates to
the receiver. In traveling route G-H the number of significant parallel
266 Abouda, El-Sallabi, and Häggman

20
S
Eigenvalue @ 10% outage probability [dB] 15 R
ULA
10 UCA
URA
UCuA
5

1 2 3 4 5 6 7 8
Eigenvalue index

Figure 3. Eigenvalues of channel correlation matrix read at 10%

outage probability in traveling C-D.

20
S
15
Eigenvalue @ 10% outage probablity [dB]

R
ULA
10 UCA
URA
UCuA
5

1 2 3 4 5 6 7 8
Eigenvalue index

Figure 4. Eigenvalues of channel correlation matrix read at 10%

outage probability in traveling G-H.
Progress In Electromagnetics Research, PIER 64, 2006 267

channels reduces to two due to the lower AS. The UCuA maintains only
one significant channel in both routes since only one eigenvalue is larger
than γS . Because the second eigenvalue in traveling route C-D, λ2 , is
higher than that in traveling route G-H, higher channel capacity can be
obtained in traveling route C-D with UCuA. The use of UCA and URA
geometries have similar impact on the MIMO channel eigenvalues in
both routes where the available significant channels in traveling routes
C-D and G-H are two and one, respectively.
The eigenvalues of the channel correlation matrix at 10% outage
probability in NLOS traveling routes, A-B and E-F, are shown in
Figs. 5 and 6, respectively. In these scenarios γS is close to γR in both
routes which means that the amplitude of channel matrix elements can
be closely modeled as Rayleigh distributed random variable. However,
there are some dominant reflected components that result in Ricean
channel with low k-factor. Using ULA, UCA or URA in traveling route
A-B results in four, three and three parallel channels with λi ≥ γS
available 90% of the time, respectively. When the UCuA is deployed,
only two significant parallel channels are available which reduces the
achievable data rate significantly. Similar observations are made in
traveling route E-F, where there are three parallel channels available
by using ULA, UCA and URA while using the UCuA results in two
parallel channels.

5.3.2. Insight View of Antenna Array Geometry Effect

It can be observed that the performance of the different antenna
array geometries depends on the number of array elements facing
the direction of wave propagation and the distance between these
elements. In the case of ULA geometry, the number of elements
facing the direction of wave propagation is eight with inter-element
spacing 0.5λ, while in the UCA and the URA geometries the number
of elements facing the direction of wave propagation are five, (half
of the circle), and three, (one side of the rectangle), respectively, with
real inter-element spacing <0.5λ and 0.5λ, respectively. For the UCuA
geometry, the inter-element spacing is 0.5λ with four elements facing
the direction of wave propagation, (one side of the cube). The UCuA
has two layers, upper layer and lower layer. The four elements in
the upper layer and the four elements in the lower layer will have
highly correlated signals because the radio wave propagation takes
place in horizontal plane in the considered propagation environment.
This will reduce the number of effective antenna elements facing the
wave propagation direction to about two. Therefore, distributing the
antenna elements in three dimensions does not seem attractive in low
elevation angle spread propagation environment. It should be noticed
268 Abouda, El-Sallabi, and Häggman

20
S
Eigenvalue @ 10% outage probability [dB] 15 R
ULA
10 UCA
URA
UCuA
5

1 2 3 4 5 6 7 8
Eigenvalue index

Figure 5. Eigenvalues of channel correlation matrix read at 10%

outage probability in traveling A-B.

20
S
15
Eigenvalue @ 10% outage probability [dB]

R
ULA
10 UCA
URA
UCuA
5

1 2 3 4 5 6 7 8
Eigenvalue index

Figure 6. Eigenvalues of channel correlation matrix read at 10%

outage probability in traveling E-F.
Progress In Electromagnetics Research, PIER 64, 2006 269

that in indoor propagation environment the three dimensional antenna

array geometry may benefit from the ceiling and back reflected signals.

5.4. Effect of ULA Azimuthal Orientation on MIMO

Channel Ergodic Capacity
In spite of the geometrical simplicity of the ULA, it has shown
superiority to other considered geometries in terms of the number of
parallel channels under different propagation scenarios. However, this
superiority is valid only under some specific array orientation angles.
In this section varying ULA orientation angle from 0 to π at the two
communications ends is considered with step sizes of π/4 and π/8 for
the BS array (φBS ), and the MS array (φM S ), respectively.

5.4.1. Ergodic Channel Capacity Analysis

The effect of ULA azimuthal orientation at the two communications
ends on the ergodic channel capacity in LOS traveling routes, C-
D and G-H, are shown in Figs. 7 and 8, respectively. The ergodic
channel capacity of SISO channel at the same average SNR, 20 dB,
is 6.52 b/s/Hz. The ergodic channel capacity of the SISO channel do
not depend on the two ends array orientation angles and it is used
as a reference in the following discussions. It is clear that the MIMO
channel capacity highly depends on the array orientation angles at the
two ends. For instance, in traveling route C-D where the AS is higher
than that in route G-H, when the ULA orientation angles φBS = π/2
and φM S = π/2, the MIMO ergodic channel capacity increase relative
to the SISO channel is about 22.79 b/s/Hz. Turning the MS orientation
angle to 0 or π reduces the relative MIMO ergodic channel capacity
gain to 8.51 b/s/Hz.
In low AS propagation scenario, like the case in traveling route
G-H, the impact of ULA azimuthal orientation on the MIMO channel
capacity performance is evident. With φBS = π/2 and φM S = π/2
the relative MIMO capacity gain is 14.48 b/s/Hz. When the ULA
orientation angles at both ends are set to 0 or π, the MIMO channel
capacity gain over the SISO channel is only 3 b/s/Hz.
Figs. 9 and 10 show the effect of ULA azimuthal orientation angles
at the two ends on MIMO channel ergodic capacity in NLOS traveling
routes A-B and E-F, respectively. Deploying the MIMO technique
results in relative high and low channel capacity gain over the SISO
channel at different ULA orientation angles. For instance, in traveling
route A-B the highest relative channel capacity gain, 29.62 b/s/Hz, is
obtained when φBS = π/4 and φM S = 3π/4. In traveling route E-F
the highest relative channel capacity gain, 26.52 b/s/Hz, is obtained
270 Abouda, El-Sallabi, and Häggman

40
Ergodic channel capacity [b/s/Hz]

35
30
25
20
15
10
5
1
3/4 1
2/4 6/8
4/8
1/4 2/8
/ 0 0
BS /
MS

Figure 7. Effect of azimuthal ULA orientation on the ergodic channel

capacity in traveling route C-D at 20 dB SNR.

40
Ergodic channel capacity [b/s/Hz]

35

30

25

20

15

10
1
3/4 1
2/4 6/8
4/8
1/4 2/8
/ 0 0
BS /
MS

Figure 8. Effect of azimuthal ULA orientation on the ergodic channel

capacity in traveling route G-H at 20 dB SNR.
Progress In Electromagnetics Research, PIER 64, 2006 271

40
Ergodic channel capacity [b/s/Hz]

35

30

25

20

15
1
3/4 1
2/4 6/8
4/8
1/4 2/8
/ 0 0
BS MS
/

Figure 9. Effect of azimuthal ULA orientation on the ergodic channel

capacity in traveling route A-B at 20 dB SNR.

40
Ergodic channel capacity [b/s/Hz]

35

30

25

20

15
1
3/4 1
2/4 6/8
4/8
1/4 2/8
/ 0 0
BS /
MS

Figure 10. Effect of azimuthal ULA orientation on the ergodic

channel capacity in traveling route E-F at 20 dB SNR.
272 Abouda, El-Sallabi, and Häggman

when φBS = π/2 and φM S = 7π/8 or φM S = π/8. However, higher

resolution of φBS may result in high relative channel capacity gain
at different φBS . Changing the orientation angles to φBS = 0 or π
and φM S = π/2 reduces the relative MIMO channel capacity gain
to 13.05 b/s/Hz in traveling route A-B and in traveling route E-F to
11.09 b/s/Hz.

5.4.2. Insight View of ULA Azimuthal Orientation Effect

It can be noticed that in the all considered traveling routes, in order to
obtain the highest ergodic channel capacity with ULA geometry, the
two communication ends should be turned to face the main direction
of the radio wave propagation. This due to the fact that the wave
propagation takes place in horizontal plane in the considered scenarios
and the resolution of the ULA array is high when it is transversal
to the wave propagation direction. Therefore, the ULA azimuthal
orientation can be thought of as a mechanical steering to direct the
array beam towards the wave propagation direction. In LOS traveling
routes the wave propagation direction is parallel to the main street
while in NLOS traveling routes the wave is coming through the street
crossing. It can be also noticed that when the ULA at BS side is
parallel to the wave propagation direction, the orientation angle at
the MS side has insignificant impact on the ergodic channel capacity
performance in both LOS and NLOS traveling routes. On the other
hand, the BS orientation angle has significant impact on the ergodic
channel capacity even when the MS is parallel to the wave propagation
direction. This is due to the fact that the ergodic channel capacity
performance depends on the spatial correlation properties at the two
communication ends. Parallel ULA to the wave propagation direction
at the BS side results in high spatial correlation and therefore, varying
the orientation angle of the MS array will not have noticeable impact
on the channel capacity.
In traveling route A-B the four building corners have substantial
impact on the channel capacity performance. For instance, in deep
shadow region the propagation scenario is dominated by the RDR
rays group. The direct diffracted component from corner C3 is the
dominant and other components are so weak with respect to it which
results in a situation with low k-factor. As we have seen, the highest
channel capacity is obtained when the two communications ends are
transversal to the wave propagation direction, in traveling route A-B,
the highest channel capacity is obtained when the array at the MS
side is transversal to corner C3 which can be considered as a secondary
source. In BS side three orientation angles result in high relative
channel capacity gain, φBS = 3π/4, φBS = π/2 and φBS = π/4.
Progress In Electromagnetics Research, PIER 64, 2006 273

C2 C3

BS array
C1 C4

MS array

Figure 11. Illustration of corner C3 role on the effect of ULA

azimuthal orientation in traveling route A-B.

Illustration of corner C3 role on the effect of array orientation in

traveling route A-B when φBS = 3π/4 and φM S = 3π/4 is shown
in Fig. 11 where the two communications ends face the diffraction
corner C3 . However, the positions of the two communications ends and
the separation distance between them have influence on the resolution
of the array orientation. In traveling route E-F, it can be noticed
that the array resolution to the orientation angles is less than that
in traveling route A-B due to separation distance. This piece of
information might be very valuable for fixed wireless communications
systems deployment. In order to achieve maximum channel capacity in
NLOS scenarios, our findings suggest taking into account the geometry
of the propagation environment when installing the antenna arrays at
both ends.

5.5. Comparison of channel capacity performance

Figs. 12, 13, 14 and 15 show the cumulative distribution function
(CDF) of the channel capacity obtained with different antenna
array geometries in traveling routes A-B, C-D, E-F and G-H,
respectively, at 20 dB average SNR. The channel capacity obtained
under the assumption of independent identical distributed (iid)
complex Gaussian elements with the same MIMO system size and at
the same average SNR is also shown in these figures. In practice, the
MS orientation angle is unlikely to be fixed to a specific direction but
the BS array most probably going to be fixed. Therefore, the channel
capacity performance of the ULA is shown for three cases; 1)- averaging
over all orientation angles at both ends, denoted as ULA, 2)- averaging
274 Abouda, El-Sallabi, and Häggman

1
iid
0.9 ULA
ULA, = /2
BS
0.8 ULA, = /2, =0
BS MS

0.7 UCA
P(capacity < abcissa)

URA
UCuA
0.6

0.5

0.4

0.3

0.2

0.1

0
0 10 20 30 40 50
Channel capacity [b/s/Hz]

Figure 12. CDFs of channel capacity obtained with different antenna

array geometries at 20 dB SNR in traveling route A-B.

0.9

0.8

0.7
P(capacity < abcissa
)

0.6

0.5

0.4

iid
0.3
ULA
ULA, BS= /2
0.2 ULA, BS= MS= /2
UCA
0.1 URA
UCuA
0
0 5 10 15 20 25 30 35 40 45 50
Channel capacity [b/s/Hz]

Figure 13. CDFs of channel capacity obtained with different antenna

array geometries at 20 dB SNR in traveling route C-D.
Progress In Electromagnetics Research, PIER 64, 2006 275

0.9

0.8

0.7
P(capacity < abcissa)

0.6

0.5

0.4 iid
ULA
0.3 ULA, = /2
BS

0.2 ULA, = /2, =0

BS MS
UCA
0.1 URA
UCuA
0
0 10 20 30 40 50
Channel capacity [b/s/Hz]

Figure 14. CDFs of channel capacity obtained with different antenna

array geometries at 20 dB SNR in traveling route E-F.

0.9

0.8

0.7
P(capacity < abcissa)

0.6

0.5

0.4 iid
ULA
0.3 ULA, = /2
BS

0.2 ULA, BS
= MS
= /2
UCA
0.1 URA
UCuA
0
0 10 20 30 40 50
Channel capacity [b/s/Hz]

Figure 15. CDFs of channel capacity obtained with different antenna

array geometries at 20 dB SNR in traveling route G-H.
276 Abouda, El-Sallabi, and Häggman

over all MS orientation angles with fixed BS array, φBS = π/2 and 3)-
fixed array orientation angle at both ends, φBS = π/2 and φM S = 0 in
NLOS traveling routes, A-B and E-F and φM S = π/2 for LOS traveling
routes C-D and G-H. It can be noticed that in these propagation
scenarios the maximum achievable channel capacity, regardless of the
antenna array geometry, is less than that of the iid channel. This is due
to the fact that the iid assumption requires rich scatter environment
which is not the case in the propagation environment under study.
It can be clearly seen that in general the ULA outperforms the other
geometries in terms of channel capacity performance even when the BS
array is not assumed fixed. However, fixing the BS array to φBS = π/2
improves the channel capacity performance significantly relative to the
case where both ends are not assumed fixed. In NLOS traveling routes
A-B and E-F the relative increase in the median channel capacity is
about 6 and 7 b/s/Hz, respectively, while in LOS traveling routes C-D
and G-H the relative increase in the median channel capacity is about
9 and 6 b/s/Hz, respectively. However, keeping the BS array fixed
to φBS = π/2 and turning the MS array to φM S = π/2 in traveling
routes C-D and G-H or to φM S = 0 in traveling routes A-B and E-
F also results in significant increase in channel capacity performance
compared to the case when only fixed BS array is assumed. This
relative increase ranges from 1.3 b/s/Hz in traveling route A-B to
4 b/s/Hz in traveling route C-D.

6. CONCLUSIONS

We have shown that antenna array geometry and ULA azimuthal

orientation have significant impact on MIMO channel properties. It
is concluded that the MIMO channel properties highly depends on
the number of antenna array elements facing the wave propagation
direction and the distance between these elements. In outdoor
microcellular environment under LOS and NLOS propagation
conditions the ULA outperforms the other array geometries in terms
of the number of significant parallel channels and the ergodic channel
capacity. Since specific array orientation at BS side is an engineering
choice, the obtained results suggest that using ULA to obtain the
maximum channel capacity, the BS array should be transversal to the
main direction of the LOS and NLOS propagation.
Progress In Electromagnetics Research, PIER 64, 2006 277

REFERENCES

1. Foschini, G. J. and M. J. Gans, “On limits of wireless

communications in a fading environment when using multiple
antennas,” Wireless Personal Commun., Vol. 6, No. 3, 311–335,
1998.
2. Telatar, I. E., “Capacity of multi-antenna Gaussian channels,”
European Trans. Tel., Vol. 10, No. 6, 585–595, 1999.
3. Shui, D. S., G. J. Foschini, M. J. Gans, and J. M. Kahn, “Fading
correlation and its effect on the capacity of multielement antenna
systems,” IEEE Trans. on Commun., Vol. 48, No. 3, 502–513,
2000.
4. McNamara, D. P., M. A. Beach, and P. N. Fletcher, “Spatial
correlation in indoor MIMO channels,” Proc. of IEEE Personal
Indoor and Mobile Radio Communications, Vol. 1, 290–294, 2002.
5. Forenza, A. and R. W. Heath Jr., “Impact of antenna geometry
on MIMO communication in indoor clustred channels,” Proc. of
IEEE Antennas and Propagation Symposium, Vol. 2, 1700–1703,
2004.
6. Saleh, A. A. M. and R. A. Valenzuela, “A statistical model for
indoor multipath propagation,” IEEE Journal on Selected Areas
in Commun., Vol. 5, No. 2, 128–137, 1987.
7. Andersen, J. B. and B. N. Getu, “The MIMO cube — a compact
MIMO antenna,” Proc. of IEEE Wireless Personal Multimedia
Commun., Vol. 1, 112–114, 2002.
8. Li, X. and Z. Nie, “Effect of array orientation on performance of
MIMO wirelss channels,” IEEE Antenna and Propagation Letters,
Vol. 3, 368–372, 2004.
9. Almers, P., F. Tufvesson, P. Karlsson, and A. F. Molisch, “The
effect of horizontal array orientation on MIMO channel capacity,”
Proc. of IEEE Vehicular Technology Conference VTC03 , 34–38,
2003.
10. Kermoal, J. P., L. Schumacher, K. I. Pedersen, P. E. Mogensen,
and F. Frederiksen, “A stochastic MIMO radio channel model
with experimental validation,” IEEE Journal on Selected Areas
in Commun., Vol. 20, No. 6, 1211–1226, 2002.
11. Abouda, A. A, N. G. Tarhuni, and H. M. El-Sallabi, “Model-
based investigation on MIMO channel capacity in main street
of urban microcells,” Proc. of IEEE Antennas and Propagation
Symposium, 313–316, 2005.
12. TR 101 112 V3.2.0 (1998-04) Technical Report: Universal Mobile
Telecommunications System (UMTS); Selection procedures for the
278 Abouda, El-Sallabi, and Häggman

choice of radio transmission technologies of the UMTS (UMTS

30.03 version 3.2.0)
13. Gans, M. J., N. Amitay, Y. S. Yeh, H. Xu, T. C. Damen,
R. A. Valenzuela, T. Sizer, R. Storz, D. Taylor, W. M. MacDonald,
C. Tran, and A. Adamiecki, “Outdoor BLAST measurement
system at 2.44 GHz: calibration and initial results,” IEEE Journal
on Selected Areas in Commun., Vol. 20, No. 3, 570–583, 2002.
14. Gesbert, D., H. Bolcskei, D. Gore, and A. Paulraj, “Outdoor
MIMO wireless channels: models and performance prediction,”
IEEE Trans. on Commun., Vol. 50, 1926–1934, 2002.
15. El-Sallabi, H. M. and P. Vainikainen, “Physical modeling of line-
of-sight wideband propagation in a city street for microcellular
communications,” Journal of Electromagnetic Waves and Appli-
cations, Vol. 14, 905–927, 2000.
16. El-Sallabi, H. M. and P. Vainikainen, “Radio wave propagation in
perpendicular streets of urban street grid for microcellular commu-
nications. Part I: Channel modeling” Progress In Electromagnetic
Research, PIER 40, 229–254, 2003.
17. Zheng, L. and D. N. Tse, “Diversity and multiplexing: A
fundamental tradeoff in multiple-antenna channels,” IEEE Trans.
on Information Theory, Vol. 49, No. 5, 1073–1096, 2003.
18. Andersen, J. B., “Array gain and capacity for known random
channels with multiple element arrays at both ends,” IEEE
Journal on Selected Areas in Commun., Vol. 18, 2172–2178, 2000.
19. Foschini, G. J., G. D. Golden, R. A. Valenzuela, and P. W. Wolni-
ansky, “Simplified processing for high spectral efficiency wireless
communication employing multi-element arrays,” IEEE Journal
on Selected Areas in Commun., Vol. 17, 1841–1852, 1999.
20. Carl Johnk, T. A., Engineering Electromagnetic Fields and
Waves, 637, John Wiley and Sons Inc., 1988.
21. Herdin, M., H. Özcelik, H. Hofstetter, and E. Bonek, “Linking re-
duction in measured MIMO capacity with dominant wave propa-
gation,” Proc. of International Conference on Telecommunications
ICT03 , 1526–1530, 2003.
22. Driessen, P. and G. J. Foschini, “On the capacity formula for
multiple-input multiple-ouput wireless channels: A geometric
interpretation,” IEEE Trans. on Commun., Vol. 47, No. 2, 173–
176, 1999.