Different Antenna Configurations in massive

mimo on favourable propagation

Abstract Asymptotic orthogonality of channel vectors, also called favorable propagation (FP), plays a key role
in massive multiple-input multiple-output (MIMO) systems which allows linear processing to achieve
optimality and maximize the information rate. Much research on the FP condition emerges under different
assumptions, most of which has been proved to satisfy the FP condition for some massive MIMO
implementations. A generic channel model without mutual coupling is proposed to study more general
implementations. Two theorems establish i) conditions on the steering matrix alone that are sufcient to
guarantee the FP condition holds, and ii) that a simple condition on the steering matrix alone is sufcient to
guarantee FP for all practical massive MIMO implementations. Then the uniform linear array (ULA), uniform
planar array (UPA) and uniform circular array (UCA) antenna congurations are studied.

Index Terms-. Favourable propagation, massive MIMO ,steering matrix, uniform linear array(ULA),
uniformplanararray(UPA),uniformcirculararray(UCA).

I. INTRODUCTION II. PRELIMINARIES

M ASSIVE multiple-input multiple-output

A. System Model

(MIMO) systems have attracted much attention in A typical massive MIMO system with one BS
5G discussion because of their potential for equipped with M antennas that simultaneously
enormous enhancements in spectral efciency with serves K single-antenna users. In general, M is
linear processing .Benets can be achieved only much larger than K, and M is often increased to
under the condition that the channel vectors innite value in theoretical analyses. We assume a
associated with different users become orthogonal narrowband at fading channel model and obtain
as the number of antennas, M, increases without the M 1 received signal vector y at the BS
bound. This condition, called favourable
propagation (FP), is one of the key assumptions in y = puGx + n
massive MIMO. Whether the actual channel
B. Favourable Propagation
satises the FP condition or not will determine the
application of massive MIMO in 5G. Several Favourable propagation- in massive mimo, the
studies have addressed the validity of the FP channel says that the channel has FP when
condition under different situations. The early normalized channel vectors become asymptotically
research on the FP condition is almost all based on orthogonal to each other.it can be expressed as
the ULA conguration. For consideration of
channels with different antenna structures, some
research extending ULA has emerged .The uniform
planar array (UPA), and the uniform circular array Practical condition that satisfies for all three
(UCA) have rst investigated and explores the FP antenna configuration
condition from a structured channel where the
channel is assumed as a combination of virtual
paths for massive MIMO regime. Here analysis is
based on fixed channels. Although it can prove the
FP condition and show some useful properties
anyway, it has obviously defects that the method
cannot examine how close and how fast the
different channel vectors become orthogonal when
increasing the number of antennas in a realistic
C. CHANNEL MODEL
channel where only the angular distribution can be
obtained, which could be important for transceiver
design and practical deployment.
 The channel model consist of a steering part and
wireless channel part, which is modelled by an
appropriate matrix. In this steering part, the
elements are the response of BS antenna
corresponding to different AoA (Angle of Arrival).
In the wireless channel part, the elements are the
complex signal gains before the signals reach the
BS antennas, as a function of the AoA according to
the channel chartecterstics.
total number of the BS antennas is then M =
MxMy. Different from the ULA case, the BS
antenna array must be specied by two AoAs,
which means more complexity in the analysis. In
this illustration, we use and to denote the
elevation and azimuth AoA, respectively.
A typical deployment where the BS uses a UPA is
depicted in Figure shown below

In matrix form representation.

G=wv

D. ANALYSES OF DIFFERENT CHANNEL VECTOR

According to the channel model the channel

response gmk can be expressed as

wml is the m-th element of steering vector wl.

Then, according to the denition of favorable
propagation i and k represent two arbitrary users Figure shows the illustration of the UPA that
with 1 i < k K. equips the BS with Mx and My antennas in x-axis
and y-axis directions, respectively. The
Theoretical analyses indicate how close corresponding antenna spacings are dx and dy, and
and how fast the inner product of different steering , represent the elevation and azimuth angles of
vectors converges to zero for a nite number of arrival, respectively
base station (BS) antennas, and prove that the FP
condition is satised for the antenna conguration Assuming the AoA are uniformly distributed on
considered for an unlimited number of BS antennas [0,2] we have the following proposition
only if the antenna spacing is not proportional to
the reciprocal of the number of antennas. The Proposition1: Letting
analysis shows that the UCA has the shortest
distance from the FP condition among the three
antenna structures, while the UPA is especially
suitable for massive MIMO systems for 5G owing
to its good FP performance and compact physical The expectation and variance of the steering
size. It is proved in this paper with simulation vectors for a ULA can be expressed as,
results .

A.UPA

The most commonly used antenna conguration in Proposition 2: Letting

practical deployment for 5G due to the advantage
of compact size a typical deployment where the BS
uses a UPA is composed of Mx and My antennas in
the planar x and y axes, respectively, and the
corresponding antenna spacings are dx and dy. The
the expectation and variance of the steering vectors uniformly distributed on the circle, the
for a UPA can be expressed neighbouring antenna spacing is

and the angle from the rst antenna to the m-th

antenna equals

B.ULA

This ULA is a typical deployment where the BS

uses a ULA is depicted in Figure below. Note that
the BS knows only the azimuth angles of the
received signals

Figure shows, An illustration of the ULA that

equips the BS with M antennas, where the antenna
spacing is d, and represents the azimuth angle of
arrival

Assuming the AoA are uniformly distributed on
[0,2)
we have the following proposition
Proposition1: Letting
Figure shows the illustration of the UCA that
equips the BS with M antennas, which are
uniformly placed on the circular array of radius r.
The neighbouring antenna spacing is d, and ,
represent the elevation and azimuth angles of
arrival, respectively.

Assuming the elevation AoA are uniformly

distributed on (90,90) and the azimuth AoA are
uniformly distributed on[0 ,2).
The expectation and variance of the steering
Proposition 3: Letting
vectors for a ULA can be expressed as,

the expectation and variance of the steering vectors

for a UCA can be expressed as,

C.UCA

The BS has M antennas, and the radius of the

array is r. Assuming the antenna elements are
III. SIMULATION RESULTS

1. The decay of the expectation of inner products of

different steering vectors versus the number of BS
antennas for different antenna arrays can be
obtained by comparing different antenna
configurations
(a) The expectation of the product of the steering
vectors.

(a). The derivative of the expectation of the

product of the steering vectors

(b) The variance of the product of the steering

vectors.

As per the simulation results all arrays increase fast

(b) The derivative of the variance of the product of
the steering vectors.
From the two graphical analysis, UCA is far
superior to the decays for the ULA and UPA, which
are almost the same with the UPA being slightly
better.
2. Derivative of the expectation and the variance of
the product of the steering vectors with an
increasing number of BS antenna for the ULA,
UPA and UCA
to converge FP condition is achieving faster with
increasing number of BS antennas for all three
antenna structures.
From the two graphical analysation UCA will be
more faster than other two antenna structures, while
ULA has the worst among them, which is a bit
lower than the UPA.
3. The expectation and the variance of the product
of the steering vectors versus M for different values
of d for a UPA

(a) The expectation of the product of the steering
vectors.
All the antenna configurations,The inner product of
different steering vectors decrease quickly with an
increasing number of BS antennas.
The inner product of different steering vectors
increases as the distance between antennas
decreases- smaller antenna spacing leads larger
inter user interference.
From the two graphical analysation ,UPA is the
most sensitive to small antenna spacing compared
with the ULA and UCA.
Simulation well confirm that the FP condition is
achievable with an unlimited number of BS
antennas only if the antenna spacing is not
proportional to the reciprocal of the number of BS
antennas for all three antenna structures.
IV. CONCLUSION
This paper is focus on the FP condition without
mutual coupling that the channel only knows the
angular distribution. .Theoretical analysis shows
how close and how fast the inner products between
different steering vectors converge to zero for
different antenna congurations under a practical
angular distribution. The analysis also shows that
the inter-user interference vanishes with an
unlimited number of BS antennas only if antenna
spacing is not proportional to the reciprocal of the
number of antennas for all the antenna structures.
Furthermore, a brief discussion comparing different
antenna congurations is also considered with
respect to the distance from FP. Finally, some
simulation results are given to support the analysis.
The distance from FP of all antenna configurations
coverage rapidly to zero, which implies massive
MIMO system may not require large number of
antennas.
For a fixed number of BS antennas, the UCA has
shortest distance from FP compared to ULA and
UPA.
But in massive MIMO systems UPA is more
suitable, since it offers comparable performance to
ULA and UCA.
REFERENCES
[1] T. L. Marzetta, Noncooperative cellular
wireless with unlimited numbers of base station
antennas, IEEE Trans. Wireless Commun., vol. 9,
no. 11, pp. 35903600, Nov. 2010.
[2] F. Rusek, D. Persson, B. K. Lau, E. G. Larsson,
T. L. Marzetta, O. Edfors, and F. Tufvesson,
Scaling up mimo: Opportunities and challenges
with very large arrays, IEEE Signal Process. Mag.,
vol. 30, no. 1, pp. 4060, Jan. 2013.
[3] M. Matthaiou, G. K. Karagiannidis, E. G.
Larsson, T. L. Marzetta, and R. Schober, Guest
editorial: Large-scale multiple antenna wireless
systems, IEEE J. Sel. Areas Commun., vol. 31, no.
2, pp. 113116, Feb. 2013.
[4] J. G. Andrews, S. Buzzi, W. Choi, S. V. Hanly,
A. Lozano, A. C. K. Soong, and J. C. Zhang, What
will 5g be? IEEE J. Sel. Areas Commun., vol. 32,
no. 6, pp. 10651082, Jun. 2014.
[5] H. Q. Ngo, E. G. Larsson, and T. L. Marzetta,
Energy and spectral efciency of very large
multiuser MIMO systems, IEEE Trans. Commun.,
vol. 61, no. 4, pp. 14361449, Apr. 2013.
[6] Tan, Weiqiang, Peter J. Smith, Himal A.
Suraweera, Michail Matthaiou, and Shi Jin,
Spectral Efciency of Multi-User mmWave
Systems with Uniform Linear Arrays and MRT. In
Proc. VTC spring, pp. 15, Jul. 2016.
[7] H. Q. Ngo, E. G. Larsson, and T. L. Marzetta,
Aspects of favorable propagation in massive
mimo, in Proc. EUSIPCO, pp. 7680, Sep. 2014.
[8] Y. Sun, L. Tian, J. Zhang, L. Wu, and P. Zhang,
On asymptotic favorable propagation condition
for massive mimo with co-located user terminals,
in Proc. WPMC, pp. 706711, Sep. 2014.