Professional Documents
Culture Documents
AbstractTraditional two-dimensional (2D) multi-input multioutput (MIMO) beamforming technologies can only adjust the
beamformers in the horizontal dimension according to horizontal
channel information. However, due to the three-dimensional
(3D) character of the real channel, 2D MIMO beamforming
technologies can not achieve the optimal system throughput.
In this paper, a 3D MIMO beamforming scheme is proposed,
which takes into account vertical beamforming. In the proposed
scheme, we first design a codebook for Vertical dimension based
on 3D MIMO channel model; then a 3D MIMO beamforming
scheme is proposed combining the proposed vertical beamforming codebook and legacy horizontal dimension beamforming
codebook. Through simulation, we evaluate the proposed 3D
MIMO beamforming scheme and compare it with former 2D
beamforming technology. Owing to the additional spatial degrees
of freedom in vertical dimension, our 3D MIMO beamforming
scheme can effectively improve the overall system performance.
I. I NTRODUCTION
wireless communication systems are expected to
provide high data rates and a better quality of wireless
signals satisfying demanding multimedia services such as
video and teleconferencing. The so-called MIMO (multi-input
multi-output) systems, in which multiple antennas are used at
both transmitter and receiver, have been proposed to achieve
these rates due to an improvement in spectrum efficiency [1].
Traditional MIMO systems are based on 2D MIMO spatial
channel models in the sense that the double directivities of
each path/subpath are expressed only in the XY-plane.
Recently, 3D MIMO has attracted substantial research attention as an important technique to improve vertical coverage
and overall system capacity [2] [3], in which the propagating
waves are assumed to arrive not only from the azimuth plane
but also the elevation plane. The 3D MIMO channel model is
presented in [4] [5].
However, existing 2D MIMO beamforming technologies
can only be made according to horizontal dimension channel
information [6] [7] [8]. But in fact the real channel is 3D
characterized, so the 2D MIMO beamforming technologies
cannot achieve the optimal system throughput without considering vertical dimension channel information. In order to
evaluate the potential for 3D MIMO system, it is of practical
interest and importance to investigate 3D MIMO beamforming
schemes, which takes into account vertical dimension channel
information in 3D space. However, there is few papers discuss
about it until now.
UTURE
hu,s,n (t) =
M
Pn SF
M m=1
h3D
u,s,n (t)
Pn SF
M
(v)
BS (n,m,AoD )
(h)
BS (n,m,AoD )
T
(v,v)
exp(jn,m )
(h,v)
rn2 exp(jn,m )
(v,h)
rn1 exp(jn,m )
(h,h)
exp(jn,m )
(v)
M S (n,m,AoA )
(h)
M S (n,m,AoA )
T
(v,v)
(v,h)
1
(v)
exp
j
k
exp
j
n,m
n,m
n,m
F
(
,
)
BS,s n,m n,m
(h)
(h,v)
(h,h)
1
FBS,s (n,m , n,m )
.
(v)
FM S,u (n,m , n,m )
m=1
(h)
1
1
(1)
(2)
Fig. 2.
Fig. 1.
[9]
B. 3D SCM Model
In 3D channel modeling, the departure and arrival angles
have to be modeled using not only the azimuth angle in XYplane, but also the elevation angle with respect to the Z axis
[2], [10], [11]. A spherical coordinate system for 3D channel
model is shown in Fig.2. In this case, the 2D channel modeling
expressed in (1) can be straightforwardly extended to the 3D
case, as shown in (2).
(v)
(h)
Where FBS,s (n,m , n,m ) and FBS,s (n,m , n,m ) are complex field patterns of the sth transmit antenna at BS for V
polarization and H polarization,respectively; the symbols n,m
and n,m denote the azimuth angle of departure (AAoD) and
elevation angle of departure (EAoD) for the mth subpath
of the nth path, respectively; rs = [xs , ys , zs ] is the vector
denoting the position of the sth transmit antenna at BS in
(v)
[2]
(h)
Vertical Codebook
Codebook
DFT-based codebook
DFT-based codebook
Spread Angles
0 180
Codebook Size N
N = 16
N =8
Feedback overhead
4 bits
3 bits
360
(
)
n,m,AoD
forming weight-vector of antenna. When the channels are high
SF
BS
= PnM
(h)
spatially correlated, every column codebook can corresponds
BS (n,m,AoD )
m=1
to an beamformed angle. The larger the codebook, the more
(v,v)
(v,h)
and finer corresponding angles. However, we need to feedback
exp(jn,m )
rn1 exp(jn,m )
more bits with lager codebook which leads more system
(h,v)
(h,h)
rn2 exp(jn,m )
exp(jn,m )
overhead. Hence we should make the best selection of the
codebook size N for vertical dimension considering tradeoff
(v)
(
)
n,m,AoA
between beamforming accuracy and feedback overhead.
MS
(h)
In fact, different from horizontal dimension, angles spread in
M S (n,m,AoA )
vertical dimension is much smaller. More specifically, in UMi
Scenario, users spread over 0 360 in horizontal dimension
exp(jkds sin(n,m,AoD )) exp(jkdu sin(n,m,AoA ))
but only 0 180 in vertical dimension. With the change of
cell radius and antenna height, the range in vertical dimension
exp(j2vn,m t).
(4)
may even be less. For this reason, there is no need to feedback
the same bits for vertical dimension as horizontal dimension. n,m,AoD and n,m,AoA denote the elevation angle of deparSimulation results with different codebook size N will be ture (EAoD) and elevationh angle of arrival (EAoA).
given in Section IV, which decides the optimum codebook
The vertical 2D PMI can also be selected from the codebook
size N for vertical dimension should be equal to 8 in UMi which is proposed in this paper according to the 2D vertical
scenario.
channel coefficient matrix given above and feedback to BS.
H44
2D
(5)
42
W2D
x1
42
x2
WV42
= H44
3D WH
x3 + n
x4
x1
x3
44
42
42
= H3D WH
+ WH
+ n,
x2
x4
(6)
TABLE II
SIMULATION PARAMETERS
Parameters
Assumption
Network layout
Traffic Model
Full Buffer
ISD
500m
Load
10 UE per sector
Carrier frequency
2.0GHz
Bandwidth
10MHz
Channel model
UE speed
3 km/h
Antenna configuration
Scheduler
Proportional Fair
UE Receiver
MMSE
Rx: Nr = 4
Fig. 3.
Fig. 4.
where H44
3D is the 3D MIMO channel matrix which we de42
is the horizontal MIMO precoding
scribed in section II. WH
42
matrix and WV is the vertical MIMO precoding matrix.
IV. S IMULATION R ESULTS
This section presents simulation results with different vertical beamforming codebook size to decide optimum codebook
size of vertical beamforming codebook and evaluates the
performance of the proposed 3D MIMO beamforming scheme
presented in Section III.
The system performance is evaluated by means of systemlevel simulations using 3GPP LTE-Advanced evaluation
methodologies [15]. We mainly consider the typical deployment scenarios UMi. Table II gives the main system level
simulation parameter configuration.
Fig.3 make comparision of cell throughout with different
vertical beamformer codebook size. We can see that the
codebook size N increases from 4 to 8, the performance is
improved, however, when larger codebook size 16 is used, The
performance is similar to the performance with codebook size
V. C ONCLUSION
In this work, we have investigated the 3D MIMO beamforming problem. A DFT-based beamforming codebook is
designed for vertical dimension and vertical codebook size
is decided according to the elevation angle distribution in
the vertical direction and simulation results. In particular, a
3D MIMO beamforming scheme is proposed to improve the
overall system capacity. By considering the additional spatial
degrees of freedom in Vertical dimension, we can transmit
more data streams at BS. Numerical results showed that our
proposed 3D MIMO beamforming scheme can reasonably
improve the overall system capacity performance compared
with conventional 2D MIMO beamforming.
ACKNOWLEDGEMENT
Fig. 5.
Fig. 6.