International Journal of Engineering and Technical Research (IJETR)

ISSN: 2321-0869 (O) 2454-4698 (P), Volume-4, Issue-1, January 2016

User scheduling for beam forming MU MISO -

OFDM Systems
Najoua Achoura, Ridha Bouallegue

separates the multiuser data streams by beamforming, i.e.

AbstractThe use of multi-user multiple-input assigning a beamforming vector to each of the Nt active
multiple-output (MU-MIMO) transmission is an attractive users. The beamforming vectors { } are selected from
strategy to increase spectral efficiency in wireless systems such multiple sets of unitary orthogonal vectors following the
as 3GPP Long-Term Evolution (LTE) and LTE-Advanced. In
beam and user selection algorithm. Equal power allocation
this work, we investigate beamforming and scheduling
strategies in the downlink of limited feedback MU-MIMO over scheduled users is considered. The received signal of the
systems. Modeling transmission under orthogonal uth scheduled user is expressed as:
frequency-division multiplexing (OFDM) using the achievable

sum rate as performance measure, numerical results show that
PU2RC achieves higher throughput and is more robust against
CSI quantization errors than the popular alternative of FZ Where we use the following notation:
beamforming if the number of users is sufficiently large. :number of transmit antennas and also number of
scheduled users; (Nt 1 vector) downlink channel;
Index Terms MIMO-OFDM, beamforming, ZF,Greedy, transmitted symbol with E[|xu|2] = 1; received symbol;
PU2RC. (Nt 1 vector) beamforming vector, The index set of
scheduled users, P :transmission power;

I. INTRODUCTION
III. MULTI-USER MISO-OFDM DOWNLINK CHANNEL
The combination of beamforming multiple input
-single-output (MISO) and orthogonal frequency A. Zero-Forcing Beamforming
multiplexing (OFDM) provides a robust communications Zero-forcing (ZF) decomposes the channel into several
link in fading environments by maximizing received signal parallel scalar channels with only additive noise, and the
strength via beam-steering, and has been incorporated in interference is removed completely by transmit beamforming
standards such as WiMAX and 3GPP Long Term Evolution. techniques[10]. Suppose a base station with Nt transmit
A MISO-OFDM system is often characterized by a single antennas transmits information to K users, each user is
base station with multiple antennas transmitting equipped with one single receive antenna. The received
simultaneously to a number of individual users. More signal by the kth user is
practical downlink algorithms are based on transmit
beamforming, including zero forcing [1][4], a
signal-to-interference-plus-noise-ratio (SINR) constraint [5],
minimum mean squared error (MMSE) [6]. These boadcast
algorithms can be combined with multiuser scheduling to ZF method transmits the signals towards the intended user
further increase the throughput by exploiting multiuser with nulls steered in the direction of the other users, i.e.,
diversity, which refers to scheduling only a subset of users ( ) .The users will receive only the desired
with good channels for each transmission [7][8]. Both
signal without any interference because of the perfect nulling.
scheduling and beamforming in a broadcast system require
In this case, the received data at the kth user can be written
channel state information (CSI) at the base station.
Unfortunately, CSI feedback from each user potentially
incurs excessive overhead because of the multiplicity of The corresponding vector equation is
channel coefficients. Therefore, this paper focuses on
downlink transmission that supports efficient CSI feedback Therefore, if the normalized transmit beamforming vector
and uses CSI for joint beamforming and scheduling of the kth user is selected

II. SYSTEM MODEL

The downlink or broadcast system is described as follows.
The base station with Nt antennas transmits data where h(y)k is the kth column of the pseudo inverse of H,
simultaneously to Nt active users chosen from a total of U denoted as H(y). Then it is shown that the interference can be
users, each with one receive antenna. The base station canceled completely. In this case, the SNR of the kth user is
| |

Najoua Achoura, National Engineering School Of Tunis, SUPCOM,

B. Channel Vector Quantization
InnovCom Laboratory In [9], authors analyzed the channel capacity with perfect
Ridha Bouallegue, High School Of Communication Of Tunis, channel knowledge at the receiver, but with limited channel
SUPCOM, InnovCom Laboratory, Carthage University

19 www.erpublication.org
 User Scheduling for beamforming MU-MISO-OFDM Systems

knowledge at the transmitter. Specifically, the optimal where , is the t identity matrix, and
beamformer is quantized at the receiver, and the quantized denotes the matrix consisting of the channel vectors of the
version is relayed back to the transmitter. Given the users selected in the first b-1 steps as
C {w ,
1 2
w B} [ ] .
quantization codebook , which is also
known a priori at the transmitter, and the channel H, the From all the orthogonal users, we select the user who
receiver selects the quantized beamforming vector to maximizes the sum rate (according to the projection
maximize the instantaneos rate, [9] properties) as

w( H ) arg max log(1 Hw j )
j V
2
Find user ab such that
(10)
where 1 n is the background signal-to-noise ratio
2

(SNR). The (uncoded) index for the rate-maximizing Set { }.

beamforming vector is relayed to the transmitter via an 3. Beamforming:
error-free feedback link. The capacity depends on the Apply the ZF transmit beamforming as previously described
beamforming codebook V and B. With unlimited feedback by using the selected user set
(B ) the w(H) that maximizes the capacity is the B. PU2RC
eigenvector of H*H, which corresponds to the maximum
Having collected quantized CSI from all U users, the base
eigenvalue.
station schedules Nt users for transmission and computes
We will assume that the codebook vectors are independent
their beamforming vectors. To maximize the throughput, Nt
and isotropically distributed over the unit sphere. It is shown scheduled users must be selected through an exhaustive
in [12], that this RVQ scheme is optimal (i.e., maximizes the
search, which is infeasible for a large user pool [5].
achievable rate) in the large system limit in which
Therefore, we adopt a simpler joint scheduling and
(B,Nt,Nr) with fixed normalized feedback B = B/Nt and
beamforming algorithm. This algorithm schedules a subset of
Nr = Nr/Nt. (For the MISO channel Nr = 1). Furthermore, the
users with orthogonal quantized channel shapes, and
corresponding capacity grows as log(Nt), which is the same furthermore applies these channel shapes as the scheduled
order-growth as with perfect channel knowledge at the users beamforming vectors. It is elaborated as follows. First,
transmitter. Although strictly speaking, RVQ is suboptimal
each member of the codebook F, which is a potential
for a finite size system, numerical results indicate that the
beamforming vector, is assigned a user with the maximum
average performance is often indistinguishable from the
performance with optimized codebooks [12], [14]. SINR. Consider an arbitrary vector, for instance , which
th
is the th member of the m orthonormal subset of the
IV. SCHEDULING codebook F. This vector can be the quantized channel shapes
of multiple users, whose indices are grouped in a set defined
In the case of K > Nt a multiuser diversity can be achieved by
performing user scheduling before applying ZF-BF. as { }
Multiuser diversity is a form of selection diversity among where is the uth users quantized channel
users; when the number of users is large, the base station can shape. can be equivalently defined as:
schedule its transmission to those users with favorable
{ |
channel fading conditions to improve the system throughput.
When full CSI is available, a better choice of beamforming }
directions can be made and it provides fairly good
Among the users in is associated with the one
performance under the zero-forcing strategy. Namely, the
providing the maximum SINR, which is feasible since the
transmitter can choose a group of users with high channel
SNRs are known to the base station through feedback. The
magnitudes and for which their channel directions are
matched to zero forcing beam directions. index and SINR of this user associated with
(m) can be written as
A. Greedy Scheduling
1. Initialization phase:
Setb=1. In the event that the vector the vector is
Let . Find the best user a1, according to its
associated with no user and the maximum SINR is set as
channel gain a1 such that
zero.
Second, the orthonormal subset of the codebook that
maximizes throughput is chosen, whose index is
Then BS knows its first selected user, and the set of selected . Thereby, the users
users will be { } associated with this chosen subset, specified by the
2. Sum rate calculation phase:
Forb=2 to Nt repeat: indices | , are scheduled for simultaneous
The objective is to find users who are orthogonal to the transmission using beamforming vectors from the (m)th
selected ones. Therefore, we project each remaining channel orthonormal subset.
vector onto the orthogonal complement of the subspace
spanned by the channels of the selected users. V. SIMULATIONS RESULTS
The projector matrix is obtained by
[ ] In Figure 1, the sum rates of ZF and PU2RC (with randomly
generated codebooks) are compared for various values of

20 www.erpublication.org
 International Journal of Engineering and Technical Research (IJETR)
ISSN: 2321-0869 (O) 2454-4698 (P), Volume-4, Issue-1, January 2016
Tfb; for each strategy, B has been optimized separately. It is showed that PU2RC can achieve significant gains in
seen that ZF maintains a significant advantage over PU2RC throughput with respect to Zero Forcing method for the same
for Nt = 4. At small value of Tfb, both schemes perform amount of CSI feedback. This paper focuses on the
similarly, but ZF maintains an advantage compared to scheduling criterion of maximizing throughput. The design
PU2RC. Intuitively, difference with perfect CSIT case in and performance analysis of PU2RC based on the criterion of
term of sum rates is understandable if we consider the impact proportional fairness is a topic under investigation. As
of the multiuser interference penalty term and the perspective, we can consider other scheduling algorithms and
non-optimality of these algorithms in the other hand. compare their performance with PU2RC and ZF algorithms.
10
REFERENCES
ZF-greedy [1] Hojae Lee, Sanghoon Lee Joint precoding selection diversity with
9.5 perfect CSIT limited feedback Digital Signal Processing, Volume 30, July
PU2RC 2014, Pages 141-153
[2] Wang Li, Yang Ke, Hu Han-ying and Cui Wei-jia Robust joint linear
Sum Rate (bps/hz)

9 transceiver design for MU-MIMO with imperfect CSI, Proc. GMC,

pp. 1-6, Oct. 2010.
[3] C. Siriteanu, S. D. Blostein, A. Takemura, H. Shin, S. Yousefi, and S.
8.5
Kuriki, Exact MIMO zero-forcing detection analysis for
transmitcorrelated Rician fading, IEEE Transactions on Wireless
Communications, vol. 13, no. 3, pp. 15141527, March 2014
8
[4] G. Dimic and N. D. Sidiropoulos, On downlink beamforming with
greedy user selection: performance analysis and a simple new
algorithm, IEEE Trans. on Signal Processing, vol. 53, pp. 385768,
7.5
100 150 200 250 300 350 400 450 500 Oct. 2005.
Feedback Total T (bits)
[5] P.drayi K. NagarajuModelling, Simulation of Multi-User Diversity
Fig1. Comparison between Greedy end PU2RC with perfect case with Channel State Information in MIMO Systems IRACST
International Journal of Computer Networks and Wireless
Communications (IJCNWC), ISSN: 2250-3501Vol.4, No5, October
2014
Figure 2 compares the throughput of PU2RC and ZF for an [6] Yan Yang, Hui Yue Efficient beamforming method for downlink
increasing SNR. The number of transmit antennas is Nt = 4 MU-MIMO broadcast channels International Journal of Electronics
and the codebook size is N = 64. As observed from this figure and Communications, Volume 69, Issue 3, March 2015, Pages
for the number of users U = 20, PU2RC achieves lower 636-643
throughput than ZF over the range of SNR under [7] M. Jung, Y. Kim, J. Lee, and S. Choi, Optimal number of users in
zeroforcing based multiuser MIMO systems with large number of
consideration (0 SNR 20 dB). Nevertheless, for larger antennas, Journal of Communications and Networks, vol. 15, no. 4,
numbers of users (U = 40 or 80), PU2RC outperforms ZF for pp. 362369, 2013
a subset of the SNRs. Specifically, the throughput versus [8] Z. Shen, R. Chen, J. G. Andrews, R. W. Heath Jr., and B. L. Evans,
SNR curves for PU2RC and ZF crosses at SNR=7 dB for U = Low complexity user selection algorithms for multiuserMIMO
systems with block diagonalization, IEEE Trans. on Signal
40 and at SNR=18 dB for U = 80. The above results suggest Processing, vol. 54, pp. 365863, Sept. 2006.
that in the practical range of SNR, PU2RC is preferred to ZF [9] W. Santipach, M.L. Honig Optimization of Training and Feedback
only if the user pool is sufficiently large. Overhead for Beamforming over Block Fading Channels IEEE Trans.
Nt = 4, N = 64 Info. Theory, August 2009
10
[10] G. Caire, N. Jindal, M. Kobayashi, and N. Ravindran, Multiuser
9
MIMO Achievable Rates with Downlink Training and Channel State
Feedback, IEEE Trans. Information Theory, Vol. 56, No. 6, pp.
8 2845-2866, June 2010

7
Sum Rate (bps/hz)

5
ZF,K=10
4 ZF,K=150
PU2RC,K=10
PU2RC,K=150
3

1
0 2 4 6 8 10 12 14 16 18 20
SNR

Fig. 2. PU2RC and ZF

VI. CONCLUSION
This paper presents asymptotic throughput scaling laws for
downlink transmission with orthogonal beamforming known
as PU2RC. In the normal SNR or noise-limited regimes, the
throughput of PU2RC is found to scale double
logarithmically with the number of users and linearly with the
number of antennas at the base station. Numerical results

21 www.erpublication.org