Adaptive Opportunistic Transmission in
MU-MIMO Downlink with Reduced Feedback
Zhao Li, Jiawei Yang, Junliang Yao
State Key Laboratory of Integrated Service Networks, Xidian University, Xi’an 710071, China
Email: {zli, jwyang}@xidian.edu.cn, jlyao@pcn.xidian.edu.cn
Abstract— In this paper we proposed two threshold-aided
adaptive opportunistic transmission strategies in multiuser
multiple-input multiple-output (MU-MIMO) downlink with
limited feedback. Threshold scheme in company with trans-
mission mode adaptation are employed to handle feedback
cost and improve system performance. In one time slot, each
mobile station (MS) carries out channel estimation, selects
appropriate transmission mode and feeds back adaptively.
With feedback information the base station (BS) schedules
one user out of candidate MSs and transmits to it. One
of the proposed strategies employs single-level threshold.
However, proper design of the threshold is difficult. Since
a tight one would result in high outage probability whereas
a loose one could not reduce feedback load effectively.
Thus the other method employs double-level threshold. In
this scheme, the primary threshold is used for feedback
reduction, the secondary one is employed to guarantee the
outage performance. In evaluating the strategies, statistical
analysis and Monte-Carlo simulation are used. Results
show that with properly designed thresholds, the proposed
schemes can greatly reduce the feedback load, achieve high
throughput and good outage performance.
Index Terms— Multiuser, MIMO, Downlink, Adaptive, Op-
portunistic, Threshold
I. INTRODUCTION
MU-MIMO is a set of advanced MIMO technologies
that exploit the availability of multiple independent mo-
bile users in order to improve communication perfor-
mance. When multiple users are simultaneously supported
by BS two major issues are under investigation, mul-
tiuser interference (MUI) elimination (precoding method
design) [1-3] and multiuser scheduling [4-7]. These two
issues are always discussed jointly and need channel
state information (CSI) feedback. In existing MU-MIMO
works, MUI elimination always induces much feedback
load and computational complexity.
In a multiuser environment with multiple independent
wireless links, at any given point in time the probability
that all of those links have low quality is small. This ad-
vantage is called multiuser diversity (MUD). Accordingly,
some works focus on opportunistic beamforming (OB) [8-
13], i.e. in each time slot only one user among multiple
candidates is selected to exploit MUD. Moreover, when
OB is under discussion MUI is no longer considered.
Obviously, all of the above techniques require BS to
know the channel information of users, which is usually
Manuscript received January 20, 2010; revised April 6, 2010; ac-
cepted May 12, 2010.
estimated at the MS and fed to BS through a finite-rate
reverse link. Since feedback information is part of system
overhead, a lot of works focus on feedback load reduction.
Most of them employ threshold to eliminate MSs whose
channel quality is below the given threshold [6-9, 12-13],
i.e. users who are not eligible for transmission would keep
silent. However, these works are usually discussed in a
simplified system model, i.e. the receiver end is equipped
with single antenna [8-9, 12] or only part of multiple
antennas is utilized [6]. As a result the advantages of
MIMO are not fully exploited. Moreover, those works
always consider fixed transmission mode. If adaptive
transmission [14] is employed, not only the communica-
tion performance can be improved, the system feedback
cost could also dynamically vary along with channel
status. In [13] both threshold scheme and transmission
mode adaptation are adopted. However, threshold design
and mode selection criterion are not elaborated. Moreover,
better performance may be achieved by modifying the
number and redesigning the value of thresholds.
In this paper we proposed two adaptive opportunistic
transmission strategies in MU-MIMO downlink. Gen-
eral MIMO scenario is considered in which both BS
and MS are equipped with multiple antennas. Threshold
scheme in company with transmission mode adaptation
are employed to handle feedback cost and improve system
performance. In one time slot, each MS carries out
transmission mode selection and adaptively feeds CSI.
With feedback information, BS schedules one user out
of multiple candidates and implements transmission.
Among the proposed strategies, the first one employs
single-level threshold, and the criterion of transmission
mode adaptation is deduced from this threshold. However,
proper design of the threshold is difficult, since a tight one
would result in high outage probability whereas a loose
one could not reduce feedback load effectively. Thus a
double-level threshold scheme is introduced subsequently.
In this strategy two thresholds are employed. The primary
one is used for feedback reduction, the secondary one is
to guarantee low outage probability.
The rest of this paper is organized as follows. In
Section II we describe the system model. In Section III
the single-level threshold strategy is introduced. Section
IV presents the double-level threshold scheme. In the
above two sections we also give the design of threshold
values and analyzed the system performance theoretically.
Section V provides the simulation results. Finally, the
JOURNAL OF COMMUNICATIONS, VOL. 5, NO. 7, JULY 2010 511
© 2010 ACADEMY PUBLISHER
doi:10.4304/jcm.5.7.511-520