1164 IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 8, NO. 3, MARCH 2009
Error Performance of Transmit Beamforming with
Delayed and Limited Feedback
Yao Ma, Senior Member, IEEE, Dongbo Zhang, Alex Leith, and Zhengdao Wang, Senior Member, IEEE
Abstract—Feedback delay can severely affect the performance
of transmit beamforming (TB) and the analytical quantification
of the performance degradation has attracted much research
interest recently. In this letter, we study the effect of delayed
and limited-rate codebook index feedback on the error rate
performance of TB systems ov er Rayleigh fading channels.
We derive closed-form expressions for the moment generating
function (MGF) and the probability density function (PDF) of
the receiver output signal-to-noise ratio (SNR) including the
effects of outdated and finite-rate feedback and further provide
accurate analytical error rate expressions, which are verified
by simulation results. The coding gain gap between the full-
rate and limited-rate feedback and the coding gain advantage of
multiple transmit antennas to the single antenna are analyzed,
for limited and delayed feedback. These results ar e simple and
concise and provide new analytical insight into the achievable
diversity and combining gains and the loss caused by feedback
delay for different system parameters and modulation formats.
Index Terms—Delayed feedback, diversity and combining
gains, error rate, limited-rate feedback, transmit beamforming
(TB).
I. INTRODUCTION
T
RANSMIT beamforming ( TB) is an effective fading
countermeasure technique in multiple antenna systems
[1]–[10]. To achieve coherent beamforming, channel state
information at the transmitter (CSIT) is r equired, and to
provide such knowledge the CSI feedback from the receiver
to the transmitter is a popular technique.
1
The CSI feedback
rate in the reverse link is often limited due to the power and
bandwidth constraints, for example, in the uplink of cellular
systems.
Besides the feedb ack rate limit, the CSI feedback delay
also degrades the TB performance. Recently, the joint effects
of outdated and limited-rate feedback have been studied for
various systems, including the error rate [6], [11], [12] and the
ergodic capacity [13], [14] for the multiple-input single-output
(MISO) TB system, the error rate and outage probab ility
of multiple-input multiple-output (MIMO) subchannels based
Manuscript received April 25, 2008; revised September 19, 2008; accepted
November 21, 2008. The associate editor coordinating the review of this letter
and approving it for publication was C. Tellambura.
Y. Ma, A. Leith, and Z. Wang are with the Department of Electrical and
Computer Engineering (ECE), Iowa State University, Ames, IA, 50011 (e-
mail: {mayao, aleith, zhengdao}@iastate.edu).
D. Zhang was with Department of ECE, Iowa State University. He is now
with QualComm, San Diego, CA.
This paper was presented in part at the IEEE GlobeCom Conference,
Washington DC, No v. 2007.
Y. Ma’s work was partially supported by the Rockwell Collins Inc., Cedar
Rapids, IA. Z. Wang’s work was supported in part by NSF Grant No. 0431092.
Digital Object Identifier 10.1109/TWC.2008.080570
1
Since the codeword index feedback is equivalent to feedback of the index
of CSI region which maximizes the transmit SNR, we use CSI feedback and
codebook index feedback interchangeably in this letter.
on transmit precoding [15] and singular value decomposition
(SVD) [16], and the spectral efficiency of th e multi-user TB
systems [8], [9], [17], to name a few. Among these works, the
effect of channel estimation error (CEE) was further analyzed
in [12], [14], [16]. Despite these results (except the conference
version of this Letter [11]), the performance loss caused by
delay and limited feed back in term s of the diversity and coding
gains has not been analytical quantified. Yet such a result
will provide insight into the perf ormance loss as a function
of various system parameters.
In this letter, we study the effects of delayed and limited
feedback on the MISO TB system, and provide an analyt-
ical quantification on the achieved error rate performance,
and the diversity gain and the coding gain. By exploiting
an accurate method for modelling the feedback delay [1 0]–
[12], [14], [15], [17], we derive concise moment generating
function (MGF) and probability density function (PDF) results
of the TB receiver output signal-to-noise ratio (SNR) and
analytical error rate expressions. These results are simple and
concise and provide new analytical insight into the achievable
diversity and combining gains and the loss caused by feedback
delay for different system parameters and modulation formats.
The coding gain gaps between the full-rate and limited-
rate f eedback and between the multiple and single transmit
antennas are analyzed . Compared to the existing literature
on the TB feedback analysis [8], [9], [12], [14], [15], [17],
our results provide some new observations, including: (1) For
channels with bandlimited fading spectra the error rate may
not monotonically degrade as the delay increases; (2) In the
case of non-zero Doppler bandwidth, the feedback delay will
reduce the TB diversity gain to one for high SNRs, that is,
no diversity enhancement but only an array gain is attained;
and (3) The p erformance loss caused by feedback delay can
be much more detrimental than the limited feedback rate for
all SNRs.
Throughout this letter, we use superscript H to represent
Hermitian transpose; E[x] and [x] are the expected value
and real part of x, respectively, and ||x|| is the Frobenius
norm of vector x.Furthermore,I
M
is the M × M id entity
matrix while 0
M×N
is the M ×N all-zero matrix. CN(µ, Σ)
denotes the complex Gaussian distribution with mean µ and
covariance matrix Σ.
II. S
IGNAL MODEL
Consider a TB system with N
t
transmit antennas and one
receive antenna. The channel vector is given b y h(t)=
[h
1
(t),...,h
N
t
(t)]
T
, with E[|h
l
(t)|
2
]=σ
2
h
. The received
signal at time t can be written as
y(t)=w
H
(t)h(t)s(t)+η(t), (1)
1536-1276/09$25.00
c
2009 IEEE
Authorized licensed use limited to: SHENZHEN UNIVERSITY. Downloaded on April 18,2010 at 17:03:09 UTC from IEEE Xplore. Restrictions apply.