1406 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO.

3, MARCH 2013
A Weighted OFDM Signal Scheme for Peak-to-Average
Power Ratio Reduction of OFDM Signals
Chang Eon Shin, Kyung Soo Rim, and Youngok Kim
AbstractIn this paper, a peak-to-average-ratio (PAPR) reduction
scheme based on a weighted orthogonal frequency-division multiplexing
(OFDM) signal is proposed to reduce the PAPR without distortion in
removing the weight at the receiver side. In the proposed scheme, a
weight is imposed on each discrete OFDM signal via a certain kind of a
bandlimited signal, and an OFDMsignal formed with the weighted discrete
data is then considered before a high power amplier (HPA), whereas the
original signal can be recovered completely at the receiver side. Meanwhile,
the time duration needed to transmit the weighted OFDM signal is the
same as the time duration for the original OFDM signal. The effectiveness
of the proposed scheme is evaluated with computer simulations. According
to numerical results, the PAPR of the weighted OFDM signal is smaller
than that of the clipping and ltering (C&F) method, and the bit-error-rate
(BER) performance of the weighted OFDM system is improved compared
with the C&F method. Here, the proposed method is simpler than the C&F
method.
Index TermsConvolution, orthogonal frequency-division multiplexing
(OFDM), peak-to-average-power ratio (PAPR), weighted data.
I. INTRODUCTION
Orthogonal frequency-division multiplexing (OFDM) is a trans-
mission technique that modulates multiple carriers simultaneously.
Although their spectra overlap, the transmitted multiple carriers can
be demodulated orthogonally, provided that correct time windowing
is used at the receiver. Since the OFDM-based system has high
spectral efciency and is robust against intersymbol interference
and frequency-selective fading channels, it has been widely chosen
for European digital audio/video broadcasting and wireless local/
metropolitan area network standards, and now, it is used in most
broadband wireless communication systems.
However, one of the major problems of OFDM-based systems is
the high peak-to-average-power ratio (PAPR) of a transmitted signal,
which causes a distortion of a signal at the nonlinear high-power
amplier (HPA) of a transmitter. Thus, the power efciency of the HPA
is seriously limited to avoid nonlinear distortion; otherwise, the high
PAPR results in signicant performance degradation. Because of the
practical importance of this problem, a number of algorithms for re-
ducing the high PAPR have been developed, such as clipping and lter-
ing (C&F) [1][3]; coding [4][7]; adaptive symbol selection, such as
selected mapping; partial transmit sequence and interleaving [8][10];
tone reservation/injection [11], [12]; active signal constellation exten-
sion [13], companding [14][16]; and others.
In this paper, a PAPR reduction scheme based on a weighted OFDM
signal is proposed to reduce the PAPR without distortion in removing
Manuscript received April 6, 2012; revised August 2, 2012 and October 17,
2012; accepted November 29, 2012. Date of publication December 11, 2012;
date of current version March 13, 2013. This work was supported by the
National Research Foundation of Korea of the Ministry of Education, Science
and Technology through the Basic Science Research Program under Grant
2010-0028130. The review of this paper was coordinated by Dr. J.-C. Chen.
C. E. Shin and K. S. Rim are with the Department of Mathematics,
Sogang University, Seoul 121-742, Korea (e-mail: shinc@sogang.ac.kr; ksrim@
sogang.ac.kr).
Y. Kim is with the Department of Electronic Engineering, Kwangwoon
University, Seoul 139-701, Korea (e-mail: kimyoungok@kw.ac.kr).
Color versions of one or more of the gures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identier 10.1109/TVT.2012.2232949
the weight at the receiver side. This method is motivated by a circular
convolution process, i.e., the modulated OFDM signal is convoluted
with a certain kind of signal for smoothing the peak of the OFDM
signal before the HPA. Here, we choose the signal to satisfy that
the Fourier transform of has no zero on the real line. The
convoluted signal can be written as a simple weighted OFDM signal.
When the discrete data {a
k
}
N1
k=0
is given, we consider weighted data
{a
k
(k)}
N1
k=0
and form an OFDM signal with this weighted discrete
data. Then, this weighted OFDM signal is the same as the given
convoluted signal.
Since weight is nonuniform, the bit-error-rate (BER) performance
could be degraded. In practice, to improve the BER performance,
we modify the weight by adding a suitable positive constant to the
original weight. The PAPR of the weighted OFDM signal with the
modied weight is smaller than that of the C&F method, and
the BER performance is improved compared with the C&F method.
The effectiveness of the proposed scheme is evaluated with computer
simulations.
In this weighted OFDM method with modied weight, the time
duration needed to transmit the weighted OFDM signal is the same as
the time duration for the original OFDM signal. Moreover, the original
discrete data can be recovered completely at the receiver side with
additional 2N complex multiplications of computational complexity
without extra cost in transmission.
The weighted OFDM scheme was introduced in [17], where the
Gaussian function, sine function, and some other functions were used
as weighted functions. In [17], when the noise is not present, the
PAPR of the weighted OFDM system with Gaussian weight is reduced
remarkably. As mentioned in the conclusion, however, the noise was
not also considered for BER performance. If the additive Gaussian
noise is considered, the BER performance of the weighted OFDM
system with Gaussian weight will be even degraded. In this paper, we
suggest the weighted OFDM system with modied weight to improve
the BER performance, and we also provide the condition for a function
to be a weight function and a mathematical reason for the merit of the
weighted OFDM system derived from a circular convolution system.
The rest of this paper is organized as follows. In Section II, the
considered system is briey described. In Section III, we provide
the weighted OFDM signal motivated by the convolution method. In
Section IV, the effectiveness of the proposed scheme is evaluated with
simulation results. Finally, the conclusions are given in Section V.
II. SYSTEM DESCRIPTION
Orthogonal multicarrier modulation is an efcient method of data
transmission over channels with frequency-selective fading. This
method has a relatively simple implementation based on the inverse
fast Fourier transform (IFFT).
The simplied block diagrams for an OFDM system with the
convolution scheme and the proposed weighted scheme are shown in
Fig. 1. As described in Fig. 1(a), the modulated data stream is carried
on the multicarriers by the IFFT, and the convolution block reduces
the PAPR of signal, which is corresponding to the weight block of the
proposed scheme, as shown in Fig. 1(b). In the following block, the
cyclic prex is added before the HPA.
For a discrete data {a
k
}
N1
k=0
, multicarrier-modulated signal x
N
(t)
on [0, NT] is represented by
x
N
(t) =
1

N
N1

k=0
a
k
e
j2f
k
t
(1)
0018-9545/$31.00 2012 IEEE
w
w
w
.
t
a
k
e
o
f
f
p
r
o
j
e
c
t
s
.
c
o
m
For Further Details-A Vinay 9030333433,0877-2261612 0
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 3, MARCH 2013 1407
Fig. 1. Simplied block diagrams for an OFDM system with (a) convolution scheme and (b) proposed weighting scheme.
where N is the number of subcarriers, T is the original symbol period,
f = 1/NT, and f
k
= kf, k = 0, . . . , N 1. The PAPR of x
N
over the time interval [0, NT] is dened by
PAPR(x
N
) =
max
0tNT
|x
N
(t)|
2
E

|x
N
(t)|
2
(2)
where E() denotes the expectation operator.
III. WEIGHTED ORTHOGONAL FREQUENCY-DIVISION
MULTIPLEXING SYSTEM
Here, we provide the weighted OFDM signal, where the weight is
derived from a suitable bandlimited signal having no zero on the real
line. This method is motivated by a convolution method.
Taking the circular convolution between the multicarrier-modulated
signal x
N
and a suitable signal having compact support, the
PAPR of the convoluted signal can be reduced. In fact, for p R
with 1 p , from Youngs inequality (see [18]), x
N

p

x
N

p
, and x
N
belongs to L
p
although x
N
L
1
, where
f
p
= (

R
|f(x)|
p
dx)
1/p
and the space L
p
= {f : f
p
< }. For
p > 1, since L
p
is more tempered than L
1
, essentially, the PAPR of the
convoluted signal can be reduced. Simultaneously, we should consider
carefully to sustain the size of the expectation of x
N
.
First, we consider the convolution method and then derive the
corresponding weighted OFDM signal.
A. Convolution Method
The Fourier transform F[f] of f is dened by
F[f]() :=

R
f(x)e
jx
dx (3)
if the integral exists. The inverse Fourier transform F
1
[F] of F is
dened by
F
1
[F](x) :=
1
2

R
F()e
jx
d (4)
provided that the integral exists. Then F
1
[F[f]] = f, when f and
F[f] are integrable, and
F [F[f]] = 2

f (5)
where

f(x) = f(x).
We consider signal as
(x) =
1 sinc(x)

2
x
2
(6)
where
sinc x =

sin x
x
, x = 0
1, x = 0
.
By direct computation, the Fourier transform:=F[] of is given by
() =

1
2

1
||

2
, ||
0, otherwise.
The signal is a bandlimited signal with bandwidth , has no zero on
the real line, and
= . (7)
For more information about , see the Appendix at the end of this
paper.
Consider the circular convoluted signal as follows:
y
N
(t) :=
1
2
x
N
(t) =
1
2

x
N
(t )()d. (8)
Taking the Fourier transform in (8), we have by (5) and (7) that
F[y
N
] =
1
2
F[x
N
]F[] = F[x
N
] (9)
where F[x
N
] and F[y
N
] are the Fourier transforms in the sense of
distribution. Since has no zero on the real line and F[x
N
]() =
(2/

N)

N1
k=0
a
k
( 2f
k
), we can recover the discrete data so
that. for k = 0, . . . , N 1, we have
a
k
=

NF[x
N
](2f
k
)
2
=

NF[y
N
](2f
k
)
2(2f
k
)
. (10)
B. Weighted OFDM System
We show that the convoluted signal in (8) can be written as a simple
weighted OFDM signal y
N
.
Observing by (5), (7) and (8) that

e
j2f
k
(t)
()d = 2(2f
k
)e
j2f
k
t
the convoluted signal in (8) can be expressed as the following weighted
OFDM signal:
y
N
(t) =
1

N
N1

k=0
a
k
(2f
k
)e
j2f
k
t
, 0 t NT. (11)
C. Weighted OFDM System With Modied Weight
The demerit of the weighted OFDM signal in (11) is the degradation
of BER performance since the weight is nonuniform. To overcome
w
w
w
.
t
a
k
e
o
f
f
p
r
o
j
e
c
t
s
.
c
o
m
For Further Details-A Vinay 9030333433,0877-2261612 1
1408 IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 3, MARCH 2013
Fig. 2. CCDF of the C&F and proposed methods for N = 128, 256, 1024.
this obstacle, we consider the modied weight with a positive constant
as follows:

(x) = (x) + / log N (12)

where is a shift parameter, and log N is obtained by experiment.
Then, =
0
. In the weighted OFDM signal in (11), we replace
weight with

for a suitable positive constant to get the weighted

OFDM signal, i.e.,
z
N
(t) =
1

N
N1

k=0
a
k

(2f
k
)e
j2f
k
t
, 0 t NT (13)
as a transmitted signal instead of x
N
in (1).
In system(13), weight

(2f
k
) is imposed on the discrete data a
k
,
k = 0, . . . , N 1, and we form an OFDM signal with the weighted
discrete data {a
k

(2f
k
)}
N1
k=0
to get weighted OFDM signal z
N
.
We transmit weighted OFDM signal z
N
for the same time duration
[0, NT] as the original OFDM signal.
We note that weight is positive on the real line; therefore, the mod-
ied weight

is positive on the real line. Since

(2f
k
) = 0 for
any k = 0, . . . , N 1, the discrete data {a
k
}
N1
k=0
can be completely
recovered.
The PAPR of the weighted OFDM signal z
N
is given by
PAPR(z
N
) =
max
0tNT
|z
N
(t)|
2
E

|z
N
(t)|
2
. (14)
In the following, we provide the simulation results showing that
the PAPR of the weighted OFDM signal with modied weight

is smaller than that of the C&F method (see Fig. 2), and the BER
performance of the weighted OFDM system with modied weight

is improved compared with the C&F method (Fig. 3). We note that as
increases, due to the modication of weight, the BER performance
is improved, whereas the complementary cumulative distribution func-
tion (CCDF) grows slightly.
In (13), we can recover discrete data {a
k

(2f
k
)}
N1
k=0
by the
conventional method of the OFDMsystem. Since by dividing the given
discrete data by

(2f
k
) we can obtain the original discrete data,
the weighted OFDM system is not expected to cause any computa-
tional complexity in recovering the original discrete data. In fact, 2N
complex multiplications are additionally needed compared with the
original OFDM method.
We note that a sufcient condition for a signal to be a proper
weight is that (2f
k
) = 0 for any k = 0, . . . , N 1. We expect that
the performance of the weighted OFDM system corresponding to
depends on the smoothness of the Fourier transform of .
Fig. 3. CCDF and BER of the C&F and proposed methods for N = 128.
IV. SIMULATION RESULTS
The performance of this proposed scheme is analyzed through
the simulations. In the simulations, 10
3
quadratic-phase-shift-keying
(QPSK)-modulated OFDM symbols were randomly generated.
Fig. 2 shows the CCDFs of the C&F method and the proposed
method for N = 128, 256, 1024. The proposed method is simulated
with a xed shift parameter = 0.03, and several C&Fs are simulated
with various clipping ratios CR = 0.8, 1.2, 1.6, respectively. As shown
in the gure, the proposed scheme can reduce the PAPR around 3 dB
for N = 128 and 2 dB for N = 1024, respectively, at the 1% of the
CCDF, compared with the C&F scheme. Note that the PAPR of the
original OFDM signal exceeds 14.8 dB for N = 128 and 16 dB for
N = 1024, respectively. In Figs. 2 and 3, since the results induced by
quadratic-amplitude modulation mapping are almost the same as those
induced by QPSK mapping, here, we provide only the results induced
by QPSK mapping.
Fig. 3 compares the C&F method with the proposed method for
CCDFs and BER performance over the additive White Gaussian noise
channel together. As shown in the gure, the BER performance and
the CCDF of the proposed method with = 0.15 are superior to those
of the C&F method for CR = 0.8, 1.2, 1.6 when N = 128 is xed.
V. CONCLUSION
A PAPR reduction scheme based on a weighted OFDM signal
has been proposed to reduce the PAPR without data distortion in
removing the weight at the receiver side in the mathematical view. To
reduce the peak of the OFDM signal, a bandlimited signal , which
is not zero on the set {2f
k
}
N1
k=0
, is introduced, and we form weight
w
w
w
.
t
a
k
e
o
f
f
p
r
o
j
e
c
t
s
.
c
o
m
For Further Details-A Vinay 9030333433,0877-2261612 2
IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, VOL. 62, NO. 3, MARCH 2013 1409

=+/ log N for a suitable positive constant . We consider

a weighted discrete data to form a weighted OFDM signal, which
is dened on the same time interval as the original OFDM signal,
before the HPA, where the weights are imposed by using signal

.
It is shown that the PAPR of this weighted OFDM method is smaller
than that of the C&F method, and the BER performance is improved
compared with the C&F method.
APPENDIX
We introduce some properties of the signal in (4).
The signal in (4) was found in the process of sampling expansion
for a bandlimited signal, which has a polynomial growth on the
real line.
Note that signal is a bandlimited signal having a polynomial decay
on the real line, i.e., there exists C > 0, such that, for any real num-
ber x, (x) (C/(1 +|x|)
2
.
For example, if f is a bandlimited signal and there exists C

> 0,
such that, for any real number x, |f(x)| C

(1 +|x|),, then the mul-

tiplication f is a bandlimited signal with nite energy; therefore,
we can apply the Shannon sampling theorem to f , and we can
obtain the sampling expansion of the signal f on the real line.
ACKNOWLEDGMENT
The authors would like to thank anonymous reviewers for their
comments, which improved the quality of this paper.
REFERENCES
[1] X. Li and L. J. Cimini, Jr., Effects of clipping and ltering on the
performance of OFDM, IEEE Commun. Lett., vol. 2, no. 5, pp. 131133,
May 1998.
[2] J. Armstrong, Peak-to-average power reduction for OFDM by repeated
clipping and frequency domain ltering, Electron. Lett., vol. 38, no. 5,
pp. 246247, Feb. 2002.
[3] Y.-C. Wang and Z.-Q. Luo, Optimized iterative clipping and ltering for
PAPR reduction of OFDM signals, IEEE Trans. Commun., vol. 59, no. 1,
pp. 3337, Jan. 2011.
[4] X. Li and J. A. Ritcey, M-sequences for OFDM peak-to-average power
ratio reduction and error correction, Electron. Lett., vol. 33, no. 7,
pp. 554555, Mar. 1997.
[5] V. Tarokh and H. Jafarkhani, On the computation and reduction of
the peak-to-average power ratio in multicarrier communications, IEEE
Trans. Commun., vol. 48, no. 1, pp. 3744, Jan. 2000.
[6] J. A. Davis and J. Jedwab, Peak-to-mean power control in OFDM,
Golay complementary sequences, and reed-muller codes, IEEE Trans.
Inf. Theory, vol. 45, no. 7, pp. 23972417, Nov. 1999.
[7] K. G. Paterson and V. Tarokh, On the existence and construction of good
codes with low peak-to-average power ratios, IEEE Trans. Inf. Theory,
vol. 46, no. 6, pp. 19741987, Sep. 2000.
[8] R. W. Bauml, R. F. H. Fischer, and J. B. Huber, Reducing the peak-
to-average power ratio of multicarrier modulation by selected mapping,
Electron. Lett., vol. 32, no. 22, pp. 20562057, Oct. 1996.
[9] S. H. Muller and H. B. Huber, OFDM with reduced peak-to-mean power
ratio by optimum combination of partial transmit sequences, Electron.
Lett., vol. 33, no. 5, pp. 368369, Feb. 1997.
[10] A. D. S. Jayalath and C. Tellambura, Reducing the peak-to-average
power ratio of orthogonal frequency division multiplexing signal through
bit or symbol interleaving, Electron. Lett., vol. 36, no. 13, pp. 1161
1163, Jun. 2000.
[11] J. Tellado, Peak to average power reduction for multicarrier modulation,
Ph.D. dissertation, Stanford Univ., Stanford, CA, Sep. 1999.
[12] S. Yoo, S. Yoon, S. Y. Kim, and I. Song, A novel PAPR reduction scheme
for OFDM systems: Selective mapping of partial tones (SMOPT), IEEE
Trans. Consum. Electron., vol. 52, no. 1, pp. 4043, Feb. 2006.
[13] B. S. Krongold and D. L. Jones, PAR reduction in OFDM via active
constellation extension, IEEE Trans. Broadcast., vol. 49, no. 3, pp. 258
268, Sep. 2003.
[14] X. Huang, J. Lu, J. Zheng, K. B. Letaief, and J. Gu, Companding
transform for reduction in peak-to-average power ratio of OFDM sig-
nals, IEEE Trans. Wireless Comuun., vol. 3, no. 6, pp. 20302039,
Nov. 2004.
[15] T. Jiang, Y. Yang, and Y. H. Song, Exponential companding technique
for PAPR reduction in OFDM systems, IEEE Trans. Broadcast., vol. 51,
no. 2, pp. 244248, Jun. 2005.
[16] X. Wang, T. T. Tjhung, and Y. Wu, On the SER and spectral analyses of
A-law companded multicarrier modulation, IEEE Trans. Veh. Technol.,
vol. 52, no. 5, pp. 14081412, Sep. 2003.
[17] H. Nikookar and R. Prasad, Weighted OFDM for wireless multipath
channels, IEICE Trans. Commun., vol. E83B, no. 8, pp. 18641872,
Aug. 2000.
[18] G. B. Folland, Real Analysis. Hoboken, NJ: Wiley-Interscience, 1999.
Adaptive-Rate Transmission With Opportunistic
Scheduling in Wireless Networks
Yufeng Wang, Ravi Sankar, and Salvatore Morgera
AbstractIn this paper, we propose an adaptive-rate transmission
scheme with opportunistic scheduling in a two-hop relay network over a
Rayleigh fading environment. We assume that only channel state informa-
tion (CSI) at receivers is available in a decentralized network, where n
source-to-destination pairs are completely independent and mhalf-duplex
relays cooperate by exchanging their selected source IDs at the beginning
of the two-hop transmission. In the limit of large nand xed m, the system
throughput of the scheme scales as (m/2) log log n. It is shown that
this is the same achievable scaling even with perfect CSI assumptions at
transmitters and full cooperation among nodes. Furthermore, it is shown
that the optimal scaling of mis (log n), under which a linear increase
in throughput with m is obtained. A closed-form delay expression of the
proposed scheme is also presented.
Index TermsCooperative relaying, opportunistic scheduling, two-hop
communication.
I. INTRODUCTION
The demand for more efcient and reliable communication net-
works has given rise to a number of innovative systems and network
architectures, such as ad hoc networks, cognitive radio (CR), relay
extensions for cellular networks, and wireless sensor networks. In their
seminal work [1], Gupta and Kumar focused on the study of system
throughput. Numerous schemes have since been proposed that apply
different assumptions to the channel state information (CSI) and levels
of cooperation among communicating nodes [2][5]. These studies
have contributed to the understanding of system throughput in wireless
networks. One common thread among these studies is a focus on
the scaling of the system throughput with xed-rate transmissions.
For some network architectures, such as ad hoc, sensor, and CR
networks, using adaptive-rate transmission may provide better system
performance, due to its ability to change transmission parameters. In
particular, the transmit power and modulation level can be adjusted
Manuscript received May 25, 2012; revised September 16, 2012 and
October 19, 2012; accepted October 26, 2012. Date of publication
November 21, 2012; date of current version March 13, 2013. The review of
this paper was coordinated by Prof. C. P. Oestges.
The authors are with the Department of Electrical Engineering, University
of South Florida, Tampa, FL 33620 USA (e-mail: ywang2@mail.usf.edu;
sankar@eng.usf.edu; sdmorgera@eng.usf.edu).
Color versions of one or more of the gures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identier 10.1109/TVT.2012.2228512
0018-9545/$31.00 2012 IEEE
w
w
w
.
t
a
k
e
o
f
f
p
r
o
j
e
c
t
s
.
c
o
m
For Further Details-A Vinay 9030333433,0877-2261612 3