IEEE SIGNAL PROCESSING LETTERS, VOL. 17, NO.

10, OCTOBER 2010

883

Joint Channel Estimation and PTS to Reduce

Peak-to-Average-Power Radio in OFDM
Systems Without Side Information
Lili Guan, Tao Jiang, Daiming Qu, and Yang Zhou

AbstractIn this letter, we propose a novel scheme to reduce the

PAPR of OFDM signals, and its key idea is that the side information of partial transmit sequence is considered as a part of channel
frequency response. Therefore, the proposed scheme does not need
to reserve bits for submitting side information, resulting in an increase in the data rate. Furthermore, we propose a novel pilot arrangement, in which pilot tones are independently inserted into
each subblock. Simulations show that the proposed scheme with
the novel pilot arrangement could offer good performances of bit
error rate and PAPR reduction.
Index TermsChannel estimation, OFDM, partial transmit sequence, peak-to-average power ratio, side information.

OFDM signals could be directly recovered via channel estimation without any knowledge of the SI at the receiver. Therefore,
the system does not need to reserve extra bits to deposit and
protect the phase rotation factors. Furthermore, we also propose a novel pilot arrangement, in which the pilot tones are independently inserted into each subblock to achieve an accurate
channel estimation.
The rest of this letter is organized as follows. We briefly describe the OFDM system and channel estimation in Section II.
In Section III, the CE-PTS method and novel pilot arrangement are proposed in detail. The simulation results are shown
in Section IV, followed by conclusions in Section V.

I. INTRODUCTION

T is well-known that orthogonal frequency division multiplexing (OFDM) has been adopted in various wireless communication standards. However, it suffers from high peak-to-average power ratio (PAPR). Therefore, many methods have been
proposed to reduce the PAPR of OFDM signals [1]. For example, one promising scheme is the partial transmit sequence
(PTS) method [2][5]. However, for the traditional PTS (T-PTS)
method, the communication system has to reserve some bits for
the transmission of the phase rotation factors as side information (SI), resulting in a decrease in the data rate. Note that, SI
only represents the information about the phase rotation factors
in this letter. Commonly, it is hard to correctly recover the SI at
the receiver, which may largely increase the bit error rate (BER).
When extra bits are employed to protect the SI for being correctly submitted, the total data rate is further degraded.
In this letter, we propose a novel method, which combines
the channel estimation and T-PTS method to reduce the PAPR
of OFDM signals, called as the CE-PTS method. Particularly,
a virtual channel frequency response is considered as the combination of the traditional channel frequency response and the
phase rotation factors of the T-PTS method. Thus, the original

Manuscript received June 28, 2010; revised August 04, 2010; accepted August 05, 2010. Date of publication August 12, 2010; date of current version
August 26, 2010. This work was supported in part by the NSF of China under
Grant 60872008, the Program for New Century Excellent Talents in University
of China under Grant NCET-08-0217, and the Research Fund for the Doctoral
Program of Higher Education of the Ministry of Education of China under Grant
200804871142. The associate editor coordinating the review of this manuscript
and approving it for publication was Dr. Teng Joon Lim.
The authors are with the Wuhan National Laboratory for Optoelectronics,
Department of Electronics and Information Engineering, Huazhong University
of Science and Technology, Wuhan 430074, China (e-mail: guanlili87@gmail.
com; [email protected]; [email protected]; [email protected]).
Digital Object Identifier 10.1109/LSP.2010.2066562

II. SYSTEM MODEL

For an OFDM system, the transmitted signal in the time domain is given by
(1)
subcarriers, and
is the input data
independent modulated data symbols
block with
.
is defined as
In general, the PAPR of the OFDM signal
where

is

the

number

of

(2)
where
represents the expectation.
Commonly, to mitigate the inter-symbol interference, pilot
tones need to be inserted to acquire the channel state information (CSI) for frequency-selective channels in OFDM systems
at the receiver, such as to estimate the frequency response of the
fading channel. For simplicity, we do not consider the CSI at
the transmitter in this letter, and the frequency domain channel
estimation based on the comb type pilot arrangement [6] is conas the locations of
sidered. Denote
, where
pilot tones and
is the number of pilot tones. Obviously,
is the pilot
, whereas,
when
. Denote
tone for
as the complementary set of
in
, representing the locations of the data symbols, and
. Obviously,
when
and
is the corresponding data symbol
.
Therefore, we have

1070-9908/$26.00 2010 IEEE

884

IEEE SIGNAL PROCESSING LETTERS, VOL. 17, NO. 10, OCTOBER 2010

(3)
For the arrangement of the comb type pilot used for channel
estimation, the pilot tones are uniformly inserted with the
between the adjacent pilots. Based on the requirespacing
satisfies
ment of channel estimation,

is obtained by taking the inverse

The time domain signal
, which is indefast fourier transform (IFFT) operation to
pendently multiplied by the phase factor
, where
is the number of the allowed phase
, we obfactors. Meanwhile, after taking IFFT operation to
. Then, the time domain OFDM signal
tain
is
(10)

(4)
is the length of an OFDM
where is the symbol period,
is the length of the maximum multipath delay,
data block,
denotes the largest integer smaller than or equal to
and
[7]. Note that,
, where
denotes the smallest
integer larger than or equal to [6]. Therefore, we have
(5)
.
where is a positive integer and
At the receiver, after taking the fast fourier transform (FFT)
operation, the received signal in the frequency domain can be
written as
(6)

Therefore, the aim of the T-PTS scheme is to find an approto minimize

priate phase sequence
the PAPR of the transmitted OFDM signals . As a result, for
the T-PTS method, when no bits are reserved to protect SI, the
bits.
SI should be sent to the receiver with
B. Proposed CE-PTS Scheme
is parFor the proposed CE-PTS method, the data block
disjoint subblocks
titioned into
instead of
. Denote
as the time domain signal
having the lowest PAPR with the phase factors
. Thus, the frequency domain data block
of
is expressed as

where

is the frequency response of the fading channel, and

is the frequency domain additive noise. Denote
as
, and we have
the estimation of

(11)

(7)

At the receiver, after taking the FFT operation, the received

signal in the frequency domain is

and
are the estimated frequency channel
where
response on the pilot tone and the data symbol, respectively.
are known at the
Since the value and the location of
from (6) and (7), i.e.,
receiver, we could obtain

(12)

(8)

Similarly,

the

received

signal

is
partitioned
disjoint subblocks
, where
into
the type of the disjoint subblocks is the same as that of the
transmitter. As a result

Then,

is obtained by linear interpolation when

, otherwise, it is obtained by linear extrapolation.
is
Therefore, the estimated data symbol

(13)

(9)

where is the index of the subcarrier in the

is a positive integer with
Denote

subblock, and
.

(14)
III. PROPOSED CE-PTS SCHEME
A. T-PTS Scheme
As we know, the pilot tones are inserted to estimate the frequency response of the fading channel. For
remains unchanged and
the T-PTS method,
is partitioned into U disjoint subblocks
. Therefore,
, where
. For simis a positive integer. Therefore,
plicity, we assume
when
, otherwise,
.

as the virtual frequency channel response, which is the combination of the traditional channel frequency response and the phase
rotation factors used in the T-PTS method.
Substituting (3) and (14) into (13), we have
(15)
where
and
are the virtual frequency channel response on the pilot tone and the data symbol, respectively.
Similarly, we have
(16)

GUAN et al.: JOINT CHANNEL ESTIMATION AND PTS

885

Fig. 1. Comparisons between the comb type pilot arrangement and the novel type pilot arrangement with

where
is the estimation of
. After linear interpolation and extrapolation, we have the estimated channel response
. Therefore, the estimated data
at the receiver is

N = 256, N = 24 and k 2 [64; 127].

TABLE I
SYSTEM PARAMETERS EMPLOYED FOR SIMULATION

(17)
Comparing (9) with (17), we find that the CE-PTS method
without
could directly recover the original data symbol
the SI. In other words, unlike the T-PTS method, the proposed
scheme does not need to reserve extra bits to deposit and protect
.
the phase rotation factors
C. Novel Type Pilot Arrangement
For the proposed method, we independently make channel
estimation for each subblock with
subcarriers and
pilot tones. For simplicity,
is assumed as a positive integer. Particularly, denote as the number of the nonzero subas the number of
carriers in front of the first pilot tone, and
subthe nonzero subcarriers behind the last pilot tone in the
block. Note that, due to the lack of pilot tones at the edge of
subblock, the channel estimation error for the edge
the
and
subcarriers may be increased via the extrapolation. In
and
is large,
other words, when the difference between
the channel estimation error may be significantly increased [8].
To illustrate the principle of the novel type pilot arrangement,
we give some intuitive illustrations as shown in Fig. 1, where
,
,
,
, and
.
Fig. 1(a) shows the locations of the pilot tones for the traditional
comb pilot arrangement. As shown in Fig. 1(b), the channel estimation error of the CE-PTS method with the traditional comb
pilot arrangement may be significantly increased due to the bad
and
effect of extrapolation since the difference between
is large (
and
). In this paper, we propose a novel
pilot tones are
comb type pilot arrangement, in which
independently inserted into each subblock with the spacing
between adjacent pilots. According to the result analyzed in [8],
for the novel pilot arrangement, we set
(18)
and
(19)

Fig. 2. Comparison of PAPR reductions between the T-PTS and CE-PTS

methods.

As shown in Fig. 1(c), 3 pilot tones are independently inserted

and
to reduce the bad
into each subblock with
subblock. Simulation results
effect of extrapolation for the
conducted in the next section show that the proposed CE-PTS
scheme with novel comb pilot arrangement can offer better BER
performance than that with the traditional comb type arrangement.
IV. SIMULATION RESULTS
To verify the performances of the proposed scheme in the
OFDM data blocks are generated
PAPR reduction and BER,
randomly, the convolutional encoder with the constraint length
7 and the solid-state power amplifier (SSPA) with the saturation
are employed. For simplicity, the set of phase
point
, i.e.,
. The system parotation factors is
rameters employed for simulation are shown in Table I.
Fig. 2 shows the complementary cumulative distribution
functions (CCDF) of the PAPR for original OFDM signals and
other signals obtained by the T-PTS and CE-PTS methods,
respectively. For the T-PTS method, the value of pilot remains
unchanged. As seen from Fig. 2, compared with original

886

Fig. 3. BER performances over a Rayleigh fading channel, U

10 T.

IEEE SIGNAL PROCESSING LETTERS, VOL. 17, NO. 10, OCTOBER 2010

= 8 and L T =

arrangement. However, the BER performance of the proposed

CE-PTS method with novel type pilot arrangement is better than
that of the CE-PTS method with traditional comb pilot arrange, the BERs of
ment. As shown in Fig. 3, when
, CE-PTS2,
, and CE-PTS1,
the Perfect SI,
are
,
and
, respectively. However, the SI of the T-PTS method may be not perfectly received. For the simulation, we set the error probability
. As shown in Fig. 3, when SNR is
of the SI as a constant
low, the BER performance mainly depends on the SNR. However, with the increase in SNR, the imperfect SI affects the BER
performance more severely than the noise. For example, when
, both the BERs of the Perfect SI,

are
. However, when
and Imperfect SI,
, the BERs of the Perfect SI,
and
are
and
, respecImperfect SI,
tively. Furthermore, the PAPR performances of the two methods
are better than those when
, which is less
when
influenced by the HPA nonlinearity, such that the BER perforare
mances of the T-PTS and CE-PTS methods when
, comparing Fig. 3 with Fig. 4.
better than those when
Obviously, for the T-PTS method, when the SI is perfectly
obtained at the receiver, its BER performance is a little better
than that of the proposed CE-PTS scheme with novel type pilot
arrangement. However, in a real OFDM system, the T-PTS
method cannot ensure that the SI could be perfectly received at
the receiver.
V. CONCLUSIONS

Fig. 4. BER performances over a Rayleigh fading channel, U

10 T.

= 4 and L T =

OFDM signals, the T-PTS method offers about 4.1 dB and 5.4
dB of the PAPR reduction at
when
and
, respectively. The proposed CE-PTS method provides
and
the PAPR reduction of 3.5 dB and 5.2 dB when
, respectively. Obviously, both the two methods could
achieve more PAPR reduction when is larger.
Fig. 3 and Fig. 4 show the BER performances of the T-PTS
and
, reand the proposed CE-PTS methods when
spectively. The Ideal curve represents the BER performance
of the original signal transmitted without the high power amplifier (HPA) and PAPR reduction. The curve labeled Perfect
SI shows the BER performance of the T-PTS method with both
the perfect SI and the traditional comb type pilot arrangement.
The Imperfect SI curve represents the BER performance of
the T-PTS method with the imperfect side information and the
traditional comb type pilot arrangement. The CE-PTS1 and
CE-PTS2 curves describe the BER performances of the proposed CE-PTS scheme with the traditional comb type pilot arrangement and the novel type pilot arrangement, respectively.
Without HPA and PAPR reduction, the OFDM system has
when
. The BER perforthe BER of
mance of the T-PTS method with perfect SI is a little better
than that of the proposed CE-PTS scheme with novel type pilot

In this letter, we proposed a novel pilot arrangement to

cooperate the proposed CE-PTS scheme, which combines the
channel estimation and the PAPR reduction in OFDM system.
Compared with the T-PTS method, the proposed CE-PTS
scheme does not need to reserve extra bits for submitting and
protecting the SI used in T-PTS, which results in the increase
of the data rate.
REFERENCES
[1] T. Jiang and Y. Wu, An overview: Peak-to-average power ratio reduction techniques for OFDM signals, IEEE Trans. Broadcast., vol. 54,
no. 2, pp. 257268, June 2008.
[2] Y. Xiao, X. Lei, Q. Wen, and S. Li, A class of low complexity PTS
techniques for PAPR reduction in OFDM systems, IEEE Signal
Process. Lett., vol. 14, no. 10, pp. 680683, Oct. 2007.
[3] T. Jiang, W. Xiang, P. C. Richardson, J. Guo, and G. Zhu, PAPR reduction of OFDM signals using partial transmit sequences with low
computation complexity, IEEE Trans. Broadcast., vol. 53, no. 3, pp.
719724, Sep. 2007.
[4] S. H. Han, J. M. Cioffi, and J. H. Lee, On the use of hexagonal constellation for peak-to-average power ratio reduction of an OFDM signal,
IEEE Trans. Wireless Commun., vol. 7, no. 3, pp. 781786, Mar. 2008.
[5] J.-C. Chen, Partial transmit sequences for peak-to-average power
ratio reduction of OFDM signals with the cross-entropy method,
IEEE Signal Process. Lett., vol. 16, no. 6, pp. 545548, Jun. 2009.
[6] N. Chen, J. Zhang, and P. Zhang, Improved channel estimation based
on parametric channel approximation modeling for OFDM systems,
IEEE Trans. Broadcast., vol. 54, no. 2, pp. 217225, Jun. 2008.
[7] Z. Du, N. C. Beaulieu, and J. Zhu, Selective time-domain filtering
for reduced-complexity PAPR reduction in OFDM, IEEE Trans. Veh.
Technol., vol. 58, no. 3, pp. 11701176, Mar. 2009.
[8] Y. H. Jung and H. Park, Pilot sharing method for uplink OFDMA
channel estimation, Electron. Exp., vol. 6, no. 14, pp. 10511056, Jul.
2009.