M. S. Hossain, S. Ahmed, E. Ullah and M. A. Islam / IJECCT 2013, Vol.

3 (2) 1

PAPR Reduction of OFDM System Through Iterative

Selection of Input Sequences
Md. Sakir Hossain1, Sabbir Ahmed2, Enayet Ullah3, Md. Atiqul Islam1
1
International Islamic University Chittagong, Bangladesh
2
Ritsumeikan University, Japan
3
University of Rajshahi, Bangladesh
[Link]@[Link]

Abstract: Orthogonal Frequency Division Multiplexing portable wireless LAN terminal etc. Thus implementation
(OFDM) based multi-carrier systems can support high of OFDM technique in a power efficient manner has drawn
data rate wireless transmission without the requirement of significant research attention in the recent past and many
any extensive equalization and yet offer excellent methods have been proposed with the objective of PAPR
immunity against fading and inter-symbol interference. reduction
But one of the major drawbacks of these systems is the A survey of PAPR reduction techniques reveals that
large Peak-to-Average Power Ratio (PAPR) of the perhaps the most widely known methods are signal
transmit signal which renders a straightforward clipping [6], block coding [7-8], selected mapping SLM
implementation costly and inefficient. In this paper, a new [9] and partial transmit sequence (PTS) [10-12]. Of them,
clipping is a very simple method to reduce PAPR . This
PAPR reduction scheme is introduced where a number of
can lower the PAPR easily by cutting away the signal
sequences from the original data sequence is generated by
above the assigned clip level. But it results in out-of-band
changing the position of each symbol and the sequence radiation and in-band distortion causing poor signal
with lowest PAPR is selected for transmission. A quality. Block coding method can reduce PAPR
comparison of performance of this proposed technique significantly. It has been used in Magic WAND (wireless
with an existing PAPR reduction scheme, i.e., the ATM network demonstrator) system because it does not
Selective Mapping (SLM) is performed. It is shown that degrade the OFDM signal and shows an additional coding
considerable reduction in PAPR along with higher effect. However, the code rate and bandwidth efficiency
throughput can be achieved at the expense of some are very low. Also, computation is exponentially increased
additional computational complexity. with the number of subcarriers. Finally, SLM and PTS are
the phase control method to reduce PAPR. SLM multiplies
Keywords – OFDM; PAPR; SLM; Iterative selection. an OFDM data by several phase sequences in parallel and
selects the data sequence of the lowest PAPR among them.
PTS divides the input OFDM data into several clusters and
phase rotation factors (or combining sequences) are
I. INTRODUCTION
multiplied to get the low PAPR signal. Although these two
OFDM is a very attractive technique for high speed data methods can reduce PAPR effectively without any signal
transmission in mobile communications due to various distortion, the side information about the phase rotation
advantages such as high spectral efficiency, robustness to must be transmitted to the receiver. Error in the side
channel fading, immunity to impulse interference, and information can cause significant BER performance
capability of handling very strong multipath fading and degradation. Furthermore, system complexity considerably
frequency selective fading without having to provide goes up because of many IFFT (inverse fast Fourier
powerful channel equalization. Considering these, it has transform) stages and the long phase optimization
already been adopted as the standard transmission processes.
technique in the wireless LAN systems [1],[2] and the In this paper, we propose a new probabilistic scheme to
terrestrial digital broadcasting system [3]. It is also being reduce PAPR based on the selection of input sequence.
considered as one of the candidate transmission techniques This proposed technique produces all possible permutation
for the next generation of mobile communications systems. of input sequence and the sequence with minimum PAPR
One drawback of OFDM technique is the larger peak to is transmitted. The simulation result show that the
average power ratio (PAPR) of its time domain signal [4], proposed method can show better PAPR reduction
[5]. The problem stems from the fact that high PAPR can performance than SLM with relatively less side
drive transmitter power amplifiers in to saturation region information.
causing inter-modulation noise and there-by inflict severe
degradation of bit error rate (BER) performance. The
simplest solution to overcome this problem is to operate II. OFDM SYSTEM MODEL
the non-linear amplifier at the linear region by taking out Let us define N symbols in OFDM are { Xn, n=0,1 2,…N-
enough power from the input, i.e., input back-off. 1}. A set of N sub-carriers, i.e., { fn, n=0,1,2….,N-1}, is
However, this method degrades power efficiency of the used for these symbols in the OFDM. The N sub-carriers
overall system and hence is not suitable for power are chosen to be orthogonal, that is, fn=n∆f in frequency
constrained handheld devices like mobile terminal and domain, where ∆f=1/NT and T is the OFDM symbol
 M. S. Hossain, S. Ahmed, E. Ullah and M. A. Islam / IJECCT 2013, Vol. 3 (2) 2

duration. The basic OFDM transmitter and receiver is are also considerably large. However, in this section Selective
shown in figure 1. The OFDM signal is expressed as Mapping (SLM) and our proposed technique will be
described in detail.
1 N 1
xt    X n e j 2f n t , 0  t  T (2.1) A. SLM Method
N n 0 A block diagram of Selected Mapping (SLM) is shown in
In practical systems, a guard interval (cyclic prefix) is Figure 3. In the SLM, the U-1 statistically independent
inserted by the transmitter in order to remove intersymbol phase sequences are generated. Symbol sequences are
interference and interchannel interference (ICI) in the multiplied by the U-1 different phase sequences whose
multipath environment. However, it can be ignored since it length is equal to the number of carriers before the IFFT
does not affect the PAPR. process. The PAPR is calculated for the U-1 phase rotated
symbol sequences and one original sequence. Then the
symbol sequence with the lowest PAPR is selected and
transmitted. The used phase sequence is transmitted as
the side information. The receiver performs the reverse
rotation to recover the data symbol. The SLM needs the
IFFT process whose number is equal to U and thus a large
amount of calculation.

Figure 1: Basic OFDM Transmitter and Receiver

The PAPR of the transmit signal s(t), defined above, is

the ratio of the maximum instantaneous power and the
average power, given by
2
max xt 
PAPR  (2.2)

E xt 
2

where E{.} denotes expectation. In some frames of OFDM
signals, large PAPR happens since the structure of the
given symbols may cause these peaks. IFFT operation
can be viewed as multiplying sinusoidal functions to the
input sequence, summing, and sampling the results. Thus Figure 2: Illustration of the effect of number of subcarriers on
PAPR when FFT size is fixed at 128.
the high correlation property of IFFT input causes the
sinusoidal functions to be arranged with in-phase form.
After summing the in-phase functions, the output might
have large amplitude.
Moreover, with the increase of number of subcarriers,
PAPR of the resulting system also increases. The reason
for this is that when the number of subcarriers is large and
they all are added in some positive or negative phases, the
resulting amplitude becomes large enough to exceed
saturation point of high power amplifier (HPA). Figure 2
shows such situation.
III. PAPR REDUCTION STRATEGIES
A number of methods are available for PAPR reduction. Figure 3: OFDM transmitter model for the SLM approach.
But each of these methods provides some advantages from
others creating some problems also. Here we introduce a B. PROPOSED METHOD
new PAPR reduction strategy that is based on iterative
Here we propose a new statistical scheme, which we shall
selection of input sequences. In this technique, different
call in this paper Iterative Selection of Input Sequence
permutations of input data are made and the permutation
with the least PAPR is selected for transmission. But in (ISIS), which can reduce PAPR to a minimal level. The
SLM, the input sequence is modified by multiplying it block diagram of the proposed system is shown in figure
with number of phase vectors and the sequence with the 4.
minimum PAPR is selected for transmission. Another
advantage of proposed technique in addition to the least
PAPR is that it does not need any multiplication.
Moreover, the state space for finding minimum PAPR is
 M. S. Hossain, S. Ahmed, E. Ullah and M. A. Islam / IJECCT 2013, Vol. 3 (2) 3

Table 1: Simulation Parameters

Modulation Method QPSK
The number of FFT points 16
The number of subscribers 8
No. of frame transmitted 512

It is found from figure 5 that SLM reduces PAPR

significantly. Figure 5 also compares the performance of
Walsh-Hadamard sequence and Golay complementary
sequence as techniques for producing phase rotation vector
for SLM method. The performances of both sequences are
almost same. An important thing that is found from this
Figure 4: Block diagram of ISIS approach
figure is that a considerable amount of more PAPR
reduction is obtained by the proposed technique than the
The proposed method can be explained as follows:
SLM. It is seen that PAPR of proposed technique is about
Step-1: Take the input sequence to be transmitted to a 2 dB lower than SLM. The reason for this result can be
buffer. explained as: IFFT actually multiplies input sequences by
Step-2: Produce all possible permutations of the buffered sinusoidal functions, sum different signals after
sequence. If the number of elements in the input sequence multiplication, take sample after summation. If we
is N then the maximum number of permutations would be rearrange the input sequence number of times and perform
N!. multiplications, summation and sampling for each
Step-3: Pass each parallel sequence through IFFT arrangement, then maximum peaks for each set of samples
separately and compute the PAPR for each sequence. Put will be different. If we choose the time domain signal with
the resulting PAPR in buffer. least maximum peak, then signal with minimum PAPR can
Step-4: Finally find the sequence for which PAPR is be achieved. For maximum number of rearrangement
minimum and transmit that sequence. minimum PAPR can be achieved. If OFDM frame size is n
Step-5: As side information, send the serial number of the then maximum number of arrangement that can be
transmitted sequence in the permutation table through a achieved is n!. Thus to get minimum PAPR,
secure channel or it may be transmitted number of times n! arrangements are needed to be used. Thus the minimum
and at the receiver majority voting technique may be PAPR can be achieved from this technique but with
applied for finding the correct sequence number. relatively more complexity.
Step-6: At the receiver, produce all possible permutation It is found from Figure-5 that for minimum PAPR, the
of the received sequence and place them in buffer. number of sequences that are needed to use is n!. The
effect of number of sequences on PAPR (i.e., how does
Step-7: Produce permutations of each buffered sequence
PAPR vary with number of sequences used) is investigated
and compare the SI-th sequence to the received sequence.
below. The effect of total number of sequences used on
If they are all the same. Then the current sequence whose proposed method’s performance is shown in Figure-6.
permutation is created is the actual desired sequence that
was altered and transmitted.
An example of this method would be: For example, if
original data sequence is [A.B,C,D], then there are 24
combinations of these four symbols and say, the sequence
[B,C,D,A] produces a minimum PAPR. The sequence
[B,C,D,A] is selected for transmission. If the index of this
sequence in the permutation table is S then S is sent as side
information. At the receiver, permutation of [B,C,D,A] is
taken and the permutation of each sequence obtained from
permutation of [B,C,D,A] is taken. Then match the
received sequence with S-th sequence of each 24 set of
sequence. The sequence for whose S-th sequence is the
received sequence (that is, [B,C,D,A]) is the original data
sequence for which [B,C,D,A] was sent from transmitter.
However, input sequence can also be changed on
binary data and recovered in the receiver in binary form.

IV. SIMULATION RESULTS

Figure 5: Comparison between SLM with two sequences and
Here we simulate figure 3 and figure 4 using Matlab and proposed ISIS method for PAPR reduction.
compare the results. Also we try to enhance the
performance of the ISIS approach. The simulation It is seen from figure-6 that with the increase of
parameters are given in Table I. number of sequences used, PAPR reduces. It is seen that
PAPR performance is same as SLM when number of
sequences used is 8 A considerable reduction of PAPR is
 M. S. Hossain, S. Ahmed, E. Ullah and M. A. Islam / IJECCT 2013, Vol. 3 (2) 4

achieved when 100 sequences are used. Using 500 V. CONCLUSION

sequences gives nearly same result as n! number of In this paper, we proposed a PAPR reduction scheme and
sequences. But using 1000 sequences shows the same compared its performance with the well known SLM
result as n!. Thus about 40 times complexity can be technique. We showed that with proposed method,
reduced using smaller number of sequences without any significant PAPR reduction can be achieved compared to
degradation of PAPR performance. SLM. In addition, the requirement of side information is
also lesser here, i.e., log2 (no number of phase rotation
vector) in SLM and only a single symbol in our scheme.
Moreover, in SLM the number of phase rotation vectors is
needed to be increased for better PAPR reduction. The
main overhead of our system is the computation
complexity which goes up with the number of possible
input sequences. But even with a small number of chosen
input sequences, our system shows similar PAPR
reduction capability compared to SLM with much less
computation overhead since there is no multiplication
involved as is the case with SLM.

REFERENCES
[1] IEEE Std. 802.11a, High-speed Physical Layer in the 5GHz Band,
1999.
Figure 6: Effect of number of total number of sequences used on [2] IEEE 802.16 WG, “SC-FDE PHY Layer System Proposal for Sub
PAPR reduction 11GHz BWA”, March 2001.
[3] P. Dambacher, “Digital Terrestrial Television Broadcasting,”
Figure 7 shows the effect of OFDM frame size on the Springer 1998.
performance of proposed technique. It is seen from figure [4] J. Tellado, “Multicarrier Modulation with Low PAR,” Kluwer
Academic Publishiers, 2000.
7 that for basic OFDM, PAPR increases for greater frame
[5] D Dardari, V. Tralli and A Vaccari, “A Theoretical
size (that is, for greater number of subcarrier). It is also Characterization of Nonlinear Distortion Effects in OFDM
seen that the proposed technique performs better for larger Systems,” IEEE Trans. on Comm., Vol. 48, no. 10, pp.1775-1764,
frame size. The reason for this is that as frame size Oct 2000.
increases the number of possible combinations also [6] X. Li and L. J. Cimini Jr., "Effects of clipping on the performance
increases. Hence the probability to find signal with lower of OFDM with transmitter diversity," in Proc. IEEE
PAPR also rises. Vehicular Technology Conferences, vol. 3, pp. 1634-1638, May
1997.
[7] T. A. Wilkinson and A. E. Jones, "Minimization of the Peak-
to- Mean Envelope Power Ratio of Multicarrier
Transmission Schemes by Block Coding," Proc. of IEEE VTC '95,
Chicago, pp.
825-829, July 1995.
[8] H. Ochiai and H. Imai, "Block Coding scheme Based on
Complementary Sequences for Multicarrier Signals," IEICE Trans.
on Fundamentals, pp. 2136-2143, Nov. 1997.
[9] A. D. S Jayalath, C. Tellambura, H. Wu “A reduced
complexity PTS and new phase sequence for SLM to reduce PAP
of an OFDM signal," IEEE VTC 2000, Vol. 3, pp. 1914 -1917,
2000.
[10] L. J. Cimini, N. R. Sollenberger "Peak-to-Average Power Ratio
Reduction of an OFDM Signal Using Partial Transmit Sequences,”
IEEE Communications Letters, Vol. 4, pp. 86-88, March 2000.
[11] L. J. Cimini, N. R. Sollenberger "Peak-to-Average Power Ratio
Reduction of an OFDM Signal Using Partial Transmit Sequences
With Embedded Side Information," IEEE GLOBECOM '00, Vol.
Figure 7: Effect of number of subcarrier on the performance of 2, pp. 746-750, 2000.
proposed technique [12] A. D. S Jayalath, C. Tellambura, "Adaptive PTS approach for
reduction of peak-to-average power ratio of OFDM
signal," Electronics Letters, Vol. 36, pp. 1226-1228 July 2000.