Peak-to-Average Power Ratio Reduction Techniques For OFDM Signals
Peak-to-Average Power Ratio Reduction Techniques For OFDM Signals
Peak-to-Average Power Ratio Reduction Techniques For OFDM Signals
Due to its advantages, OFDM modulation has been employed Thirdly, these techniques can be classified as deterministic or
in many wireless applications such as Wireless Personal Area, probabilistic schemes [13]. In deterministic techniques such as
Local Area and Metropolitan Area Networks, Digital Audio Peak Clipping, the PAPR of the signal is kept below a
and Video Broadcasting [3]. It is also being used as a prefixed threshold level. In this process, the OFDM signal
modulation technique for IEEE 802.20, IEEE 802.16 and gets distorted. In probabilistic techniques, several OFDM
3GPP-LTE [3]. A simple one tap equalizer is required at the signals are generated and the one having smallest PAPR is
receiver side as the effect of ISI is eliminated by introducing transmitted. This method includes PTS, SLM, TI and TR
cyclic prefix (CP) [2]. techniques.
Despite its advantages, some problems still need to be Lastly, the PAPR reduction techniques can be classified
resolved in the design of OFDM systems. One of the depending on whether these transform the original OFDM
prominent problems is high Peak-to-Average Power Ratio signal before or after the multi carrier modulation at the
(PAPR) of the transmitted OFDM signals. The transmit transmitter [9]. Techniques like coding, SLM and PTS fall
signals in an OFDM system can have high peak values in the under the first category and different types of companding
time domain as many subcarriers are added due to IFFT techniques belong to the second type as they change the
operation at the transmitter. Therefore, OFDM systems have a formation of the signal after multi carrier modulation.
high PAPR as compared with single-carrier communication
40
International Journal of Computer Applications (0975 – 8887)
Volume 96– No.22, June 2014
max xL [n]
In the remaining paper, the distribution of PAPR is analyzed 2
based on the characteristics of OFDM signals. Then various
PAPR xL [n] 0 n NL 1
techniques to suppress PAPR in OFDM systems are described (5)
2
and reviewed. Various criterions for the choice of appropriate E xL [ n ]
PAPR reduction technique are also discussed in this paper.
2.3 Complementary Cumulative
2. OFDM MODEL AND PAPR
Distribution Function (CCDF) of PAPR
2.1 OFDM System Model The CCDF can be used to estimate the bounds of the PAPR
OFDM is a block transmission scheme. At the transmitter and is generally used as a performance indicator for most of
side, a block of N data symbols represented as the PAPR reduction schemes.
X X k , k 0,1,...., N 1 is first converted from serial to
As per probability theory, the CCDF represents the
parallel stream. Each symbol modulates separately one of the probability that a random variable X will be having a value
subcarriers taken from fk , k 0,1,....., N 1. greater than or equal to real valued number x [14]. Central
limit theorem suggests that the real and imaginary parts of the
To ensure, the subcarrier frequencies, f k k f , are separated discrete OFDM signal follow Gaussian distributions for
sufficiently large number of subcarriers, assuming that each of
equally with subcarrier spacing f 1 NTs . Here Ts is the
imaginary and real parts have zero valued mean and a
original symbol period and NTs denotes the OFDM block variance of 0.5. This implies that the amplitude of an OFDM
period. signal follow Rayleigh distribution.
The OFDM signal is generated by the summation of the The CCDF of the PAPR denotes the probability that the
orthogonal and modulated subcarriers. Hence the complex PAPR of the transmitted OFDM symbols exceeds a preset
baseband OFDM signal in continuous time domain can be threshold value.
represented as [3],
For an OFDM system with N subcarriers, in which sampling
N 1 is done at Nyquist rate, the CCDF of PAPR represented as
X e
1 2 fk t
x (t ) k
, 0 t NTs (1) [14]
N k 0
Pr ob PAPR 1 1 e
N
X e
1 n
as well as uncorrelated. The above expression does not hold
x[ n] k
N
, n 0, 1, ......., N 1. (2)
N k 0
when oversampling is applied and when the number of
subcarriers is not large as the assumption of Gaussian
The discrete time OFDM signal is needs to be sampled at distribution does not hold in that case.
more than Nyquist rate for better approximation of the
continuous time signal. Let it be oversampled by a factor of L For an OFDM system with sufficiently large number of
subcarriers, the preceding expression does not hold good. For
i.e. the sampling frequency fs L / Ts . The discrete time such a system, an approximate expression of CCDF can be
OFDM signal can then be expressed as, given as [16]
nk
Pr ob PAPR 1 1 e
N 1 2.8 N
j 2
X e
1 (7)
xL [ n ] k
LN
, n 0, 1, ......., LN 1. (3)
N k 0
Figure 1 represents the distribution of the PAPR of the OFDM
It is to be noted that the time-domain samples of the OFDM signal with 256 number of subcarriers and different values of
signal which is L-times oversampled, are actually LN-point oversampling factor L. As evident from the figure, the PAPR
IFFT of the data block with ( L 1) N zero-padding. An does not increase much beyond L 4.
oversampling factor of L 4 is enough to successfully
represent the peak value of the continuous time OFDM signal 3. PAPR REDUCTION TECHNIQUES
[16]. Several techniques have been proposed till date in the
literature for the reduction of PAPR in OFDM systems. PAPR
2.2 Peak-to-Average Power Ratio reduction technique should be appropriately selected as per
The PAPR of an OFDM signal x(t ) in continuous time the system requirements, as no single technique may be
suitable for all systems.
domain is defined as the ratio of maximum instantaneous
power to its average power 3.1 Amplitude Clipping and Filtering
The idea of clipping technique is to clip the peak of the OFDM
max x(t )
2
0 t NTs signal below a prefixed threshold level [5]. The output of the
PAPR[ x(t )]
(4) amplitude clipper can be expressed as
2
E x (t )
x, x μ
C x (8)
where E
denotes the expectation operator.
μ, x >μ
The PAPR of the OFDM signal after L times oversampling is where μ is the prefixed threshold level selected for clipping
given by and is a positive real number.
41
International Journal of Computer Applications (0975 – 8887)
Volume 96– No.22, June 2014
42
International Journal of Computer Applications (0975 – 8887)
Volume 96– No.22, June 2014
Table 1. Performance Comparison of some coding represented by many equivalent points in the expanded
Techniques constellation. This results in extra degrees of freedom and
hence can be utilized for reducing PAPR. This method is
PAPR Reduction (dB) named so because the process of replacing a signal point in
N n R the original constellation by a new point in the expanded
CBC SBC MCBC CC
4 1 3/4 3.56 3.56 - 3.56 constellation is same as applying a subcarrier of appropriate
1 7/8 2.59 2.52 - 3.66 frequency and phase to the original OFDM signal.
8
2 3/4 2.67 3.72 2.81 (R=3/4) Unlike TI, it is essential to send side information along with
1 15/16 2.74 1.16 - the transmit signal in case of TR.
2 7/8 2.74 2.52 - 3.74
16 4. CRITERION FOR PAPR REDUCTION
3 13/16 2.74 - - (R=3/4)
4 3/4 2.74 2.98 3.46 TECHNIQUES
1 31/32 1.16 0.55 - PAPR reduction can be achieved only at the expense of
2 15/16 1.16 1.16 - bandwidth, data rate, BER performance, complexity and/or
3 29/32 2.75 - - power efficiency. An efficient trade off of these conflicting
32 -
4 7/8 2.50 2.51 - factors must be made before selecting an appropriate PAPR
5 27/32 2.75 - - reduction technique as follows:
8 3/4 2.75 3.00 3.45
4.1 Decrease in PAPR
Obviously this factor is the most important point in selecting
3.4 Selective Mapping and Partial an appropriate PAPR reduction technique. But care must be
Transmit Sequence taken to avoid other unwanted associated drawbacks such as
distortion introduced by the techniques.
In SLM, the transmit side produces a set of OFDM signals,
where all signals represent the same information as the 4.2 Increase in Power of Transmit Signal
original signal. Then the one having the lowest PAPR is
Certain techniques require an increase in the average power of
transmitted [12]. However this method is more
the transmit signal after using PAPR reduction method. TR
computationally complex as more than one IFFT blocks are
and TI need more signal power at the transmitter side. This
required at the transmitter. In addition, the useful data rate is
necessitates a HPA with large linear operating range resulting
suffered as additional overhead bits will be required to get
in degraded BER performance.
back the original data at the receiver. In [24], modified SLM
is suggested to reduce the computational complexity and to 4.3 Effective Loss in Data Rate
decrease the amount of side information required to be Some PAPR reduction techniques decrease the effective data
transmitted. rate of the OFDM signal. These include block coding, SLM
In PTS, the original input data block is divided into multiple and PTS. In block coding, parity check bits are required to be
disjoint sub-blocks. All subcarriers belonging to each sub- added. In SLM and PTS, transmit side information is needed
block are scaled by a phase factor. All these phase factors are to be sent to enable the receiver to bring back the signal to its
statistically independent. Subsequently, the signal with the original constellation size. When channel coding is also
lowest PAPR is selected for transmission. There are different applied, further loss of data rate takes place.
methods for dividing the data blocks into multiple sub-blocks.
These include pseudorandom partition, interleaved partition 4.4 BER Performance Degradation
and adjacent partition. Pseudo-random partitioning is the best This important factor is closely linked to the increase in power
among these methods. An improvement over existing PTS in of the transmit signal. In some techniques, if the transmit
MIMO-OFDM systems was proposed recently in [25], in signal power is fixed; there might be rise in BER at the
which the authors used constant modulation algorithm (CMA) receiver. In some techniques such as SLM and PTS, if the side
to solve the phase optimization problem. The resulting information is not received properly, the entire data block may
method has less computational complexity without any be lost. This will also degrade the BER performance at the
degradation in error performance compared with existing receiver.
methods.
4.5 Computational Complexity
Similar to SLM, the PTS also suffers from computational Generally, more complex techniques perform better as far as
complexity and low data rate. However PTS is less complex reduction in PAPR is concerned.
than SLM. Moreover it performs better than SLM so far as
PAPR reduction is concerned. However PTS requires more 4.6 Bandwidth Expansion
bits for sending the side information than SLM [26]. The bandwidth of any communication system is a valuable
and scarce resource. Those PAPR reduction techniques should
3.5 Tone Reservation and Tone Injection be preferred which required less bandwidth.
Both these schemes reserve a subset of subcarriers (also called
tones) for the generation of a PAPR reducing signal [14], [15].
These reserved tones are not meant for data representation.
The common block diagram of TR and TI is shown in figure
2.
In TR, the goal is to determine signal c which must be added
to the original signal x to decrease PAPR.
In TI, the constellation size is extended in such a manner that
every signal point in the original constellation can be Fig. 2. Block diagram of TR/TI
43
International Journal of Computer Applications (0975 – 8887)
Volume 96– No.22, June 2014
A good PAPR reduction scheme should offer minimum of [8] T. Jiang, Y. Yang, and Y. Song, “Exponential
PAPR for the smallest possible level of BER. In Table 2, companding transform for PAPR reduction in OFDM
some PAPR reduction techniques are compared. systems,” IEEE Trans. Broadcasting, vol. 51, no. 2, pp.
244–248, Jun. 2005.
Table 2. Performance Comparison of some PAPR Reduction [9] J. Hou, J. H. Ge and J. Li, “Trapezoidal companding
Techniques scheme for peak to average power ratio reduction of
OFDM systems,” Electron. Letters, vol. 45, no. 25, pp.
1349-1351, Dec. 2006.
Power Compl- Extra band- BER
rise exity width degradation [10] S. S. Jeng, and J. M. Chen, “Efficient PAPR reduction in
required OFDM systems based on a companding technique with
Clipping No Low No Yes trapezium distribution,” IEEE Trans. Broadcasting, vol.
Coding No Low Yes No 57, no. 2, pp. 291–298, June 2011.
Compand- No Low No No [11] A. E. Jones, T. A. Wilkinson, and S. K. Barton, “Block
ing Coding Scheme for Reduction of Peak to Mean Envelope
PTS/SLM No High Yes No Power Ratio of Multicarrier Transmission Scheme,”
TR/TI Yes High Yes No Elect. Lett., vol. 30, no. 22, pp. 2098–2099, Dec. 1994.
[12] A. Alavi, C. Tellambura, and I. Fair, “PAPR reduction of
OFDM signals using partial transmit sequence: An
5. CONCLUSION optimal approach using sphere decoding,” IEEE Trans.
OFDM has been proved to be a suitable technique for wireless Communications Letters, vol. 9, no. 11, pp. 982–984,
communication due to its spectral efficiency and resistance to Nov. 2005.
channel impairments. However it suffers from high PAPR.
Despite two decades of continuous and intensive research, this [13] S. H. Wang and C. L. Pir, “A low complexity PAPR
problem stands in the way of successful implementation of reduction scheme for SFBC MIMO-OFDM systems,”
practical OFDM based communication systems. Many PAPR IEEE Signal Processing Letters, vol. 16, no. 11, pp. 941–
reduction techniques have been reported and proposed so far. 944, Nov. 2009.
In this paper, these techniques are described and compared. [14] D. W. Lim, S. J. Heo and J. S. No, “An overview of
Most of these techniques can achieve low PAPR at the cost of PAPR reduction schemes for OFDM signals,” Journal of
bandwidth efficiency, more complexity and poor error Communications and Networks, vol. 11, no. 3, pp. 229–
performance. Hence no single technique can be considered as 239, June 2009.
best technique. PAPR reduction technique should be
appropriately selected as per the system requirements after [15] D.-W. Lim, H.-S. Noh, J.-S. No, and D.-J. Shin, “Near
making optimum tradeoff between different requirements. In optimal PRT set selection algorithm for tone reservation
addition compatibility with existing standards should also be in OFDM system,” IEEE Trans Broadcast., vol. 54, no.
taken into account. 3, pp. 454–460, Sept. 2008.
44
International Journal of Computer Applications (0975 – 8887)
Volume 96– No.22, June 2014
[23] T. Jiang and G. X. Zhu, “OFDM peak-to-average power Information,” IEEE Communications Letters, vol. 5, no.
ratio reduction by complement block coding scheme and 6, pp. 239–241, Jun. 2001.
its modified version,” in The 60th IEEE Vehicular
Technology Conference 2004 Fall, Los Angeles, USA, [26] S. Khademi, and A. J. Veen, “Constant modulus
pp. 448–451, Sept. 2004. algorithm for PAPR reduction in MIMO OFDM/A,”
IEEE Signal Processing Letters, vol. 20, no. 5, pp. 531-
[24] H. Lee and S. W. Golomb, “A new construction of 64- 534, May 2013.
QAM Golay complementary sequences,” IEEE Trans.
Inf. Theory, vol. 52, no. 4, pp. 1663–1670, Apr. 2006. [27] S. H. Muller and J. B. Huber, “A comparison of peak
power reduction schemes for OFDM,” in Proceeding of
[25] H. Breiling, S. H. Muller-Weinfurtner, and J. B. Huber, IEEE Global Telecommunications Conference, Nov.
“SLM Peak-Power Reduction without Explicit Side 1997, vol. 1, pp. 1–5.
IJCATM : www.ijcaonline.org 45