 Open Access
 Total Downloads : 1813
 Authors : Kokil Dhiman, Tarun Gulati
 Paper ID : IJERTV1IS6106
 Volume & Issue : Volume 01, Issue 06 (August 2012)
 Published (First Online): 30082012
 ISSN (Online) : 22780181
 Publisher Name : IJERT
 License: This work is licensed under a Creative Commons Attribution 4.0 International License
PAPR Reduction in OFDM Systems
Kokil Dhiman1 and Tarun Gulati2
1 Assistant Professor, Dev Bhomi Engineering College, Himachal Pradesh, INDIA
2 Assistant Professor, Maharishi Markandeshwar University, Mullana (Ambala), INDIA
Email:
Abstract: Orthogonal Frequency Division Multiplexing (OFDM) is a special form of multicarrier modulation which is particularly suited for transmission over a dispersive channel. It offers several advantages, because of which it is widely used in recent years. However, it has few limitations with the key one is peaktoaveragepower ratio (PAPR), which determines the systems power efficiency. High PAPR is one of the major drawbacks in OFDM systems, which causes a significant level of signal distortions, when the modulated signals are amplified through high power amplifiers (HPAs). In particularly, high PAPR increases the complexity of Analogue to Digital (A/D) and Digital to Analogue (D/A) converters and also reduces the efficiency of RF High Power Amplifier (HPA). Therefore, PAPR reduction is very essential. For this purpose, a ZadoffChu matrix Transform (ZCT) based precoding and postcoding techniques are quantitatively evaluated in this paper in terms of its parameters q and r. From the analysis, it is inferred that an optimum selection of q and r can lead to a significant level of reduction in PAPR.
Keywords: ZadoffChu matrix Transform; OFDM conventional; WHTOFDM; HPA; Peak to Average Power Ratio and BER.

INTRODUCTION
OFDM is a key multicarrier modulation technique [1], which provides high spectral efficiency, low implementation complexity [2], less vulnerability to echoes and nonlinear distortion [3]. Due to these advantages of the OFDM system, it is vastly used in various communication systems. However, it has a practical limitation of high peaktoaverage power ratio [4]. A large PAPR increases the complexity of the analog todigital and digitaltoanalog converter and reduces the efficiency of the radio frequency (RF) power amplifier [5]. There are a number of techniques dealing with the problem of PAPR. The key technologies are: constellation shaping, phase optimization, nonlinear companding transforms [6], tone reservation [7] and tone injection (TI), clipping and filtering [8], partial transmit sequence [9] and precoding based techniques. However, these techniques do PAPR reduction at the expense of increase in transmit signal power, bit error rate [7], data rate loss and computational complexity. Therefore, two novel PAPR reduction techniques, namely, ZadoffChu matrix Transform (ZCT) precoding based PAPR reduction technique and ZCT postcoding based PAPR reduction technique for OFDM systems were proposed in this paper. In the proposed schemes, the reshaping of the ZCT is carried out one way for precoding and another way for postcoding. For precoding, the reshaping of the ZCT matrix is row wise and is applied before the IFFT; however, for postcoding, the reshaping of the ZCT matrix is done column wise and it is applied after the IFFT.

SYSTEM MODEL
A. OFDM System Model
To design the proposed ZCT precoding and postcoding system for reducing PAPR, consider the block diagram of an OFDM system shown in Figure 1. In Figure 1, an OFDM signal consists of N subcarriers that are modulated by N complex symbols selected from a particular QAM constellation.
Figure 1: Block diagram of general OFDM system.
In Figure 1, the baseband modulated symbols are passed through serial to parallel converter which generates complex vector of size N. This complex vector of size N can be expressed as
X = [XO, X1, X2, X3 . XN1] (1)
X is then passed through the IFFT block of size N Ã— N IFFT matrix. The resulted complex baseband OFDM signal with N subcarriers can then be written as
to an electromagnetic signal of constant amplitude, whereby cyclically shifted versions of the sequence imposed on a signal result in zero crosscorrelation with

N k=O k
xn = 1 IN1 dX e
j2rrkn N
n = 0, 1, . . . , N 1.
(2)
one another at the receiver. The sequence exhibits the useful property that cyclically shifted versions of itself are
After paralleltoserial conversion, a cyclic prefix (CP)
with a length of Ng samples is appended before the IFFT output to form the timedomain OFDM symbol, s = [s0, . . , sN+Ng1]. The useful part of OFDM symbol does not include the Ng prefix samples and has duration of Tu seconds. The samples (s) are then amplified, with the amplifier characteristics is given by function F and the
orthogonal to one another, provided, that is, that each cyclic shift, when viewed within the time domain of the signal, is greater than the combined propagation delay and multipath delayspread of that signal between the transmitter and receiver. This is expected to reduce PAPR by significant level. Mathematically, ZadoffChu sequences can be defined for a sequence of length L as:
output of amplifier produces a set of samples denoted by y.
At the receiver front end, the received signal is applied to a
z(k) = e
j2rrr k2
L 2 +qk for L even
(9)
matched filter and then sampled at a rate Ts = Tu/N. After dropping the CP samples (Ng), the received sequence z, assuming an additive white Gaussian noise (AWGN) channel, can be expressed as
z = F(Wd) + (3)
Where, the noise vector consists of N independent and normally distributed complex random variables with zero mean. Subsequently, the sequence z is fed to the fast Fourier transform (FFT), which produces the frequency domain sequence r as
ej2rrr(k(k+1)+qk) for L odd
L
2
Where, k = 0, 1, 2 L1, q is any integer and r is any
integer relatively prime to L. From the above equation, q and r comes in the numerator of the exponential power fraction. This means, that their variation can lead to a significant change in the output. This is the key focus of the research reported in this paper. This ZCT sequence can be applied in two forms: precoding and postcoding, as discussed below.
r =WH z (4)
Where, kth element of r is given by
C. ZCT PRECODING BASED OFDM SYSTEM
The block diagram of ZCT precoding based OFDM
1 N1
j2rrkn
(5)
system is shown in Figure 2. In the ZCT precoding based
rk = N In=O zne N k=0, 1, 2,, N1
Finally, the estimated symbols vector d can be obtained from r. It is to be noted that the demodulation is performed based on the assumption of perfect symbol timing, carrier frequency, and phase synchronization. One of the main disadvantages of OFDM systems is the high PAPR of transmitted signal due to the combination of N modulated subcarriers. The PAPR for a continuous time signal x(t) is defined as:
2
OFDM system, the baseband modulated data is passed
through S/P convertor which generates a complex vector of size N that can be written as X = [X0, X1, X2 XN1]T. ZCT precoding is then applied to this complexvector which transforms this complex vector into new vector of length N that can be written as Y=[Y0, Y1, Y2YN1]T; where, R is a ZCT based rowwise precoding matrix of size L=N*N. By reordering,
k = mN + l (10)
PAPR =
max{lx(t)l }
E{lx(t)l2}
0 t Tu
(6)
And the matrix R with row wise reshaping can be written
The PAPR for discrete time signals can be estimated by as oversampling the vector d by a factor L and computing NL
point IFFT. The PAPR in this case is defined as:
R =
rOO
rO(N1)
(11)
PAPR =
max{lx(n)l2}
E{lx(n)l2}
n = 0,1, . . , NL – 1 (7)
r(N1)O r(N1)(N1)
Where, R is a N*N, ZCT complex orthogonal matrix with
The denominator in above equation represents the average power per OFDM symbol at the amplifier input, which is denoted as Pin. PAPR, in quantitative terms, is usually expressed in terms of Complementary Cumulative Distribution function (CCDF) for an OFDM signal, and is
length L2 =N*N. By letting, q=1 and r=1, the ZCT for even L can be written as rk = exp[(j*pi*k2) / L2] . Accordingly, precoding X gives rise to Y as follows:
Y =RX (12)
Or,
mathematically given by
P(PAPR > PAPRO) = 1 – (1 – ePAPRO )N (8)
N1
Ym = L rm,lXl m = 0,1,2, . . . , N – 1
l=O
(13)
Where, PAPR0 is the clipping level. This equation can also
be read as the probability that the PAPR of a symbol block exceeds some clip level PAPR0.
Where, rm,l means the mth row and lth column of precoder matrix. Therefore, the complex baseband OFDM signal with N subcarriers with ZCT precoding is given by
N1
1
j2nmn
L1
jnl2
j2nml
(14)
B. ZADOFFCHU SEQUENCES
In general, a ZadoffChu sequence is a complexvalued sequence which, when applied to radio signals, gives rise
xn
= L e(

N m=O
N ) ejnm2 L Y . e L2 e L
l
l=O
The above equation can be expressed as xn the IFFT of constellation data Xl premultiplied with ZCT matrix. The PAPR of this signal is same as before.
Where, C is a ZCT column wise postcoding matrix of size L2 =N*N. For q = 1 and r =1, the complex baseband ZCT postcoding based OFDM signal with N subcarriers can be written as:
wm = e
jrrm2
L I [e
2 L1
l=O
jnl2
. xl]. e
j2rrml
L
(18)
Figure 2: Block diagram of an OFDM system with ZCT precoding approach.

ZCT POSTCODING BASED OFDM SYSTEM
The block diagram of ZCT postcoding based OFDM system is shown in Figure 3. In the ZCT postcoding based OFDM system, the baseband modulated data is passed through S/P convertor which generates a complex vector of
size N that can be written as X = [X , X , X X ]T. Then,
m = 0,1, . . , N – 1
Where, wm is the IFFT of constellation data Xl premultiplied with quadratic phase and IFFT postcoded, and then alternated with Â±1. The PAPR of this ZCT postcoding based OFDM signal is same as before.
Figure 3: Block diagram of an OFDM system with ZCT postcoding approach.
0 1 2 N1
IFFT is performed to this complex vector which transforms this complex vector into new vector of length N that can be written as y = IFFT{X} = [y0, y1, y2 yN1]T. Similarly, re ordering as in precoding to have:
k = m + lN (15)
And, ZCT matrix C with column wise reshaping is written as
cOO cO(N1)


IMPLEMENTATION
To do the PAPR analysis and to evaluate performance of ZCT precoding and postcoding based OFDM systems, the following key steps are undertaken.

To do the PAPR analysis of ZCT precoding and postcoding based OFDM, firstly binary data is
C =
c(N1)O c(N1)(N1)
(16)
generated randomly.

The generated binary data is then converted into
In other words, the N2 point long ZadoffChu sequence fills the precoding matrix columnwise. The matrix C is N*N, ZCT complex orthogonal matrix with length L2
=N*N. By letting, q=1 and r=1, the ZCT for even L can be written as ck=exp[j*pi*k2/L2] . Accordingly, postcoding y gives rise to w as follows:
w =C.y (17)
symbols and is then modulated by MQAM (where
M=4, 16, 64, 256). The ZCT matrix is calculated for the length of the data.

The serial data is then converted into parallel data and IFFT is performed. The ZCT matrix (Transform) is applied before and after the IFFT, respectively for precoding and postcoding. The PAPR is then calculated and compared.

The parallel signal is then converted to serial data and is passed through a multipath channel with AWGN noise added.

The received signal is again converted from serial to parallel data. ZCT inverse matrix is applied before and after the FFT. The final demodulated signal is parallel data, which is then converted into serial data. The serial data is then demodulated to retrieve back the signal.


RESULT & DISCUSSION
To evaluate the performance of ZCT precoding and postcoding based OFDM systems, extensive simulations have been performed in Matlab. The simulations are performed for different values of q and r. To show the PAPR analysis of ZCT precoding and postcoding based OFDM systems, data is generated randomly and then modulated by MQAM (where M = 4, 16, 64, 256). The order of FFT and IFFT is taken as 64. The results obtained are shown in Figures 4 to 11, for various combinations of q and r; whereas, q= 1 & 7 and r =2, 13, 31, 53, 73, 101. The key conclusions drawn from the results are discussed below:

From Figures 4 to 11, it can be interpreted that the results obtained for ZCT precoding and postcoding are different. The PAPR is more reduced in case of ZCT precoding for the same value of q and r.

From Figures 4 to 11, for ZCT precoding, the PAPR is more reduced when value of r is increased from 2 to 101, for any value of q. ZadoffChu sequences is directly proportional to the exponential raised to power r and product of rk2 for even L. Therefore, the increase
in value of r will yield greater output and so the PAPR will increase as observed from Figures 4 to 11.

As the constellation size increases, i.e., from M = 4, 16, 64, 256; there is significant reduction in PAPR for the same values of q and r. This is clearly observed in Figures 4 and 11.

It can also be seen that ZCT postcoding has quick and sharp roll off than ZCT precoding. Also, there is less irregularity in ZCT postcoding compared to ZCT precoding.

It can also be seen that ZCT precoding curves overlap for small time and some of these are intersecting each other.
From the above discussion, the value of q and r significantly changes the PAPR. Also, for same values of q and r, the variation in constellation order also significantly changes the PAPR. It is also found that ZCT precoding is more effective that ZCT postcoding. Therefore, the selection of value of q r, M and ZCT precoding/postcoding depends upon the level of PAPR reduction required. Here in this paper, it is preferred to choose a small value of q and large value of r for ZCT post coding and both small values of q and r for ZCT pre coding. It is preferred to choose ZCT post coding because of its sharp rolloff; however, at the expense of less reduction in PAPR or requirement of large values of r if results comparable to ZCT precoding needs to be achieved.
CASE I: Constellation size = 4
Figure 4: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 4 and with ZCT precoding & postcoding. The values are: q=1 and r=2, 13, 31, 53, 73, 101.
Figure 5: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 4 and with ZCT precoding & postcoding. The values are: q=7 and r=2, 13, 31, 53, 73, 101.
CASE II: Constellation size = 16
Figure 6: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 16 and with ZCT precoding & postcoding. The values are: q=1 and r=2, 13, 31, 53, 73, 101.
Figure 7: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 16 and with ZCT precoding & postcoding. The values are: q=7 and r=2, 13, 31, 53, 73, 101.
CASE III: Constellation size = 64
Figure 8: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 64 and with ZCT precoding & postcoding. The values are: q=1 and r=2, 13, 31, 53, 73, 101.
Figure 9: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 64 and with ZCT precoding & postcoding. The values are: q=7 and r=2, 13, 31, 53, 73, 101.
CASE IV: Constellation size =256
Figure 10: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 256 and with ZCT precoding & postcoding. The values are: q=1 and r=2, 13, 31, 53, 73, 101.
Figure 11: CCDF plots of PAPR of OFDM with QAM modulation for constellation size 256 and with ZCT precoding & postcoding. The values are: q=7 and r=2, 13, 31, 53, 73, 101.


CONCLUSION
This paper presented the results obtained on applying a ZCT precoding and postcoding for the purpose to reduce PAPR in OFDM systems with MQAM modulation, where as M=4, 16, 64, 256. It is observed from the graphs that the response of both ZCT pre coding and postcoding is different for r>1 and q>1. The PAPR significantly varies with the values of r and q. For any particular value of r and q, the PAPR reduction is more for ZCT precoding than for ZCT precoding. From the results, it is preferred to choose a small value of q and large value of r for ZCT post coding and both small values of q and r for ZCT precoding. It is preferred ZCT post coding because of its sharp rolloff; however, at the expense of less reduction in PAPR or requirement of large values of r if results comparable to ZCT precoding needs to be achieved.

REFERENCES

E. A.Dalakta, A. A.Dweik & C. Tsimenidis (2012). Efficient BER Reduction Technique for Nonlinear OFDM Transmission Using Distortion Prediction. IEEE Transactions On Vehicular Technology. Vol. 61, NO. 5, pp. 23036.

H.B. Jeon, J. S. and D. J. Shin (2011). A LowComplexity SLM Scheme Using Additive Mapping Sequences for PAPR Reduction of OFDM Signals. IEEE Transactions On Broadcasting. Vol. 57, NO. 4, pp. 86675.

Ghassemi and T A. Gulliver (2007). SLMBASED RADIX FFT FOR PAPR REDUCTION IN OFDM SYSTEMS. IEEE International Conference on Signal Processing and Communications (ICSPC 2007). pp 103134.

C. P. Li, S. H. Wang and C. L. Wang (2010). Novel Low Complexity SLM Schemes for PAPR Reduction in OFDM Systems. IEEE Transactions On Signal Processing. Vol. 58, NO. 5, pp 291621

I. Baig and V. Jeoti (2010). PAPR Reduction in OFDM Systems: ZadoffChu Matrix Transform Based Pre/PostCoding Techniques. Second International Conference on Computational Intelligence, Communication Systems and Networks.

H. Liang, Y. R. Chen, Y. F. Huang, C. H. Cheng (2009). A Modified Genetic Algorithm PTS Technique for PAPR Reduction in OFDM Systems. Proceedings of the 15th AsiaPacific Conference on Communications (APCC 2009). pp 182185.

H. Li, T. Jiang and Y. Zhou (2011). An Improved Tone Reservation Scheme With Fast Convergence for PAPR Reduction in OFDM Systems. IEEE Transactions On Broadcasting. Vol. 57, No. 4, pp. 90206.

L. Wang and C. Tellambura (2005). A Simplified Clipping and Filtering Technique for PAR Reduction in OFDM Systems. Signal Processing Letters, IEEE . Vol.12, No.6, pp. 453456.

P. Varahram and B. M. Ali (2011). Partial Transmit Sequence Scheme with New Phase Sequence or PAPR Reduction in OFDM Systems. IEEE Transactions on Consumer Electronics. Vol. 57, No. 2, pp. 36671.