Professional Documents
Culture Documents
Journal of Optical Communications New Hybrid PAPR Reduction Techniques For OFDM Based Visible Light Communication Systems PDF
Journal of Optical Communications New Hybrid PAPR Reduction Techniques For OFDM Based Visible Light Communication Systems PDF
net/publication/333194205
A New Hybrid PAPR Reduction Technique for OFDM based Visible Light
Communication Systems
CITATIONS READS
4 120
4 authors, including:
M.I. Dessouky
Minoufiya University
295 PUBLICATIONS 1,927 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Emad Hassan on 14 September 2017.
of OFDM signals and the limited dynamic range of LEDs after the IFFT stage. The obtained results show that the
devices. High PAPR deteriorates the VLC systems perfor- DCT precoding scheme can further improve the BER and
mance in terms of BER, so PAPR reduction is very impor- PAPR performances of OFDM-based VLC systems.
tant in OFDM-based VLC systems [21, 22]. There are many The rest of this paper is organized as follows: Section 2
techniques were proposed in literature to reduce the introduces the OFDM-based VLC system model using
PAPR and to improve the power efficiency of RF commu- Hadamard transform and clipping. Section 3 presents the
nication systems [23–25]. However, since VLC differs from Hadamard transform combined with clipping technique
RF communication in many aspects such as system con- with DC bias. Section 4 proposes the OFDM-based VLC
straint, baseband signal format and nonlinearity charac- system model using DCT precoding and clipping. In
teristic of front-end devices, it is not obvious how the Section 5, we present the Monte Carlo simulations and
PAPR is related to the illumination-to-communication results and discussion. The conclusions are given in
conversion efficiency. Section 6.
In this paper, we propose a hybrid technique to
reduce the high PAPR of OFDM-based VLC systems called
Hadamard Transform Combined with Clipping (HTCC) 2 System model
technique. The effect of using HTCC technique on BER
performance of VLC systems is also presented in this The system model of OFDM-based VLC system treated in
paper. High signal clipping ratio will introduce some this paper is shown in Figure 1. The original data are
deleterious side effects, such as the degrading of BER firstly processed by serial-to-parallel (S/P) and mapped
and the growth of in-band radiation and out-of-band onto M-level QAM constellation before being transformed
radiation. The authors in [26, 27] demonstrated that the by the Hadamard block which reduces the correlation of
Hadamard transform reduces the PAPR without improv- the input data sequences. Hermitian symmetry is
ing the system BER. Therefore, combining the clipping imposed with the IFFT to obtain real value signal. When
techniques with Hadamard transform as in the proposed applying IFFT, we do oversampling to make the distribu-
technique will reduce the PAPR without improving the tion of the amplitude of the discrete IFFT output signals
BER performance. We also evaluate the optimum clipping close to that of continuous signals, and after the parallel-
threshold at which we can reduce the PAPR and simulta- to-serial (P/S) operation, the cyclic prefix (CP) is added to
neously, improve the BER of VLC systems. eliminate the inter-symbol interference. The OFDM signal
In this paper we also study the effect of using DCT enters the clipping block to reduce the PAPR. The OFDM
precoding matrix to transform the transmitted data before time domain signal must be both real and positive, so DC
being processed by IFFT stage. Then, the clipping is biased is introduced. Intensity modulation (IM) is
applied to further reduce the PAPR of the OFDM signal employed at the transmitter. The forward signal y(t)
Clipping
DC bias Channel
h(t)
AWGN
w(t)
drives the LED which in turn converts the magnitude of ^ M after N-point IFFT may be expressed
The signal X
the input electric signals y(t) into optical intensity. At the as:
receiver side, direct detection (DD) is employed. The PIN
photo detector (PD) converts the incoming optical power ^ ^ M = IFFTðHM XM Þ
xM = IFFT X (5)
into the amplitude of an electrical signal. CP is removed,
Inverse Hadamard transform is performed after the FFT
and the M-QAM demodulation is performed. Finally, data
signals are successfully received after the P/S conversion. 3.2 The proposed clipping method
The received OFDM signal propagated through the wire-
less optical channel with impulse response h(t) is given by: The simple and effective technique to reduce PAPR is
clipping algorithm. The clipping algorithm is that the
yr ðt Þ = RxðtÞ#hðtÞ + wðtÞ (1)
signal is limited to the set threshold when the peak of
R is the photodetector responsivity, # denotes the circu- the input signal exceeds a certain threshold; otherwise,
lar convolution and w(t) is the additive white Gaussian the signal passes through directly. Clipping operation can
noise (AWGN) which added in the electrical domain. This be represented as:
is the model commonly used in wireless infrared commu- 8
nication systems where the main impairment is due to <− C ð^x < − CÞ
yC = ^ x ð−C≤^ x ≤ CÞ (6)
high level ambient infrared radiation. The ambient sig- :
C ð^
x > CÞ
nals are mainly at DC and can be filtered out, however
they cause shot noise in the detector, which is accurately where ^xis the IFFT output signal which is real value, C is
modeled as AWGN. In this paper we are considering the the clipping threshold and yC denotes the output signal
performance in a flat channel therefore we assume that after clipping block. In the clipping algorithm, the clip-
R = hðtÞ = 1. ping ratio (CR) is a very important parameter, which is
used to determine the clipping threshold C above. It is
defined as
3 The proposed hybrid Hadamard C
CR = 20log pffiffiffi dB (7)
combined with clipping
P
technique where P is the average power of an OFDM symbol. Since
about a half of the clipped samples are Negative real and
3.1 Hadamard transform the minimum value is –C at least, we add bias of C to
each OFDM sample to make the transmit signal be posi-
Hadamard transform is used to reduce the autocorrelation tive. With the above operations, the dynamic range of the
between the input data sequence and the PAPR of the OFDM amplitude can be suppressed to less than 2C. Moreover,
signal [15]. Let HM be the Hadamard transform matrix of M thanking to clipping without filtering the bias can be a
orders, the elements of Hadamard matrix is 1 or –1, the constant for every OFDM symbol. Therefore, the receiver
second order of Hadamard matrix can be written as does not need to estimate the bias value in every OFDM
symbol [27].
1 1 The steps of the proposed hybrid technique as shown
H2 = (2)
1 −1 in Figure 1 can be summarized as follows:
The Hadamard matrix of M order can be given as: Step 1: The sequence X from the QAM mapper is trans-
formed by the Hadamard matrix as in eq. (4).
HM=2 HM=2 Step 2: The transformed sequence is converted to time
HM = (3)
HM=2 − HM=2 domain using IFFT as in eq. (5).
where the matrix complementary of HM=2 is − HM=2 . Step 3: Set the CR and the OFDM symbol x ^n enters the
The data sequence after M-QAM is given by clipping block.
^ M after the
X = ½X1 , X2 , X3 , . . . , XM T , the new sequences X Step 4: Calculate the clipping level, C, using eq. (7) and
Hadamard transform can be represented as: generate the sequence yC by clipping x ^n using eq. (6).
Step 5: Finally add DC bias of C to each sample of OFDM
^ M = HM XM
X (4) symbols to make the transmitted signal positive.
3.3 PAPR of OFDM-based VLC As shown in Figure 2, the main idea of this new
technique is that the transmitted data is first transformed
The PAPR of the continuous time OFDM signal is defined by the DCT matrix. Then, the transformed data are pro-
as the ratio of the maximum instantaneous power to the cessed by the IFFT unit. In order to produce a real output
average power. The PAPR of discrete time VLC-OFDM from IFFT stage, the input of the IFFT must be a
signal^xn is defined as: Hermitian symmetric structure.
DCT is a real transform which converts the modu-
max
j^x j2 lated data,XM into the new transform-domain signal
Δ 0≤n≤N n
PAPRðdBÞ = 10log10 (8) using K × K DCT matrix F which is given as [30]
E½j^
xn j2
(
p1ffiffiffi n = 0, 0 ≤ m ≤ K − 1
where E [⋅] denotes statistical expectation. Fn, m = pffiffiffiffiffiffiffiffi h K i
The Complementary Cumulative Distribution Function 2=K cos ð2m 2K + 1Þnπ
1 ≤ n ≤ K − 1, 0 ≤ m ≤ K − 1
(CCDF) of the PAPR is one of the most frequently used (10)
performance measures for PAPR reduction techniques, it is
where n and m represent the row and column entries,
defined as [28]:
respectively. The new sequences X^ M − DCT after the DCT
CCDF = ProbðPAPR > PAPR0 Þ (9) precoding matrix can be represented as:
where PAPR0 is the given threshold value. CCDF denotes ^M − DCT = FXM
X (11)
the probability that the PAPR of a data block exceeds the
threshold value. Following the same steps as in Section 3 and applying
DCT precoding before IFFT stage, we can get a good
PAPR reduction and good BER performance as shown in
4 The proposed DCT precoding next section.
DC bias Channel
h(t)
AWGN
W(t)
Table 1: Simulation parameters. technique, the PAPR of OFDM-based VLC system can be
improved by 6.5 dB, 7.5 dB, 8.2 dB and 9.2 dB with
Modulation type QAM CR = 1.8, 1.5, 1.3 and 1, respectively at CCDF = 10−3.
OFDM symbols , The BER performance of the OFDM-based VLC sys-
No. of data subcarriers tem with Hadamard transform only and with proposed
FFT size technique using different clipping ratio is shown in
Guard interval samples times Figure 4. The original OFDM-based VLC system is
Total subcarriers
included for comparison. At BER = 10–3, the system
with Hadamard transform can achieve about 0.25 dB
improvement. We also note that, the BER is improved
by about 1 dB at a clipping ratio of 1.3 when comparing
system, we add bias values equal to the absolute value of to the original system to satisfy BER of 10−3. This
the minimum value. improvement becomes 2 dB when the clipping ratio
changed to 1.5 and 1.8.
Figure 3 shows the CCDF performance comparison Figure 5 shows the CCDF performance comparison
between the original OFDM-based VLC system, the between the original and the proposed OFDM-based
OFDM-based VLC system with Hadamard transform only VLC system using DCT precoding and different clipping
and with proposed technique using different clipping ratio. We can see that at CCDF = 10–3, compared with
ratio. We can see that at CCDF = 10–3, compared with conventional system, about 2.8 dB PAPR reduction may
original system, about 0.5 dB PAPR reduction may be be obtained when applying DCT precoding without clip-
obtained using Hadamard transform. It is also clear ping. It is also clear that, as the CR value is smaller, the
that, as the CR value is smaller, the reduction effect of reduction effect of PAPR is better. For example, using the
PAPR is better. For example, using the proposed proposed technique, the PAPR of OFDM-based VLC
100
10–2
10–3
10–4
2 4 6 8 10 12 14
PAPR0 dB
Figure 3: The CCDF performance comparison between the original and the proposed OFDM-based VLC system using different clipping ratio.
100
10−1
10−2
Bit Error Rate
10−3
10−5
0 5 10 15
Eb/No, dB
Figure 4: BER performance of OFDM-based VLC system with and without proposed technique.
100
10−2
10−3
10−4
0 2 4 6 8 10 12 14
PAPR0 dB
Figure 5: The CCDF performance comparison between the original and the proposed OFDM-based VLC system using DCT and different
clipping ratio.
system can be improved by 7 dB, 8 dB, 8.7 dB and 9.7 dB DCT precoding using clipping ratio of 1.8. It is clear that,
with CR = 1.8, 1.5, 1.3 and 1, respectively at CCDF = 10−3 by replacing Hadamard transform block by DCT precoding
Figure 6 presents the CCDF performance comparison block we can achieve further improvement in terms of BER.
between the conventional, the OFDM-based VLC system
with Hadamard transform and OFDM-based VLC system
with DCT precoding using clipping ratio of 1.8. We can 6 Conclusion
see that at CCDF = 10–3, compared with conventional sys-
tem, about 0.75 and 2.8 dB PAPR reduction may be This paper proposed new hybrid techniques to reduce the
obtained when applying Hadamard transform and DCT PAPR and improve the BER performance of OFDM-based
matrix transform, respectively without clipping. It is also VLC systems. The proposed techniques combined the
clear that, the reduction effect of PAPR become better Hadamard transform or DCT matrix transform with clip-
when applying clipping especially with smaller CR value. ping method. The optimum clipping threshold at which
The BER performance of DCT-OFDM-based VLC sys- we can reduce the PAPR and simultaneously improve the
tem with DCT precoding and different clipping ratio is BER of VLC systems was also evaluated. The performance
shown in Figure 7. The conventional OFDM-based VLC of conventional OFDM-based VLC system, the OFDM-
system is included for comparison. At BER = 10–3, the based VLC system with Hadamard transform only and
system with DCT precoding can achieve about 1.25 dB with proposed techniques using different clipping ratio
improvement in SNR compared with conventional sys- was evaluated and compared in terms of CCDF and BER.
tem. We also note that, the BER is improved by about It was demonstrated that, the OFDM-based VLC system
2 dB at a clipping ratio of 1.3 when comparing to the with Hadamard transform combined with clipping tech-
conventional system. This improvement becomes 2.5 dB nique has the lowest PAPR and good BER performance
when the clipping ratio changed to 1.5 and 1.8. compared to conventional system. The CCDF and BER are
Figure 8 shows the performance comparison between further improved when replacing Hadamard transform
the conventional, the OFDM-based VLC system with block by DCT precoding block due to the excellent energy
Hadamard transform and OFDM-based VLC system with compaction property of the DCT transform.
100
Conventional OFDM based VLC
OFDM based VLC with Hadamard
DCT precoded OFDM based VLC
OFDM based VLC with Hadamard and CR = 1.8
10−1 DCT precoded OFDM based VLC with CR = 1.8
CCDF
10−2
10−3
10−4
5 6 7 8 9 10 11 12 13 14
PAPR0 dB
Figure 6: The CCDF performance comparison between the conventional and the proposed OFDM-based VLC system and DCT VLC OFDM using
clipping ratio = 1.8.
100
10−1
10−2
Bit Error Rate
10−3
Figure 7: BER performance of DCT OFDM-based VLC system with and without proposed technique.
100
10−1
10−2
Bit Error Rate
10−3
10−5
0 5 10 15
Eb/No, dB
Figure 8: BER performance comparison of OFDM-based VLC system with FFT and DCT using proposed technique.