INTRODUCTION
The emerging of 4G has led to the necessity of looking for an attractive
technique for wireless communication system that can promise high data
rate and high spectral efficiency. The 4G systems are expected to provide
bandwidth higher than 20 Mbps and to accommodate a significantly increased
amount of traffic, so sufficient frequency resources will be required
(Mishra, 2007). Orthogonal Frequency Division Multiplexing (OFDM) is proving
to be a possible multiple access technology to be used in 4G. It is a
very attractive technique for the transmission of high bit rate data in
a radio environment. But OFDM comes with its own challenges like high
Peak to Average Power Ratio (PAPR), linearity concerns and phase noise.
OFDM is becoming the chosen multiplexing technique for wireless communications.
OFDM is actually a combination of both modulation and multiplexing (Mishra,
2007). A particularly, attractive feature of OFDM systems is that they
are capable of operating without a classic channel equalizer, when communicating
over dispersive transmission media, such as wireless channels, while conveniently
accommodating the time and frequency domain channel quality fluctuations
of the wireless channel (Hanzo and Keller, 2006). A major disadvantage
of an OFDM system is the large dynamic range signal (also referred as
PAPR) due to the summation of a large number of subcarriers coherently.
As the number of subcarriers N increases, the maximum possible peak power
becomes N times the average power (Li and Cimini, 1998). To reduce the
PAPR, several techniques have been proposed. Examples of distortion techniques
are clipping, peak windowing and peak cancellation (Van Nee and Prasad,
2000). There is a trade off between power control and modulation adaptation.
If the value of SER is small, so the transmitted power can be maintained
and a high order modulation scheme is used, or the power can be increased
and the order of modulation scheme reduced accordingly.
MATERIALS AND METHODS
In Matlab simulation, OFDM symbol would be assumed with subcarriers of
length 128 and a guard interval with duration of 1/4 of the symbol duration.
Four modes of algorithm MPC were tested. Five, four and three modulation
levels were proposed apnd adopted to be used in these modes. The modulation
or mapping schemes that would be used in this simulation were BPSK, 4QAM,
16QAM, 64QAM and 256QAM. Input data could be mapped onto symbols using
anyone of the above mentioned constellations, in this simulation the high
order modulation schemes could be used to map the input data stream at
low SNR values to enhance the data rate and improve the performance of
SER. If the digital OFDM signals were clipped directly, the resulting
clipping noise would all fall inband and could not be reduced by filtering.
To address this aliasing problem, in this simulation, OFDM signal could
be oversampled by a factor of 4. In addition, to reduce the implementation
complexity, the complex valued baseband OFDM signal was modulated up to
a carrier frequency equal to 1/4 of the sampling frequency. Then, the
real valued bandpass samples, x, were clipped at amplitude A as follows:
Table 1: 
Simulation parameters 

Table 1 shows the values of parameters were used in
this simulation model. A simple AWGN channel is used as channel model,
the channel bandwidth provided in this simulation can support two data
rates vary from 125 Mbps (when using BPSK as a mapping scheme), to 1 Gbps
(when using 256QAM as a mapping scheme), which meet the bandwidth requirement
of 4G systems.
The variation of the envelope of a multicarrier signal can be defined
by the Peak to Average Power Ratio (PAPR) which is given by:
The large peaks occur with low probability because the signal amplitude
is approximately Rayleigh distributed when the number of subcarriers is
large. To evaluate the performance, the Complementary Cumulative Distribution
Function (CCDF) of PAPR for OFDM signal x, that is:
is calculated. CCDF can be interpreted as the probability that the PAPR
of an OFDM signal exceeds some clip level PAPR0. The goal of the modulation
adaptation algorithm used in this simulation is to improve the performance
of SER and enhance the data rate through enabling the high order modulation
schemes to be used to map the input data onto symbols at low SNR values.
In this simulation, all modulation levels can be adopted for each symbols
based on the SER value which is specified for each modulation scheme at
any value of SNR values, or based on SNR value which is allocated for
each modulation scheme to maintain the SER value less than 10^{4}.
The power control algorithm used in this simulation is very simple and
used for power adjustment after each symbol transmission.
RESULTS AND DISCUSSION
Four modes of MPC were tested with five, four and three modulation levels.
The simulation results will be presented as the (CCDF) of the PAPR of
the OFDM signals.
Case 1: The modulation adaptation is based on SER: If the value of SER
is small, the high order modulation scheme will be selected, but if there is
increasing in the value of the SER, the decision will be made to select the
low order modulation scheme. The expected result is reduction in PAPR, enhancing
the data rate and all modulation levels can be applied at all SNR values depend
on the value of SER, without degrade the performance of SER. Figure
1 shows the simulation results of all above mentioned modes.
The best reduction in PAPR was noticed with the mode MPC (BPSK, 4QAM,
16QAM, 64QAM and 256 QAM) as shown in Table 2.
It is easy to note that the data rate is enhanced dramatically at low
SNR values as shown in Table 3.
At SNR equals to 8 dB which is the threshold of BPSK, 13% of the transmitted
symbols were mapped with the high order modulation scheme 256QAM, 20%
with 64QAM, 18% with 16QAM and 20% of them with 4QAM. This means 71%
of the transmitted symbols were mapped at 8 dB with modulation schemes
have order higher than the BPSK. Also at 12 dB which is the threshold
of 4QAM, 63% of the transmitted symbols were mapped with modulation schemes
have order higher than 4QAM. At 18 dB, 55% of the transmitted symbols
were mapped with modulation schemes have order higher than 16QAM. So
it was clear how the proposed algorithm can decrease the PAPR and enhance
the data rate at low SNR values without degrading the target SER. As shown
in Fig. 2, the SER performance of the tested MPC modes
with five and four modulation schemes is improved. It is easy to note
this improvement in SER performance especially at low SNR values compared
to SER performance in the normal and clipped OFDM with 256QAM.
As in Table 3 and 4 show how the
proposed algorithm MPC (BPSK+4+16+128+256QAM) mode allows the high order
modulation schemes to be applied to map the input data onto symbols at
low SNR values in order to enhance the data rate without degrade the SER
as shown in Fig. 2 and reduce the PAPR.
In Fig. 3, the reduction in PAPR using the mode MPC
(4+16+64QAM) was 4.9 dB compared to normal OFDM with 64QAM and 3.5 dB
compared to normal OFDM with 4QAM.

Fig. 1: 
CCDF of PAPR for normal OFDM with 256QAM and the
proposed modes: MPC (BPSK+4+16+64+256QAM), MPC (BPSK+4+16+128+256QAM)
and MPC (4+16+64+256QAM) 
Table 2: 
The obtained PAPR reduction 

Table 3: 
The percentage of transmitted symbols of MPC (BPSK+4+16+64+256QAM)
that were mapped using various orders of modulation schemes at different
SNR values 

Table 4: 
The percentage of transmitted symbols of MPC (BPSK+4+16+128+256QAM)
that were mapped using various orders of modulation schemes at different
SNR values 

It was found at SNR equal to 8 dB,
58% of the transmitted symbols were mapped using modulation schemes with
order higher than 4QAM. But at 18 dB, 82% of the transmitted symbols
were mapped using 64QAM.

Fig. 2: 
SER performance of the normal OFDM with 256QAM and the proposed
modes: MPC (BPSK+4+16+64+256QAM), MPC (BPSK+4+16+128+256QAM) and
MPC (4+16+64+256QAM) 

Fig. 3: 
CCDF of PAPR for normal OFDM with 64QAM and the proposed
mode MPC (4+16+64QAM) 
The SER performance of the mode MPC (4+16+64QAM), normal and clipped
OFDM with 64QAM are illustrated in Fig. 4. Although
the OFDM signals are clipped, it is easy to note the improvement in SER
performance especially at low SNR values with increment in the data rate.
Case 2: The modulation adaptation is based on SNR: The same decision
technique used in earlier will be made here, but the modulation adaptation
here is based on SNR.
Figure 5 shows the obtained reduction in PAPR is 2.7
dB when applying the mode MPC with 256QAM and 8.8 dB when the mode MPC
with 4QAM was used.

Fig. 4: 
SER performance of the normal OFDM with 64QAM and the
mode MPC (4+16+64QAM) 
Fig. 5: 
CCDF of PAPR for normal and MPC OFDM system with 64QAM
and 256QAM based on SNR value 
Table 5: 
Comparison between the required transmission time of
the normal OFDM and the proposed algorithm MPC at different SNR values 

It was obvious the bit rate is increasing as the value of SNR is increasing,
but the cost of keeping the symbol error rate under 10^{4}, is
low bit rate at low SNR as shown in Fig. 6.
The symbol error rate as shown in Fig. 7 was kept under
certain value, which is 10^{4} by applying the proposed algorithm
MPC OFDM with (2+4+16+64+256QAM) modulation schemes.

Fig. 6: 
Bit rate per second of MPC OFDM system (2+4+16+64+256QAM) 

Fig. 7: 
SER performance of the mode MPC (4+16+64+256QAM) based
on SNR value 
Table 5 shows the time required to transmit the data
(5 Gbps as an example) using the proposed mode MPC with five modulation
schemes is less than the time required using the normal OFDM at different
SNR values.
By carrying out comparison between the proposed algorithm in case 1 and
case 2, it is easy to note the two proposed algorithms introduce a good
reduction in PAPR, but the advantage of the proposed algorithm in case
1 over what was proposed in case 2 is the possibility of using the high
modulation levels at the low SNR values and this will reflect in enhancing
the bit rate at low SNR values, but the disadvantage is increasing the
SER also at low SNR values. But the proposed algorithm in case 2 can decrease
the SER at low SNR values and keep it under certain threshold, but it
can not promise high data rate at low SNR values.

Fig. 8: 
The power spectral density of the normal OFDM and MPC
OFDM signal with 256QAM 
It was clear how the proposed algorithm MPC in this simulation can attenuate
the outofband noise as shown in Fig. 8.
CONCLUSION
A new algorithm is proposed in this study which is in fact a hybrid
of different techniques. This algorithm employs the modulation adaptation
and power control techniques together with clipping technique which called
MPC. The modulation adaptation in this algorithm was classified into two
cases. The first one was based on SER and the other was based on SNR.
Both of two cases introduced a good PAPR reduction, but the algorithm
used in first case could increase the bit rate at low SNR by using the
high order modulation schemes at the low SNR values. The second case showed
the ability to reduce the SER at low SNR values. In both cases the result
was OFDM system has small PAPR and high data rate with a good SER performance.