New binary particle swarm optimization on dummy sequence insertion method for nonlinear reduction in optical direct detection orthogonal frequency division multiplexing system

R E S E A R C H A R T I C L ENew binary particle swarm optimization on dummy sequence insertion method for nonlinear reduction in optical direct-detection orthogonal frequency division

1 23

Journal of Optics

ISSN 0972-8821

J Opt

DOI 10.1007/s12596-019-00512-6

New binary particle swarm optimization

on dummy sequence insertion method for nonlinear reduction in optical direct-detection orthogonal frequency division multiplexing system

Lap Maivan & Thang Nguyentrong

1 23

use only and shall not be self-archived

in electronic repositories If you wish to

self-archive your article, please use the

accepted manuscript version for posting on your own website You may further deposit the accepted manuscript version in any

repository, provided it is only made publicly available 12 months after official publication

or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article

on Springer's website The link must be

accompanied by the following text: "The final publication is available at link.springer.com”.

R E S E A R C H A R T I C L E

New binary particle swarm optimization on dummy sequence

insertion method for nonlinear reduction in optical

direct-detection orthogonal frequency division multiplexing system

Lap Maivan1 •Thang Nguyentrong2

Received: 28 June 2017 / Accepted: 22 January 2019

 The Optical Society of India 2019

Abstract In the paper, a novel new binary particle swarm

optimization method based on dummy sequence insertion

is proposed and experimentally demonstrated in the IM-DD

optical orthogonal frequency division multiplexing

(OOFDM) system This technique can mitigate

nonlinear-ity of OOFDM system without any channel side

informa-tion Experimental results demonstrate that compared to

the original scheme, the improvement in the receiver

sen-sitivity by the proposed scheme is 1.9 dB and 3.2 dB with

launch powers of 2 dBm and 8 dBm, respectively, at the

BER of FEC 3.8 9 10-3 after transmission over 100-km

standard single-mode fiber At a complementary

cumula-tive distribution function of 10-4, the PAPR of OFDM

signal can be reduced about 2.8 dB by using the proposed

scheme, while the receiver-side hardware is the same as the

origin

Keywords Particle swarm optimization (PSO) Dummy

sequence insertion (DSI) Optical fiber communication 

Orthogonal frequency division multiplexing (OFDM)

Introduction

Recently, orthogonal frequency division multiplexing (OFDM) has been applied in optical communication sys-tem [1 5] Due to the high PAPR of OFDM, nonlinearity noise in the nonlinear optical components (such as Mach– Zehnder modulator, fiber, etc.) will cause performance degradation in optical communication system Therefore, reduction in PAPR of OFDM is very necessary in optical OFDM system Many methods have been paid more attention, such as selective mapping (SLM), partial trans-mitting sequence (PTS), clipping, companding transform technique, precoding, spreading codes, and dummy sequence insertion (DSI) However, the SLM [6] and the PTS [7] methods will increase the amount of computation

at the transmitter and receiver The clipping [8] could be an effective technique for PAPR reduction The OFDM signal can be clipped either at the Nyquist sampling rate or at an oversampling rate Clipping the Nyquist sampled signal does not cause out-of-band noise, since all the noise gen-erated by clipping falls in-band Clipping an oversampled signal produces less in-band noise, but the out-of-band noise will increase, it generally causes the out-of-band radiation of the clipped power, and the bandpass filter is required to suppress the out-of-band radiation The prob-lem of this scheme is the significant PAPR regrowth due to the bandpass filtering The companding transform tech-nique [9,10] has the advantages of simple implementation and low computational complexity, and it has better per-formance than clipping method In the OFDM system, to reduce the PAPR of an OFDM signal, the ideal case is to make the envelope of the OFDM signal constant However,

it is difficult to implement the ideal case by the available companding transform techniques due to the limitation of the BER [11] In addition, precoding [5] and spreading

& Lap Maivan

lapmv@hpu.edu.vn

Thang Nguyentrong

nguyentrongthang@tlu.edu.vn

1 Electronic and Electrical Engineering Department, Haiphong

Private University, Haiphong 180000, Vietnam

2 Faculty of Energy Engineering, Thuyloi University,

Hanoi 115070, Vietnam

123

J Opt

https://doi.org/10.1007/s12596-019-00512-6

Author's personal copy

codes [4] can also reduce the PAPR and improve BER

performance, but at the receiver, precoding or

de-spreading process must be done The DSI [12,13] method

can be used to insert dummy sequence into the

transmis-sion data block for PAPR reduction before the IFFT stage

In the DSI method [12], complementary sequence and

combination of the correlation sequence are considered as

dummy sequence, which will be discarded at the receiver;

thus, the side information is not necessary Moreover,

compared with the conventional PTS or SLM method, the

BER performance of the method is better in the case of the

errors in the side information about the phase rotation But,

the only limitation is that the computation is high In order

to solve complex computation problems, various heuristic

approaches have been adopted by researches, such as

genetic algorithm, tabu search and PSO PSO [14, 15] is

one of the optimization techniques The origin version of

the PSO [14] operates in continuous space, and the binary

PSO (BPSO) [15] operates on discrete binary variables

The PSO technique has been used in many fields; one of

them was used for PAPR reduction in wireless OFDM

system [13, 16,17] During utilization and research PSO

and BPSO, some researches have shown that standard PSO

and BPSO cannot converge well [18] To overcome this

problem with BPSO, a novel NBPSO [18] is proposed

In this paper, a novel NBPSO based on DSI method is

proposed and experimentally demonstrated for PAPR

reduction in the IM-DD optical OFDM system The

NBPSO scheme can assign a suboptimal dummy sequence

to enhance the performance of the DSI method so as to

reduce the PAPR of the IM-DD OOFDM system

System model

In the OFDM system, the baseband OFDM signal is given

by

sðtÞ ¼ 1ffiffiffiffi

N

p XN1

k¼0

X

N1

i¼0

where x is the data symbol, Df is the carrier spacing of

IFFT, and N is the number of subcarriers

The PAPR of the OFDM signal can be defined as

PAPR¼maxjsðtÞj

2

where E{•} denotes the expectation operation E {|s(t)|2} is

equal to the variance r2, and the symbols are zero mean

The statistics for the PAPR of an OFDM signal can be

given in terms of its complementary cumulative

distribu-tion funcdistribu-tion (CCDF) The CCDF of PAPR is defined as

the probability that the PAPR of the OFDM symbols exceeds a given threshold PAPR0 The CCDF for an OFDM signal is expressed as

Dummy sequence insertion method The DSI method [18] can reduce the PAPR by inserting a dummy sequence in the subcarriers of the OFDM system Dummy sequence is used for only PAPR reduction without any channel information At the receiver, dummy sequence can be discarded after FFT stage It is different from the conventional PTS and SLM methods Therefore, the DSI method can greatly reduce the complexity of the receiver And it is independent of the dummy sequence error Fig-ure1illustrates the structure of the DSI data In this paper, DSI with method 3 of Ref [13] is used, and 16QAM format

is considered Here, the total number of subcarriers in the OFDM signal is M, and the number of subcarriers reserved for the dummy sequence is L; therefore, there are N = M

-L subcarriers available for data transmission And the number of dummy sequence bit is K = 4xL

In Fig.1, after the IFFT input, the output signal can be expressed as

y tð Þ ¼ IFFT x s½ t

where y = [y1, y2,…, yM]t, x = [x1, x2,…, xN]t is the trans-mission data sequence and s = [s1, s2,…, sL]tis the inserted dummy sequence IFFT (z) implies the inverse fast Fourier transform of z, and [•]t

is a transpose operation

The PAPR of the OFDM signal with DSI method can be defined as

PAPR¼maxjyðtÞj

2

Meanwhile, Eq (5) can be written in vector form and expressed in decibels as

PAPR¼ 10 log10 maxjyðtÞj

2

meannjyðtÞj2o : ð6Þ

Fig 1 DSI data block using the complementary sequence

NBPSO scheme based on the DSI method

The DSI method is used to search out the dummy sequence

that minimizes the PAPR of an OFDM signal Based on

Eq (6), the PAPR reduction using DSI method can be

modeled as a constrained optimization problem It is given

by

Minimize PAPR subject to s; PL đ7ỡ

where PL is the total power limitation for the inserted

dummy sequence

Meanwhile, to search out the global optimal dummy

sequence so as to minimize the PAPR of the optical OFDM

signal, the NBPSO scheme is adopted in the optical OFDM

system The fitness function in this case can be shown in

Eq (7), and it can be expressed as

fđsỡ Ử 10 log10 maxjyđtỡj

2

meannjyđtỡj2o : đ8ỡ

The NBPSO scheme based on DSI method is shown in

Fig.2 It will be described in detail as follows:

Step 1 Initialization state

Firstly, the iteration counter t is reset, i.e., t = 0, and P

particles are randomly generated as [Zj(0), j = 1, 2,Ầ, P],

where Zj(0) = [zj,1(0), zj,1(0),Ầ, zj,K(0)]t, and zj,k(0) denotes

the kth bit of jth particle at t = 0 A vector of K bits

rep-resents the position of a particle and signifies a probably

desired dummy sequence Then, the initial velocities of all

particles are set to zero, such as [Vj(0), j = 1, 2,Ầ, P] = 0,

where Vj(0) = [vj,1(0), vj,2(0),Ầ, vj,K(0)]t

Secondly, modulate and evaluate the fitness function of

each particle in the initial population based on Eq (8), and

we set p_bestj = Zj(0) and Ffnj= Fj(0), j = 1, 2,Ầ, P,

where p_bestj= (p_bestj1, p_bestj2,Ầ, p_bestjK) and Ffnj

register the individual best position for the jth particle and

its fitness value of the jth particle at t = 0

Thirdly, we find the best fitness value F

bestfn-= min([Ffnj, j = 1, 2,Ầ, P]) registering the fitness values

of all initial particles, and set the particle of Fbestfn as the

global best g_best, which has an fitness value Fbestfn

Finally, set the initial values of the inertia weight w and

constants c1and c2, which are used in velocity updating

Step 2 Iteration counter updating state Update the generation counter as t = t?1

Step 3 Velocity updating state

In the NBPSO scheme, each vjk represents the proba-bility of bit zjk taking the value 1, and vjk must be con-strained to the interval [0.0, 1.0] By defining a function S(vjk) of the kth element in the jth particle, it is updated according to the following equation:

S vjk

 

Ử 2x sigmoid vjk

 

 0:5

đ9ỡ with Sigmoidđvjkỡ Ử 1

1ợevjk, and vjk (t ? 1) = w.vjk (-t) ? c1.rand().(p_bestjk Ờ zjk) ? c2.rand().(g_bestk Ờ zjk), where c1 and c2 are positive constants, rand() is a quasi-random number selected from a uniform distribution in [0.1, 1.0], w is the inertia weight, and vjkis limited in the range of [- vmax, vmax]

Step 4 Position updating state Based on the updated velocities, the position of each particle will be changed by the following equation:

If randđỡ\S v jkđtợ 1ỡ

then zjkđtợ 1ỡ

Ử exchange zjkđ ỡt 

elsezjkđtợ 1ỡ Ử zjkđ ỡ:t đ10ỡ Step 5 Individual best updating state

Each particle is evaluated by fitness function on the renewed position If there is Fj(t) \ Ffnj, j = 1, 2,Ầ, P, update individual best as P_bestj= Zj(t) and Ffnj= Fj (-t) and go to the global best updating state

Step 6 Global best updating state Search for the minimum fitness value Fmin from Ffnj,

j = 1, 2,,Ầ, P, where min is the index of particle with minimum fitness, i.e., min

Ợ {1, 2,Ầ, P} If Fmin\ Fbestfn, then the global best is updated as g_best = Zmin(t) and

Fbestfn = Fmin Step 7 Stop criteria checking state

If the stop criteria are satisfied, the procedure comes to a stop, or else goes to the Step 2

In this paper, the use of global model in NBPSO is considered, and the parameters in Eq (9) are set as the same as Ref [18] Usually, vmaxis set to be 6, c1= c2= 2, and the weight w is decreasing linearly from 0.6 to 0.2 The number of particle (NP) is 20, and the length of bit in each particle (P) is 32 The number of iteration (T) is considered

to be 30, which is the stopping criteria

Fig 2 NBPSO scheme based

on the DSI method

J Opt

123

Author's personal copy

Experimental setup and results

Experimental setup

Figure3 shows the experimental setup of the NBPSO

based on DSI method in the IM-DD optical OFDM system

In this experiment, three types of signal are used, such as

the original OFDM signal, the DSI method signal, and the

NBPSO based on the DSI signal The number of OFDM

subcarriers is 256 Among these subcarriers, 184 are used

for data and 16 for dummy sequence insertion The length

of CP is 1/8 of OFDM symbol duration corresponding to 32

samples in each OFDM symbol In the experiment, the

number of OFDM symbols per frame is 256 and training

sequence per OFDM frame is 1

The pseudo-random binary sequence (PRBS) is

con-verted into parallel data by S/P converter, and then the

NBPSO based on DSI method is processed for PAPR

reduction The algorithm of NBPSO includes a loop: a dummy sequence is added to the end of the each parallel data, and then they are mapped onto 16QAM After that, GI and Hermitian constraints are added Then, the data symbol

is passed through the IFFT, and PAPR is calculated After the NBPSO method implemented, the signal with lowest PAPR is obtained A complex-valued time-domain wave-form is produced; meanwhile, the CP is added to mitigate the ISI In addition, a training sequence (TS) can be used for channel estimation and symbol synchronization The electrical baseband OFDM signals are generated by offline MATLAB and uploaded into a commercial arbitrary waveform generator (AWG) A continuous-wave generated

by an external cavity laser (ECL) at 1556.26 nm is fed into

a MZM biased at 2.4 V, and the OFDM signals generated

by an AWG are injected into the MZM to generate optical OFDM signals The PRBS length is 94208 The sampling rate of the AWG is 5 G samples/s, and the peak-to-peak

Fig 3 Experimental setup for the IM-DD OFDM system with the

NBPSO based on DSI method AWG arbitrary waveform generator,

VOA variable optical attenuator, ECL external cavity laser, PC

polarization controller, MZM Mach–Zehnder modulator, EDFA Erbium-doped fiber amplifier, PD photodiode, TDS real-time/digital storage oscilloscope, and LPF low-pass filter

voltage of the signals is 1 V The half-wave voltage of the

MZM is 4 V The driving amplitude (Vpp) of OFDM

signals is 2 V, and the output power of the ECL is 14.5

dBm The output power of optical OFDM signals is 2 dBm

The optical OFDM signals are amplified by the first EDFA

(EDFA-1) before transmitting over 100-km SMF After

transmission over 100-km SSMF, the second EDFA

(EDFA-2) and a variable optical attenuator (VOA) are

adopted for controlling the received power The optical

OFDM signals are converted into electrical wave signals

via a PIN photodiode with a 3 dB bandwidth of 10 GHz,

and then passed through the LPF with 3 dB bandwidth of

2 GHz The electrical OFDM signal is captured by a 20 G

samples/s real-time digital storage oscilloscope

(TDS-6804B) The waveform recorded by TDS is processed by

offline MATLAB as the same as the original OOFDM

system The channel estimation and the symbol

synchro-nization are realized by using training sequence (TS) The

TS is similar to Park’s method [19] The symbol

syn-chronization is realized by Chen et al [20], and the channel

estimation is calculated by linear interpolation [21]

Finally, BER performances of the received signals are

calculated

Experiment results and discussion

The net bit rate of data signal is 6.36 Gbps, and the net bit

rate of DSI signal is 0.55 Gbps In this way, the

trans-mission efficiency is calculated as follows: subcarriers for

data/(subcarriers for DSI ? subcarriers for data) (%) = 92/

(8 ? 92)% = 92%

Figure4 shows the CCDF versus PAPR of OFDM sig-nal, NBPSO based on DSI sigsig-nal, and DSI signal with the threshold of 12 dB At the CCDF of 10-4, the PAPR of the OFDM signal with the NBPSO based on DSI method is reduced by 2.8 dB and 1.4 dB, compared with that of the original OFDM and the DSI method with PAPR threshold

of 12 dB, respectively

The BER performance of original OFDM signal, DSI signal, and NBPSO based on DSI signal with 2 dBm of fiber launch power after transmission over 100-km SMF is shown in Fig 5 At the BER of FEC 3.8 9 10-3, the received optical power is about - 4.8 dBm for that with NBPSO based on DSI method, - 3.7 dBm for that with the DSI method, and - 2.9 dBm for original OFDM signal The received sensitivity of OFDM signal with the NBPSO based on DSI method can be improved by 1.1 dB and 1.9 dB when compared with that of the DSI method and original OFDM, respectively

The BER performance of OFDM signals with received power is shown in Fig.6 At the BER of FEC 3.8 9 10-3, the received optical power of NBPSO based on DSI signal, the DSI signal, and original signal is about - 6.8, - 4.8, and - 3.6 dBm, respectively The received sensitivity with NBPSO based on DSI method can be improved by 2 dB and 3.2 dB when compared with the case of DSI method and original OFDM, respectively

Figure7 shows BESR via launch power of optical OFDM signals at received optical power of - 2 dBm after transmission over 100-km SMF At the same launch power, BER performance of the proposed technique is the best and BER performance of original is the worst As the launch power increases from 2 to 8 dBm, BER of all optical

Fig 4 Complementary cumulative distribution function (CCDF)

versus peak-to-average power ratio (PAPR) of OFDM signals

Fig 5 BER curves of OFDM signals at 2 dBm launch power after transmission over 100-km SMF

J Opt

123

Author's personal copy

OFDM signal is decreasing Meanwhile, the BER of the

proposed technique decreases faster than that of the other

techniques The BER of all optical OFDM signal is the min

value as the launch power is about 8 dBm, so that the

launch power of 8 dBm is the optimal launch power

Moreover, with the launch power increasing from 8 to 11

dBm, the BER of all optical OFDM signal is improving,

but the BER performance of the proposed technique is

better than that of the other techniques The experimental

results indicate that the proposed technique can resist

nonlinear and it is better than other techniques

Conclusion

To sum up, this paper has proposed and experimentally demonstrated a novel NBPSO based on DSI method in the IM-DD-OOFDM system The novel proposed technique is used to reduce the fiber nonlinear effect through reducing the PAPR of the OFDM signal The experimental results show that at the launch power of 8 dBm, the received sensitivity with NBPSO based on DSI method can be improved by 2 dB and 3.2 dB when compared with that of DSI method and original OFDM, respectively Meanwhile,

at the CCDF of 10-4, the PAPR of OFDM signal with the proposed technique is reduced by 1.4 and 2.8 dB, com-pared with that of the DSI method and the original OFDM, respectively At a BER of FEC limitation 3.8 9 10-3, the received power with proposed technique is more sensitive than that of the original OFDM Thus, by using the pro-posed technique, it can reduce the fiber nonlinear effect efficiently, while the receiver-side hardware remains as similar as the origin

References

1 W Shieh, X Yi, Y Ma, Y Tang, Theoretical and experimental study on PMD supported transmission using polarization diver-sity in coherent optical OFDM systems Opt Express 15, 9936–9947 (2007)

2 W Shieh, H Bao, Y Tang, Coherent optical OFDM: theory and design Opt Express 16, 841–859 (2008)

3 B.J.C Schmidt, A.J Lowery, J Armstrong, Experimental demonstrations of electronic dispersion compensation for long-haul transmission using direct-detection optical OFDM J Light-wave Technol 26, 196–203 (2008)

4 L Maivan, J He, M Chen, F Mangone, L Chen, New hybrid peak-to-average power ratio reduction technique based on carrier interferometry codes and companding technique for optical direct-detection orthogonal frequency division multiplexing sys-tem Opt Eng 53, 086104 (2014)

5 W.A.N.G Zhong-peng, C.H.E.N Fang-ni, W.U Ming-wei, C.H.E.N Ming, T.A.N.G Jin, C.H.E.N Lin, Experimental evaluation of the BER performance in optical OFDM system based on discrete Hartley transform precoding Optoelectron Lett 10, 224–227 (2014)

6 R.W Ba¨uml, R.F.H Fisher, J.B Huber, Reducing the peak-to-average power ratio of multicarrier modulation by selected mapping Electron Lett 32, 2056–2057 (1996)

7 L.J Cimini, N.R Sollenberger, Peak-to-average power ratio reduction of an OFDM signal using partial transmit sequences IEEE Commun Lett 4, 86–88 (2000)

8 D Kim, G.L Stuber, Clipping noise mitigation for OFDM by decision-aided reconstruction IEEE Commun Lett 3, 4–6 (1999)

9 X.B Wang, T.T Tjhung, C.S Ng, Reduction of peak-to-average power ratio of OFDM system using a companding technique IEEETrans Broadcast 45, 303–307 (1999)

10 X.B Wang, T.T Tjhung, C.S Ng, Reply to the comments on

‘reduction of peak-to-average power ratio of OFDM system using

Fig 6 BER curves of OFDM signals at 8 dBm launch power after

transmission over 100-km SMF

Fig 7 BER via launch power of OFDM signals after transmission

over 100-km SMF

a companding technique’’ IEEE Trans Broadcast 45, 420–422

(1999)

11 G Ren, H Zhang, Y Chang, H Zhang, in International

Con-ference on Communications, Circuits and Systems Proceedings

(ICCCAS) A self companding transform to reduce

peak-to-av-erage power ratio in OFDM based WLANs, Guilin, China, 2006,

pp 1142–1146

12 H.-G Ryu, J.-E Lee, J.-S Park, Dummy sequence insertion

(DSI) for PAPR reduction in the OFDM communication system.

IEEE Trans Consum Electron 50, 89–94 (2004)

13 S.W Kim, H.S Byeon, J.K Kim, H.G Ryu, in Fifth

Interna-tional Conference on Information, Communication and Signal

Processing An SLM-based real-time PAPR reduction method

using dummy sequence insertion in the OFDM communication,

Bangkok, Thailand, 2005, pp 258–262

14 J Kennedy, R.C Eberhart, in Proceedings of IEEE International

Conference on Neural Networks Particle Swarm Optimization,

Perth, WA, 1995, pp 1942–1948

15 J Kennedy, R.C Eberhart, in IEEE International Conference on

Computational Cybernetics and Simulation A discrete binary

version of the particle swarm algorithm, Orlando, FL, 1997,

pp 4104–4108

16 S.H Lee, H.-L Hung, in International Conference on

Broad-band, Wireless Computing, Communication and Applications

(BWCCA) Particle swarm optimization on DSI method for PAPR reduction in OFDM systems, Fukuoka, 2010, pp 639–642

17 J Taghipour, A.A Parandoosh, V.T Vakili, in 20th Telecom-munications Forum TELFOR An agent based particle swarm optimization for PAPR reduction of OFDM systems, Serbia, Belgrade, 2012, pp 839–842

18 H Nezamabadi-pour, M Rostami-shahrbabaki, M.M Farsangi, Binary particle swarm optimization: challenges and new solu-tions J Comput Soc Iran Comput Sci Eng 6, 21–32 (2008)

19 B Park, H Cheon, C Kang, D Hong, A novel timing estimation method for OFDM systems IEEE Commun Lett 7, 239–241 (2003)

20 M Chen, J He, Z Cao, J Tang, L Chen, W Xian, Symbol synchronization and sampling frequency synchronization tech-niques in real-time DDO-OFDM systems Opt Commun 326, 80–87 (2014)

21 M Chen, J He, L Chen, Real-time optical OFDM long-reach PON system over 100 km SSMF using a directly modulated DFB laser J Opt Commun Netw 6, 18–25 (2014)

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

J Opt

123

Author's personal copy