Analysis of Nonlinear Signal Interference to Optimize the Coherent Optical Transmission System

IOSR Journal of Electronics and Communication Engineering (IOSR-JECE)
e-ISSN: 2278-2834,p- ISSN: 2278-8735.Volume 12, Issue 1, Ver. I (Jan.-Feb. 2017), PP 24-31
www.iosrjournals.org
DOI: 10.9790/2834-1201012431 www.iosrjournals.org 24 | Page
Analysis of Nonlinear Signal Interference to Optimize the
Coherent Optical Transmission System
Abdul Gafur1
1
Department of Electronic and Telecommunications Engineering, International Islamic University Chittagong,
Bangladesh
Abstract: In this paper we analyze the power of Nonlinear Signal Inference ( NLSI
p ) and optimize the coherent
optical transmission system (COTS) based on PM-16QAM higher modulation format for different baud rates
considering commercially available silica fiber (SF). Theoretically, this work follows up four Gaussian Noise
(GN) models depending on multiple identical spans, non-Nyquist WDM system, non-identical spans and multi-
channels to confirm the better accuracy of these models. The most fitted model is then used to optimize the
COTS achieved by analyzing the Amplified Spontaneous Emission noise power (PASE) to determine the optimum
Optical Signal to Noise Ratio (OSNR). Matlab R2011b simulation tool is run here to generate the OSNR and Bit
Error Rate (BER) performance curves for the COTS. The OSNR has been simulated considering input
transmitting power per span, span length, dispersion and fiber loss coefficient. Results show that, the OSNR
penalty is around 3dB and 6dB when line rate is multiplied with the factor of 2xbit rate, 4x bit rate respectively
for same modulation and demodulation formats in the COTS.
Keywords: OSNR, PM-16QAM, GN, SF, BER
I. Introduction
Usually laser coherence is utilized in coherent optical transmission system. Here optical signal carrying
capacity and transmission rate are the key factors to upgrade in the existing optical network. On the other hand,
capacity depends on higher line rate and spectral efficiency of the channels [1]. The desired optical link
performance based on higher line rate and narrower channel spacing for each non-identical channel in
wavelength division multiplexing technique can be achieved [2]. Similarly, optical link set up and processing
takes long time for submarine network where terrestrial network is easier, not more difficult but it is little bit
challenging to improve and optimize the capacity of the network [3].
Thus, to achieve the significant OSNR, higher Baud rate and more spectral efficient modulation techniques are
required to enhance the performance of the COTS. For this, higher channel power, lower fiber loss coefficient and lower
nonlinearity factors are expected for higher OSNR [1]. Like submarine link, for higher line rate, COTS plays an important
job for lower cost information transmission because of DWDM and EDFA in the transmitter and the chain of the link
respectively. Consequently, optimization is an important issue to upgrade and minimize the higher cost of the existing
optical network considering higher modulation formats [4]. The performance of COTS is also obtained through the
improvement of the nonlinear optical signal propagation.In this regard, we have considered four GN models on multiple
identical spans, non-Nyquist WDM system, non identical spans and multi-channels.
To get the best performance of COTS, the power of NLSI is the dominating factor which occurs for
non-identical channel propagation in the link. That’s why we follow the algorithm of scheduling used for
nonlinear propagation where linear routing and wavelength choice finding are the great problems [5]. As a
result, accuracy and simple simulation process are required to obtain the best performance of the link. In this
case, non-linear Schrodinger equation complies with these factors. For this situation, it is very hard to find out
the power spectral density of NLSI and simulation process takes longer time. Moreover, the commercial optical
simulator software is expensive and simulation running output takes longer time [6]. Consequently, GN model is
the way and method to determine the power of nonlinear signal interference ( NLSI
p ) [7-9]. In GN model, we
control the total number of spans causing variation to the link length randomly [10].
In this way, we set the input transmitting power ( TXP ), NLSI
p and ASE
P for each span to obtain the optimum
OSNR performance for the COTS. After each span we confirm the OSNR which is calculated after total number
of spans.
For span length, we take SF commercial fibers [11] – [12] for 50 numbers of spans.
It is an important issue to consider the values of OSNR for the receiver section. Firstly, the receiver
part cannot identify the signal for low values of OSNR. Secondly, to reach the maximum point parameters,
wavelength goes through generation and regeneration in the link. Thirdly, OSNR values estimate the
performance of the system and make an assurance of error free transmission in the link. Fourthly, BER is of

Analysis of Nonlinear Signal Interference to Optimize the Coherent Optical Transmission System
crucial value and quality of propagation of the optical signal which is dependent on OSNR values. Considering
all these factors, it can be said that OSNR is an important parameter and it depends on number of optical links,
topology of the network, symbol rate of the higher modulation formats and the target of lower BER to design
the optical network efficiently [13].The optical link model used for the simulation is shown in Fig. 1. The total
number of channels in the transmitter section is divided into odd and even number of channels to implement
polarization multiplexing. The number of channels in the WDM comb is 27 channels with the 14th
channel
considered as the Centre.
Every channel has a channel bandwidth, channel power and line rate respectively. The modulated
optical signal goes through the optical filter [14] then the combiner combines the channels. The specification
data for SF are shown in Table 1. The EDFA is connected after a span of 100-km. A noise figure of 5dB
compensates the link loss depending on the fiber loss coefficient and span length.The pNLSI is determined at the
input of the EDFA and the PASE is observed at the output of the EDFA [15] in the optical link of the COTS as
shown in Fig. 1 [9].In this condition 27 WDM comb are considered where 193.4145 THz is the center channel
frequency within the range from 192.7645 THz to 194.0645 THz. These channel frequencies create odd and
even number of channels with 0.05THz spacing .The simulation is performed for 5000 km silica fiber link
distance considering 128Gb/s and 256Gb/s .We analyze the GN models for different modes and conditions to
predict the accuracy among the models. The power spectrum of the ASE noise is considered to obtain the
optimum OSNR for the COTS.
Fig. 1: Model of optical link.
2. Analysis of nonlinear signal interference (NLSI)
In this work we follow four Gaussian non-linear (GN) models to determine the accuracy and
convergences of the models in optical coherent transmission. Then we choose very close and tight fit
convergence of GN model to determine the optimum ONSR for different line rate ( sR ) in the COTS.
The power NLSI
p depends on Gaussian nonlinear (GN) signal spectrum considering an optical noise bandwidth
of 12.5GHz. The expression for NLSI
p in the case of multichannels is presented below [19]
n
s
sCheff
CheffsNLSI
B
R
RNL
PLNp 3
2
22
2
2
32
)log(
27
8
1


 (1)
In the above expression, s
N ,  , ffeL , Ch
P , 2 , Ch
N , s
R and n
B are the number of span, non-linearity
coefficient, effective length , channel power , dispersion parameter of fiber ,number of channels, line rate of the
channel and optical noise bandwidth of COTS respectively.
We also follow the GN model for non identical span and mathematical expression are shown below to calculate
the NLSI
p : [10,20]
s
ch
s
ss
snssn
s
s
s
snssn
s
s
s
s
s
ninin
N
n
ichnchnch
N
nn
L
n
n
n
L
n
neff
N
n
nnNLSI
GGG
egeg
LBp
,,,
1
,,,
2-
1-
1
6-3
2
,
1
2
)-2.(.
..
.
27
16
∑
∏∏
∑
,,
2












(2)

The effective length is calculated for using the following equation [15]:

2/]-1[ ).-( sL
eff eL  (3)
Where α is the fiber loss coefficient and sL is the span length of the COTS.
The power spectral density for nth
and ith
channels are given in [17] is used in the model as below,
s
nch
nch
R
P
G ,
,  (4)
s
ich
ich
R
P
G ,
,  (5)
Where for nth
and ith
channels, the channel power is nchP ,
and ichP ,
respectively. The symbol ni is used to
maintain the center frequency for the WDM comb based on 0 and 1 condition when n=i and
in ≠ respectively.
When n =i and the symbols
snin ,, and snii ,, are used in model to select the strategy of the channel [10], [20]
ss
ss
ss
ss
s
nn
ichnchichnchnn
nn
ichnchichnchnn
nin
BBff
BBff
,2
1-
,,,,,2
1-2
,2
1-
,,,,,2
1-2
,,
)2(4
)]2/--[)2(asinh(
-
)2(4
)]2/-[)2(asinh(



 

(6)
1-
,2
,
21-
,2
2
,,
]2[2
)]2[
2
asinh(
ss
ss
s
nn
ichnn
nii
B



 (7)
Where 2 and ch
B are the second order fiber dispersion parameter and -3dB channel bandwidth respectively.
For the
th
i channel, the center channel is denoted as ichf , in the comb.
For the model accuracy, the NLSIp expression is for the case of Non-Nyquist and presented below [5]
  












f
R
chsaeff
aeff
WDMeff
nNLSI
s
NRL
L
GL
Bp
22
,2
2
,2
32
2
2
1
asinh.
...
27
8
3



(8)
where aeffL , and WDM
G are the asymptotic effective length and PSD of the transmission in the super channel of
comb. In the model, asymptotic effective length is used in the following way:

1
, aeffL (9)
The NLSIp analysis is for the case of multiple identical spans model [9] is shown below
12
221
22
221
22
2
212
2
)-()-(42-
∞
-∞
2121
∞
∞-
2
))-)(-(2(sin
))-)(-(2(sin
.
)-()-(4-2
-1
.
)-()()(
27
64
212
2
4
∫∫
dfdf
Lffff
LffffN
ffffj
ee
fffGfGfGBp
s
ss
ffffLjL
TxTxTxnNLSI
ss






(10)

In Eq. (10), GTx is the PSD which depends on channel power and line rate. The frequencies (f1+f2-f) for four
wave mixing (FWM) are the elements frequency of f [16].
II. OSNR And Bit Error Rate (BER) Performance Evaluation
Based on analytical analysis of nonlinear signal interference, the numerical results are simulated in
Fig.2 (a) and Fig.2 (b) considering four GN models using Matlab R2011b tool. These results confirm the
convergence and accuracy among the models.
0.016 0.018 0.02 0.022 0.024 0.026 0.028 0.03 0.032
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
x 10
-3
Rs
(THz)
pNLSI
pNLSI
1
pNLSI
2
pNLSI
3
pNLSI
4
Fig. 2 (a): Non linear signal interference for four GN models considering power of nonlinear signal interference
( NLSI
p ) vs. line rate (Rs) with SF.
10 11 12 13 14 15
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
x 10
-3
Pch
(dBm)
pNLSI
pNLSI
1
pNLSI
2
pNLSI
3
pNLSI
4
Fig. 2 (b): Non-linear signal interference for four GN models considering power of nonlinear signal interference
( NLSI
p ) vs. channel power (Pch) with SF.
Fig.2 (a) represents the power of nonlinear signal inference versus line rate of the COTS. This result is
simulated for SF with 27 channels, 32 GHz channel bandwidth (Bch), 50GHz channel spacing (∆f), center
frequency (fc) 193.4145 THz, channel power (Pch) 10 dBm, number of span (Ns) is 1, span length (Ls) 100km,
optical noise bandwidth (Bn) 12.5GHz. In this figure, Black, Blue, Magenta and Red lines carry the information
for 1
NLSI
p , 2
NLSI
p
,
3
NLSI
p
and
4
NLSI
p
respectively. It is found that the power of nonlinear interference is going
down with the increment values of Rs. Magenta and Blue lines are fitted very well. Magenta, Blue and Red lines
are seemed to be very closed for line rate from 0.028THz to 0.032THz. On the other hand, Black line shows
very small nonlinear signal inference in comparison with Blue and Magenta lines. It is found that Blue line
shows very good accuracy compared to other three models.

Fig.2 (b) is simulated for channel power (Pch) and power of nonlinear signal interference (pNLSI) based on all
parameters used in Fig.2 (a).In this results, Black, Blue, Magenta and Red lines are used for 1
NLSI
p , 2
NLSI
p
,
3
NLSI
p
and
4
NLSI
p respectively. The nonlinear signal interference power is going up with the increase of channel
power in the system. It is found that Blue, Magenta and Red lines show the best fitting curve for channel power
10dBm to 15dBm.The Black line performs the good matching for 10 dBm to 11 dBm channel power. It is
selected that the Blue line performs the good accuracy. The optimum OSNR [7] [18] depends on span
transmitting input power ( TXP ), ASEP and 2
NLSI
p as expressed below:
n
B
o
fhF
s
A
ASE
P  (11)
)
3-
ch
3
(
OSNR
2 BP
ASE
P
s
N NLSI
p
P
TX
TX

 (12)
In Equ.(11)
s
A , F , h and
o
f are the span loss, noise figure, plank constant and center frequency of the
channel of comb respectively. In this simulation we put noise figure 5 dB, where span loss depends on span
length and fiber loss coefficient per kilometer of SF. In Equ. (12),
s
N is the number of spans. It is used 50
spans in our simulation process.
To measure the Bit Error Rate (BER), OSNR is required where OSNR determined from 2
NLSI
p , ASE
P and TXP
respectively. For PM-16QAM modulation format, 16 GBaud and 32 GBaud produced 128 Gb/s and 256 Gb/s
line bit rates respectively.
SNR is required to determine the BER. SNR is related with OSNR and BER is expressed below [9].
OSNR
R
B
SNR
s
n
)( (13)
BER= )(SNR (14)
 is the modulation format that operates the bit rates and upgrade the system. The modulation format for PM-
16QAM used in [9]
)( 10
1
8
3
16QAM)-(PM
BER SNRerfc (15)
In Equ. (15) erfc is the complementary error function.
III. Optimization Of OSNR For COTS
Based on theoretical analysis, the simulation is done with SF commercial optical fiber. This result is
with 27 channels, 32 GHz channel bandwidth (Bch), 50GHz channel spacing (∆f), center frequency (fc) 193.4145
[THz], channel power (Pch) 15 dBm, number of span(Ns) is 50, span length (Ls) 100km, optical noise bandwidth
(Bn) 12.5GHz.
Fig.3. (a): OSNR [dB] versus PTX [dBm] based on linear and nonlinear propagation considering 16GBaud and
32GBaud line rates of the COTS of SF

Fig.3. (b): PTX [dBm] versus BER based on linear and nonlinear propagation considering 16GBaud and
32GBaud line rates of the COTS of SF
In Fig.3 (a) OSNR [dB] versus TXP [dBm] where dotted lines represent the linear propagation; solid
lines show the non-linear propagation. Blue line is for 16Gbaud and Red line is for 32GBaud line rates. Case for
nonlinear signal propagation, the optimum OSNR is 13.77dB and 16.78dB when TXP is -2.44dBm and 0.5 dBm
considering 16GBaud and 32GBaud respectively. For the result's confirmation, it is found that the optimum
penalty between linear and nonlinear propagation is around 1.76dB [21]. It is also confirmed that the lines
overlap for 16GBaud and 32GBaud respectively considering linear propagation. It is also found that OSNR
changed corresponding to the TXP dBm by 1dB linearly. For the same modulation format, the optimum OSNR
difference is 3dB following the factor of 2xbit rate and 4xbit rate. Fig. 3(b) is the plot of BER for PM-16QAM
modulation format. In this case it is assumed that all parameters used here is the same as used previously for
Fig.3 (a).
Table1. Specification of optical fiber
Fiber Types α [dB/km] γ [1/W/km] D [ps/nm/km]
SF 0.18 0.9 21
Fig.4. (a): Considering line rate 16GBaud, contour plot of OSNR versus span length [km] and dispersion
[ps/nm/km] parameter of SF.

Fig. 4(b): allowing for line rate 16GBaud, contour plot of OSNR [dB] versus fibre loss [dB/km] and dispersion
[ps/nm/km] parameter of SF
Fig. 4. (c): Considering line rate 32GBaud, contour plot of OSNR [dB] versus span length [km] and dispersion
[ps/nm/km] parameter of SF
Fig. 4(d): Considering line rate 32GBaud, contour plot of OSNR [dB] versus fibre loss coefficient [dB/km] and
dispersion [ps/nm/km] parameter for SF

Fig.4 (a) is the contour plot of OSNR versus span length [km] and dispersion [ps/nm/km] parameter of SF. Fig.4
(b) shows contour plot of OSNR [dB] versus fibre loss coefficient [dB/km] and dispersion parameter [ps/nm/km] of SF. Both
plots are for 16GBaud and 50 numbers of spans. with 27 channels, 32 GHz channel bandwidth (Bch), 50GHz channel spacing
(∆f), center frequency (fc) 193.4145 THz, channel power (Pch) 10 dBm, span length (Ls) 100km, optical noise bandwidth (Bn)
12.5GHz. It is found that the optimum OSNR is 11.59dB for Fig.4 (a) and Fig.4 (b) when dispersion 21[ps/nm/km], span
length100km and fibre loss coefficient 0.18[dB/km]. On the other hand Fig.4(c) and Fig.4 (d) are the contour plot of OSNR
versus span length [km] and dispersion [ps/nm/km] parameter, OSNR [dB] versus fibre loss [dB/km] and dispersion
[ps/nm/km] parameter of SF considering 32GBaud. Here, it is found that the optimum OSNR is 14.38dB when dispersion
21[ps/nm/km], span length100km and fibre loss coefficient 0.18[dB/km]. The OSNR penalty between 16Gbaud and
32GBaud is 2.8dB when TXP =-10dBm to-3dBm.
IV. Conclusion
In conclusion, this paper describes a detailed evaluation on four GN models successfully. These four
GN models, 2
NLSI
p is the most fitted with better accuracy than the other models considering same symbol rate
and channel power. When the channel power (Pch) 15dBm is considered for 128Gb/s and 256Gb/s data rates, the
optimum OSNR is 13.77dB and 16.78dB when TXP =-2.44dBm and 0.5dBm respectively. Considering channel
power (Pch) of 10 dBm,, the optimum OSNR is 11.59dB and 14.38dB for a dispersion parameter (D) of
21[ps/nm/km], fiber loss coefficient (α) 0.18 [dB/km] and span length (Ls) 100km. For the same transmission
format it is found that, the OSNR variation is the factor of around 3dB and 6dB with respect to the 2xbit rate and
4xbit rate respectively.
References
[1] John D. Downie, “112 Gb/s PM-QPSK transmission systems with reach lengths enabled by optical fibers with ultra-low loss and
very large effective area,” 2012 Photonics West Conference, pp. paper 828403, January 2012
[2] V. Bobrovs, A. Udalcovs, R. Parts, and I. Trifonovs, “Evaluation of Nonlinear Effect Impact on Optical Signal Transmission over
Combined WDM System,” PIERS Proceedings, Taipei, Mar. 2013.
[3] N. B. Pavlovic and R. Nogueira, “Optimization of Extreme Optical Links for Network Upgrades,” ICTON 2012.
[4] Eduardo S. Rosa1,Victor E. S. Parahyba1, Júlio C. M. Diniz, Vitor B. Ribeiro1 and Júlio C. R. F.Oliveira, “Nonlinear Effects
Compensation in Optical Coherent PDM-QPSK Systems,” Journal of Microwaves, Optoelectronics and Electromagnetic
Applications, Vol. 12, No. 2, December 2013.
[5] Pontus Johannisson and Erik Agrell, “Modeling of Nonlinear Signal Distortion in Fiber-Optical Networks”, posted on arXiv,
www.arxiv.org, paper identifier: 1309.4000v1, 16 sep. 2013.
[6] Kerry Hinton, J. C. Li,Peter M. Farrel, Wayne V. Sorin, “A New Design Technique for Optical Links”, WU1, 15.30 - 15.45, 2009.
[7] Pierluigi Poggiolini,“The GN Model of Non-Linear Propagation in Uncompensated Coherent Optical Systems,” J. Lightw.
Technol., vol. 30, no.24, Dec. 2012.
[8] P. Poggiolini, A. Carena, V. Curri, G. Bosco, and F. Forghieri, “Analytical modeling of nonlinear propagation in uncompensated
optical transmission links,” IEEE Photon. Technol. Lett., vol. 23, no. 11, pp. 742–744, Jun. 2011.
[9] A. Carena, V. Curri, G. Bosco, P. Poggiolini, and F. Forghieri, “Modeling of the impact of nonlinear propagation effects in
uncompensated optical coherent transmission links,” J. Lightw. Technol., vol. 30, no. 10, pp. 1524–1539, May 2012.
[10] Pierluigi Poggiolini, Gabriella Bosco, Andrea Carena, Vittorio Curri, Yanchao Jiang and Fabrizio Forghieri, “A Detailed Analytical
Derivation of the GN Model of Non-Linear Interference in Coherent Optical Transmission Systems”, posted on arXiv,
www.arxiv.org, paper identifier :1209.0394, (2012).
[11] A. Nespola, S. Straullu, A. Carena, G. Bosco, R. Cigliutti, V. Curri, P. Poggiolini, M. Hirano, Y. Yamamoto, T. Sasaki, J.
Bauwelinck, K. Verheyen, F. Forghieri, “Extensive Fiber Comparison and GN-model Validation in Uncompensated Links using
DAC-generated Nyquist-WDM PM-16QAM Channels,” Opt. Fiber Commun. Conf. (OFC/NFOEC) Technical Digest, OSA, 2013.
[12] Vittorio Curri, Pierluigi Poggiolini, Gabriella Bosco, Andrea Carena, and Fabrizio Forghieri, “Performance Evaluation of Long-
Haul 111 Gb/s PM-QPSK Transmission Over Different Fiber Types,” IEEE Photon. Technol. Lett., vol. 22, no. 19, Oct. 2010.
[13] www.MapYourTech.com,“COMBATING OSNR IN DWDM LINKS,” a power point presentation; http://www. slideshare.
net/sanmap/techniques-to-combat-osnr-in-dwdm-links
[14] A. Carena, G. Bosco, V. Curri, P. Poggiolini, F. Forghieri, “Impact of the Transmitted Signal Initial Dispersion Transient on the
Accuracy of the GN-Model of Non-Linear Propagation”, Eur. Conf. Opt. Commun. (ECOC), paper Th.1.D.4, 2013.
[15] P. Poggiolini, G. Bosco, A. Carena, R. Cigliutti, V. Curri, F. Forghieri, R. Pastorelli, S. Piciaccia, “The LOGON Strategy for Low-
Complexity Control Plane Implementation in New-Generation Flexible Networks,” Opt. Fiber Commun. Conf. (OFC/NFOEC),
OSA,2013.
[16] Abdul Gafur, Md. Razu Ahmed, Md. Badiuzzaman and Abu Riduan Md Foisal, “OSNR Improvement of Coherent Uncompensated
Optical Transmission Systems for Various Commercial Optical Fiber Types,” ICEEICT Proceedings, 10-12 April, 2014, Dhaka,
Bangladesh.
[17] G. Bosco, A. Carena, R. Cigliutti, V. Curri, P. Poggiolini, “ Next generation Terabit Optical networks: theory, simulation and
experiments,” Poster-session , DET-day, http://www.det.polito.it/focus/det_day/session_1_14_15_15_00.
[18] Vittorio Curri, Andrea Carena, Pierluigi Poggiolini, Gabriella Bosco, and Fabrizio Forghieri, “Extension and validation of the GN
model for non-linear interference to uncompensated links using Raman amplification,” Optical Express, Vol. 21, No. 3,11, February
2013.
[19] Robert Maher, Tianhua Xu, Lidia Galdino, Masaki Sato, Alex Alvarado, Kai Shi, Seb J. Savory, Benn C. Thomsen, “Spectrally
Shaped DP16QAM SuperChannel Transmission with MultiChannel Digital BackPropagation,” Scientific Reports 5, Nature
Publishing Group, Article number: 8214 doi:10.1038/srep08214, Feb. 2015.
[20] P. Poggiolini, G. Bosco, A. Carena, V. Curri, F. Forghieri, “The GN-Model of Fiber Non-Linear Propagation and its Applications,”
J. Lightw. Technol., vol. 32, no. 4, pp. 694-721, April 2014.
[21] G. Bosco, A. Carena, R. Cigliutti, V. Curri, P. Poggiolini, F. Forghieri, “Performance Prediction for WDM PM-QPSK Transmission
over Uncompensated Links”, in Proc. of OFC 2011, Los Angeles (USA), paper OThO7, Mar. 2011.

Analysis of Nonlinear Signal Interference to Optimize the Coherent Optical Transmission System

More Related Content

What's hot (19)

Similar to Analysis of Nonlinear Signal Interference to Optimize the Coherent Optical Transmission System (20)

More from IOSRJECE (20)

Recently uploaded (20)

Analysis of Nonlinear Signal Interference to Optimize the Coherent Optical Transmission System