J. Opt. Commun.

2017; aop

Shaymaa Riyadh Tahhan, Mudhafar Hussein Ali and Abdulla Khudhiar Abass*

Characteristics of Dispersion Compensation for 32

Channels at 40 Gb/s Under Different Techniques
https://doi.org/10.1515/joc-2017-0121 system at different bit rates with dispersion compensa-
Received July 16, 2017; accepted November 6, 2017 tion using optical phase conjugation and FBG compensa-
tion; they found that in dispersion-managed fiber
Abstract: In this study, a simulation analysis of 40-Gb/s
method, pre-compensation scheme is better than all
long-haul (120 km) dense wavelength-division multi-
other methods for long optical communication, consider-
plexer system (DWDM) of 32 channels with 50 GHz spa-
ing that the FBG compensation technique is the next best
cing is conducted for duobinary return-to-zero
technique and provides better eye opening [3]. Gupta
modulation formats. Pre- and post-dispersion compensa-
et al. observed that duobinary coding schemes perform
tion schemes utilizing dispersion compensating fiber
better with FBG at long distances of up to 4000 km [4].
(DCF) and fiber Bragg grating (FBG) are analyzed to
Attenuation losses can be compensated by utilizing
compare these schemes at high bit rate. Quality factor
an erbium-doped fiber amplifier (EDFA) [5]. The position
(Q) and bit error rate (BER) values are used to measure
of EDFA with FBG should be considered because it plays
the response of the system as a function of input power
an important role in enhancing system performance. If
for both compensation schemes. The pre-FBG scheme
FBG is used at the transmitter side for dispersion com-
shows a better performance in terms of Q-factor that
pensation (pre-scheme), then better performance (mini-
reached 11.5 at 0 dBm input signal power.
mum BER) has been recorded as compared with
Keywords: dispersion compensation fiber, fiber Bragg compensation at the receiver side (post-scheme) [6]. The
grating, pre-compensation scheme, post-compensation previous work provided a simulation study on less than
scheme 32 transmitters for 40 Gb/s DWDM, where non-return-to-
zero (NRZ) or return-to-zero (RZ) modulation format was
used in the transmitters. The challenge in this work is to
1 Introduction increase the number of transmitters to 32 and utilize the
duobinary modulation format, and use FBG in the chan-
The DWDM technique is employed in communication nel and examine whether it performs better than DCF
systems to increase the capacity of carrying information. when the number of the transmitters increases.
Attention to different details should be considered in the Comparative study between two fibers of two different
design of DWDM systems. The generation and propaga- channel schemes using the duobinary transmitter was
tion of multiple bit streams overlapping in the time simulated based on the study of Sharan et al. [7].
domain create broadening that causes chromatic disper-
sion, which results in errors in signal detection [1]. This
phenomenon has been successfully eliminated or com- 2 Duobinary RZ transmitter
pensated by several studies. For DWDM of 32 channels, a
DCF was employed within a post-compensation scheme
modulation format
design at different bit rates (10, 20, and 40 Gbps) by Kaur
Duobinary modulation is a three-level format. Two mod-
et al. in 2015 [2]. Each carrier could be modulated inde-
ulation types are used in this modulation scheme,
pendently and transmitted simultaneously via DWDM.
namely, amplitude and phase modulations, which con-
Tripathi et al. studied the performance of a duobinary
tribute in reducing pulse broadening. Thus, duobinary
*Corresponding author: Abdulla Khudhiar Abass, Laser Engineering modulation has low sensitivity to chromatic dispersion
Branch, Laser and Optoelectronics Department, University of and nonlinear intrachannel effects. In duobinary coding,
Technology, Baghdad, Iraq, E-mail: [email protected] a π-phase shift occurs when the “0”s are odd between
Shaymaa Riyadh Tahhan, Laser and Optoelectronics Engineering
two groups of “1”s [1].
Department, College of Engineering, Al-Nahrain University,
Baghdad, Iraq, E-mail: [email protected] The same configuration is used for both 40 Gb/s and
Mudhafar Hussein Ali, Network Engineering Department, College of 10.6 Gb/s duobinary transmitters. Duobinary procedure is
Engineering, Al-Iraqia University, Baghdad, Iraq used to create the duobinary NRZ signal combined with NRZ

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
2 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s

generator and duobinary pulse generator. Two LiNb Mach– the average power launched per span. Equation (1) is
Zehnder modulators (MZMs) are employed in this format. related to the Q-factor in dB as a function of the number
The first MZM is driven by the NRZ generator whose output of spans N [11], as shown as follows:
is connected to the second MZM, which is driven by a sine
" #
wave signal generator with a frequency of 40 GHz and 2. PaDC.Rd . R R− 1
  
phase of −90o. The resultant bandwidth of the duobinary QdB ðN Þ = 20.Log vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , (1)
u 2N.R2d ðFG − 1Þ.hv. 2P
 a
.B
u
e
signal is half the bandwidth of the NRZ format [8]. Figure 1 u DC
t Rsens 2 qffiffiffiffiffiffiffiffiffiffiffiffiffi
Rsens 2
shows the principle of duobinary modulation [9]. + 7 + 7

where Pa is the power difference between the power in

the 1s and 0s; Rd is the responsivity of the photodiode;
3 Wavelength-division multiplexer DC is the duty cycle of the signal for an NRZ signal, the
(WDM) system duty cycle is unity; R is the extinction ratio; F is the
amplifier noise figure; G is the amplifier gain; hv is the
As briefly described in the Introduction, WDM system photon energy; Rsens is the sensitivity of the receiver,
transports a number of high-capacity signals from the which is directly proportional to the thermal noise of
input port (transmitter) to the output port (receiver) via the PIN diode. Thus, the Q-factor will enhance by 3 dB
single optical channel (fiber). The number of channels to when the launched power is increased by 3 dB; however,
be used in WDM depends on the optical bandwidth in the high launched power increases the effect of nonlinearity.
transmission link, the spacing between the channels, and The best model that can emulate the actual system is to
the optical bandwidth required by each channel. The last double stage the optical amplifier [9].
two factors indicate the spectral efficiency S, which is
0.8 bits/s/Hz for 40 Gb/s and 50 GHz channel spacing.
This efficiency is extremely high; thus, duobinary mod-
ulation is preferable to allow a practical implementation 4 Optical fiber channel and
of the proposed work [9].
dispersion compensation
The number of channels of WDM is limited by the
EDFA bandwidth and channel frequency spacing. “Actual schemes by DCF
WDM standards have defined a frequency grid to place
the channels at a fixed frequency spacing of 200 GHz, By using the residual dispersion, the dispersion of DCF
100 GHz, and 50 GHz, allowing for 16, 32, and 64 general value can be estimated over a wide range of wavelengths.
cases of the maximum channel count, respectively” [10]. The residual dispersion is measured after the DCF at the
The span length is another factor that degrades the receiver. The total residual dispersion (Dres) of the link is
output gain. The gain is directly dependent on the span given by [12].
loss, which is a function of the accumulated noise of the
optical amplifier; thus, the Q-factor is directly related to Dres = DTF LTF + DDCF LDCF , (2)

Figure 1: Schematic of the principle of optical duobinary, building blocks, and electrical-to-optical conversion in MZM [9].

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s 3

where DTF is the dispersion of the total fiber, and LTF is its in Figure 2. Post-compensation scheme is when the DCF is
length. In our work, zero dispersion is required after the placed after the SMF (Figure 2(a)), whereas the pre-com-
DCF (LDCF). Thus, pensation scheme is when the DCF is placed before the
SMF (Figure 2(b)). These two schemes are investigated in
LDCF = − DTF LTF =DDCF . (3)
this study. The third scheme, symmetrical scheme, is when
The condition for simultaneous dispersion and dis- the DCF is placed before and after the SMF (Figure 2(c)).
persion slope compensation is expressed as

SDCF =DDCF = STF =DTF . (4)

5 Dispersion compensation
To achieve perfect dispersion compensation, the rela-
tive dispersion of both fibers should be the same. The
with FBG
relative dispersion is the ratio of dispersion slope (S) to
The compensating fiber component has some grating with
dispersion (D).
different refractive indices, which provide different velocities
RDS = S=D (5) for various modes of light signals to compensate the delay
between two different modes [14]. The Bragg wavelength
The common type of dispersion in single-mode opti-
varies along the grating length such that different frequency
cal fiber is the chromatic dispersion, which can be
components of the incident beam are reflected [13].
expressed as [13]:
λB = 2 n
Λ (9)
1 d2 β 2π c ps
 
d  
D= =  − 2 β2 units : =km ,
dλ vg dλ dω λ nm The grating dispersion Dg is expressed as
(6) 2n
Dg = , (10)
where λ is the operating wavelength, vg is the group c ðΔλÞ
velocity, ω is the angular frequency, and β is the propa-
where n is the average mode index, Δλ is the grating
gation constant of a mode. If the order of the propagation
bandwidth, and c is the velocity of light. The compensa-
is m, then
tion fiber has a high negative dispersion compared with
DCF. In the first assumption, a fixed effective refractive
 m 
d β
βm = ðm = 0, 1, 2 . . .Þ. (7)
dωm ω = ωo index was considered; the bandwidth is optimized to
obtain the best results.
The chromatic dispersion slope can be expressed as
follows:

dD 2π c 2π c
 

dλ
= 3 2β2 −
λ
β3 . (8) 6 Simulation setup
λ
Three different schemes are available for dispersion com- OptiSystem 7.0 simulator provides an environment iden-
pensation by reposition of DCF in the channel, as shown tical to actual optical communication systems. The

Figure 2: DCF schemes: (a) post-com-

pensation; (b) pre-compensation; and
(c) symmetrical.

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
4 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s

solution of the nonlinear Schrodinger equation utilizing Table 1: Simulation parameters of optisystem 7.0.
the split-step Fourier method represents the basis of this
simulation. The proposed 32-channel DWDM system Name Value Units
includes a transmitter, optical fiber, and an optical recei- Simulation window Set bit rate
ver, as depicted in Figure 3. In this work, the central Reference bit rate
frequency of the input signal is changed from 193.1 THz Bit rate     Bits/s
to 194.65 THz with a step of 50 GHz. The simulation para- Time window .e- S
Sample rate .e +  Hz
meters are specified in Table 1.
Sequence length  Bits
Samples per bit 
Numbers of samples ,
6.1 Transmitter section

Duobinary transmitter is a three-level modulation. compare the two methods of dispersion compensation for
Duobinary generator is fed by 40 GB/s, 27 − 1 pseudo- duobinary modulation. However, 120 km of SMF is used
random bit sequence generator and binary NRZ modula- in FBG schemes.
tion. Two LiNb MZMs with an extinction ratio of 30 dB are
used. The first MZM modulates the CW laser input signal
amplitude, whereas the second MZM modulates the phase
over the entire input signal by a sine wave signal genera- 6.2.1 Pre-compensation DCF
tor with a phase of −90°. The simulation setup of the
duobinary transmitter is shown in Figure 4, and the resul- At 10 km, DCF, with the specifications shown in Table 2,
tant signal at 0 dBm input power is shown in Figure 5. is connected to the SMF of 50 km. An optical amplifier
with a gain of 5 dB is utilized to compensate the attenua-
tion losses. The last is followed by another optical ampli-
6.2 Fiber section fier of 11 dB gain, whereas the noise figure of both
amplifiers is the same and equal to 4 dB. Figure 6(a)
At 100-km, SMF is connected to the duobinary transmitter shows the simulation of the pre-compensation connec-
via a span. The specifications are shown in Table 2. The tions. The lengths and dispersion values are calculated to
dispersion of SMF is compensated by post- and pre-DCF obtain a zero total dispersion. The optical fiber channel is
in the first approach and then by post- and pre-FBG to connected to the receiver with a span of N = 2 to obtain a

Figure 3: Schematic of simulation

setup: (a) pre-compensation
and (b) post-compensation schemes.

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s 5

Figure 4: Simulation arrangement of duobinary modulation format.

total length of 120 km, which is an extremely long haul

according to [15]. Values in this work indicate traditional
synchronous digital hierarchy target distances.

6.2.2 Post-compensation DCF

The same connection in the previous pre-compensation

scheme is used. The only difference is replacing the DCF
with SMF and the optical amplifier with each other, as
shown in Figure 6(b).

6.2.3 Dispersion compensation with FBG

In the second approach, an ideal dispersion compensation

with FBG is used to cancel the dispersion. The used FBG in
the post-compensation scheme with a bandwidth of 0.9 nm
and a negative dispersion of −1945 ps/nm is connected to a
120 km SMF with an optical amplifier has 30 dB gain in
between and directly connected to the receiver, as shown
Figure 5: Duobinary output signal.
in Figure 7(a). While in the pre-compensation scheme, the

Table 2: Specifications of SMF and DCF.

Fiber Attenuation Dispersion Dispersion slope Effective core Nonlinear refractive index
(dB/km) (ps/km·nm) (ps/km·nm) area (µm) (n) × − (cm/W)

SMF .  .  .

DCF . − −.  

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
6 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s

Figure 6: Simulation connection

of DCF in (a) pre-compensation
and (b) post-compensation
schemes.

Figure 7: Simulation connection

of FBG in (a) pre-compensation
and (b) post-compensation
schemes.

duobinary transmitter is directly connected to the channel detected by a PIN detector with responsivity of 1 A/W. A low
via an ideal dispersion compensation with FBG, which has pass filter with a cutoff frequency of 60 GHz is used to filter
a bandwidth of 1.75 nm and a negative dispersion of the detected signal. Finally, a 3R generator is combined with
1925 ps/nm followed by 120 km SMF. An optical amplifier the BER analyzer to record the Q-factor and BER. Figure 8
of 12 dB gain is used to amplify the modulated signal before shows the receiver section.
the receiver, as shown in Figure 7(b).

6.2.4 Receiving section 7 Simulation results

and discussion
The received signal is filtered by utilizing fourth-order Bessel
optical filter with an optimized bandwidth of 47 GHz, which Thirty-two transmitters with duobinary modulation are
is added after the demultiplexer. The received signal is then connected by DWDM to study the performance of systems

Figure 8: Receiver section.

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s 7

that use DCF and FBG in dispersion compensation. The signal power leads to an increase in the nonlinear effects,
system performance is examined for 40 Gb/s with chan- and the signal-to-noise ratio is degraded. Thus, at a high
nel spacing of 50 GHz, in which the input frequency input signal power of 15 dBm, the minimum Q-factor.
range varies from 193.1 GHz to 194.65 GHz. A second- The performance of the 32-channel duobinary sys-
order optical Gaussian filter with a bandwidth of 50 GHz tem for pre-compensation and post-compensation
is used with each transmitter before connecting it with schemes with FBG is shown in Figure 10(a) and 10(b),
DWDM to eliminate crosstalk. The input signal power is respectively. The pre-compensation scheme with FBG
swept from −20 dBm to 15 dBm with a step of 5 dBm exhibits a better Q-factor compared with the post-com-
to evaluate the influence of the input signal power on pensation scheme with FBG at an input signal power
system performance. Figure 9(a) and 9(b) shows the per- ranging from −20 dBm to 10 dBm. By contrast, Q-factor
formance of the 32-channel duobinary system in terms of is degraded to 4.7 at a high input signal power of
Q-factor as a function of time (bit period) for pre-com- 15 dBm for the pre-compensation scheme with FBG,
pensation and post-compensation schemes with DCF, whereas it recorded a higher value of 7.6 for the post-
respectively. The input signal power ranging from compensation scheme with FBG at the same input sig-
−20 dBm to 15 dBm with a step of 5 dBm is examined. nal power.
The results show almost the same peak Q-factor value Figure 11 shows the comparison among the four
of 10 at small input signal power for both schemes. For schemes. The results show the Q-factor values versus
the pre-compensation scheme with DCF, the Q-factor the input signal power ranging from −20 dBm to
value is degraded earlier compared to the post-compen- 15 dBm. Although the post-compensation scheme
sation scheme with DCF with an increasing input signal with FBG shows the lower Q-factor compared with
of more than 5 dBm. However, the increment of the input the other schemes, the pre-compensation scheme with

Figure 9: Q-factor versus time (bit period) for (a) pre-compensation scheme with DCF and (b) post-compensation scheme with DCF.

Figure 10: Q-factor versus time (bit period) for: (a) pre-compensation scheme with FBG and (b) post-compensation scheme with FBG.

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
8 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s

BER to −5, degraded the Q-factor to 4.7, and worsened the

eye shape.
In order to investigate both of the BER and the
Q-factor for the pre-compensation scheme, the input
power is fixed at −5 dBm and the number of loops is
increased from 1 to 6 with a step of 1 loop as illu-
strated in Figure 13. The increasing in the number of
loops represented increasing in the cascading of fiber
length, amplifier stage and dispersion compensating
stage. The results show that the Q-factor is degraded
from 11.3 to 10.4 and the BER got worse from −26.69 to
−25.34 as the number of loops increased from 1 to 6
loops.
Figure 11: Q-factor versus input signal power for four schemes.

FBG exhibits the best Q-factor among all other

schemes at input signal power interval from −20 dBm
to 10 dBm. The post-compensation scheme with FBG
demonstrates a stable Q-factor at low and high input
signal power.
From the results, the pre-compensation scheme with
FBG clearly represents the optimum scheme among the
others. The minimum log of BER and Q-factor versus the
input signal power are shown in Figure 12, as well as the
eye diagram illustrated at different signal powers. The
minimum log of BER and Q-factor exhibit an acceptable
value of −29 and 11.5, respectively, with input signal Figure 13: Min. log. of BER and Q-factor versus no. of loops for pre-
power ranging from −20 dBm to 5 dBm. According to the compensation with FBG at P = −5 dBm.
results, good dispersion compensation is achieved in
terms of signal quality and eye shape; in addition, the
edge neat graph is symmetrical. Moreover, the increment 8 Conclusion
of input signal to 15 dBm increased the minimum log of
In this study, 32 channels of 40 Gb/s DWDM system with
50 GHz spacing over a transmission distance of 120 km
were simulated using duobinary RZ modulation format
in the presence of nonlinear refractive index (n2). The
input frequency of the first channel was set to 193.1 THz.
The power of the CW laser was swept from −20 dBm to
15 dBm with a step of 5 dBm. Two different dispersion
compensation schemes were utilized, namely with DCF
and FBG. The pre- and post-compensation schemes with
DCF and FBG were compared in terms of Q-factor and
BER. The pre-compensation scheme with FBG exhibited
the best Q-factor among all other schemes at an input
signal power interval from −20 dBm to 10 dBm. Thus,
FBG is preferable because it is cost effective and
requires less number of optical amplifiers compared
with DCF, in addition to its ease of fabrication accord-
Figure 12: Pre-compensation scheme with FBG and input signal ing to the required specification of users. Thus, FBG
power. could be used instead of DCF.

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM
 S. R. Tahhan et al.: Dispersion Compensation for 32 Channels at 40 Gb/s 9

References linearity with under-compensated dispersion. In: ICP 2012-3rd

International Conference on Photonics, 2012: 210–14.
8. Sharan L, Shanbhag AG, Chaubey VK. Design and simulation of
1. Agrawal G. Fiber-optic communication systems, 3rd ed., Vol. 6.
modified duobinary modulated 40 Gbps 32 channel DWDM
New York: John Wiley & Sons, Inc., 2002.
optical link for improved non-linear performance. Cogent Eng
2. Kaur M, Sarangal H. Simulative investigation of 32 × 10, 32 × 20
[Internet]. 2016;3(1):1–12. Available from: http://dx.doi.org/10.
and 32 × 40 Gb/s DWDM systems with dispersion compensating
1080/23311916.2016.1256562
fibers. Int J Signal Process Pattern Recognit. 2015;8(8):127–34.
9. Azcoaga AB. 40 Gb/s optical transmission systems. Ph. D.
3. Tripathi DK, Singh P, Shukla NK, Dixit HK Performance study in
Thesis, Technical University of Denmark, 2003.
dispersion compensation techniques with Duobinary format at
10. ITU-T Recommendation. Optical interfaces for multichannel
different bit rates. In: ICPCES 2nd International Conference on
systems with optical amplifiers. Vol. G 692, 1998.
Power, Control and Embedded Systems. 2012: 1–5.
11. Videcrantz CJ. Optical amplification and processing in high capa-
4. Gupta MK, Agarwal J, Dhingra A, Singh G. Performance
city photonic networks. Ph. D. Thesis, Technical University of
analysis and optimization of 40 Gbps transmission system
Denmark, 2009.
over 4000 km with FBG. In: Proceedings of the International
12. Wandel M, Kristensen P. Fiber designs for high figure of merit
Conference on Recent Cognizance in Wireless
and high slope dispersion compensating fibers. J Opt Fiber
Communication & Image Processing. New Delhi: Springer,
Commun Reports. 2006;3(1):25–60.
2016: 759–65.
13. Gnanagurunathan G, Rahman FA. Comparing FBG and DCF as
5. Abass AK, Abdul-Razak MJ, Salih MA. Gain characteristics for
dispersion in the long haul narrowband WDM systems.
C-band erbium doped fiber amplifier utilizing single and dou-
Wireless and Optical Communications Networks (WOCN), IFIP
ble-pass configurations: A comparative study. Eng Tech J.
International Conference 2006: 4 pp.-pp. 4.
2014;32(9):2165–73.
14. Zaman HU, Salam KMA A comparative study of different dis-
6. Sharma A, Singh S, Sharma B. Investigations on dispersion
persion compensation techniques in long haul communication.
compensation using fiber braggs grating. Int J Comput Appl.
In: 1st International Conference on Computer & Information
2013;73(2):34–43.
Engineering, 2015: 26–27.
7. Sharan L, Chaubey VK Design and simulation of long-haul
15. ITU-T Recommendation G.691. Optical interfaces for single channel
32 × 40 Gb/s Duobinary DWDM link in the presence of non-
STM-64 and other SDH systems with optical amplifiers, 2006.

Brought to you by | La Trobe University

Authenticated
Download Date | 1/7/18 4:34 PM