See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/259181285

Suppression of Four Wave Mixing Crosstalk in Four 10 Gbps Channels Over a

50 km Dispersion Shifted Fiber Using Unequal Channel Spacing

Conference Paper · October 2009

CITATIONS READS

8 69

2 authors:

Bassem K. Abd Elrazak Moustafa H. Aly

Arab Academy for Science, Technology & Maritime Transport Arab Academy for Science, Technology & Maritime Transport
2 PUBLICATIONS   10 CITATIONS    263 PUBLICATIONS   395 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optical Communications View project

Optical logic gates View project

All content following this page was uploaded by Moustafa H. Aly on 07 January 2014.

The user has requested enhancement of the downloaded file.

 Suppression of Four Wave Mixing Crosstalk in Four
10 Gbps Channels Over a 50 km Dispersion Shifted Fiber
Using Unequal Channel Spacing
Bassem K. Abd ElRazak (enbassem@yahoo.com) and
Moustafa H. Aly, Member OSA (drmosaly@gmail.com)

College of Engineering and Technology,

Arab Academy for Science, Technology and Maritime Transport, Alexandria, Egypt.

Abstract— Four wave mixing (FWM) is a nonlinear effect in showed that it is possible to reduce FWM wave generation
optical fibers employing high bit rate channels over long distance using different unequal channel spacing algorithms in WDM
dispersion shifted fibers (DSFs) in wavelength division systems [4]-[7].
multiplexing (WDM) systems. This paper presents a study of the
effectiveness of using unequal channel spacing to suppress four-
In this paper, a comparison of the transmission of four 10-
wave mixing crosstalk in four channels 10 Gbps per channel
WDM system. The study is conducted using VPI Transmission
Gbps channels over 50 km of repeaterless DSF with both
Maker simulator. equally and unequally channel spaced schemes is reported.
The results show that using an equal channel spacing scheme
Index Terms— Four-wave mixing, dispersion shifted fibers, with minimum frequency spacing even as small as 50 GHz
channel spacing, and wavelength division multiplexing systems. will result in greatly reducing the generation of FWM
products at the channel frequencies and improving the overall
I. INTRODUCTION performance of the system.
In WDM systems, the need of transmitting many channels
over high speed/long-haul systems can be limited by variety
II. THEORY OF FWM
of impairments including the four-wave mixing (FWM)
crosstalk [1]. The need to reducing channel spacing, In a multi-channel system, the signal channel suffers from
increasing the bit rate, and the use of dispersion-shifted fibers FWM, which generates various combinations of different
(DSFs) near the zero-dispersion wavelength region as the channel frequencies and causes crosstalk degradation. In such
transmission media, all results in the reduction of the phase- systems, at any particular channel frequency, there will be a
mismatch naturally provided by the fiber dispersion, which number of FWM waves generated from various combinations
enhances the efficiency of generating FWM waves [2]. In of interacting signals whose frequencies satisfy [9]
WDM systems, the transmission of more than 2 channels
results in the generation of these FWM waves [3]. The more f FWM = f i + f j − f k (1)
channels being transmitted the more mixing products will
occur. In an equally channel spaced WDM system, many of
where, i, j, and k Є [1, N] and k ≠ {i, j}.
these mixing products generated by FWM fall at the channel
frequencies [3]. This nonlinear effect is considered the main
Considering all possible permutations, N co-propagating
non-linear impairment in optical fibers and it severely
optical signals will give rise to M new optical signals as
degrades the overall WDM system performance. For this
M=N2(N-1)/2.Some of these new frequencies fall onto the N
reason, the need to eliminating such effect has been the
original channels, while others are found in other new
driving force for researchers over many years to find methods
frequency locations. Those FWM signals, which overlap with
to help in the reduction and elimination of this major
the original ones, are considered as crosstalk and will
problem.
interfere with the normal operation of the WDM channels. A
well-known formula used for FWM estimation was originally
Different methods like unequal channel spacing, dispersion
derived in [9] and was later reformulated to include the
management, and using different modulation techniques have
phase-matching dependent efficiency in [2]. This formula,
been proposed as a solution of this impairment [3]-[7].
which has been widely used in recent years to evaluate the
Unequal channel spacing is an appealing method in terms of
FWM induced crosstalk in WDM systems, can be written as
eliminating the generation of FWM products at the channel
[2]
frequencies for WDM systems [4]-[8]. Previous experiments

ICCTA 2009, 17-19 October 2009, Alexandria, Egypt 228

 2 allow an adequate amount of rejection of the undesired
1024π 6  Dx1111Leff 
p ijk = 4 2 2   pi p j pk e −αLη (2)
n λ c  Aeff 
 required Bun (optical bandwidth for unequal spacing) can be
where Pijk is the power of the lightwave resulting from FWM
at the frequency f ijk. pi, pj, and pk, represent, respectively, the
input power of the frequencies: fi, fj, and fk. n is the fiber
refractive index, Aeff is the effective mode area of the fiber, D
is the degeneracy factor (D=3 for i=j, D=6 for i≠j ), x1111 is
the third-order nonlinear susceptibility and α is the fiber loss
coefficient, Leff is the effective length defined as
Leff = (1 – e-αL)/α. The FWM efficiency parameter, η, is a
function of phase mismatch, ∆β and is given by [2]
α2  4e −αL sin 2 (∆βL / 2) 
η= 1 +  (3)
α 2 + ∆β 2   channels) and n (minimum number of slots between
(1 − e−αL ) 2  channels), there are several optimum solutions [5], [8].
∆β may be expressed in terms of signal frequency differences
as [2]
2πλ2 λ2 dDc
∆β = β ijk + β k − β j − β i = ∆f 2 ( Dc + ∆f ) (4)
c 2c dλ
where Dc is the fiber dispersion, and ∆f is the channel
spacing. It is clear that, for sufficiently low fiber chromatic
dispersion, ∆β ≈ 0, and the efficiency η ≈ 1, which will result
in a high FWM crosstalk and ultimately degrading the system
performance.
III. ALGORITHM
During the fifties, a design methodology had been
proposed to reduce the effect of third order intermodulation
interference in radio systems. It had been shown that the use
of this design is possible to be applied in optical fibers to
transmit signal channels in WDM systems and reducing FWM
degradation and ultimately improving the overall performance
[4]-[7]. The algorithm used depends on reducing the channel
allocation design to an integer linear programming (ILP)
problem of finding optimum Golomb rulers (OGR), by
dividing the available optical bandwidth into equal slots of
bandwidth, ∆f, large enough to avoid overlap between spectra
in adjacent slots. In terms of slot numbers, Eq. (1) becomes
nijk = ni + n j − nk k ≠ i, j (5)
If nijk does not coincide with any of the channel slot numbers
for any choice of i, j, k, no FWM wave generated by the
signals through the nonlinear effect is created in any of the
channel slots. An important condition in designing the
unequal channel spacing scheme is that minimum frequency
spacing ∆fc = n·∆f between the channels must be provided to
ICCTA 2009, 17-19 October 2009, Alexandria, Egypt
channels. A lower bound to the total optical bandwidth
found from the condition
 
 −1 
N
Bun ≥ 1 + 2  Beq (6)
 n 
 
 
where Beq = (N – 1)·∆f is the total optical bandwidth of a
conventional WDM system with the channels equally spaced
by ∆fc.
Generally, for any value of N (number of transmitted
One important factor that must be considered is the
bandwidth expansion. When the number of channels N is
small, the minimum number of slots n can be chosen small as
well. This leads to a small affordable expansion in the
bandwidth.
IV. MODEL AND SIMULATION
In this paper, the suggested model is used to experiment the
effectiveness of the above algorithm in transmitting four
10 Gbps channels over 50 km repeaterless DSF, both with
equal and unequal channel spacing with a minimum channel
spacing ∆fc = 50 GHz to compare their performance. Slot
width ∆f is 25 GHz.
In order to illustrate the effectiveness of the suggested
system in suppressing FWM waves, the system is simulated
with the experimental setup block diagram shown in Fig. 1. It
consists of 4 channels 2 mW each, modulated by Mach-
Zehnder modulator. The modulator is driven by a 10 Gbps
NRZ 231 - 1 pseudorandom bit sequence generator. A 50 km
of near-zero DSF with a fiber loss of 0.2 dB/km and a
dispersion of 0.2 ps/nm·km is used. The zero dispersion
frequency fo = 193.1 THz and is located at the center between
channels 2 and 3 to provide a worst case.
Figure 1 Block diagram of the used system.
229
Figure 2(a) shows the assigned frequencies of the four shows the improvement in the eye diagram with the unequal
channels with the equal channel separations of 50 GHz. The channel spacing. This is due to the generation of FWM
channels are assigned the frequencies 193.025, 193.075, products outside the channel bandwidth and in the spaces
193.125, and 193.175 THz. The total equal bandwidth is between the channels but with just some depletion due to the
150 GHz. Figure 2(b) shows the assigned frequencies with the power transferred to the mixing products generated in other
unequal channel spacing. The minimum channel spacing is slots.
∆fc = 50 GHz. The channels are assigned the frequencies
192.9625, 193.0625, 193.1375, and 193.1875 THz. The Equal spacing Unequal spacing
channel separations from left to right are 100, 75 and 50 GHz,
respectively. The total unequal bandwidth is 225 GHz. The
minimum number of slots between any two channels is n = 2.
It is clear that the bandwidth expansion is 1.5. This means
that unequal channel spacing bandwidth will be 75 GHz
greater than the equal channel spacing bandwidth, which is an
affordable price compared to the expensive new transmitters
and receivers when a different modulation technique is used
instead, especially when the system is employing a relatively
small number of channels.
(a) (b)
Figure 3 Comparison of the fiber output spectrum for (a) equal
channel spacing and (b) unequal channel spacing. Both at the output of
50 km DSF (D=0 ps/nm.km) for four 10 Gbps channels with 2 mW
launched power each.

Equal spacing Unequal spacing

Figure 2 Schematic of the assigned frequencies for the four

transmitted channels with (a) equal channel spacing and (b) unequal
channel spacing.

Figure 3 shows a comparison of the measured spectrum at the

output of the 50 km DSF with a zero dispersion frequency
fo = 193.1 THz when four 2-mW channels are launched at the
input of the fiber for both equal channel spacing (channel
spacing = 50 GHz) shown in Fig. 3(a), and unequal channel
spacing shown in Fig. 3(b). It can be seen that for unequal
channel spacing, the FWM waves are generated in the spaces
between the channels and outside the received bandwidth.
Moreover, the power of the FWM waves is greatly depleted
in the unequal channel spacing compared to the equal
spacing.
The eye diagram of the received four channels is shown in
Fig. 4. It shows a degradation in the eye diagram and
complete spread out of the “1” level with the equal channel
spacing shown in Fig. 4(a) due to the FWM crosstalk caused
by the interference between the transmitted channels and the
generated FWM waves at the channel frequency. Figure 4(b)
(a) (b)
Figure 4 Eye diagram of the four channels after 50 km transmission for
(a) equal channel spacing (degraded) and (b) unequal channel spacing
(improved).
Figure 5 displays the power depletion of the generated FWM
products as they travel along the optical fiber. The lines are
the measured FWM products outside the channel frequency
bandwidth. The line with circles (labeled signal) is the
measured power at channel 3 at frequencies fo + 0.5∆fc and
fo + 0.75∆fc, for equal and unequal channel spacing,
respectively. The line with crosses (labeled highest) is the

ICCTA 2009, 17-19 October 2009, Alexandria, Egypt 230

FWM products measured at fo + 3.5∆fc. The line with
triangles (labeled upper) is the FWM products measured at
fo + 2.5∆fc. The line with rhombuses (labeled lower) is the
FWM products measured at fo - 3.5∆fc. The line with squares
(labeled lowest) is the FWM products measured at fo - 4.5∆fc.
∆fc = 50 GHz is the minimum channel spacing.

Comparing the FWM products generated with equal channel

spacing, Fig. 5(a), and that generated with unequal channel
spacing, Fig. 5(b), a noticeable depletion is found in the lower
and the upper FWM products. Also, an increase in the power
of the FWM products is noticed in the lowest and highest
FWM products. This is due to the equal distribution of the
FWM products in unequal channel spacing in the slots
between the transmitted channels and outside the channels
bandwidth unlike equal channel spacing where FWM
products are generated at the channel frequency, resulting in a
high level of interference and crosstalk. The line with circles
is the output power in dB versus transmission distance
measured at channel 3. (b)
Figure 5 FWM products versus transmission distance for (a) equal
channel spacing (b) unequal channel spacing.

(a) (b)
Figure 6 Fiber output spectrum for (a) equal channel spacing (b)
unequal channel spacing, both with channel 3 turned off.

From Fig. 6, it can be seen that for the equal spacing, there
are FWM waves generated at channel 3. On the other hand,
for the unequal spacing, channel 3 is free of FWM products
and the FWM products generated outside the received
(a) bandwidth are depleted.

In order to further investigate the effectiveness of the unequal
channel algorithm in generating all the mixing products
outside the channel slots, the simulation is repeated but with
channel 3 turned off, and the FWM products generated by the
other transmitted channels will be measured at channel 3.
This is displayed in Fig. 6
Figure 7 shows the FWM products versus the transmission
distance. It is clear that at channel 3, the FWM products (the
green line with small circles) are greatly depleted in the
unequal channel spacing. This means that for the equal
spacing with channel 3 turned off, the other transmitted
channels generate FWM products at channel 3 but for the
unequal spacing channel 3 is free of FWM products.

ICCTA 2009, 17-19 October 2009, Alexandria, Egypt 231


(a)
(b)
Figure 7 FWM products versus the transmission distance for (a)
equal channel spacing and (b) unequal channel spacing, both with
channel 3 turned off.
Figure 8 summarizes the obtained results. It shows a
comparison of the FWM products power measured at channel
3 for both equal and unequal channel spacing. The FWM
products power (due to the transmission of the other three
channels) at channel 3 at the fiber output is 22 dBm less than
equal spacing.
ICCTA 2009, 17-19 October 2009, Alexandria, Egypt
Figure 8 FWM products power measured at channel 3 for both equal
and unequal spacing for NRZ modulation.
It is worth noting that, the nonlinear refractive index, n2, in
DSFs is 2.6×10-20 m2/W, while in this part of the simulation
we used a worst value of 3.2×10-20 m2/W. So, one can expect
better results than the previously obtained results when n2 =
2.6×10-20 m2/W is used.
V. CONCLUSION
In this paper, a study of the effectiveness of using unequal
channel spacing to suppress the FWM crosstalk in four
channels 10 Gbps per channel WDM system with minimum
channel spacing reduced to 50 GHz is reported. It has been
demonstrated that the unequal channel spacing algorithm is
effective in suppressing FWM products and generating them
in the spaces between channels and outside the received
bandwidth, resulting in decreasing FWM crosstalk and
improving the overall system performance. In order to further
investigate the effectiveness of the unequal channel algorithm
in generating the mixing products outside the channel slots,
the simulation was repeated but with channel 3 turned off.
The results showed a reduction of 22 dBm in the FWM power
measured at channel 3 when unequal channel spacing is
employed. This method is effective with relatively small
number of channels being deployed in the fiber compared to
other methods but on an affordable expense of bandwidth
expansion.
VI. REFERENCES
[1] Andrew R. Chraplyvy, “Limitations on lightwave communications
imposed by optical-fiber nonlinearities,” J. Lightwave Technol., vol. 8,
no.10, pp. 1548-1557, 1990.
232
[2] N. Shibata, R. P. Braun, and R. G. Waarts, “Phase-mismatch
dependence of efficiency of wave generation through four-wave
mixing in a singlemode optical fiber,” IEEE J. Quantum Electron.,
vol. 7, pp. 1205-1210, 1987.
[3] R. W. Tkach, A. R. Chraplyvy, F. Forghieri, A. H. Gnauck, and R. M.
Derosier, “Four-photon mixing and high-speed WDM systems,”
J.Lightwave Technol., vol. 13, no. 5, pp. 841-849, 1995.
[4] Vrizlynn L. L. Thing, P. Shum, and M. K. Rao, “Channel allocation
algorithm for WDM systems,” Optics Express 1327, vol. 11, no. 11,
pp. 1322-1327, 2003.
[5] F. Forghieri, R. W. Tkach, and A. R. Chraplyvy, "WDM systems with
unequally spaced channels," J. Lightwave Technol., vol. 13, no. 5, pp.
889-897, 1995.
[6] M. Derosier, “Repeaterless transmission of 8 channels at 10 Gb/s over
137 km (11 Tb/s-km) of dispersion-shifted fiber using unequal
channel spacing,” IEEE Photon. Technol. Lett., vol. 6, no. 11, pp.
1374-1376, 1994.
[7] F. Forghieri, R. W. Tkach, A. R. Chraplyvy, and D. Marcuse,
“Reduction of four-wave mixing crosstalk in WDM systems using
unequally spaced channels,” IEEE Photon. Technol. Lett., vol. 6, no.
6, pp. 754-756, 1994.
[8] J. B. Shearer, “Some new optimum Golomb rulers,” IEEE Trans.
Inform.Theory, vol. IT-36, no. 1, pp. 183-184, 1990.
[9] K. 0. Hill, D. C. Johnson, B. S. Kawasaki, and R. I. MacDonald, “CW
three-wave mixing in single-mode optical fibers,” J. Appl. Phys., vol.
49, pp. 5098-5106, no. 10, 1978.

ICCTA 2009, 17-19 October 2009, Alexandria, Egypt 233

View publication stats