[ieee 2012 ninth international conference on wireless and optical communications networks - (wocn) -...
TRANSCRIPT
Comparison of Metaheuristic Approaches for
Impairment Aware Transparent Optical Networks
Urmila Bhanja1, Debajyoti Mishra
2
1Assistant Professor,
2Senior Lecturer.
Dept. of Electronics and Communication Engineering, Indira Gandhi Institute of Technology, Sarang, Orissa,
759146, India.
E-mail: [email protected], [email protected] Abstract- In this paper two metaheuristic approaches are
compared while solving the dynamic routing and wavelength
assignment (DRWA) problem in wavelength division multiplexing
(WDM) optical network. In this problem, the effect of Four Wave
Mixing (FWM) and amplifier spontaneous emission (ASE) noise
are incorporated while solving the problem. The two
metaheuristic approaches used in this work are evolutionary
programming algorithm (EP) and simulated annealing (SA).
These metaheuristic techniques use a common objective function
that is designed to reduce the effect of FWM noise and ASE noise
on a requested lightpath. The FWM crosstalk products and the
static FWM noise power per link are pre computed in order to
reduce the set up time of a requested lightpath, and stored in an
offline database. These are retrieved during the setting up of a
lightpath and evaluated online taking the dynamic parameters
like cost of the links into consideration.
Keywords-Evolutionary programming algorithm; Simulated
annealing; Impairment aware DRWA problem; Four wave mixing;
ASE noise.
I. INTRODUCTION
An important feature of optical networks is their capability to
deliver data with a low bit error rate (BER) over long distances.
However during the routing of signals, as the transmission
distance increases the optical signal undergoes various physical
impairments that include noise generated in optical amplifiers,
nonlinear crosstalk appearing in WDM or dense wavelength
division multiplexing (DWDM) systems due to the fiber
nonlinear effects like cross phase modulation (CPM) and four
wave mixing (FWM), inter symbol interference because of
fiber chromatic dispersion (CD) and polarized mode dispersion
(PMD), etc. These impairments affect the optical networks and
their effects increase with an increase in the propagation
distance. The quality of signals in an optical network, which is
measured by the BER, is therefore dependent on the network
state. Unlike linear impairments, non-linear impairments affect
not only each optical channel individually, but they also cause
disturbance and interference among them. One such
impairment is FWM, in which signals at different wavelengths
interact resulting in generation of new signals called the FWM
components. Some of these FWM components interfere with
the original signals and degrade their quality [13]. The number
of FWM components generated increases with the increase in
the number of users. FWM severely degrades the network
performance if the input power is large and/or the channel
spacing is narrow [13]. Although, nonlinear impairments like
self phase modulation (SPM) and cross phase modulation
(XPM) that come into play only at high data rates, are
considered by some of the authors, the effect of four wave
mixing (FWM), which may degrade the signal quality in a
WDM or a DWDM system even at moderate powers and bit
rates, has received very little attention while addressing the
dynamic routing and wavelength assignment (DRWA)
problem. However, very few papers consider nonlinear effect
like FWM on a DRWA problem [4, 5, 8, 9, 14].
As reported by Azodolmolky et al., there are very few papers
available that use metaheuristics to solve the impairment aware
DRWA problem [2]. Tan et al. have reported an FWM aware
wavelength assignment using Ant colony optimization (ACO)
[14]. They have tried to correlate the input signal power with
FWM crosstalk power. Marsden et al. have proposed a DRWA
algorithm assuming that the quality of service (QoS) of
lightpaths are degraded mainly due to the FWM generated
noise [8]. The same authors have also reported a mechanism to
reduce the set up time for a lightpath in this FWM aware
DRWA algorithm by pre computing the FWM crosstalk power
[9]. However in this work, the authors have assumed the link
lengths to be the same. The main objective of the work
reported in this paper is to compare two FWM aware
metaheuristic approaches to investigate the wavelength routed
optical network design for a non ideal physical layer. The
proposed algorithms such as evolutionary programming
algorithm (EP) [4, 5] and simulated annealing approach
consider the ASE noise and the crosstalk components due to
the FWM effect while minimizing the connection blocking
probability. Simulated annealing (SA) has been used
extensively for modeling many of the optimization problems in
optical network. However, the potential of SA for solving an
impairment aware DRWA problem is not explored yet in the
literature. Therefore, in this work FWM aware SA for the
DRWA problem is developed and compared with EP [4, 5] in
terms of network performance metrics such as mean blocking
probability and mean execution time. Unlike an earlier
approach reported in [8, 9], this work is focused on a network
with variable link lengths, which makes the problem more
complex. In order to reduce the set up time of a lightpath, the
FWM crosstalk products and the static partial powers of all the
FWM terms are calculated offline and are stored in a database.
During the online evaluation of signal quality, the FWM
crosstalk power for the corresponding links are retrieved from
978-1-4673-1989-8/12/$31.00 ©2012 IEEE
the database and then the quality of signal is evaluated
dynamically taking the link length into consideration [4, 5]. In
the proposed approach, the signal quality measured in terms of
BER, is also evaluated, in addition to finding a wavelength
continuous path. If the BER is found to be below a threshold
value then the call request is accepted.
II. FOUR WAVE MIXING CROSSTALK IN A MULTICHANNEL
SYSTEM
FWM is one of the nonlinear effects in optical fiber, in which
three signals of different wavelengths interact to generate a
fourth wavelength that can interfere with the information signal
and result in performance degradation [8, 13].
III. SYSTEM AND PROBLEM MODEL
A. Problem definition
For the impairment aware DRWA problem the lightpath
requests are assumed to arrive at the network dynamically
according to a Poisson process with an average arrival rate of
Λ. A lightpath request is specified by three attributes: S, D, and
Th, which respectively represent the source node, the
destination node, and the holding time, for the request. The
source and the destination for each request are uniformly
randomly distributed. The holding time for a lightpath request
are assumed to be exponentially distributed with fixed mean,
T_hold=E[Th]. During this holding time duration, the network
resources are reserved for the connection request. Once the
holding time expires, the network resources are released to be
used later by other incoming lightpath requests. The network
load is defined as: Network load= Λ ×T_hold, the mean
blocking probability, and the mean execution time, are defined
as (Number of requests blocked ⁄Total number of requests
processed), (Total simulation time⁄ Total number of requests)
respectively [6].
B. The network model
It is assumed that the reduced version of NSFNET used to
illustrate the proposed algorithm can be modeled as a graph
G(V, E), where V is the set of nodes, representing N
wavelength routing nodes (WRNs), and E is the set of fiber
links, representing physical connectivity between the nodes.
Fig.1 shows the network model assumed in this work [10]. The
components present in a wavelength routing node (WRN)
include a cross connect switch (XCS), optical power taps for
monitoring signal, and a pair of EDFAs on either side of the
XCS for signal amplifications.
In the proposed model, the WRN also contains a transmitter
array and a receiver array that helps in adding or dropping a
local signal at any of the wavelengths at the node [10]. The
WRN’s are connected through non-zero dispersion shifted
optical fibers (NZDSF). It is assumed that there are no in-line
amplifiers in the network. In this work, the effect of signal
leaks in the optical cross connect switches and the effect of
non-ideal filtering at the demultiplexers, are neglected. In this
model the effect of chromatic dispersion (CD) and polarized
mode dispersion (PMD) are neglected.
Fig.1. Architecture of a wavelength routing node (WRN) [10]
.
C. The routing and wavelength assignment model
The routing and wavelength models assumed in this work are
identical to that used in [1, 3, 4, 5, 12].
D. Online signal power and noise power evaluation module
The calculation of received signal power, FWM crosstalk
power, and ASE noise power along a lightpath during a call
admission step is dynamic in nature. The signal power and the
ASE noise power depend on the number of links traversed by
the lightpath during the call admission phase and the associated
link cost. The FWM crosstalk power depends on the number
of different signals present in a link and the length of the
associated links that a lightpath traverses during the call
admission phase. The signal power, the ASE noise power at the
output of the kth intermediate node, and the FWM crosstalk
power for the multichannel system are expressed for the xth
individual as in [8, 10].
In a multichannel system, each lightpath traverses H hops or
links until it reaches its destination node. The accumulated
FWM crosstalk power at the destination node, PDNx(fr+fl-fm), is
the sum of all the crosstalk components generated in the links
traversed by the lightpath for the xth individual or solution and
is given by
1
( ) ( , , )r l
H
DNx r l m rlmx r l m
c f f fm
P f f f P f f f=
+ − =∑∑∑∑ , (1)
The FWM crosstalk power per link at the kth node for the xth
individual due to the three co-propagating signals at
wavelengths , ,r l mλ λ λ is given by [8,9],
2 2 ( 1, ) 2 2( , , , ) ( ( ) ( ) ( ) ( ) ( ))/9.0r L k k
rlmx r l m rlm r l m eff in tap mx dm sw outP k D PPPe L G k L L k L k L kG kαλ λ λ η γ − −=
(2)
The effective length Leff for the kth node is expressed as
-αL(K -1,K)
eff
(1- e )L =
α (3)
In the above equations, L(k-1,k) denotes the length of the link
between the (k-1)th node and the kth node. ( , )sigx i
p k λ
represents the signal power of kth node at wavelength λi,
( 1, )sigx i
p k λ− represents the signal power of (k-1)th node at
wavelength λi, ( 1, )fx
L k k− represents the fiber loss between
the (k-1)th and the kth node, and Gin(k) and Gout(k)
respectively represent the gain of the EDFAs at the input and
output of the kth node for any wavelength, ( )dmL k represents
the demultiplexer loss at the kth node, ( )swL k represents the
switch loss at the kth node, ( )mxL k represents the multiplexer
loss at the kth node, tap
L represents the tap loss,
( , )asex ip k λ represents the ASE noise power of kth node at
wavelength λi, ( 1, )asex ip k λ− represents the ASE noise power
of (k-1)th node at wavelength λi, h is Planck’s constant, iν is
the optical frequency at iλ , spη represents the spontaneous
emission factor for the EDFAs, Bo represents the optical
bandwidth, and Drlm represents the degeneracy factor in the
presence of the frequencies fr, fl, and fm.
E. Online bit error rate evaluation model
Optical signal received at the destination node in the presence
of ASE noise and FWM crosstalk power can be expressed as
( ) ( ) ( )( ) cos(2 ( )) ( ) ( )R x i ase x fwm x
E t A t t E t E tπν φ= + + + (4)
The first term in (4) represents the signal component at
frequency νi for the xth individual, A is the signal amplitude,
and φ(t) is its phase. The second and the third terms
respectively represent the received ASE noise power and the
FWM crosstalk power at the receiver node for the xth
individual. The photodetector is a square law device and hence
the received lightwave after photodetection produces a
photocurrent given by,
1 1 1
1
( ) 1( ) ( ) ( ) ( ) ( )
( ) ( )
( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
d x x sig ase x sig fwm x ase ase x fwm fwm x
fwm ase x th shot x
i t i t i t i t i t i t
i t i t i t
− − − −
−
= + + + +
+ + +
(5)
Equation (5) represents the signal current received for bit ‘1’
after photo detection, along with the beat noise components.
Equations (6) and (7) below represent the corresponding noise
variances.
The first term in (5) represents the signal component and the
rest of the terms represent the beat noise components for the
xth individual. The last two terms represent the thermal noise
current and the shot noise current respectively. The combined
noise can be modeled as a zero mean Gaussian random process
with a variance given by,
1 1
2 2 2 2 21
2 2
( ) ( ) ( ) ( ) ( )
( )
sig ase ase ase sig fwm fwm ase
thermal shot
x x x x x
x
σ σ σ σ σ
σ σ
− − − −= + + +
+ +
(6)
In (6), σ2
sig-ase(x) represents signal-ASE beat noise components,
σ2ase-ase(x) denotes ASE-ASE beat noise, σ
2sig-fwm1(x) denotes
signal-FWM beat noise, σ2
fwm1-ase(x) represents FWM-ASE
beat noise in the presence of signal, σ2
thermal represents thermal
noise variance, σ2
shot(x) represents shot noise variance for the
xth individual or solution and are expressed as in [7,13]. Equation (7) below represents the variance due to the beat
noise components at the receiver when bit ‘0’ is received by
the photodetector.
0
2 2 2 2 2
0( ) ( ) ( ) ( )ase ase ase fwm thermal shotx x x xσ σ σ σ σ− −= + + + (7)
In (7), σ2ase-fwm0(x) represents ASE-FWM beat noise variance
for bit ‘0’ for xth individual. These are expressed as in [7, 13]. The receiver BER due to the xth individual can be expressed
as,
1( ) ( ) ( )
1 0
( )0.25[ ( ) ( )]
2 ( ) 2 ( )
x th x th x
x
i t i ierfc erfc
x xξ
σ σ
−= +
(8)
The receiver BER is evaluated with a fixed decision threshold
ith(x). By suitably selecting the threshold, one can minimize the
BER. In this work, the threshold value is fixed at i1(x)(t)/2 [10].
IV. THE PROPOSED ALGORITHMS
This section briefly explains the mechanism of the proposed
evolutionary programming algorithm (EP) [3, 4, 5] and
simulated annealing (SA).
A. Chromosome representation
A chromosome represents a route or a path encoded from
source to destination as described in [3, 4, 5].
B. Population initialization and fitness calculation
The population that consists of a single individual is generated
randomly so as to satisfy the given constraints within the
threshold time of T1.
The fitness function for the problem is formulated as follows
[4, 5]:
1
2 2
( ), ( 1)
1 ( , )
2 2 2 2
1 1
[ ( ) ( )
( ) ( ) ( )]
xk
x x gx j gx j ijx shot sig ase
j i j E
ase ase thermal sig fwm fwm ase
f W C H x x
x x x
σ σ
σ σ σ σ
−
+ −= ∈
− − −
= + + +
+ + + +
∑ ∑
(9)
In the above equation, fx represents the fitness value of the xth
chromosome or path. In this work, the objective is to minimize
the fitness function. In (9) the first term represents the total cost
of a path, the second term represents the total number of hops
in the path, and the rest of the terms represent different noise
variances for the lightpath. Wx represents the free wavelength
factor for the lightpath, xk is the length of the xth chromosome
or path, and ( )x j
g and ( 1)x j
g + respectively represent the gene
of the jth locus and the (j+1)th locus of the xth chromosome
Wx, The free wavelength factor, is set to one if the wavelength
continuity constraint is satisfied for all the links of a
chromosome; else, it is set to 0.
C. Mutation
The mutation operation used in the FWM aware DRWA
algorithm is identical to that described in [3, 4, 5] and the
threshold time is fixed at T2 during this process. The threshold
times T1 and T2 attempt to reduce the execution time of the
algorithm during initialization and the mutation process.
D. Wavelength assignment algorithms
The wavelength assignment techniques investigated in this
work are the Random technique, and an FWM aware priority
based wavelength assignment technique [15, 4].
V. THE PROPOSED SIMULATED ANNEALING APPROACH
A. Initialization
The initial temperature Temp is set to 10,000 and the cooling
rate ∆ is set to 0.95.
B. Chromosome generation
A chromosome is generated as in [3, 4].
C. Energy function calculation
It is identical to the fitness function of EP.
D. Selection
In this step the difference between the energy functions
corresponding to the two chromosomes are found out. If the
second chromosome or solution is better than the first, then the
second one is selected. If it is worse than the first solution, the
solution is accepted with a probability of exp(-∆E/Temp). Here
∆E represents the difference in energy functions corresponding
to the two solutions and Temp is the temperature.
E. Termination
Steps are repeatedly executed till the final temperature Temp
becomes 0.000001. With each iteration, the temperature is
updated as Temp = ∆*Temp.
VI. THE ALGORITHM IMPLEMENTATION
The proposed algorithms are simulated using Microsoft Visual
C++ on an Intel Core i3 processor (2.4 GHz clock and 3 GB
RAM). The reduced version of 14-node NSF network topology
is used and the cost values are assigned as reported in [6]. The
threshold times T1 and T2 are fixed at 0.5 sec and 1.5 sec,
respectively. The hop count bound, h0, is kept at 4 [3]. The hop
count bound tries to minimize the ASE noise and FWM
crosstalk noise as the lightpath now traverses limited number
of hops and hence, minimizing the mean blocking probability.
VII. COMPUTATIONS OF CROSSTALK PRODUCTS AND POWERS
A.Offline
The FWM crosstalk power is calculated partially for all the
possible combinations of active lightpaths and stored in an
offline database. As soon as a request arrives, a wavelength
continuous path is found and then on each link in the path, the
wavelengths of the co-propagating signals are identified. The
partial crosstalk power corresponding to these wavelengths is
then retrieved from the offline database and is used in the
online evaluation of the lightpath [4].
Equations (10) and (11) below show the FWM power and the
phase mismatching factor for three different wavelengths at the
kth node respectively.
2 2 2( , , , ) ( ( ) ( ) ( ) ( ) ( ))/9.0rlmx r l m rlm r l m in tap mx dm sw outP k D PPPG k L L k L k L k G kλ λ λ γ= (10)
2 2
0 00 0
2( )( )( ( )( )[( ) ( )])
2
cr m l m c r l
dDf f f f D f f f f
c c d
πλ λβ
λ= − − + − + − (11)
B. Online
In the proposed work, a simple search technique is used for the
faster retrieval of data from the offline database. At any
instant, the corresponding partial FWM power and the phase
mismatching factor for the existing combination of wavelengths
in a link are retrieved from the offline database. The FWM
efficiency, the FWM crosstalk power per link, and the total
accumulated crosstalk power for the lightpath, are computed
online as expressed below in (12), (13) and (14) for any request.
It is assumed that the total number of hops a lightpath traverses
is denoted by ‘H’ [4].
The system parameters and the values used in the model are as
referred in [7, 13].
2 2
2 2 2
4 sin ( . / 2)1
(1 )
L
L
e L
e
α
α
α βη
α β
−
−
= +
+ − (12)
2 2 ( 1, ) 2 2( , , , ) ( ( ) ( ) ( ) ( ) ( ))/(9.0)L k k
rlmx r l m rlm r l m eff in tap mx dm sw outP k D PPPe L G k L L k L k L kG kαλ λ λ η γ − −=
(13)
1
( ) ( , , )r l
H
DNx r l m rlmx r l m
c f f fm
P f f f P f f f=
+ − =∑∑∑∑ (14)
VIII. RESULTS AND DISCUSSIONS
This section compares the simulation results exhibited by EP
and SA. As the proposed algorithms are randomized
approaches, each of the experiments are run ten times and the
average is estimated to plot the results. The results are also
analyzed statistically to estimate the margin of error for each of
the experiments. The proposed FWM aware assignment
technique tries to minimize the effect of FWM crosstalk. In this
work, Random and FWM aware priority based wavelength
assignment techniques are considered as these two techniques
are found to give better network performance compared to
other assignment techniques [4]. Fig.2 depicts the mean
blocking probability for Random wavelength assignment
technique and for FWM aware wavelength assignment
technique for a total of 50,000 requests. As expected the FWM
aware priority based wavelength assignment technique
provides the lowest mean blocking probability for both the type
of algorithms EP and SA. The mean blocking probability of the
order of 10-2
even with the estimated margin of error of value
±9.59527519 x 10-4
and ±7.17498651 x 10-2
for 95% of time
obtained by EP and SA respectively when integrated with the
FWM aware priority based wavelength assignment technique
and are found to be better compared to that of heuristic [8] and
metaheuristic approaches [14] at network load of 90 Erlang.
50 60 70 80 90 100 11010
-3
10-2
10-1
100
Network Load (Erlangs)
Me
an
Blo
ck
ing
Pro
ba
bil
ity
Random EP
Random SA
FWM aware EP
FWM aware SA
Fig.2. Mean blocking probability for different network loads
Fig.3 depicts the mean execution time for Random wavelength
assignment technique and for FWM aware wavelength
assignment technique for a total of 50,000 requests. It is
observed from the experiment that FWM aware wavelength
assignment technique for both EP and SA can be used for real
time application for a network load less than 90 Erlang [11]. The margin of error for EP and SA was estimated and found to
be ±0.0098998108 and ±0.0076812385 respectively for a
network load of 80 Erlang for 95% of time. Fig.4 below depicts
the fitness convergence curve for the FWM aware EP and SA. The estimated margin of error was found to be ±0.001568204
and ±0.0093206151 for the EP and SA respectively at the point
of convergence for 95% of time.
50 60 70 80 90 100 1100
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Network Load (Erlangs)
Me
an
Ex
ec
uti
on
tim
e (
se
co
nd
)
FWM aware EP
FWM aware SA
Random EP
Random SA
Fig.3. Mean execution time for different network loads
1 2 3 4 5 6 7 8
102.82
102.83
102.84
102.85
102.86
102.87
Generations
Avera
ge f
itn
ess s
co
re
FWM aware EP
FWM aware SA
Fig.4. The fitness convergence curve
IX. CONCLUSION
Two different metaheuristic approaches are compared in terms
of network performance metrics such as mean blocking
probability and mean execution time. The FWM aware EP is
found to give better results compared to the FWM aware SA.
REFERENCES
[1] Ahn, C. W. and Ramakrishna, R.S., “A Genetic Algorithm for
Shortest Path Routing Problem and the Sizing of Populations,”
IEEE Transactions on Evolutionary Computation, vol.6, no.6,
2002, pp.566-579.
[2] Azodolmolky, S. et al. “A survey on physical layer impairments
aware routing and wavelength assignment algorithms in optical
networks,” Computer Networks, vol.53, no.7, 2009, pp.926- 944.
[3] Bhanja, U., Roy, R., Mahapatra, S., “An evolutionary
programming algorithm for finding constrained optimal disjoint
paths for multihop communication networks,” International Journal
of Metaheuristics, vol.1, no.2, 2010, pp.132-155.
[4] Bhanja, U., Mahapatra, S., Roy, R., FWM aware evolutionary
programming algorithm for transparent optical networks, Photonic
Network Communications, vol.23, no.3, 2012, pp.285-299.
[5] Bhanja, U., Roy, R., “Impairment aware evolutionary
programming algorithm for transparent optical networks,” IEEE
Conference on ICBEIA, 2011, Kuala Lumpur, Malaysia.
[6] Bisbal, D. et al. “Dynamic Routing and Wavelength Assignment in
Optical Networks by Means of Genetic Algorithms,” Photonic
Network Communications, vol.7, no.1, 2004, pp.43-58.
[7] Kaur, G., Singh, M.L., Patterh, M.S., “Theoretical investigations of
the combined effect of fiber nonlinearities, amplifier and receiver
noise in a DWDM transmission system,” Journal of Optical and
Fiber Communications Research, vol.6, no.1-10, 2009, pp.1-10.
[8] Marsden, A., Maruta, A., Kitayama, K., “Routing and Wavelength
Assignment Encompassing FWM in WDM Lightpath Networks,”
ONDM, 2008, pp.128-133.
[9] Marsden, A., Maruta, A., Kitayama, K., “Reducing the lightpath
establishing time of FWM-aware dynamic RWA for wavelength-
routed optical networks,” Photonic Network Communications,
vol.18, 2009, pp.183-190.
[10] Ramamurthy,B.,Datta,D.,Feng,H.,Heritage,J.P.,Mukherjee,B.,
“Impact of transmission impairments on the teletraffic performance
of wavelength-routed optical Networks,” Journal of Lightwave
Technology, vol.17, no.10, 1999, pp.1713-1723.
[11] Ramjee, R., Murakami, K., Buskens, R.W., Lin, Y-J., La porta,
T.F., Design,Implementation, and Performance of a Cluster Mobile
Switching Center, Wireless Multimedia Network Technologies, R.
Ganesh, K. Pahlavan, Z. Zvonar Ed., The International Series in
Engineering and Computer Science, Springer, USA, 2006.
[12] Sahu, P. P., “A new shared protection scheme in optical network,”
Current Science, vol.91, no.9, 2006, pp. 1176-1183.
[13] Singh, S. P., Kar, S., Jain, V.K., “Effect of four wave mixing on
optimal placement of optical amplifier in WDM star networks,”
Fiber Integrated Optics, 2006, pp.111- 140.
[14] Tan, S.C., Abbou., F.M., Ewe, H.T., Four wave mixing aware
wavelength assignment using ant-based algorithm,” Journal of
Applied Sciences,vol.7, no.23,2007, pp.3796-3800.
[15] Zang, H., Jue, J.P. and Mukherjee, B., “A Review of Routing and
Wavelength Assignment Approaches for Wavelength-Routed
Optical WDM Networks,” optical Network Magazine, vol.1, no.1, 2000, pp.47-60.