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

net/publication/323319404

Routing, spectrum, and core assignment in SDM-EONS with MCF: Node-arc

ILP/MILP methods and an efficient XT-aware heuristic algorithm

Article in Journal of Optical Communications and Networking · March 2018

DOI: 10.1364/JOCN.10.000195

CITATIONS READS

88 852

3 authors:

Mingcong Yang Yongbing Zhang

University of Tsukuba University of Tsukuba
19 PUBLICATIONS 250 CITATIONS 105 PUBLICATIONS 1,177 CITATIONS

SEE PROFILE SEE PROFILE

Qian Wu
Hosei University
13 PUBLICATIONS 165 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mingcong Yang on 24 October 2018.

The user has requested enhancement of the downloaded file.

Yang et al.
Routing, Spectrum, and Core Assignment
in SDM-EONs With MCF: Node-Arc
ILP/MILP Methods and an Efficient
XT-Aware Heuristic Algorithm
Mingcong Yang,
Abstract—In this paper, we focus on the static routing,
spectrum, and core assignment (RSCA) problem in space-
division multiplexing (SDM)-based elastic optical networks
(EONs) with multi-core fiber (MCF). In RSCA problems, it is
a challenging task to control the inter-core interference,
called inter-core crosstalk (XT), within an acceptable level
and simultaneously maximize the spectrum utilization. We
first consider XT in a worst interference scenario (i.e., XT-
unaware), which can simplify the RSCA problem. In this
scenario, we formulate the RSCA problem using a node-
arc-based integer linear programming (ILP) method in
which the numbers of both variables and constraints are
greatly reduced compared with previous ILP methods,
thereby leading to a significant improvement in conver-
gence efficiency. Then, we consider the XT strictly (i.e.,
XT-aware) and formulate the problem using a mixed integer
linear programming (MILP) method, which is an extension
of the above node-arc-based ILP method. It is more suitable
for different XT thresholds and/or geographically large net-
works, in that it has a higher degree of generalizability.
Finally, we propose an XT-aware-based heuristic algorithm.
The simulation results demonstrate that our heuristic algo-
rithm achieves higher spectrum efficiency, higher degree of
generalizability, and higher computational efficiency than
the existing heuristic algorithm(s).
Index Terms—ILP/MILP optimization model; Multi-core
fiber; Routing, spectrum and core assignment; Space
division multiplexing; XT-aware.
I. INTRODUCTION
W ith the development and increasing popularity of
cloud computing, video-on-demand (VoD), internet
of things (IoT), and other emerging Internet services, net-
work traffic is growing at an extremely rapid rate [1].
However, the growth in the transmission capacity of stan-
dard single-mode fiber (SMF) has dramatically slowed
because the transmission capacity per fiber is close to
the nonlinear Shannon limit of the existing SMF [2,3].
To satisfy the strongly increasing future traffic demand,
the space-division multiplexing (SDM) technique has
Manuscript received November 13, 2017; revised December 23, 2017;
accepted December 28, 2017; published February 21, 2018 (Doc. ID 313350).
The authors are with the Graduate School of Systems and Information
Engineering, University of Tsukuba, Tsukuba, Japan (e-mail:
ymc6642664@gmail.com).
https://doi.org/10.1364/JOCN.10.000195
1943-0620/18/030195-14
VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 195
Yongbing Zhang, and Qian Wu
recently been proposed, and it is considered to be a prom-
ising technology for overcoming the physical limitations
of SMF by using multi-core fiber (MCF), multi-element fi-
ber (MEF), multi-/few-mode fiber (MMF/FMF), and their
combinations [4]. However, some new challenges, such as
network planning and resource optimization, should be ad-
dressed. A typical problem of resource assignment is the
routing, spectrum, and core assignment (RSCA) problem
in space-division multiplexing elastic optical networks
(SDM-EONs) with MCF. In contrast to the routing and
spectrum assignment (RSA) problem in EONs with SMF,
SDM-EONs with MCF have introduced some new features
that make the RSCA problem more challenging.
In the traditional WDM systems with SMF, several paral-
lel WDM systems can increase system capacity, but they also
increase system cost and energy consumption, leading to the
result that the cost and energy consumption remain the
same per bit [5]. However, to satisfy the potential increasing
demands for communications capacity, transmission equip-
ment is asked to reduce both energy and cost per bit.
Compared with the parallel WDM systems, component inte-
gration is the crucial feature of SDM systems, which can
significantly reduce system cost and energy consumption.
In SDM systems, integrated components [5,6] in the spatial
domain, including transponders, optical amplifiers, and net-
working elements such as optical switches, can share system
components among spatial channels to reduce both capital
(CAPEX) and operational (OPEX) expenditures.
Another feature of SDM-EONs with MCF is core switch-
ing. Three different SDM switching strategies have been
proposed [7–9]: 1) independent switching (InS), where all
frequency slots (FSs) and spatial modes can be independ-
ently directed to any output port; 2) joint switching (JoS),
where all spatial modes/cores are treated as a single entity,
while FSs can be freely switched by the wavelength-
selective switch; and 3) fractional JoS (FJoS), where a few
groups of spatial modes and all FSs can be independently
switched to all output ports. The core switching can miti-
gate the constraint of spectrum continuity compared to
RSA since there are multiple cores that can be used on a
fiber link. In other words, a lightpath from a source to
its destination can have a larger routing space. In this pa-
per, the InS strategy is considered, because it is the most
Journal © 2018 Optical Society of America
196 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018
flexible switching scheme and leads to better network per-
formance.
The third feature in SDM systems is the inter-core inter-
ference of MCF, called the inter-core crosstalk (XT), which
becomes a crucial constraint for SDM-EONs with MCF
that may severely affect the propagation quality. In this pa-
per, we only consider the XTs between two adjacent cores,
which is a general assumption for the RSCA problem [3,10–
12]. That is, the XT occurs only when the same FSs are
assigned to adjacent cores in a common fiber link to serve
different transmission requests. A simple example of XT in
a 7-core MCF is shown in Fig. 1. For simplicity, we assume
that the transmission requests are assigned to only three
cores (i.e., cores 1, 2, and 3) and that core 1 (or core 3) and
core 2 are adjacent. The strength of the XT on an FS f in a
core c is related to the number of active (i.e., the same f has
been assigned to other transmission requests) adjacent
cores of core c. For example, because the same FSs (i.e.,
the 1st and 2nd FSs) on the adjacent cores of core 1 (e.g.,
core 2) have not been assigned to other requests, there is no
occurrence of XT for R:1. For R:2 and R:6, because the same
6th FS has been assigned to R:4 and their cores are adja-
cent, there is an occurrence of weak XT for R:2 and R:6 on
the 6th FS. Moreover, because the XTs are affecting each
other on the adjacent cores, there is an occurrence of strong
XT for R:4 on the 6th FS because it is affected by R:2 and
R:6 at the same time. For a transmission request, only
when the XT on each assigned FS for the transmission re-
quest is less than an acceptable XT threshold, the signals
can be successfully transmitted on these assigned FSs.
To minimize the XT and achieve a dense core arrange-
ment to enable high-capacity and long-distance transmis-
sions, how to suppress the XT has become a primary focus
in MCF research [13–19]. Takenaga et al. developed a
trench-assisted multi-core fiber (TA-MCF) in [19]. In TA-
MCF, XT is directly related to the mode-coupling
coefficient, which can be estimated without the need for
numerical simulations.
According to the equations proposed in Ref. [20], the XT
between two adjacent cores c and c0 can be obtained by the
following equation:
Fig. 1. Example of inter-core crosstalk in a 7-core MCF.
Yang et al.
1 − exp−2hL
XTcc0  ; (1)
1  exp−2hL
2
where h  2k r
βωtr represents the mean XT increase per
unit length of the fiber and is also regarded as the
power-coupling coefficient in coupled-power theory [20].
The parameters k, r, β, and ωtr represent the coupling co-
efficient, bend radius, propagation constant, and core pitch,
respectively [17,21,22]. L represents the fiber length.
Equation (1) can also be obtained from the equations pro-
posed in Refs. [4,18,22] by setting the number of active ad-
jacent cores equal to 1 (i.e., for core c, only core c0 is active).
The other approaches to calculate the XT between two
adjacent cores can also be used in this paper.
Furthermore, Hayashi et al. [17] claimed that “The mean
XT from multiple cores to one core can be represented as a
sum of the mean crosstalk from each of the multiple cores
to the single core.” Therefore, on a given f th FS, we assume
that the XT of core c can be calculated as follows:
X
XTc  XTcc0 : (2)
c0 ∈Ac
In Eq. (2), the XT between two adjacent cores XTcc0 is cal-
culated using Eq. (1), where Ac represents the set of active
adjacent cores of core c; in other words, for each core c0 ∈ Ac ,
the same f th FS on core c0 has been assigned to serve the
other transmission requests. Generally, the parameters h,
m, and L are fixed for a given MCF. Thus, the XT of core c on
the f th FS depends only on Ac . Therefore, in addition to the
development of MCF on the physical layer, we can also
reduce the XT by avoiding spectrum overlaps between ad-
jacent cores, that is, reducing the number of elements in Ac .
A. Related Works
In contrast to the RSA problem with SMF, which should
satisfy the constraints of spectrum continuity, contiguity,
and non-overlapping, the XT constraint should be taken
into account in the RSCA problem with MCF when as-
signed the spectral resources, i.e., FSs. There are mainly
three approaches, as shown in Fig. 2, when considering
the XT constraint.
The first approach, called XT-avoid, is to avoid spectrum
overlaps between adjacent cores in the fiber links while al-
locating FSs and cores to different transmission requests
[23]. As shown in Fig. 2(a), the same FSs cannot be as-
signed to different transmission requests that traverse
through adjacent cores in the common links. The second
approach, called XT-WC, is to consider the XT in a worst
interference scenario [18,21]. As shown in Fig. 2(b), the
blank FSs are the empty FSs (i.e., have not been assigned
to any transmission request), but for the XT-WC approach,
they are considered to be occupied to evaluate the XT. The
third approach, called XT-aware, is to compute the XT
strictly depending on the interference of a core with other
active adjacent cores that share the same FS and link
[3,10–12], as shown in Fig. 2(c).
Both the XT-avoid and the XT-WC approaches can
simplify the RSCA problem and significantly reduce the
Yang et al.
Fig. 2. Examples of three approaches corresponding to the XT
constraint in a 7-core MCF. (a) XT-avoid approach. (b) XT-WC
approach. (c) XT-aware approach.
complexity of the problem. However, the XT-avoid approach
prevents the use of the same FSs on the adjacent cores of a
fiber. Thus, even though the XT-avoid approach can com-
pletely prevent the XT from occurring, the efficiency of
the spectrum multiplexing of MCF will be decreased.
Furthermore, in the XT-WC approach, all the adjacent
cores of a target core are treated as active cores.
Therefore, for a given XT threshold, the maximum trans-
mission reach will be bounded significantly, and it is
shorter than what the optical signal can actually reach.
This makes the XT-WC approach over-conservative since
the feasible lightpath (i.e., the lightpath whose length is
shorter than the bounded maximum transmission reach)
space may become significantly small. In an extreme case,
for a geographically large network, due to the limitation of
a strict XT threshold, it is possible that no feasible light-
path between a source-destination pair can be found by
the XT-WC approach even though such a lightpath actually
exists (i.e., the length of the shortest path between a node
pair still longer than the bounded transmission reach).
VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 197
Muhammad et al. [21] introduced an arc-path-based ILP
method [24] based on the XT-WC approach to solve the
static RSCA problem. For each transmission request, the
XTs for all the candidate paths (in a worst interference
scenario) between the source and destination are pre-
computed, and the candidate path whose XT is less than
the XT threshold can be selected to be the feasible light-
path for this transmission request. To improve the conver-
gence efficiency of the ILP method, they further bounded
the candidate paths into k-shortest paths. They also pro-
posed a heuristic algorithm called SPSA for the problem.
Zhang et al. [18] also introduced an arc-path-based ILP
method based on the XT-WC approach to solve the static
RSCA problem. In the pre-processing stage, they pre-
computed the maximum transmission reach of the light-
path (considering a worst interference scenario) based on
the given XT threshold. Then, in their arc-path-based
ILP method, they added a constraint to ensure that only
when the length of a pre-calculated candidate path is
shorter than the maximum transmission reach, the candi-
date path can be selected as a lightpath to serve the trans-
mission request. A heuristic algorithm called ARSCA-SP
was also proposed in their work.
In contrast, the XT-aware approach can always perform
better than both the XT-avoid and the XT-WC approaches,
and it leads to an optimal solution. It strictly calculates the
XT depending on the active adjacent cores that share the
same FS on a common link along the lightpath. Therefore,
for a strict XT threshold, the XT-aware approach can pre-
vent the same FSs from being assigned to adjacent cores to
ensure that the XT of the lightpath (for a given transmis-
sion request) is less than the XT threshold. For a relaxed
XT threshold, the same FSs on adjacent cores are allowed
to be multiplexed more times to achieve higher spectrum
efficiency. However, in the XT-aware approach, how to cal-
culate the XT is a key and challenging task. To the best of
our knowledge, no efficient method for XT calculation
considering the XT-aware approach has been proposed in
the previous literature for the static RSCA problem.
B. Key Contributions
In this paper, we focus on the static RSCA problem in
SDM-EONs with MCF. The key contributions of this work
are summarized as follows.
We first introduce a node-arc-based ILP optimization
model considering the XT based on the XT-WC approach.
In contrast to previous arc-path-based ILP methods
[18,21] that need to pre-compute several candidate paths
for each transmission request during the pre-processing
stage, our proposed ILP method can simultaneously find
an optimal combination of routing paths and assign corre-
sponding FSs and cores for each transmission request in a
node-arc manner [25,26]. In this way, the numbers of both
variables and constraints are less than the ones in the
previous methods, leading to much higher convergence
efficiency. Our simulation experiments show that our pro-
posed node-arc-based ILP method achieves a significantly
198 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018 Yang et al.

shorter execution time compared with the previous
arc-path-based ILP methods.
Second, we introduce a MILP optimization model consid-
ering the XT strictly based on the XT-aware approach. In
contrast to the XT-WC-based ILP methods, XT is calculated
depending on the actual interference between cores on a
common link and therefore ensures a complete feasible
routing path space for each transmission request; conse-
quently, a strictly optimal solution can be obtained by
our MILP optimization model.
Furthermore, the static RSCA problem is an extension of
the static RSA problem, which is known to be NP-hard [25].
This problem is difficult to solve in an acceptable execution
time by the ILP/MILP methods for large-scale problem
instances. For this reason, we propose an XT-aware-based
heuristic algorithm for achieving scalability. In our heuristic
algorithm, a strict XT-affected FS check mechanism is pro-
posed to achieve the XT-aware approach. The simulation ex-
periments show that our proposed heuristic algorithm can
obtain solutions that are quite close to the optimal solutions
compared to the previous heuristic algorithms. To the best of
our knowledge, our work is the first work that consider the
XT-aware approach for the static RSCA problem.
C. Organization of This Paper
The remainder of this paper is organized as follows. In
Section II, we present our node-arc-based ILP/MILP opti-
mization models for the static RSCA problem based on the
XT-WC and XT-aware approaches. In Section III, we
propose a novel heuristic algorithm based on the XT-aware
approach. In Section VI, we present the results and
evaluate the performance. Section V concludes the paper.
II. PROBLEM FORMULATION
R: Transmission request matrix, r  fsr; dr; λr g ∈ R,
where sr and dr are the starting and ending
nodes of r, respectively, and λr is the required
FSs of r.
Li : Set of links that start or end at node i.
xtecc0 : The XT between core c and core c0 on link e based on
Eq. (1) by setting L equal to the length of link e.
Θm : Inter-core crosstalk threshold of the given
modulation format m.
F: Set ofPtotal required FSs (in the worst case,
jFj  r∈R λr ).
M: A large number.
2) Variables:
r ∈ f0; 1g:
lec A binary variable that is equal to 1 if core c
on link e is selected to serve transmission
request r and is otherwise equal to 0.
nir ∈ f0; 1g: A binary variable that is equal to 1 if the
lightpath that is used to serve transmission
request r passes through node i and is
otherwise equal to 0.
αr ∈ Z : An integer variable that denotes the starting
FS index of the lightpath that is used to serve
transmission request r.
βr ∈ Z : An integer variable that denotes the ending
FS index of the lightpath that is used to serve
transmission request r.
0
φrr ∈ f0; 1g: A binary variable that takes a value of 0 if
the ending FS index βr0 of the lightpath that
is used to serve transmission request r0 is
smaller than the starting FS index αr of
the lightpath that is used to serve transmis-
sion request r.
F max ∈ Z : An integer variable that indicates the
number of maximum required FSs for the
network.

3) Objective Function:
In this section, we first present a node-arc-based ILP
optimization model based on the XT-WC approach. Then,
Minimize F max : (3)
we present an extended (node-arc-based) MILP optimiza-
tion model based on the XT-aware approach. Finally, we 4) Constraints:
compare our optimization models with the existing arc-
a) Cost Constraint:
path-based ILP optimization models. The spectral super-
channel policy [27,28], which is created by allocating βr ≤ F max ∀ r ∈ R: (4)
sub-channels in contiguous FSs on just one of the available
fibers/cores, is considered in this paper. Additionally, we as- The constraint in Eq. (4) ensures that the maximum FS
sume that the modulation format m used for transmission index F max is not smaller than the ending FS index of the
is given beforehand. lightpath for any transmission request r.
b) Route Selection Constraints:
A. Node-Arc-Based ILP Optimization Model Based X
r ≤ 1 ∀ r ∈ R; e ∈ E;
lec (5)
on XT-WC Approach c∈C

X X
1) Parameters:
r  1 ∀ r ∈ R;
lec (6)
c∈C e∈Lsr
V: Set of network nodes.
E: Set of network links. X X
C: Set of cores on each MCF. r 1
lec ∀ r ∈ R; (7)
Nc: Set of adjacent cores of core c. c∈C e∈Ldr
Yang et al. VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 199

XX B. Extended MILP Optimization Model Based on

r − 2nr  0 ∀ r ∈ R; i ∈ V : i ≠ sr; dr;
lec i (8) XT-Aware Approach
c∈C e∈Li

The strict XT-aware MILP model is an extension of the

X j
node-arc-based ILP model in Section II.A. The point is that
r ≤ nr  nr
2 lec ∀ r ∈ R; e : i; j ∈ E:
i
(9) we should calculate the XT on each assigned FS f for each
c∈C
transmission request r strictly. Here, we introduce some
The constraint in Eq. (5) ensures that there is at most additional variables and constraints, as follows.
one core that can be selected on a link e for any transmis- 1) Variables:
sion request r. The constraint in Eq. (6) ensures that for a
transmission request r, the first link of a lightpath starts oecf
r ∈ f0; 1g: A binary variable that is equal to 1 if the f th

from node sr. Thus, the sum must be equal to one. The FS of core c on link e is assigned to serve transmission
constraint in Eq. (7) ensures that for a transmission re- request r and is otherwise equal to 0.
quest r, the last link of a lightpath ends at node dr. tef
r : A continuous variable that indicates the XT on link e
Thus, the sum must be equal to one. The constraint in at the f th FS for transmission request r.
Eq. (8) ensures that for any intermediate node traversed
2) Constraints:
by a lightpath, there is one link that should end at the node
and one link that should start from the node. The con- X ecf
or − λr lec
r  0 ∀ r ∈ R; e ∈ E; c ∈ C; (14)
straint in Eq. (9) ensures that if the lightpath for a trans-
f ∈F
mission request r traverses a link e:i; j, it must also
traverse nodes i and j. X 
c) Spectrum Assignment Constraints: jFj oecf
r − 1 ≤ f − αr ∀ r ∈ R; e ∈ E; f ∈ F; (15)
c∈C
0
φrr  φrr0  1 ∀ r; r0 ∈ R (10) X 
jFj oecf
r − 1 ≤ βr − f ∀ r ∈ R; e ∈ E; f ∈ F: (16)
c∈C

βr0 − αr ≤ jFjφrr0  2 − lec

r − lr0  − 1
ec
(11) The constraints in Eqs. (14)–(16) ensure the relationship
∀ r; r0 ∈ R : r ≠ r0 ; e ∈ E; c ∈ C ecf
r , αr , β r , and the additional variable or . In other
between lec
words, when core c on link e is selected to serve transmis-
sion request r (lec r  1), the FSs between the starting
βr  αr  λr − 1 ∀ r ∈ R: (12) FS index αr and ending FS index βr should be assigned
to serve transmission request r on core c of link e
(oecf
r  1, f ∈ αr ; βr ).
The constraints in Eqs. (10) and (11) ensure that the FSs
of two lightpaths that share a common core on a link do not 3) Crosstalk Constraints
overlap. In other words, when two lightpaths share a core For the strict XT-aware approach, the XT constraint in
on a common link, the assigned FSs of one transmission Eq. (13) for the above node-arc-based ILP method is not
request can be either before or after the FSs of the other available here, and two new constraints are introduced:
transmission request. Thus, the spectrum non-overlapping
constraint is satisfied. The constraint in Eq. (12) ensures X X 0
xtecc0 orec0 f − tef ecf
r ≤ M · 1 − or 
that the ending FS index βr is equal to the starting FS in- c0 ∈N c r0 ∈R (17)
dex αr plus the number of required FSs minus one for each
transmission request r, which ensures the constraint of ∀ r ∈ R; e ∈ E; c ∈ C; f ∈ F;
spectrum contiguity. In addition, αr and βr are assigned
X ef
as belonging to each transmission request r, which tr − Θm ≤ 0 ∀ r ∈ R; f ∈ F: (18)
potentially ensures the spectrum continuity. e∈E

d) Crosstalk Constraints:
The constraint in Eq. (17) ensures that if the f th FS of
XX core c on link e is assigned to serve transmission request r
r ≤ Θm
xtecc0 · lec ∀ r ∈ R; c ∈ C: (13)
(oecf
r  1), then the XT of transmission request r on link e at
c0 ∈N c e∈E
the f th FS (tef r ) should be greater than the sum of the
XT due to the active adjacent cores on the common link
The constraint in Eq. (13) ensures that for each trans- e that share the same f [based on Eq. (2)]. For example,
mission request r, the XT of the lightpath that is selected when
P oecf
Pr  1, the constraint in Eq. (17) becomes
0
e ec f ef 0
by the route selection constraints in Eqs. (5)–(9) should be c0 ∈N c r0 ∈R xtcc0 or0 ≤ tr . If an active adjacent core c of
less than the XT threshold where, based on the XT-WC core c (in N c ) on the common link e is excited to serve an-
0
approach, the worst interference scenario is considered; other transmission request r0 (orec0 f  1), then the XT be-
that is, we set Ac [see Eq. (2)] equal to N c . tween these two cores (xtcc0 ) should be summed. Then, tef
e
r
200 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018
should be greater than the sum. In another case, when
oecf
r  0, the constraint in Eq. (17) is deactivated due to
the larger value jFj on the right-hand side of the constraint.
In other words, tef
r can be 0 in this case. The constraint in
Eq. (18) ensures that for each transmission request r, the
XT of the selected lightpath should be less than the XT
threshold on each assigned FS. Overall, according to the
constraints above, we have incorporated the XT-limit con-
straint into the MILP formulation based on the strict
XT-aware approach.
C. Analysis of our ILP/MILP Models
In this subsection, we summarize our proposed node-
arc-based ILP/MILP optimization models and compare
them with the previous arc-path-based ILP optimization
models [18,21]. The major differences are stated as follows.
First, our proposed ILP/MILP methods, which belong to
the node-arc category, decide the optimal routes and allo-
cate FSs and cores for all the transmission requests simul-
taneously, whereas the ILP methods in Refs. [18,21], which
belong to the arc-path category, only allocate FSs for the
lightpath services but do not find routes for the lightpath
services. As a result, several candidate paths between each
source–destination pair should be pre-calculated as the in-
put parameters of the ILP optimization models. If the XT
constraint is not taken into account, the optimal solutions
can be obtained using our node-arc-based ILP method. In
contrast, only when all the routing paths for each transmis-
sion request are considered is it guaranteed that the
optimal solutions can be obtained using the existing
arc-path-based ILP methods in Refs. [18,21]. If only a
few (e.g., the k-shortest) pre-calculated candidate paths
are considered for each transmission request, a very effi-
cient or even optimal solution may also be found, but we
cannot claim that the current solutions are the optimal sol-
utions if all the routing paths for each transmission request
are not considered in the optimization process.
Second, the previous ILP methods [18,21] consider the
XT in a worst interference scenario (XT-WC), and the fea-
sible lightpath space may become smaller (see Section I.A)
for a strict XT threshold, particularly in geographically
large networks. In an extreme case, some transmission re-
quests may not be satisfied, since in a worst interference
scenario, the XT may be greater than the XT threshold,
even for the shortest path of these transmission requests.
In contrast, our proposed MILP method considers the XT
strictly (i.e., XT-aware approach). Therefore, the optimal
solution can be obtained.
Next, we examine the convergence efficiency of the optimi-
zation models by counting the numbers of variables and con-
straints of the optimization models. Note that in some cases,
adding constraints in an optimization model will not reduce
but rather improve the convergence speed of the model.
However, the numbers of variables and constraints are still
a metric that is worth considering [25]. For all the optimiza-
tion models, we assume that the number of cores of a multi-
core fiber, denoted by jCj, is equal to an invariant constant.
Yang et al.
For our ILP optimization model (Section II.A), the dom-
inant numbers of variables and constraints are OjRj2 
0
(according to φrr ) and OjRj2 · jEj [according to the con-
straint in Eq. (11)], respectively, where jRj and jEj re-
present the total numbers of transmission requests and
links in the network, respectively.
For the ILP optimization model in Ref. [21], the number
of variables is bounded by OjRj2 · jEj, while the number of
constraints is bounded by the larger of OjRj2 · jEj and
OjRj · jPj, where jPj represents the number of pre-
calculated candidate routing paths per transmission
request. When all the possible routing paths are consid-
ered, the ILP method in Ref. [21] can safely ensure finding
the optimal solution, but an upper bound for jPj can be up
to O2jNj  [24]. Meanwhile, when only the k-shortest paths
are considered, an optimal solution may not be found, but
the jPj can be significantly reduced to k. For the ILP opti-
mization model in Ref. [18], the numbers of both variables
and constraints are bounded by OjRj2 · jPj2 .
Finally, for our MILP optimization model (Section II.B),
the dominant number of variables is the larger one be-
0
tween OjRj2  (according to φrr ) and OjRj · jEj · jFj (accord-
ecf
ing to or ), and the dominant number of constraints is the
larger one between OjRj2 · jEj [according to the constraint
in Eq. (11)] and OjRj · jEj · jFj [according to the constraint
in Eq. (17)], where jFj is the total number of required FSs,
which is the objective of the optimization model. Because it
is difficult to find an exact number for jFj before the opti-
mization model isP solved, we set it as its upper bound,
which is equal to r∈R λr , where λr is the traffic demand
units in FS for transmission request r. Here, we set λ as
the average number of traffic demand units in FS for
each transmission request; thus, we can derive the upper
bound for the dominant numbers of both variables and
constraints as OjRj2 · jEj · λ.
Overall, we can observe that our ILP optimization model
has considerably smaller numbers of variables and
constraints. However, in our MILP optimization model,
to achieve the XT-aware approach, we need to mark the sta-
tus of each FS on each core and link, which inevitably in-
troduces more variables and constraints. We summarize
the convergence efficiencies of the four optimization models
in Table I.
We also present the total numbers of columns (variables)
and rows (constraints) given by the Gurobi ILP solver [29]
when solving the optimization models for a six-node nine-
link simple network [21] [Fig. 4(a) in Section IV] in Table II,
where jRj is set to be 15 (a transmission request between
each node pair), and λr is set as a random number from 1–3
(FSs). The number of cores of a multi-core fiber jCj is set to
TABLE I
COMPARISON AMONG FOUR OPTIMIZATION MODELS
Model # Variables # Constraints
Our ILP model OjRj 2
OjRj2 · jEj
Our MILP model OjRj2 · jEj · λ OjRj2 · jEj · λ
ILP model in Ref. [21] OjRj2 · jEj OjRj2 · jEj or OjRj · jPj
ILP model in Ref. [18] OjRj2 · jPj2  OjRj2 · jPj2 
Yang et al. VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 201
8
TABLE II >
>
> ar
<
If FS f on core c of link e
COMPARISON AMONG FOUR OPTIMIZATION MODELS FOR THE N6S9
NETWORK (BY GUROBI SOLVER) Ue; c; f   has been assigned to request ar :
>
>
>
: None Otherwise
Model Columns Rows Non-Zeros
Our ILP model (Section II.A) 736 6,285 31,185
Our MILP model (Section II.B) 35,836 33,823 449,578 • A table of the current XT of previously assigned trans-
ILP Model in Ref. [21] (All paths) 15,623 23,239 95,490 mission requests ar on their assigned FSs, which is
ILP Model in Ref. [21] (jPj  5) 14,671 23,205 94,026 denoted as follows:
ILP Model in Ref. [21] (jPj  3) 13,831 23,175 92,796 8
ILP Model in Ref. [18] (All paths) 44,674 99,560 375,536 >
> The current XT of ar on FS f
<
ILP Model in Ref. [18] (jPj  5) 21,586 47,850 180,240
XTar; f   : If f is assigned to ar :
ILP Model in Ref. [18] (jPj  3) 8,026 17,550 65,820 >
>
:
None Otherwise

be 3. To decide the numbers of variables and constraints The pseudocode for our strict XT-affected FS check
of our MILP optimization model, we considered the mechanism is shown in Algorithm 1. The inputs of the al-
worst
P case for jFj; in other words, we set jFj equal to gorithm are the pre-assigned routing path p  ∪︀e; c con-
λ
r∈R . Three different cases of pre-calculated candidate
r taining links and cores along the path and the pre-assigned
paths per node pair P are considered, which represent FSs, that is, the starting FS f s and the ending FS f e. The
all the routing paths (for each transmission request) in output of the algorithm is a Boolean number, where 1 rep-
the n6s9 network topology, 5-shortest paths, and 3-shortest resents that the transmission request can be transmitted
paths. successfully on these pre-assigned FSs and 0 otherwise.
As shown in Table II, our proposed ILP optimization
model (Section II.A) has significantly fewer rows and Algorithm 1: Strict XT-Affected Frequency Slot Check
columns than previous ILP optimization models, even Mechanism
though we have bound the number of the pre-calculated Input: p  ∪︀e; c, f s , f e
candidate paths jPj equal to 3. These results are consistent Output: 1 or 0
with the analytical results shown in Table I. 1: for frequency slot f in range f s ; f e  do
2: Set XT self
f equal to 0 as the XT of the new transmis-
sion request itself on f
III. HEURISTIC ALGORITHM FOR RSCA PROBLEM 3: for (link e, core c) in routing path p do
4: for each adjacent core c0 of core c do
In this section, we present our XT-aware heuristic algo- 5: Set ar equals to Ue; c0 ; f , that is, the previously
rithm, which contains two components: a strict XT-affected assigned transmission request that occupies the
FS check mechanism and a RSCA component. f on core c0 of link e
6: if ar ≠ None then
7: The current XT of previously assigned
A. Strict XT-Affected Frequency Slot Check transmission request
Mechanism XT ar;f Δxtc;c0 ;e
XT self 0
8: f   Δxtc; c ; e
9: if XTar; f  > XT -threshold then
Our strict XT-aware FS check mechanism works as fol-
10: return 0
lows. When a new transmission request arrives, we should
11: end if
check the increased XTs of the previously assigned
12: end if
transmission requests to ensure that they are still less
13: end for
than the XT threshold, because the XTs of previously as-
14: end for
signed transmission requests will increase if the same
15: if XT self
f > XT -threshold then
FS(s) in an adjacent core are assigned to the new transmis-
16: return 0
sion request. Moreover, since XT between two adjacent
17: end if
cores is mutually affected, we should also ensure that 18: end for
the XTs on each assigned FS of the new transmission 19: return 1
request itself are less than the XT threshold, which are
related to the number of active adjacent cores on the same We present a simple example in Fig. 3. For simplicity, we
FS (used to serve the previously assigned transmission consider a simple chain network, where each link has a
requests). 3-core MCF. We assume that there are some FSs already
We use the following two tables in the strict XT check successfully assigned to other transmission requests ar
mechanism. (shown in green), that is, for each ar, the XTs on their as-
signed FSs (i.e., XTar; f ) are less than the XT threshold.
• A usage table represents the FS usage per core and link, For instance, we can observe that ar2 shares the same
which is denoted as follows: FSs (i.e., f 1 and f 2 ) with ar1 and ar3 on a common link
202 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018
e12 ; therefore, the XT of ar2 on f 1 (i.e., XTar2 ; f 1 ) can be
calculated by Eq. (2), which equals the sum of
Δxtc1 ; c2 ; e12  [see Eq. (1), where L equals the length of link
e12 ] and Δxtc2 ; c3 ; e12 . In the same way, XTar2 ; f 2  can also
be calculated as Δxtc1 ; c2 ; e12   Δxtc2 ; c3 ; e12 . As we have
mentioned above, XTar2 ; f 1  and XTar2 ; f 2  are assumed
to be less than the XT threshold to ensure a successful
transmission.
We now show how our XT-aware mechanism works.
Here, a new transmission request arrives with a require-
ment of four FSs from source node 1 to destination node 3.
We assume that the transmission request is pre-assigned
to core 2 starting from the f 3 FS on each link. We need to
check, for example, whether we can successfully assign the
FSs, shown in red, to the new transmission request.
First, we check the increased XTs of the previously as-
signed transmission requests due to the new transmission
request to ensure that they are still less than the XT
threshold (lines 4–10). For example, since the previously
assigned transmission request ar1 shares the same f 3 on
adjacent cores of the common link e12 with the new trans-
mission request, the XT of ar1 on the f 3 FS will increase by
Δxtc1 ; c2 ; e12 . Thus, we should ensure that the sum of
XTar1 ; f 3  (the current XT stored in the XT table) and
Δxtc1 ; c2 ; e12  is less than the XT threshold. Similarly,
we should ensure that the XT for ar4 (f 3 on common link
e23 ) is less than the XT threshold (line 15). Moreover, for
ar3, because it shares the same f 3 FS with the new trans-
mission request on two common links (i.e., e12 and e23 ), the
XT of ar3 will increase twice [by Δxtc2 ; c3 ; e12  and
Δxtc2 ; c3 ; e23 , respectively] and should also be less than
the XT threshold. We should repeat this process on each
pre-assigned FS of the new transmission request. If any
of the increased XTs of the previously assigned transmis-
sion requests are greater than the XT threshold, the pre-
assigned FSs and cores cannot be successfully assigned
to the new transmission request.
Then, we check the XT of the new transmission request
itself on each pre-assigned FS to ensure that it can be
Fig. 3. Example of our strict XT check mechanism.
Yang et al.
successfully transmitted. As can be seen in Fig. 3, the
XT of the new transmission request itself on the f 3 (XTself f3 )
is the sum of Δxtc1 ; c2 ; e12 , Δxtc2 ; c3 ; e12 , Δxtc1 ; c2 ; e23 ,
and Δxtc2 ; c3 ; e23 . Using the same approach, the XTs on
the f 4 , f 5 , and f 6 can be calculated to be Δxtc1 ; c2 ; e12 ,
0, and 0, respectively. Then, we should check these
XTs, and if any of them is greater than the XT threshold,
the new transmission request cannot be successfully
transmitted.
B. Routing, Spectrum, and Core Assignment
Algorithm 2: Routing, Spectrum, and Core Assignment
Input: Transmission request matrix
R (r  fsr; dr; λr g ∈ R)
1: Sort R in descending order by the required FSs λr
2: Calculate the k-shortest routing paths for each pair of
nodes based on the K-SP algorithm [30]
3: for each r in R do
4: Set the starting FS of r equal to 1 as f s
5: for each ksp in KSPr do
6: if FSs between f s ; f s  λr − 1 are available on at
least one core (the core is selected according to
the predefined reducing XT algorithm in
Ref. [31]) of each link along the ksp then
7: Check XTs based on Algorithm 1
8: if Algorithm 1 returns 1 then
9: Assign the FSs between f s ; f s  λr − 1, light-
path ksp and the selected cores to serve trans-
mission request r
10: Update usage table Ue; c; f  and XT ta-
ble XTr; f 
11: break (Move to the next transmission
request)
12: end if
13: end if
14: end for
15: f s  1
16: return to line 5
17: end for
For a given transmission request matrix R, we first sort
the transmission requests in descending order according to
the traffic demand and then determine the routing, the
spectrum, and the core assignment for the transmission re-
quests r ∈ R one by one, starting from the one with the
largest traffic demand. For two transmission requests with
the same demand, we first serve the one whose shortest
path has more hops because it is difficult to assign avail-
able consecutive FSs for a transmission request with more
hops due to the spectrum contiguity and continuity
constraints.
The algorithm works as follows. First, we pre-calculate
the k-shortest routing paths for each transmission request
r as a set of KSPr in a pre-processing phase based on the
K-SP algorithm [30]. Then, for each r in sorted R, the start-
ing FS f s will start from the first FS. For each ksp in KSPr ,
if the FSs between f s ; f s  λr − 1 are available on at least
one core (the core is selected according to the predefined
Yang et al.
reducing XT algorithm in Ref. [31]) of each link along
the ksp (i.e., satisfying the contiguity and continuity
constraints), check the XT-limit constraint based on
Algorithm 1. If the XT-limit constraint is also satisfied
(Algorithm 1 returns 1), assign the FSs between
f s ; f s  λr − 1, the routing path ksp and the cores to the
transmission request r. However, if all the ksp in KSPr
are not feasible on the current f s , we increase the starting
FS f s by one and repeat the above processes until there is at
least one ksp in KSPr that is feasible. Finally, we update
the usage table Ue; c; f  and XT table XTar; f  and move
to the next transmission request. The pseudocode of our
Routing, Spectrum, and Core Assignment Algorithm is
shown in Algorithm 2.
IV. SIMULATIONS AND PERFORMANCE EVALUATIONS
In this section, we evaluate the performance of our ILP/
MILP methods and our heuristic algorithm. Two types of
MCFs are considered in the simulation experiments, which
are a 3-core MCF (for small-scale problem instances) and a
7-core MCF [10,22] (for large-scale problem instances). The
parameters r, β, k, and ωtr of the MCFs are set as 50 mm,
4 × 106 m−1 , 4 × 10−4 , and 4.5 × 10−5 m, respectively
[10,21,22]. The simulation experiments were performed
on Microsoft Windows 10 using a computer with an Intel
Xeon 8-core 3.5 GHz CPU and 64 GB of memory. We used
the Gurobi Optimizer 7.0.1 [29] to solve the optimization
models, and for all the heuristic algorithms, we used the
same k-SP algorithm proposed in Ref. [30], where k is
set to be 3. Moreover, in the simulation experiments, a unit
FS (i.e., the grid granularity) is assumed to be 12.5 GHz
bandwidth [30,32], and two neighboring optical paths
are separated by guardband that occupies 12.5 GHz
bandwidth (i.e., 1 FS) [33].
A. Simulation Experiments of Small-Scale
Instances
We first evaluate the performance of our ILP/MILP
methods and our heuristic algorithm in small-scale
Fig. 4. (a) Six-node nine-link (n6s9) simple network [21]. (b) 14-node 21-link NSF network [34]. (c) 28-node 34-link EON network [35].
VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 203
instances (simple network topology, 3-core MCF, small
set of offered demand requests, and so forth).
1) Convergence Efficiency of our ILP/MILP Optimization
Models: We employ the existing ILP optimization models
proposed in Refs. [18,21] as benchmarks to evaluate the
convergence efficiency of our ILP/MILP optimization mod-
els. For our MILP optimization model, we set jFj to be an
upper bound as the sum of all the required bandwidth (FSs)
for all transmission requests.
For the existing two arc-path-based ILP optimization
models, we consider three different cases of pre-calculated
candidate paths, which are all the possible routing paths
(AP), 5-shortest paths (5SP) and 3-shortest paths (3SP).
Moreover, the two existing ILP methods in Refs. [18,21]
and our ILP method consider the XT in a worst interference
scenario. To ensure that the candidate path space will not
be bounded by the XT threshold (see Section II.C), we con-
sider a simple and geographically small six-node nine-link
(n6s9) network, as shown in Fig. 4(a). In the n6s9 network,
the XTs (worst case) of all the paths are less than the XT
threshold (set as −30 dB in this part of the simulation
experiments [18,21]).
According to the statements above, the solutions ob-
tained by our node-arc-based ILP/MILP methods and the
existing arc-path-based ILP methods (AP) can guarantee
the optimality. A fair comparison should be considered
among these four methods, but for reference, we also evalu-
ate the performance of the existing ILP methods in the
cases of 3SP and 5SP.
The number of transmission requests jRj is set to be 15
(a transmission request between each node pair). The traf-
fic demand ranges considered are from 1–3, 2–4, and to a
maximum of 5–7 units of FSs. For each of the traffic
demand ranges, we generate 10 different transmission re-
quest matrices R. The criterion for the convergence effi-
ciency measurement is the “runtimes” that are given by
AMPL/Gurobi [29].
Figure 5 shows the convergence efficiencies of the four
optimization models in the n6s9 network. The x axis shows
different ranges of traffic demands. The y axis shows the
execution time. The results are averaged by 10 different
transmission request matrices R.
204 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018
Fig. 5. Convergence efficiency comparison among the four optimi-
zation models for the n6s9 network.
As shown in Fig. 5, the execution time increases with in-
creasing traffic demand range for all the methods because
the feasible solution spaces of the optimization models
expand with the increase in required FSs, leading to
longer execution time. We can observe that our ILP
method can solve the problem in less than 1 s, and even
compared with the cases of 3SP and 5SP of the existing
ILP methods, it requires less execution time. However,
because our MILP method consider the XT strictly
(i.e., XT-aware), it takes more execution time. The
results are consistent with the theoretical analysis in
Tables I and II.
2) Generalizability of our MILP Method and Heuristic
Algorithm: To evaluate the generalizability of our MILP
method and our heuristic algorithm, we consider the pre-
vious n6s9 network and let the XT threshold decrease from
−30 dB to −40 dB (the different modulation formats [36]
and the XT margin [37], which can be set by the network
operators, will lead to different XT thresholds). The de-
crease of the XT threshold can also be viewed as the geo-
graphical size of the network becomes larger, but the XT
threshold remains the same, where the number of trans-
mission requests jRj is fixed as 15 (between each node pair)
and the required FSs are randomly selected between 1 and
5 FSs. For each XT threshold, we generated 10 different
transmission request matrices R, and the results are
averaged as shown in Fig. 6.
As shown in Fig. 6, there are no points for the two
existing XT-WC-based heuristic algorithms when the XT
threshold is reduced to −34 dB, because the two existing
XT-WC-based heuristic algorithms become unable to as-
sign the lightpaths to some transmission requests. That
is, in the worst interference scenario, the XTs of the short-
est path for these transmission requests become greater
than −34 dB (e.g., the transmission request between node
pair [1,6]); thus, these transmission requests will be
blocked. In contrast, our XT-aware-based MILP method
and heuristic algorithm can assign the FSs adaptively to
ensure successful transmissions by avoiding assigning
Yang et al.
Fig. 6. Generalizability of different approaches for the n6s9
simple network.
the same FSs to adjacent cores on some common links
along the lightpaths. Thus, as shown in Fig. 6, the required
FSs become large as the XT threshold decreases. We list
the number of blocked transmission requests by the
existing XT-WC-based heuristic algorithms for each XT
threshold in Table III as follows.
3) Spectrum Efficiency of our Heuristic Algorithm in
Small-Scale Problem Instances: In this paragraph, we
evaluate the spectrum efficiency achieved by our heuristic
algorithm in small-scale problem instances compared with
the heuristic algorithms proposed in the previous works
[18,21]. The solutions obtained based on our ILP optimiza-
tion model are set as the benchmarks. Because in geo-
graphically small networks, the solutions obtained based
on our ILP optimization model are also the most optimal
solution (see Section IV.A.1), but our ILP method requires
the least execution time. In this part, the XT threshold is
fixed to be −30 dB.
Figure 7 presents the results of the spectrum efficiency
of each approach in the n6s9 network, in which the x axis
shows the numbers of transmission requests jRj, and the
source node and destination node for all transmission re-
quests are randomly selected. The required FSs for each
transmission request are considered as a random number
from 1 to 5 (FSs). For each jRj, we generate 10 different
transmission request matrices R, and the results—
maximum required bandwidth (FSs) in the entire
network—are averaged, as shown on the y axis.
As shown in Fig. 7, in small-scale problem instances, our
proposed heuristic algorithm performs better than the two
existing heuristic algorithms, and the solutions are closer
to the optimal solutions.
TABLE III
NUMBER OF BLOCKED TRANSMISSION REQUESTS BY THE XT-WC-
BASED HEURISTIC ALGORITHMS IN THE N6S9 NETWORK
Θ (dB) −30 −32 −34 −36 −38 −40
Number of Blocked r 0 0 1 3 7 12
Yang et al. VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 205

Fig. 7. Spectrum efficiency comparison among different

approaches for the n6s9 simple network.

B. Simulation Experiments of Large-Scale

Instances

We evaluate the performance of our heuristic algorithm

in large-scale problem instances, i.e., two realistic geo-
graphically large network topologies [i.e., the NSF network
and the EON network as shown in Figs. 4(b) and 4(c)], a
7-core MCF [10,22] and a large set of transmission requests.

1) Spectrum Efficiency of our Algorithm in Large-Scale

Problem Instances: First, we evaluate the spectrum effi-
ciency achieved by our heuristic algorithm. The numbers
of transmission requests jRj are considered as 100 up to
1000 per 100 step (shown on the x axis), and the source
node and destination node for all transmission requests
are randomly selected. The required FSs for each transmis-
sion request are considered as a random number from 1 to
10 (FSs). We generate 10 different transmission request
matrices R, and the results—maximum required FSs—
are averaged. Because it is very difficult for the ILP/
MILP methods to find the effective solutions for large-scale
problem instances in an acceptable time, we use the cut-set
(CS) approach—a cut separates the network with N nodes
into two disjoint induced sub-graphs. All the traffic de-
mands between those two disjoint sub-graphs are carried
by the links that compose the cut, and the traffic demands
are assigned to the cores and links evenly—to obtain the
lower bounds of the problem. Additionally, for a fair com-
parison, the XT threshold is set as −20 dB to ensure that
the pre-calculated candidate path space of the existing XT-
WC-based heuristic algorithms in Refs. [18,21] will not be
bounded by the XT threshold.
Figure 8 presents the results of the spectrum efficiency of
each heuristic algorithm. From Fig. 8, we can observe that
in large-scale problem instances, our heuristic algorithm
also performs better than the two existing heuristic algo-
rithms, and the solutions are closer to the lower bounds.
2) Computational Efficiency of our Heuristic Algorithm
in Large-Scale Problem Instances: In this paragraph,
we evaluate the computational efficiency achieved by our
Fig. 8. Spectrum efficiency comparison among different
approaches for the NSF network and the EON network.
heuristic algorithm. Figure 9 shows the computation time
of the heuristic algorithms in the simulation experiments
above (Section IV.B.1).
As shown in Fig. 9, as the number of transmission re-
quests jRj increases, the computation time of the existing
SPSA algorithm in Ref. [21] increases exponentially. In
contrast, the computation time of our XT-aware heuristic
algorithm increases almost linearly as jRj increases. The
ARSCA-SP algorithm in Ref. [18] achieves the highest com-
putational efficiency, because it only considered one routing
path (i.e., the shortest path) for each node pair. However, as
shown in Fig. 8, the ARSCA-SP algorithm has the worst
spectrum efficiency.
3) Generalizability of our Heuristic Algorithm in Large-
Scale Problem Instances: In this paragraph, we evaluate
the generalizability of our heuristic algorithm. The XT
thresholds are considered from −21 dB to −32 dB per
−1 dB step. The number of transmission requests jRj is
fixed as 1000, and the required FSs for each transmission
request are considered as a random number from 1 to 10
(FSs). For each XT threshold, we generate 10 different
transmission request matrices R, and the results are
averaged as shown in Fig. 10.
206 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018 Yang et al.

Fig. 9. Computational efficiency comparison among different

Fig. 10. Generalizability of different heuristic algorithms for the
heuristic algorithms for the NSF network and the EON network.
NSF network and the EON network.

As shown in Fig. 10, the results of the ARSCA-SP remain

the same because the ARSCA-SP algorithm does not con- Meanwhile, we can observe that the required FSs of our
sider multiple candidate paths (i.e., all the transmission heuristic algorithm increase in this case because our
requests are connected by the shortest path). heuristic algorithm starts to avoid assigning the same
FSs to adjacent cores to ensure successful transmissions.
Moreover, for the 14-node 21-link NSF network, we can The similar tendency can also be observed for the 28-node
observe that when the value of the XT threshold is large 34-link EON network. Furthermore, we list the percentage
(−21– −25 dB), for the XT-WC-based SPSA algorithm, all of blocked transmission requests by the XT-WC-based
the pre-calculated candidate paths (e.g., k-shortest paths) algorithms in Table IV, as follows.
are almost feasible. This leads to a relatively small differ-
ence between our heuristic algorithm and the SPSA algo- In conclusion, for large-problem instances, our heuristic
rithm. However, when the value of the XT threshold is algorithm can also cope well with different XT thresholds,
reduced to −26 dB, for the SPSA algorithm, its feasible can- which is more suitable for geographically large networks
didate path space will be significantly bounded. In con- (see Fig. 10); better solutions (see Fig. 8) can be obtained
trast, our XT-aware-based heuristic algorithm is not based on our heuristic algorithm within acceptable time
notably affected, which will result in a greater difference (see Fig. 9).
in performance between the two heuristic algorithms.
Moreover, when the XT threshold is further reduced TABLE IV
(−27 dB– − 32 dB), the XT-WC-based algorithms become AVERAGE PERCENTAGE OF BLOCKED TRANSMISSION REQUESTS BY
THE XT-WC-BASED HEURISTIC ALGORITHMS IN THE NSF
unable to find even one feasible candidate path for some
transmission requests because under such a circumstance, NETWORK AND THE EON NETWORK
even the shortest paths of these transmission requests Θ (dB) −26 −27 −28 −29 −30 −31 −32
become infeasible due to the strict XT threshold.
NSF-14 1.1 8.7 23.8 40.7 56.6 70.9 81.5
Therefore, these transmission requests cannot be satisfied EON-28 0.0 16.5 31.9 44.0 62.6 71.4 80.2
by the XT-WC-based SPSA and ARSCA-SP algorithms.
Yang et al.
V. CONCLUSION
In this paper, we address the static RSCA problem in
SDM-EONs with MCF. We propose a node-arc-based ILP
method that assigns the lightpath, the cores, and the cor-
responding spectrum simultaneously for each transmission
request. In our ILP method, we consider the XT in a worst
interference scenario similar to previous works. Our analy-
sis shows that our ILP method achieves the minimum
magnitude of required variables and constraints. The
simulation results show that our ILP method can increase
dozens or more than one hundred times in terms of conver-
gence efficiency compared with the existing ILP method.
We also propose two XT-aware-based approaches that con-
sider XT strictly: an MILP method, which is the extension
of our ILP method for small-scale problem instances, and a
heuristic algorithm. The simulation results show that the
two XT-aware-based approaches can cope well with the
cases of different XT thresholds and/or different geogra-
phical sizes of the networks. Moreover, the simulation re-
sults demonstrate that our heuristic algorithm can obtain
effective solutions within an acceptable time even for large-
scale problem instances (in 40 s for 1000 requests in 28-
node 34-link EON network), which are more approximate
to the optimal solutions or the lower bounds than the
existing approaches.
REFERENCES
[1] Cisco, “White Paper: Cisco VNI Forecast and Methodology,
2015-2020,” 2016 [Online]. Available: http://www.cisco.com.
[2] P. J. Winzer and D. T. Neilson, “From scaling disparities
to integrated parallelism: a decathlon for a decade,” J.
Lightwave Technol., vol. 35, pp. 1099–1115, Mar. 2017.
[3] H. Tode and Y. Hirota, “Routing, spectrum, and core and/or
mode assignment on space-division multiplexing optical
networks [Invited],” J. Opt. Commun. Netw., vol. 9, no. 1,
pp. A99–A113, Jan. 2017.
[4] G. M. Saridis, D. Alexandropoulos, G. Zervas, and D.
Simeonidou, “Survey and evaluation of space division
multiplexing: from technologies to optical networks,” IEEE
Commun. Surv. Tutorials, vol. 17, no. 4, pp. 2136–2156,
Nov. 2015.
[5] P. J. Winzer, “Spatial multiplexing in fiber optics: the 10× scal-
ing of metro/core capacities,” Bell Labs Tech. J., vol. 19,
pp. 22–30, Sept. 2014.
[6] J. M. Rivas-Moscoso, B. Shariati, A. Mastropaolo, D. Klonidis,
and I. Tomkos, “Cost benefit quantification of SDM network
implementations based on spatially integrated network
elements,” in European Conf. on Optical Communication
(ECOC), Dec. 2016.
[7] M. M. Dan, P. D. Colbourne, A. D’Errico, N. K. Fontaine, Y.
Ikuma, R. Proietti, L. Zong, J. M. Rivas-Moscoso, and I.
Tomkos, “Survey of photonic switching architectures and
technologies in support of spatially and spectrally flexible op-
tical networking [Invited],” J. Opt. Commun. Netw., vol. 9,
no. 1, pp. 1–26, Jan. 2017.
[8] B. Shariati, M. M. Dan, D. Klonidis, I. Tomkos, and J. M.
Rivasmoscoso, “Comparison of cd(c) ROADM architectures
for space division multiplexed networks,” in Optical Fiber
VOL. 10, NO. 3/MARCH 2018/J. OPT. COMMUN. NETW. 207
Communications Conf. and Exhibition, Mar. 2017,
paper Th2A.45.
[9] B. Shariati, J. M. Rivas-Moscoso, M. Dan, S. Ben-Ezra, D.
Klonidis, L. Velasco, and I. Tomkos, “Impact of spatial and
spectral granularity on the performance of SDM networks
based on spatial superchannel switching,” J. Lightwave
Technol., vol. 35, no. 13, pp. 2559–2568, July 2017.
[10] Y. Tan, R. Zhu, H. Yang, Y. Zhao, J. Zhang, Z. Liu, Q. Qu, and
Z. Zhou, “Crosstalk-aware provisioning strategy with dedi-
cated path protection for elastic multi-core fiber networks,”
in 15th Int. Conf. on Optical Communications and
Networks (ICOCN), IEEE, Sept. 2016, pp. 1–3.
[11] Y. Tan, H. Yang, R. Zhu, Y. Zhao, J. Zhang, Z. Liu, Q. Qu, and
Z. Zhou, “Distance adaptive routing, core and spectrum
allocation in space division multiplexing optical networks
with multi-core fibers,” in Asia Communications and
Photonics Conf., Optical Society of America, Nov. 2016,
paper AF2A.159.
[12] H. Tode and Y. Hirota, “Routing, spectrum and core
assignment on SDM optical networks,” in Optical Fiber
Communications Conf. and Exhibition (OFC), IEEE,
Mar. 2016, pp. 1–3.
[13] K. Imamura, Y. Tsuchida, K. Mukasa, R. Sugizaki, K. Saitoh,
and M. Koshiba, “Investigation on multi-core fibers with large
Aeff and low micro bending loss,” Opt. Express, vol. 19, no. 11,
pp. 10595–10603, May 2011.
[14] T. Hayashi, T. Nagashima, O. Shimakawa, T. Sasaki, and E.
Sasaoka, “Crosstalk variation of multi-core fibre due to fibre
bend,” in 36th European Conf. and Exhibition on Optical
Communication (ECOC), IEEE, Sept. 2010, pp. 1–3.
[15] T. Hayashi, T. Taru, O. Shimakawa, T. Sasaki, and E.
Sasaoka, “Low-crosstalk and low-loss multi-core fiber utiliz-
ing fiber bend,” in Optical Fiber Communication Conf. and
Exposition and the National Fiber Optic Engineers Conf.
(OFC/NFOEC), IEEE, Mar. 2011, pp. 1–3.
[16] T. Hayashi, T. Taru, O. Shimakawa, T. Sasaki, and E.
Sasaoka, “Ultra-low-crosstalk multi-core fiber feasible to
ultra-long-haul transmission,” in National Fiber Optic
Engineers Conf., Optical Society of America, Mar. 2011,
paper PDPC2.
[17] T. Hayashi, T. Taru, O. Shimakawa, T. Sasaki, and E.
Sasaoka, “Design and fabrication of ultra-low crosstalk and
low-loss multi-core fiber,” Opt. Express, vol. 19, no. 17,
pp. 16576–16592, Aug. 2011.
[18] L. Zhang, N. Ansari, and A. Khreishah, “Anycast planning in
space division multiplexing elastic optical networks with
multi-core fibers,” IEEE Commun. Lett., vol. 20, no. 10,
pp. 1983–1986, Oct. 2016.
[19] K. Takenaga, Y. Arakawa, S. Tanigawa, N. Guan, S. Matsuo,
K. Saitoh, and M. Koshiba, “Reduction of crosstalk by
trench-assisted multi-core fiber,” in Optical Fiber
Communication Conf., Optical Society of America, Mar. 2011,
paper OWJ4.
[20] K. Takenaga, Y. Arakawa, S. Tanigawa, G. Ning, S. Matsuo,
K. Saitoh, and M. Koshiba, “An investigation on crosstalk in
multi-core fibers by introducing random fluctuation along
longitudinal direction,” IEICE Trans. Commun. E,
vol. E94-B, no. 2, pp. 409–416, Feb. 2011.
[21] A. Muhammad, G. Zervas, D. Simeonidou, and R.
Forchheimer, “Routing, spectrum and core allocation in flex-
grid SDM networks with multi-core fibers,” in Int. Conf. on
Optical Network Design and Modeling, IEEE, May 2014,
pp. 192–197.
208 J. OPT. COMMUN. NETW./VOL. 10, NO. 3/MARCH 2018
[22] A. Muhammad, G. Zervas, and R. Forchheimer, “Resource
allocation for space-division multiplexing: optical white box
versus optical black box networking,” J. Lightwave Technol.,
vol. 33, no. 23, pp. 4928–4941, Dec. 2015.
[23] K. Morita and K. Hirata, “Dynamic spectrum allocation
method for reducing crosstalk in multi-core fiber networks,”
in Int. Conf. on Information Networking (ICOIN), IEEE,
Jan. 2017, pp. 686–688.
[24] W. DGrover, Mesh-Based Survivable Networks: Options and
Strategies for Optical, MPLS, SONET, and ATM Networking,
Upper Saddle River, NJ: Prentice Hall, Aug. 2003.
[25] A. Cai, G. Shen, L. Peng, and M. Zukerman, “Novel node-arc
model and multi-iteration heuristics for static routing
and spectrum assignment in elastic optical networks,” J.
Lightwave Technol., vol. 31, no. 21, pp. 3402–3413, Nov. 2013.
[26] Y. Wang, X. Cao, Q. Hu, and Y. Pan, “Towards elastic and fine-
granular bandwidth allocation in spectrum-sliced optical net-
works,” J. Opt. Commun. Netw., vol. 4, no. 11, pp. 906–917,
Nov. 2012.
[27] D. Siracusa, F. Pederzolli, P. S. Khodashenas, and J. M. Rivas-
Moscoso, “Spectral vs. spatial super-channel allocation in
SDM networks under independent and joint switching
paradigms,” in European Conf. on Optical Communication,
Oct. 2015, pp. 1–3.
[28] P. S. Khodashenas, J. M. Rivas-Moscoso, D. Siracusa, F.
Pederzolli, B. Shariati, D. Klonidis, E. Salvadori, and I.
Tomkos, “Comparison of spectral and spatial super-channel
allocation schemes for SDM networks,” J. Lightwave
Technol., vol. 34, no. 11, pp. 2710–2716, June 2016.
[29] “Gurobi v7.0.1,” 2016 [Online]. Available: http://www.gurobi.
com.
View publication stats
Yang et al.
[30] K. Christodoulopoulos, I. Tomkos, and E. Varvarigos, “Elastic
bandwidth allocation in flexible OFDM-based optical net-
works,” J. Lightwave Technol., vol. 29, no. 9, pp. 1354–1366,
May 2011.
[31] S. Fujii, Y. Hirota, H. Tode, and K. Murakami, “On-demand
spectrum and core allocation for reducing crosstalk in multi-
core fibers in elastic optical networks,” J. Opt. Commun.
Netw., vol. 6, no. 12, pp. 1059–1071, Dec. 2014.
[32] ITU-T, “G.694.1:Spectral grids for WDM applications: DWDM
frequency grid,” Dec. 2012. Available: https://www.itu.int.
[33] M. Klinkowski and K. Walkowiak, “A heuristic algorithm for
routing, spectrum, transceiver and regeneration allocation
problem in elastic optical networks,” in Int. Conf. on
Transparent Optical Networks, July 2012, pp. 1–4.
[34] B. G. Bathula and J. M. H. Elmirghani, “Constraint-
based anycasting over optical burst switched networks,”
J. Opt. Commun. Netw., vol. 1, no. 2, pp. A35–A43,
July 2009.
[35] L. Velasco, A. Jirattigalachote, M. Ruiz, P. Monti, L. Wosinska,
and G. Junyent, “Statistical approach for fast impairment-
aware provisioning in dynamic all-optical networks,” J. Opt.
Commun. Netw., vol. 4, no. 2, pp. 130–141, Feb. 2012.
[36] J. Perelló, J. M. Gené, A. Pagès, J. A. Lazaro, and S. Spadaro,
“Flex-grid/SDM backbone network design with inter-core
XT-limited transmission reach,” J. Opt. Commun. Netw.,
vol. 8, no. 8, pp. 540–552, Aug. 2016.
[37] W. Klaus, B. J. Puttnam, R. S. Luís, J. Sakaguchi, J. M.
Mendinueta, Y. Awaji, and N. Wada, “Advanced space
division multiplexing technologies for optical networks
[Invited],” J. Opt. Commun. Netw., vol. 9, no. 4, pp. C1–C11,
Apr. 2017.