Optical Switching and Networking 14 (2014) 274281

Contents lists available at ScienceDirect

Optical Switching and Networking

journal homepage: www.elsevier.com/locate/osn

Fragmentation assessment based on-line routing

and spectrum allocation for intra-data-center networks
with centralized control
Shanguo Huang n, Yu Zhou n, Shan Yin, Qian Kong, Min Zhang, Yongli Zhao,
Jie Zhang, Wanyi Gu
State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications,
Beijing 100876, PR China

a r t i c l e in f o

abstract

Article history:
Received 16 March 2014
Received in revised form
17 April 2014
Accepted 13 May 2014
Available online 9 June 2014

As is well known, elastic optical network is quite attractive due to its great spectral
efficiency and flexibility, which can allocate appropriate size spectrum based on efficient
modulation modes such as Orthogonal Frequency Division Multiplexing (OFDM) modulation, etc. These characters are able to better meet the needs of the intra-data-center
optical interconnection. To properly analyze and operate elastic optical networks with
highly dynamic traffic of the intra-data-center, considering the continuity constraint and
the contiguity constraint, efficient methods are required for the on-line routing and
spectrum allocation (RSA) issue. In this paper, we propose two methods (AFD and CSL) for
assessing the status (spectrum fragmentation) of the network, and present two algorithms
(BSPR and HSR) which use the proposed assessment value to perform on-line RSA applied
to intra-data-center networks which adopt the Software-Defined Network (SDN) control
mode. Simulation results reveal that the proposed approaches allow solving the on-line
RSA problem more efficiently with lower blocking probability of the networks.
& 2014 Elsevier B.V. All rights reserved.

Keywords:
Elastic optical network
Software-defined network (SDN)
On-line
RSA
Intra-data-center

1. Introduction
In the past few years, cloud adoption is not only an
emerging technology, but also an established networking
solution which is widely accepted and deployed by the
telecom industry. According to statistics, the traffic volumes
generated in total are largeover 10 GB per server per day in
an intra-data-center network (intra-DCN) [1]. The Cisco
global cloud index forecasts that global data center IP traffic
will reach 644 exabytes per month in 2017 (nearly triple of
the traffic in 2012) [2]. The quality of service (QoS)

n
Corresponding authors. Tel.: 86 10 61198106;
mobile: 86 10 13693578265.
E-mail addresses: shghuang@bupt.edu.cn,
shaxiaoziningyi@bupt.edu.cn, shaxiaoziningyi@163.com (S. Huang).

http://dx.doi.org/10.1016/j.osn.2014.05.011
1573-4277/& 2014 Elsevier B.V. All rights reserved.

provisioning is the users' major concern, which depends

on the communication efficiency between servers. Alloptical switching technique can just cater to the intra-DCN's
high efficient transport requirement. Compared with the
rigid spectrum grid networks realized with the traditional
WDM technology, spectrum-efficient and scalable optical
transport network architecture called SLICE which is based
on optical orthogonal frequency division multiplexing (OOFDM) technology is better spectrum efficiency and flexibility [3,4].
In elastic optical networks, the available spectrum of a
fiber link is divided into a set of spectral slots with substantially smaller spectral width (spectral slots widths are 6.25,
12.5 and 25 GHz) than a traditional wavelength (50 GHz ITUT WDM grid). Instead of occupying a whole wavelength,
connections are placed on series of contiguous subcarriers

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

according to the bandwidth demand. Moreover, ultra high

bit-rate super channels at 400 Gb/s sand 1 Tb/s will be
allowed in future elastic optical networks.
Flexible and efficient features of elastic optical network
(EON) are double-edged swords. EON face some important
challenges, although they have a great advantage than
traditional WDM networks. Due to the character of OOFDM technology and the absence of spectrum converters,
the allocated slots must be contiguous in the spectrum
(contiguity constraint), and the same slots must be used in
all links of the routing path (continuity constraint) [5].
These two constraints let EON a main challenge being
the so-called spectrum fragmentation effect, which refers
to the condition that, as the highly dynamic set up and
remove traffic over a intra-DCN, the available spectral
resources become highly fragmented, potentially leading
to blocking situations.
At present, spectrum fragmentation has attracted a lot
of attention in DCN. Some novel spectrum defragmentation algorithm has been proposed [68]. Generally, the
defragmentation is accompanied by mass of light path or
spectrum migrations, which is not a kind of suitable
solution in highly dynamic intra-DCN. Therefore, how to
reduce the extent of fragmentation at first as much as
possible and choose a better way to set up a connection
when the network has already been fragmental, which can
be summarized as efficient on-line routing and spectrum
allocation (RSA) problem is urgently needed to be solved
in the intra-DCNs.
In [9], the authors proposed two RSA methods for
super-channels accounting for sub-carriers dynamism,
which are similar to the first-fit and best-fit used in the
WDM networks. The authors of [10] introduce a serviceoriented spectrum assignment framework and two algorithms to solve the RSA problem. In [10], spectrum is
divided into some blocks for different set of services
according to the chosen properties, and RSA are done in
every block just as [9]. Both of them did the RSA one step
by step without some criterion but just satisfy the bandwidth demand, while we may need to consider the routing
and spectrum allocation together which way will make
RSA more optimal. Therefore, we need an assessment
value to evaluate the status (spectrum fragmentation) of
the network, which can be used to choose different
routing and spectrum allocation together. Also in very
recently, the papers [11,12] focus on the static RSA problem, and Integer Linear Programming (ILP) formulations
are presented to minimize the utilized spectrum.
The authors of [13] introduced a concept called Utilization Entropy (UE) to assess the degree of spectrum
fragmentation in flexible grid optical networks. The
authors of [14] presented the Spectrum Compactness
(SC), which is calculated by analyzing the times of changing among the occupied and vacant spectrum blocks on a
link. In [15], the authors proposed an assessment method
based the fragmentation of a set of paths between every
pair of nodes in the graph. However, some influence
factors were less considered in design of the assessment
methods in above work, such as the spectrum usage, the
spectrum continuity constraint and the non-uniform sample of links. In this paper, we present two fragmentation

275

assessment schemes of the whole network, which can be

more effective and accurate to reflect the status of the
network, and then use them on the on-line routing and
spectrum allocation.
From the perspective of intra-DCN control, the authors
of [16] describes the implemented control-plane which is
consist of GMPLS (Generalized Multi-Protocol Label
Switching), PCE (Path Computation Engine) and a resource
allocation tool called Sub-wavelength Assignment Engine
(SLAE) for the Time-Shared Optical Network (TSON). In
[17], the authors demonstrate the star all-optical intraDCN using extended OpenFlow-based software-defined
networking. Recently, in [18], a network-driven transfer
mode is proposed for cloud operations of data center with
a Software-Defined Network (SDN) controller and an
Application-Based Network Operations (ABNO) controller
working together. A GMPLS-enabled FBON is discussed in
[19], especially its distributed reservation approach. And
for the first time, the authors investigate the backward
reservation collision problem in an FBON, which would
dominate the blocking performance in networks with light
load or serving highly dynamic traffic. In highly dynamic
intra-DCN, centralized control mode can effectively reduce
the failure of the request which is because of the conflict of
resource access with distributed control mode, and it will
monitor resource of the whole network more efficiently
which is more conductive to the inter-DCN connection. In
[20] and [21], the authors respectively proposed routing
algorithm and protection algorithm for multi-layer and
multi-domain optical networks, which are playing guiding
roles for the research of the connection control among the
data centers. In fact, the survivability in optical network
should also be considered [22], which is one of our future
research points. We provide a feasible all-optical intraDCN architecture based on SDN control (centralized control). The architecture of the intra-DCN is based on the
previous EON based on O-OFDM. First, we need to extend
the messages between the controller and the racks. Then,
an OF agent is integrated on every rack, which can
translate the messages which are from SDN controller into
control orders for the O-OFDM ToR. The architecture and
the SDN control model are illustrated in Fig. 1.
The rest of the paper is organized as follows. In Section
2, two fragmentation assessment schemes are proposed. In
Section 3, two on-line RSA algorithms based on the
fragmentation assessment in detail are presented. In
Section 4, the performance of considered on-line RSA
formulations are compared, and then the performance of
our formulation over two realistic network topologies are
introduced. Section 5 concludes the paper.
2. Proposed spectrum fragmentation assessment
schemes
In the last section, we mentioned the continuity constraint and the contiguity constraint in elastic optical
networks. If the spectrum is not continuous among connected links in a precomputed path, it cannot be used for
the path. On the other hand, if the available slots are just
enough for bandwidth demand but discrete, they also may
not be provided for set up traffic. Therefore, the status of

276

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

Fig. 1. A feasible all-optical intra-DCN architecture based on SDN control.

the contiguity slots in the same position among connected

links are importantly influence factors of the network
fragmentation in EON (Table 1).

2.1. Available fragment degree

In this paper, we present a new concept called Available Fragment Degree (AFD) to indicate the availability of
the spectrum fragments among various resources. The
available fragment degree value is normalized between 0
and 1, which can be used as an indicator for assessing and
comparing the degree of resource fragmentation. A higher
available fragment degree indicates the spectrum fragmentation blocks can be allocated wider and more bandwidth demands.
As a kind of collection of the connected links, path is
taken as the foothold to calculate the AFD. Meanwhile, we
consider that the selection of path's length is of great
importance and closely related with the size of network
(the number of nodes); so we select all paths which are
consisted of Np connected links as a set, where Np can be
calculated as follows:
N p round up

jVj
log2

We define the AFD_path which indicates the AFD

among Np connected links. The spectrum resource of the
path is achieved by performing bitwise AND operation
among the links. Define a network as G{V,E}, where V is
the set of bandwidth variable switching (BVS) nodes and E
represents the set of fiber links connecting two nodes in V.
Some notations and variables are introduced in Table 1,
and we use |X| to represent the number of the elements
in the set of X. The calculation procedure of AFD_path is
introduced as follows.
(1) Get the spectrum usage status of the path
The spectrum occupation on a link is expressed by a
bits array sei s1ei ; s2ei ; ; sjSj
ei . By applying bitwise
AND operation among n connected links, the spectrum usage status of a path can be calculated as

follows:
spk \ ei A pk sei

(2) Calculate the probability of provisioning available

bandwidth for bandwidth demand r among path pk.
For bandwidth demand r (the number of required
contiguity slots), we first find out that how many start
slot's position can afford r contiguity slots. Then, the
value by dividing by the maximum number of the
positions (assuming all slots are vacant) is normalized.
The probability of provisioning available bandwidth
demand r can be calculated as follows:

pk ; r

jiSj 1r 1 spi;ik r  1

jSj  r 1

(3) Update the distribution probability of bandwidth

demands after a bandwidth demand r has been
accepted.

In different networks, the probability of kinds of

bandwidth demands may be different, such as one network's traffic mostly request one slot but another network's traffics mostly request three slots. For different
highly dynamic network, we need to base the assessed
value on the probabilities of different bandwidth demands.
With the whole traffics' number of a highly dynamic
network grows all the time, and for the sake of the
normalized, the probabilities of different bandwidth
demands are initialized to the same value as 1/|R|. Then,
every time the network controller receives a bandwidth
demand r, probabilities are updated as follows:


jRj
1 1=log2


Q
jRj
1 P r;old =log2

P r;new Q P r;old

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

277

Table 1
Notations and variables.
Symbol

Definition

V
E
D
P
K
R
Gc
Ev
sd,td
S
se
sv

Set of nodes, index v

Set of fiber links, index e
Set of demands, index d
Set of precomputed paths, index p
Number of slots in a link
Set of bandwidth demand, index r
Set of connected graphs, which are consist of Np different links, index g
Set of fiber links, which is connected with node v
Source and destination of nodes of demand d
Set of frequency slots, index s
Frequency slots on the link e
Frequency slots of node v, the result by performing bitwise AND operation among frequency slots of all the links
which connect with node v
Frequency slots of path p, the result by performing bitwise AND operation among frequency slots of all the links
which are on path p
Frequency slots of path p (from s to d), the result by performing bitwise AND operation among frequency slots
of all the links which are on path p
The usage status of the ith slot in the collection x, if the slot is available the value equals to 1, else 0

sp
sp(s,d)
six
Pr
p;r
v;r
Nv

P r0 ;new

The
The
The
The

distribution probability of bandwidth demand r

probability of provisioning bandwidth demand r over sp
probability of provisioning bandwidth demand r over sv
degree of a node (the number of the connected links with node v)

1  P r;new
P0
1  P r;old r ;old

Above all, the available fragment degree of a path can

be achieved as follows:
AFD_path k 8 r pk ;r P r

The available fragment degree of the network (AFD_network) is presented to indicate the AFD of the network
and it can be calculated by expression (8).
AFD_network

1
AFD_path k
jKj 8 k

follows:

vk ; r

r  1
jiSj 1r 1 si;i
vk
jSj r 1

10

Third, update the distribution probability of bandwidth

demands after a bandwidth demand r being accepted, and
the calculation method of the probability is same to the
AFD's.
Finally, the continuity of spectrum between links which
is linked node v can be achieved as follows:
CSL_node k 8 r vk ;r P r

11

The available fragment degree of the network (AFD_network) is presented to indicate the AFD of the network
and calculated as follows:

2.2. Continuity of spectrum among links

CSL_network

1
CSL_node k
jVj 8 k

12

Although the AFD assessment scheme proposed is

reasonable and scientific, because of that |P| is about the
square of the |E|, the cost (CPU instructions) of the calculation may be high. It may greatly suit a small-scale network,
and it will bring much cost for a large-scale network.
Actually, the cost can be greatly reduced by doing precomputation such as just updating the AFD_paths which are
related to the RSA. However, we proposed another quick
assessment scheme to indicate the continuity of the spectrum between links, which is called Continuity of Spectrum among Links (CSL). As the intersection of links, node
is taken as the foothold to calculate the CSL. The calculation
procedure of CSL_node is great similar with AFD_path, so it
is introduced as follows in brief.
First, the conjunct spectrum usage status of the links
around a node can be calculated as follows:

In this section, two on-line RSA algorithms are proposed based on fragmentation assessment (AFD and CSL,
we adopt AFD in the description of algorithms). The aim of
the algorithms is trying to make the network accept more
demands (keep the network in an optimal status) by
choosing the route and the allocation position of the
spectrum in the light path together. In order to simplify
the expression of the algorithm, the concept of channel is
introduced in [23]. The definition of channels can be
described as contiguous n slots, and the value of them are,

svk \ ei

i k
cix;n \ nk
i sx

A Ev sei

Second, the probability of provisioning available bandwidth for bandwidth demand r over svk is calculated as

3. On-line RSA algorithms

The above formula shows that in a collection x (se, sp or

sv), if the contiguous n slots from the position i are

278

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

Table 2
BSPR algorithm.
1:
2:

While receive a demand for r slots do

Get all nodes and the links which can afford enough contiguous slots for the demand, then
create a new graph
Find at most k shortest paths according to the K-shortest paths algorithm
For each calculated path Do
Initialize the data containers Vc and Vt (container of values which is consist of the value of AFD_network, start
slot position and path point. The subscript c for compare, t for temp
The initialize value for both containers is {  1, 0, NULL})
For each channel with exact r slots do
Calculate the AFD_network assuming the channel among the path has been located,
and use Vt record the value of AFD_network, the start slot position and path point
Compare the data containers Vt with Vc
If the value of AFD_network in Vt is bigger
Use the three data in Vt to replace the corresponding data in Vc
Else if the values of AFD_network in Vt and Vc are equal
Compare the start slot position. If the start slot position in Vt is farthest away from
the slots center (the value is equal to half of the number of possible start slots
position in a link), use the three data in Vt to replace the corresponding data in Vc
End for
End for
Compare every final Vc for different calculated path as above method, the final result with
biggest value of AFD_network and start slot position which is farthest away from the
slots center, is used to response to the demand
End do

3:
4:
5:

6:
7:
8:
9:
10:
11:
12:

13:
14:
15:

16:

Table 3
HSR algorithm.
1:
2:
3:
4:
5:

6:
7:
8:
9:
10:
11:
12:

13:
14:
15:

16:

While receive a demand for r slots do

For each channel with exact r slots do
Get all nodes and the links which each has the available corresponding channel, then
create a new graph
Find at most k shortest paths according to the K-shortest paths algorithm
Initialize the data containers Vc and Vt (container of values which is consist of the value
of AFD_network, start slot position and path point. The subscript c for compare, t for
temp. The initialize value for both containers is {  1, 0, NULL})
For each calculated path Do
Calculate the AFD_network assuming the channel among the path has been located,
and use Vt record the value of AFD_network, the start slot position and path point
Compare the data containers Vt with Vc
If the value of AFD_network in Vt is bigger
Use the three data in Vt to replace the corresponding data in Vc
Else if the values of AFD_network in Vt and Vc are equal
Compare the length of paths, if the length of the path in Vt is shorter than the
length of the path in Vc, Use the three data in Vt to replace the corresponding data
in Vc
End for
End for
Compare every final Vc with different start slot position, leave the one which meet the
conditions as follow: First, it has the biggest value of AFD_network; Furthermore, its
start slot position is farthest away from the slots center (the value is equal to half of the
number of possible start slots position in a link), the final result is used to response to
the demand
End do

all available, the channel cix;n is available otherwise it is

equal to 0.
3.1. Best spectrum position routing (BSPR) algorithm
RSA should be considered together to optimize the
carrying capacity of the network. When the network
controller receives a demand d, first it gets information
for all nodes and the links which can afford enough
contiguous slots for the demand to constitute a new graph
(topology), which way can increase the success probability

of the spectrum allocation (without by this way, although

the routing is successful, the spectrum allocation on the
route may be failed). The details of the BSPR algorithm are
shown in Table 2.
3.2. Hierarchical spectrum routing (HSR) algorithm
When the spectrum fragmentation degree is high, the
spectrum of the traditional k shortest paths we used in the
algorithms may be unable to afford enough resource. To
reduce the blocking probability of the networks, we first

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

279

Fig. 2. Network topologies for the simulations.

Table 4
All kinds of assessment schemes.
Time
complexity

Problems

UE

O(|S|n|E|)

SC

O(|S|n|E|)

Only care the change of adjacent slots, but

ignore the spectrum usage
The spectrum continuity between links has not
been considered
More computation time is spent in large-scale
networks
Cannot be used to reflect the spectrum
fragmentation among long path

AFD O(|S|n|E|2)
CSL

O(|S|n|V|)

4. Numerical simulation
The NSFNET network (14 nodes and 21 links) and the
COST239 network (11 nodes and 26 links) are chosen for
the simulations to evaluate the performance of the proposed algorithms in this paper (as shown in the Fig. 2). In
the simulations, each link has 80 slots spectrum resource
with equal bandwidth and the bandwidth demands
(required slots for connection, R {1, 2, 3, 4}) follow
Poisson distribution (the mean of Poisson distribution
2.5, and ignore the value bigger than 4 and smaller
than 1). Meanwhile the traffic time model also follows
Poisson distribution. In this paper, we assume that the
guard-band has been included in the bandwidth demand
in its forming for simplicity.

4.1. Effectiveness of the spectrum fragmentation assessment

schemes

Fig. 3. Calculated values of different assessment schemes.

get the links which can afford available channel at the

same position to constitute a new graph, after that the
routing of the traffic is implemented. The above method
operates at every reasonable layer (contiguity spectrum
block), and the procedure of HSR algorithm is shown in
Table 3.

Fig. 3 shows the calculated values of different spectrum

fragmentation assessment schemes (UE, SC, AFD and CSL)
under different resource utilization in the above two topologies. Although these four assessment schemes have certain
relevance to the resource utilization, they have difference
and characteristics. The value of UE scheme has a prominent
change when the resource utilization reaches 50%, which is
because that UE scheme mainly counts the changing frequency of the slots' state but ignores the utilization of the
slots. In terms of normalized, the values of UE scheme are
small, and SC scheme has no normalized processing. Simulation results demonstrate that AFD and CSL can effectively

280

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

Table 5
Average running time of the algorithms.
Traffic load
(Erlang)

Network
topology

indicate both the spectrum fragmentation and the change of

the resource utilization in elastic optical networks.
In highly dynamic network, the time complexity is also
a point needed to be paid attention to when we build a
spectrum fragmentation assessment schemes. As more
comprehensive factors are considered in AFD scheme,
AFD scheme has a high time complexity. Different from
AFD scheme which considers the spectrum fragmentation
on paths, CSL pays more attention on the spectrum
fragmentation of the link pairs around nodes. The time
complexity and existing problems of all mentioned assessment schemes are shown in Table 4.
4.2. Performance of the BSPR and HSR Algorithms
The simulations are designed to evaluate the proposed online RSA algorithms with CSL assessment. First-fit slot assignment with K-shortest paths routing (KRFS, KNp as in the
proposed algorithms in this paper) is used for evaluation of
on-line RSA algorithms. Fig. 4 compares the performance of

KRFS

BSPR

HSR

20

NSFNET
COST239

0.195
0.205

1.19
1.126

5.792
5.789

240

NSFNET
COST239

0.187
0.175

0.365
0.637

0.852
1.358

Time complexity

Fig. 4. Impact of proposed algorithms. (a) Compares the performance of

the proposed on-line RSA algorithms with KRFS algorithm in terms of
blocking probability in NSFNET network. (b) Compares the performance
of the proposed on-line RSA algorithms with KRFS algorithm in terms of
blocking probability in COST239 network.

Average running time of the

algorithms (ms)

O
O
O
(Tk kCTf) (Tk kCTcsl) (CTk kCTcsl)

the proposed on-line RSA algorithms with KRFS algorithm in

terms of blocking probability in NSFNET network and
COST239 network. As shown, using the algorithms proposed
in this paper reduce the blocking probability, in which BSPR is
better than KRFS, and HSR algorithm outperforms than others.
The load capacity of the COST239 is better than NSFNET just
because that the average of the former's nodes degree is
higher than the latter's.
In above two networks, the efficiency of HSR algorithm is
better than that of BSPR algorithm, which is because that
HSR algorithm more considers of the contribution of spectrum allocation than routing. Because BSRP algorithm a little
more considers the routing attribute when choose a route,
the blocking probability is only reduced a little than the KRFS
algorithm in a network which the average of nodes degree is
high (such as COST239). Meanwhile, for HSR algorithm, the
high average of nodes degree means more routing choices,
which leads to lower blocking probability.
In addition to the influences of the algorithms for the
blocking probability, the time efficiencies of the algorithms
are also what we are concerned with, which are shown in
Table 5. Tk is the time cost to running K-shortest paths
routing; k is the path number of the K-shortest paths
result; C represents the number of possible channel; Tf is
the time cost to make sure the path and channel can be
used for the demand; Tcsl stands for the time cost to
running the CSL spectrum fragmentation Assessment
Schemes. We run the Java Project with Eclipse on a PC
(3.0 GHz dual-core processor, 4 GB RAM). From the result,
we can draw conclusions as follows:
(1) The algorithms spend main time for running K-shortest paths routing, which means that the more times to
running K-shortest paths routing the more time the
algorithms will spend.
(2) Although BSPR and HSR algorithms spend more time
than KRFS algorithm, the time cost of them is
acceptable.
(3) In the different stages of the network traffic load state,
the different algorithms can be used for improving the
network performance.

5. Conclusions
In this paper, two assessment concepts called AFD and
CSL are proposed to represent the availability of the

S. Huang et al. / Optical Switching and Networking 14 (2014) 274281

spectrum fragments in EON. Then two algorithms which

use the assessment value to perform on-line RSA are
presented. The simulation results reveal that the proposed
algorithms can solve the on-line RSA problem more
efficiently with lower blocking probability, especially for
the network with high average of nodes degree such as the
intra-data-center networks.

Acknowledgments
This work is supported partly by the National Basic
Research Program of China (973 Program) (Nos.
2010CB328204 and 2010CB328202), the National Natural
Science Foundation of China (Nos. 61331008 and
61205058), the Hi-Tech Research and Development Program of China (863 Program) (No. 2012AA011302), the
Program for New Century Excellent Talents in University
(NCET-120793), the Beijing Nova Program (No. 2011065),
and the Fundamental Research Funds for the Central
Universities.
References
[1] S. Kandula, S. Sengupta, A. Greenberg, et al., The nature of data
center traffic: measurements & analysis, in: Proceedings of the 9th
ACM SIGCOMM conference on Internet measurement conference,
ACM, 2009, pp. 202208.
[2] Cisco Global Cloud Index: Forecast and Methodology, 20122017.
http://www.cisco.com/c/en/us/solutions/collateral/service-provi
der/global-cloud-index-gci/Cloud_Index_White_Paper.html.
[3] M. Jinno, H. Takara, B. Kozicki, et al., Spectrum-efficient and scalable
elastic optical path network: architecture, benefits, and enabling
technologies, Commun. Mag., IEEE 47 (11) (2009) 6673.
[4] Z. Cao, J. Yu, F. Li, L. Chen, Q. Shu, Q. Tang, L. Chen, Energy efficient
and transparent platform for optical wireless networks based on
reverse modulation, IEEE J. Sel. Areas Commun. 31 (2013).
[5] Y. Wang, X. Cao, Y. Pan, A study of the routing and spectrum
allocation in spectrum-sliced elastic optical path networks, INFOCOM, in: 2011 Proceedings IEEE, 2011, pp. 15031511.
[6] K. Sone, X. Wang, S. Oda, G. Nakagawa, Y. Aoki, I. Kim, P. Palacharla, T.
Hoshida, M. Sekiya, J.C. Rasmussen, First demonstration of hitless
spectrum defragmentation using real-time coherent receivers in
flexible grid optical networks, in: Proceedings of European Conference and Exhibition on Optical Communication (ECOC), 2012,
paper Th.3.D.1.
[7] Ke Wen, Yawei Yin, Geisler D.J., Shuo Chang, Yoo S.J.B., Dynamic ondemand lightpath provisioning using spectral defragmentation in
flexible bandwidth networks. in: Proceedings of European Conference and Exhibition on Optical Communication (ECOC), 1822
September 2011, pp. 13.

281

[8] W. Ju, S. Huang, B. Guo, et al., Node handling capacity based

spectrum fragmentation evaluation scheme in flexible grid optical
networks, Optical Fiber Communication Conference, Optical Society
of America, OW3A, 2013, p. 6.
[9] N. Sambo, F. Cugini, G. Bottari, et al., Routing and spectrum assignment for super-channels in flex-grid optical networks, in: European
Conference and Exhibition on Optical Communication, Optical
Society of America, 2012.
[10] K. Song, J. Zhang, Y. Zhao, et al., Service-oriented spectrum assignment algorithms in flexible bandwidth optical networks, in: Asia
Communications and Photonics Conference, Optical Society of
America, 2012.
[11] X. Yu, Y. Zhao, J. Zhang, B. Mukherjee, J. Zhang, X. Wang, , Static
routing and spectrum assignment in co-existing fixed/flex grid
optical networks, in: Proceedings of Optical Fiber Communication
Conference (OFC), 2014, paper W2A.59.
[12] K. Walkowiak, M. Klinkowski, B. Rabiega, et al., Routing and
spectrum allocation algorithms for elastic optical networks with
dedicated path protection, Opt. Switch. Netw. 13 (2014) 6375.
[13] X. Wang, Q. Zhang, I. Kim, P. Palacharla, M. Sekiya, Utilization
entropy for assessing resource fragmentation in optical networks,
in: Proceedings of Optical Fiber Communication Conference (OFC),
2012, paper OTh1A.2.
[14] X. Yu, J. Zhang, Y. Zhao, T. Peng, Y. Bai, D. Wang, X. Lin, Spectrum
compactness based defragmentation in flexible bandwidth optical
networks, in: Proceedings of Optical Fiber Communication Conference (OFC), 2012, paper JTh2A.35.
[15] A.N. Patel, P.N. Ji, Y.K. Huang, et al., Distance-adaptive virtual
network embedding in software-defined optical networks, in:
OptoElectronics and Communications Conference held jointly with
2013 International Conference on Photonics in Switching (OECC/PS),
2013 18th, IEEE, pp. 12.
[16] B.R. Rofoee, G. Zervas, Y. Yan, et al., Demonstration of low latency
Intra/Inter Data-Centre heterogeneous optical Sub-wavelength network using extended GMPLS-PCE control-plane, Opt. Express 21 (5)
(2013) 54635474.
[17] P.N. Ji, D. Qian, K. Sethuraman, et al., First demonstration of real-time
all-optical software-defined intra-data center star network using
OFDM and burst switching, in: OptoElectronics and Communications Conference held jointly with 2013 International Conference on
Photonics in Switching (OECC/PS), 2013 18th, IEEE, pp. 12.
[18] L. Velasco, A. Asensio, J.L. Berral, et al., Towards a carrier SDN: an
example for elastic inter-datacenter connectivity, Opt. Express 22 (1)
(2014) 5561.
[19] B. Guo, J. Li, Y. Wang, et al., Collision-aware routing and spectrum
assignment in GMPLS-enabled flexible-bandwidth optical network,
J. Opt. Commun. Netw. 5 (6) (2013) 658666.
[20] L. Guo, J. Wu, W. Hou, et al., Hybrid protection algorithms based on
game theory in multi-domain optical networks, Opt. Fiber Technol.
17 (6) (2011) 523535.
[21] S. Huang, W. Lian, X. Zhang, et al., A novel method to evaluate
clustering algorithms for hierarchical optical networks, Photonic
Netw. Commun. 23 (2) (2012) 183190.
[22] L. Guo, J. Wu, W. Hou, et al., Multicast protection algorithms based
on aggregated logical topology in survivable multi-domain optical
networks, Opt.: Int. J. Light Electron Opt. 123 (6) (2012) 521526.
[23] L. Velasco, M. Klinkowski, M. Ruiz, et al., Modeling the routing and
spectrum allocation problem for flexgrid optical networks, Photonic
Netw. Commun. 24 (3) (2012) 177186.