Received August 20, 2021, accepted September 14, 2021, date of publication September 28, 2021, date of current

version October 6, 2021.

Digital Object Identifier 10.1109/ACCESS.2021.3115998

Impact of Amplification and Regeneration

Schemes on the Blocking Performance and
Energy Consumption of Wide-Area
Elastic Optical Networks
ASTRID LOZADA 1 , FELIPE CALDERÓN 2 , JOSÉ NÚÑEZ KASANEVA 3 ,
DANILO BÓRQUEZ-PAREDES 4 , RICARDO OLIVARES 1 , ALEJANDRA BEGHELLI 5,

NICOLÁS JARA 1 , (Member, IEEE), ARIEL LEIVA 2 , (Member, IEEE),

AND GABRIEL SAAVEDRA 3 , (Member, IEEE)
1 Department of Electronic Engineering, Universidad Técnica Federico Santa María, Valparaíso 2390123, Chile
2 School of Electrical Engineering, Pontificia Universidad Católica de Valparaíso, Valparaíso 2362804, Chile
3 Department of Electrical Engineering, Universidad de Concepción, Concepción 4070409, Chile
4 Faculty of Engineering and Sciences, Universidad Adolfo Ibáñez, Santiago 7941169, Chile
5 Department of Electronic and Electrical Engineering, University College London, London WC1E 7JE, U.K.

Corresponding author: Astrid Lozada (astrid.lozadav@sansano.usm.cl)

This work was supported in part by DI-PUCV 039.382/2021, in part by USM PI_LII_2020_74, in part by USM PIIC 007/2021, in part by
the Agencia Nacional de Investigación y Desarrollo (ANID) Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT)
Iniciación 11190710, in part by ANID FONDECYT Iniciación 11201024, in part by ANID Doctorado Nacional/2021-21211075, and
in part by ANID Magister Nacional/2021-22210736.

ABSTRACT This paper studies the physical layer’s impact on the blocking probability and energy
consumption of wide-area dynamic elastic optical networks (EONs). For this purpose, we consider five
network configurations, each named with a network configuration identifier (NCI) from 1 to 5, for which the
Routing, Modulation Level, and Spectrum Assignment (RMLSA) problem is solved. NCI 1-4 are transparent
configurations based on all-EDFA, hybrid Raman/EDFA amplifiers (with different Raman gain ratio 0R ),
all-DFRA, and alternating span configuration (EDFA and DFRA). NCI 5 is a translucent configuration
based on all-EDFA and 3R regenerators. We model the physical layer for every network configuration to
determine the maximum achievable reach of optical signals. Employing simulation, we calculate the blocking
probability and the energy consumption of the different network configurations. In terms of blocking,
our results show that NCI 2 and 3 offer the lowest blocking probability, with at least 1 and 3 orders of
magnitude of difference with respect to NCI 1 and 5 at high and low traffic loads, respectively. In terms of
energy consumption, the best performing alternatives are the ones with the worst blocking (NCI 1), while
NCI 3 exhibits the highest energy consumption with NCI 2 0R = 0.75 following closely. This situation
highlights a clear trade-off between blocking performance and energy cost that must be considered when
designing a dynamic EON. Thus, we identify NCI 2 using 0R = 0.25 as a promising alternative to reduce
the blocking probability significantly in wide-area dynamic EONs without a prohibitive increase in energy
consumption.

INDEX TERMS Elastic optical networks, resource assignment, translucent configuration, transparent
configuration.

I. INTRODUCTION crunch of legacy wavelength division multiplexing (WDM)

Elastic Optical Network (EON) architectures were proposed networks [1], [2]. EONs divide the frequency spectrum into
as a new paradigm to overcome the potential capacity small slot units (Frequency Slot Units - FSU), usually of 12.5
or 6.25 GHz spectral width [3], which are flexibly allocated
The associate editor coordinating the review of this manuscript and to the different traffic demands [4]. In this way, spectral
approving it for publication was Bijoy Chand Chand Chatterjee . resources are managed more efficiently than current fixed

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 9, 2021 134355
 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

grid WDM networks. When dynamically operated (connec- • impact evaluation of different amplification schemes or
tion requests are established on-demand and released after 3R regeneration on the network performance, assuming
transmission [5]), EONs also have the potential of improving a given placement technique [9], [24]–[27].
resource utilization in contrast to static operation (resources In the first case, strategies that selectively upgrade EDFA-
permanently allocated to each connection [6]). In this work, based line optical amplifiers to HFA to reduce network
we focus on dynamic EON architectures. cost (in terms of the required number of 3R regenerators)
One of the main tasks dynamic EON operators must solve [20], [22] or blocking probability [19] have been proposed
is resource allocation. That is, finding a path, a modulation in fixed-grid WDM network scenarios. In EON scenarios,
format suitable for the path length, and a portion of the various algorithms have been proposed to solve the regen-
optical spectrum on the path that meets the contiguity and erator placement problem aiming to reduce power consump-
continuity constraints, known as the Routing, Modulation tion [18] or reducing blocking probability [21]. Additionally,
Level, and Spectrum Assignment (RMLSA) problem [4]. a solution for the regenerator assignment problem in dynamic
Usually, the performance of RMLSA solutions for dynamic EONs was proposed in [23], aiming to minimize the number
networks is contrasted in terms of the blocking ratio [6]. of 3R regenerators and frequency slots used.
Thus, an RMLSA solution is better than another if it consis- In the second case, previous studies demonstrate that
tently exhibits a lower blocking ratio for different scenarios. the use of HFA reduces spectral occupancy (i.e., more
In wide-area dynamic EON, there are two primary sources connections can be accommodated with the same capac-
of blocking: a lack of capacity to establish connections and ity) compared to pure EDFA in static EON architectures
the inability to provide an acceptable quality-of-transmission [24], [25]. In dynamic EON scenarios, it has been shown
(QoT) to the network connection, despite capacity avail- that the blocking probability can be decreased using 3R
ability [5], [7]–[9]. The former case arises when there are regenerators [9], [26]. However, the addition of regeneration
effectively no available slots to meet the requested bandwidth devices exhibits diminishing returns in terms of blocking
or having available slots these cannot be used due to spectrum gain [27].
fragmentation, which occurs when the available slots are To the best of the authors’ knowledge, previous research
isolated and they are neither continuous along the optical has not compared the performance of pure amplification
path nor contiguous on the spectrum domain [10], [11]. The schemes and the use of 3R regenerators in dynamic EON
latter case arises when the length of the route is such that architectures. Neither have they evaluated the trade-off
the accumulation of physical layer impairments (PLI), such between the performance and cost of both cases. Such
as amplified spontaneous emission (ASE) noise from optical research can be beneficial for network operation and design
amplifiers and nonlinear distortions from Kerr nonlinearity, by providing elements that help the decision-making related
degrade the signal-to-noise ratio (SNR) and thus, the bit- to what devices to deploy to increase optical reach in elastic
error-rate (BER), beyond acceptable limits. As in [9], we term optical networks.
these types of blocking as capacity blocking and reach block- In this work, we study the impact of using different
ing, respectively. amplification schemes and 3R regeneration on the blocking
Standard approaches to solve the RMLSA problem estab- probability, energy consumption, and spectral and energy
lish end-to-end all-optical communication (i.e., transpar- efficiency of dynamic EONs on two network configurations:
ent) relying only on erbium-doped fiber amplifiers (EDFAs) transparent and translucent. The transparent network configu-
to compensate losses between source and destination ration used different amplification schemes (all-EDFA, HFA,
nodes [5], [7], [8], [12]–[16]. In these cases, reach blocking or all-DFRA) without 3R regenerators, while the translucent
has been the central problem in wide-area networks [5], one uses only 3R regenerators and EDFAs. The comparative
[7]–[9]. This situation can be mitigated by increasing the study models the physical layer for all network configu-
maximum transmission reach of optical signals. Most alter- rations determining optical signals’ maximum transmission
natives include equipping the network with distributed fiber reach. This information is then fed to an RMLSA algorithm
Raman amplifiers (DFRA) or hybrid Raman/EDFA fiber operating in a dynamic scenario. As a case study, we quantify
amplifiers (HFA) [17], and 3R (re-amplification, re-timing, the network performance of the different configurations in the
and re-shaping) signal regeneration in some (or all) network NSFNet and UKNet topologies.
nodes [9], [18]. There have been other proposals to extend The remainder of this paper is structured as follows:
optical transmission reach [19], [20], but they do not out-
Section II describes the physical layer model used in this
perform the use of HFA [20], or may not be practical to
work. Next, Section III presents the methodology used to
implement [19].
Previous works studying amplification schemes and regen- compare the different network configuration scenarios, which
eration in optical networks have either focused on: includes the performance metrics, the RMLSA algorithm
• 3R regenerators/HFA placement or assignment tech- and, the network simulator tool. Numerical results are then
niques for improving network performance or minimiz- presented in Section IV. Finally, conclusions and remarks are
ing network cost [18]–[23], provided in Section V.

134356 VOLUME 9, 2021

A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

II. PHYSICAL LAYER MODEL optoelectronic conversion. As shown in Fig. 2, in this

This section describes the amplification and regeneration configuration, all-EDFA is used in each span.
network configurations studied throughout this work and the The configurations selected for this study were cho-
model used to calculate the maximum achievable reach of sen for the following reasons: NCI 1 because EDFA is
optical signals. the most used technology for the deployment of opti-
cal links [28], NCI 2 and 3 due to the benefits pre-
A. NETWORK CONFIGURATIONS sented by Raman amplification in hybrid schemes and to a
We consider five different network configurations, each asso- greater extent individually, in terms of noise accumulation
ciated with a network configuration identifier (NCI). Con- [17], [29], NCI 4 as an intermediate alternative between
figurations 1-4 correspond to transparent configurations, and NCI 1 and 3 with improved blocking rate without signifi-
configuration 5 is translucent. cant energy consumption increase, and NCI 5 with all-EDFA
amplification in order to avoid combining two elements with
a: TRANSPARENT CONFIGURATIONS high energy consumption and costs such as regenerators and
DFRA [18]–[20].
In transparent configurations, information is transmitted from
source to destination entirely in the optical domain, imply-
B. OPTICAL TRANSMISSION REACH MODEL
ing that all intermediate operations, such as switching and
amplification, are optical. Fig. 1 (a) shows a schematic of the We assumed that signals are degraded by ASE and non-
transparent link configuration: the signal is optically trans- linear interference (NLI) noise to calculate the maximum
mitted through N spans with switching occurring at inter- distance achievable by any given modulation format. Other
mediate nodes if required. Each span is a segment (typically fiber impairments, such as chromatic and polarization mode
between 80 and 100 km long) made of standard single-mode dispersion, are assumed to be compensated for using digi-
fiber (SSMF) plus one amplification point. Although this tal signal processing [30]. Furthermore, we assume that all
figure only shows one switching point (OXC box in the network components are carefully designed to avoid any
middle of the two depicted spans), a signal can be switched additional sources of signal degradation.
at several points along the path or not switched at all if an We evaluate the signal QoT of each optical path in terms
optical fiber link directly connects the source and destination of its received SNR [30]:
nodes. Optical amplification for one span can have one of the Pin
SNR = , (1)
following configurations: PASE + ψNLI P3in
• NCI 1 (all-EDFA): at the end of each span an EDFA
where Pin is the input signal power, PASE is the ASE noise
is placed to compensate for the losses, as shown power over the signal bandwidth, and ψNLI is the NLI
in Fig. 1 (b). coefficient.
• NCI 2 (HFA): a Raman pump provides distributed
The total ASE noise at the receiver was calculated
Raman gain GR , followed by an EDFA with gain GE , using [31]:
recovering the residual loss. The total gain of the HFA
is G = GR GE . This configuration is shown in Fig. 1 (c), PASE = Ns Feq hv1ref (G − 1), (2)
characterized by the Raman gain ratio (0R ) correspond-
where Ns is the number of spans, Feq is the equivalent
ing to the percentage of the total gain (in dB) compen-
amplifier noise figure, h is Planck’s constant, v is the carrier
sated by the Raman amplifier. Note that the cases when
frequency, 1ref is the signal bandwidth and G is the amplifier
0R = 0 and 0R = 1 correspond to all-EDFA and
gain.
all-DFRA, respectively.
For each amplifier configuration Feq takes a different
• NCI 3 (all-DFRA): a Raman pump is used to provide
value. We denote by FEDFA , FDFRA , and FHFA [32] the equiv-
gain in each span, as shown in Fig. 1 (d).
alent noise figure of the EDFA, DFRA, and HFA amplifiers,
• NCI 4 (alternating spans): one span uses EDFA and the
respectively. The right column of Table 1 shows the values
one immediately after uses DFRA, as shown in Fig. 1 (e).
used for FEDFA , FDFRA and FHFA . FEDFA is used in NCI 1 and
NCI 5 for each span and in NCI 4 for spans using EDFA. FHFA
b: TRANSLUCENT CONFIGURATION is used in NCI 2 for each span, depending on 0R value. FDFRA
In translucent communications, data is transmitted from is used in NCI 3 in each span and NCI 4 in spans using DFRA.
source to destination using at least one optoelectronic con- In every network configuration (NCI 1 - 5), the amplifiers
version device. In our case, we use 3R regeneration devices were set to compensate for the loss of one span and include a
for increasing the optical reach in intermediate nodes, as dis- 2 dB of margin, thus G = exp(αL) · 1.58, where α is the fiber
played in Fig. 2. Therefore, the fifth and last configuration is attenuation coefficient, and L is the span length. The values
as follows: of α and L can be found in the left column of Table 1. Finally,
• NCI 5 (translucent): the information transmitted from for calculating the maximum distance achievable by optical
source to destination is regenerated in an interme- signals, it is assumed that links are made of identical fiber
diate node using a 3R regenerator equipped with spans.

VOLUME 9, 2021 134357

 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

FIGURE 1. Transparent configuration. (a) Transparent link configuration. (b)-(e) Span configuration for different amplification configurations
(NCI 1-4).

TABLE 1. Fiber parameters for maximum achievable reach calculation for

every NCI.

FIGURE 2. Translucent link configuration (NCI 5).

The NLI coefficient, ψNLI , is given by the product

1 span
Ns ψNLI [30], [33] assuming incoherent NLI accumulation,
1 span
where ψNLI is the NLI coefficient in a single span. The
ψNLI coefficients were estimated using the conventional GN
model [30] when EDFA amplification is used (NCI 1, 4 and 5) The maximum achievable reach (MAR) for each modula-
and a closed-form approximation from [29] for the configura- tion format was defined as the maximum distance that the
tions using HFA (NCI 2) and DFRA (NCI 3 and 4). The right optical signal can be transmitted at optimum power before
column of Table 1 lists the value of ψNLI for 1 span for the exceeding the target BER. Based on that target BER, the SNR
different network configurations. We denote by ψNLI −EDFA , thresholds for each modulation format were determined:
ψNLI −DFRA and ψNLI −HFA the NLI coefficients when EDFA, 8.5, 15.5, and 21 dB for PM-QPSK, PM-16-QAM, and
DFRA, and HFA amplification is used, respectively. The PM-64-QAM, respectively. The calculated MAR values for
value ψNLI −EDFA is used in each span in NCI 1 and NCI 5 and the available modulation formats are listed in Table 2. The
only in the spans using EDFAs in NCI 4. The value ψNLI −HFA values reported have been rounded down to the closest mul-
is used in each span in NCI 2, according to the value of 0R . tiple of the span length. The MAR represents the distance
Finally, the value ψNLI −DFRA is used in each span in NCI 3, constraint in the modulation level selection in the RMLSA
and in the spans using DFRA in NCI 4. Since the value of algorithm. All fiber parameters used for the calculations of
the NLI coefficient depends on the optical bandwidth used, the maximum achievable reach used throughout this work are
the calculation of it considered a fully-loaded C-band with an listed in Table 1 where parameters αp , β2 , γ , R, and gR are
optical bandwidth of 4 THz divided in 320 FSUs of 12.5 GHz the pump attenuation coefficient, group velocity dispersion
each. coefficient, nonlinear coefficient, symbol rate, and Raman
PM-QPSK, PM-16-QAM, and PM-64-QAM are the avail- gain coefficient, respectively.
able modulation formats within the flexible transceivers, each
with a target pre-FEC BER of 4.7 · 10−3 using 6.25% over- III. METHODOLOGY
head [34]. In this case, the bit rates allowed by each modu- In this section, we describe the methodology used for the
lation format on an optical channel with a bandwidth of one comparison of NCI 1-5, described in Section II. The net-
FSU are 46, 92, and 140 Gb/s for PM-QPSK, PM-16-QAM work performance metrics are described in Subsection III-A
and PM-64-QAM, respectively. (blocking probability, energy consumption, and spectral and

134358 VOLUME 9, 2021

A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

TABLE 2. Maximum achievable reach (in km) for various modulation amplifiers and the regenerators over the observation time:
formats, using BER = 4.7 · 10−3 and 320 FSUs optical bandwidth for X X
every NCI.
ENCI = PeAMP T + PeRGN τ, (4)
span connection

where the first term on the right side of the equation corre-
sponds to the energy consumed by the optical amplifiers for
a given NCI during the observation time T , and the second
term is the energy consumed by all connections using a
regeneration device.
energy efficiency). The RMLSA algorithm and the simula- In Eq. (4), PeAMP corresponds to the electrical power con-
tion scenario are explained in subsections III-B and III-C, sumed by an optical amplifier, T is the cumulative work time,
respectively. PeRGN is the electrical power consumed by a regenerator for a
single connection, and τ is the holding time of the regenerated
A. PERFORMANCE METRICS connection. Note that the amplifier power consumption is
We define three performance metrics for evaluating all ampli- time-independent due to the fully loaded system assumption
fication/regeneration configurations on dynamic elastic opti- described in the physical layer model. The ENCI value in
cal networks. Eq. (4) is expressed in Joules [J].
The power consumption of optical amplifiers depends on
1) BLOCKING PROBABILITY the network configuration selected. We use the model for
The network must be designed so that the blocking proba- HFA presented in [19], [35] which describes the behavior
bility does not exceed a threshold value. In wide-area elastic of all configurations studied herein. We calculate the power
optical networks, a request can be blocked due to one of the consumption of an optical amplifier for a single span using:
 
following situations: First, lack of available resources (FSUs) 1 1 NR ln(GR )
PeAMP = Nch Pch 1 − + , (5)
on the selected path. Second, the selected route has enough ηEDFA GE ηR gR Leff
FSUs to accommodate the request, but the optical signal
where ηEDFA and ηR are the power conversion efficiency for
cannot reach the destination since it does not meet the quality
EDFA and Raman amplifiers, respectively; Nch and Pch are
of transmission (QoT) required. We term these two blocking
the number and power for the channels amplified by the
events as Capacity Blocking (CB) and Reach Blocking (RB),
EDFA; and NR , gR , and Leff are the Raman pump multiplicity,
respectively.
Raman gain coefficient, and effective amplification length,
Blocking events are recorded as follows. First, resource
respectively. The first term on the right side of Eq. (5) corre-
availability is checked. If resources are not available,
sponds to the electrical power required to pump the EDFA
the request is considered to be blocked due to the lack of
section with a GE gain value. The second term is equiva-
capacity, and QoT is not evaluated. Once resource avail-
lent for the Raman amplifier with a gain GR . In all-EDFA
ability has been confirmed, QoT is evaluated. If the QoT
(NCI 1) and all-DFRA (NCI 3) configurations, the second
requirements are not met, the request will be blocked and the
and first terms of the right side of Eq. (5) are equal to 0,
event will be recorded as a reach blocking event. In this way,
respectively.
the overall blocking probability (B) can be computed as the
The power consumption of 3R regenerator devices varies
sum of these two blocking events:
depending on the data rate of the regenerated signal. Here we
assume that a regenerator consists of a variable bandwidth
B = CB + RB. (3) receiver and a transmitter in a back-to-back configuration.
The model used to describe the electrical power consumption
2) ENERGY CONSUMPTION of a regenerator as a function of the bit rate [36]–[38] is as
The main drawback of the network configurations used follows:
to extend the reach of an optical connection described in
PeRGN = K1 × Bsd + K2 , (6)
Section II is their high energy consumption, leading to a
trade-off between blocking probability and energy consump- where K1 is a power scaling factor related to the transmission
tion. This work analyzes the energy consumed by each bit rate and the required DSP, Bsd is the bit rate of a connection
NCI configuration considering the optical amplifiers in the request between nodes s and d, and K2 is a constant consumed
transparent configurations and 3R regenerators and optical power related to the operation of the optical source and analog
amplifiers in the translucent case. We assume that the energy components. The values used for the constant values K1 and
consumed by network elements different from amplifiers and K2 depend on the studied transceiver, and different values
regenerators (e.g., optical transceivers and switches) remains are observed within the literature. The study from [36] is
constant during network operation. commonly used within the EON literature. In this case K1
The total energy consumed by a given network config- and K2 correspond to 1.683 W/Gb/s and 91.3 W, respectively
uration, ENCI , is the sum of the energy consumed by the [36]–[40], representing a power consumption of 260 W for

VOLUME 9, 2021 134359

 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

a 100 Gb/s transmission. However, these values correspond With each attempted route, the algorithm determines the mod-
to a study from 2010 [40]. Thus the advances in transceiver ulation format that requires the lowest number of FSUs while
manufacturing are not considered. We additionally consider having an optical reach greater than or equal to the length of
a CFP digital coherent optical (DCO) transceiver, where K1 the selected path. Next, the spectrum allocation is performed
and K2 are equal to 0.105 W/Gb/s and 21.5 W [41], [42], by executing the First-Fit (FF) algorithm, which exhibits
respectively. That presents a maximum power consumption a good trade-off between performance and computational
of just 32 W for 100 Gb/s transmission, eight times lower complexity [9]. If the connection cannot be established after
than the power consumption of devices from 2010. attempting all paths, the request is rejected (for NCI 1-4),
or Stage B is activated to attempt to establish a translucent
3) SPECTRAL AND ENERGY EFFICIENCY connection using one regenerator in the path (for NCI 5).
We study the efficiency of each network configuration In Stage B, it is assumed that all network nodes are
using the network spectral efficiency (SE) and the energy equipped with the same number of 3R regenerators and that
efficiency (EE) metrics. Both metrics, SE and EE, are associ- they are used only if a connection request cannot reach the
ated with the cost incurred to operate the network for a target destination due to limited MAR or if the capacity on the path
throughput. SE indicates how well the spectral resources of is not continuous or contiguous. The algorithm attempts to
the network are utilized and EE how much energy is required establish a translucent connection in one of the K shortest
to operate the network. paths, in increasing order of length. In that case, the path
SE is defined as the ratio between the total network is divided into two segments as described in [9], using the
throughput and the allocated spectrum during an observation regenerator allocation algorithm based on the First Longest
period [43]. The network throughput is calculated as the Reach Regenerator (FLR) policy [47]. If the node between
sum of the bit rates of all established connections during the both segments has at least one regenerator available, then
observation time T . The allocated spectrum is the sum of the algorithm allocates resources for the first segment. If this
the required bandwidth for all connections. allocation is successful, then the resource assignment for
the second segment is attempted. It is assumed that the regen-
∀(s,d) Bsd · τ
P
SE = P , (7) eration device can convert the modulation format and change
∀(s,d) 1fsd · τ the spectral position if required. If the connection between the
where 1fsd is the optical bandwidth in Hz allocated to con- nodes s and d cannot be established using these two segments
nection (s, d), equal to the number of FSUs required by in the candidate path, the algorithm attempts establishing the
connection (s, d) multiplied by the spectral width of an FSU. connection using a new pair of segments. This procedure
EE is the energy required to achieve a given network is repeated until the connection can be established, or all
throughput during an observation time T . It is expressed segment combinations in the candidate path have been unsuc-
in J/s, and it represents the energy-per-bit, a fundamental cessful. If the translucent connection cannot be established
unit to measure energy efficiency in digital communications in the shortest path, Stage B is re-started considering the
[44], [45]. It was calculated as follows: following shortest path until every path in K − SPs,d has been
ENCI evaluated. If none of these attempts is successful, then the
EE = P , (8) connection request is finally rejected.
∀(s,d) Bsd · τ
where ENCI is the energy consumption of amplifiers and C. SIMULATION SCENARIO
regenerators of a given network configuration (Eq. (4)). The network performance was evaluated using an event-
driven simulator built in C++. The operation of a flexible
B. RMLSA ALGORITHM optical network was simulated considering as inputs: the
The RMLSA algorithm used was presented in [9]. The network NCI, the network topology, the FSU capacity per
algorithm establishes connections taking into account a link, the RMLSA algorithm, the number of regenerators per
given BER threshold. Fig. 3 shows a high-level descrip- node, the set of bit rates and modulation formats supported
tion of the main steps executed by the RMLSA algorithm. by nodes, the optical reach table (MAR) of different network
In it, connection requests generated in transparent network configurations and the number of FSUs required by different
configurations (NCI 1-4) are processed executing Stage modulation formats and bit rates (Table 3). Table 4 lists the
A only, and connection requests generated in the translu- values of all the input data parameters used in the simulation.
cent network configuration (NCI 5) are processed executing Each event in the simulator represents a connection request
Stages A and B. or release. Each connection request is defined by a source-
Upon receiving a connection request between nodes s destination pair and a bit rate. All possible source-destination
and d with a bit rate of Bs,d , the algorithm attempts to pairs generate traffic according to an ON-OFF traffic model,
establish a transparent connection (Stage A). The algorithm that assumes limited traffic sources, in which a maximum of
looks for available capacity in the candidate K shortest paths one connection is established between each node pair at any
(K − SPs,d ) in increasing order of length. These routes have time. That is, if successfully established, the connection is
been previously calculated using the Dijkstra algorithm [46]. active during the ON time. During the ON period, the source

134360 VOLUME 9, 2021

A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

FIGURE 3. RMLSA flowchart.

transmits at the requested bit rate. After this time, a con- TABLE 3. Spectrum requirements in terms of FSUs for each bit rate and
modulation format pair.
nection release event is generated, and the simulator releases
the corresponding allocated resources. The connection is then
inactive during the OFF time, after which a new connection
request event for this specific source-destination pair is gen-
erated. Each time a new request event is generated, its bit rate
is uniformly selected from the set of available bit rates listed
in Table 4.
The ON and OFF periods are assumed to be expo-
nentially distributed, with mean values denoted by τ
and tOFF , respectively. The network traffic load, denoted
by %, is given by:
τ
%= . (9)
τ + tOFF
For each value of traffic load studied, the simulator output
data (number of connections established, number of connec-
tions rejected, and utilization of FSU and regenerators) is
then used to calculate the blocking probability, the energy
consumption, and the spectral and energy efficiency of each
network configuration.
FIGURE 4. NSFNet network topology with the lengths of every optical link
(in km).
IV. RESULTS
The performance of the different network configurations
NCI 1-5 was evaluated employing simulation. We use Raman
gains coefficient of 0.25, 0.50 and 0.75 for NCI 2 con- were carried out using the UKNet network topology with
figuration, and 1, 3, 5, and 10 3R regenerators per node shorter links compared to NSFNet. UKNet presents a net-
for NCI 5 scenario. Network simulations were run using work diameter of 1000 km. Overall, the same trend in
the NSFNet topology, shown in Fig. 4. The values of the the results for NSFNet was observed in the UKNet. Due
input data required for the network simulator are summarized to space constraints, in this paper, we present the results
in Table 4. For every network configuration, the blocking obtained for the NFSNet topology only. Data supporting the
probability, the energy consumption, and the spectral and results of both network topologies can be found in https://
energy efficiency were evaluated. Additionally, simulations iro-team.gitlab.io/IA3R_ext.pdf.

VOLUME 9, 2021 134361

 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

FIGURE 5. Blocking probability for the network for all network configurations.

TABLE 4. Simulation parameters. It can be seen that although a higher number of regenerators
reduces the blocking probability, the addition of more than
three devices per node offers diminishing returns, and the
improvement in the blocking ratio is marginal. For example,
considering a traffic load of 0.3, increasing the number of
regenerators per node from 3 to 10 only decreases the block-
ing probability from 3.0 · 10−2 to 1.5 · 10−2 .
NCI 4, corresponding to the alternate span configuration,
offers a reduced blocking probability than NCI 1 and NCI 5.
For example, considering a traffic load of 0.3, the blocking
probability using NCI 4 was 2.93 · 10−4 , almost three orders
of magnitude less than NCI 1 and one order of magnitude
lower than NCI 5 (average value of 2.45·10−2 ). This behavior
is obtained due to the increased reach for all modulation
A. BLOCKING PROBABILITY formats compared to NCI 1 and 5, as seen in Table 2.
Fig. 5 shows the network blocking probability for all the NCI 2 and 3 present a significant improvement in the
studied network configurations as a function of the traffic blocking probability compared to the other studied config-
load. urations. For NCI 2, 3 different Raman gain ratios (0R ) were
NCI 1, based on all-EDFA amplification, exhibits the worst studied: 0.25, 0.50 and 0.75. It can be seen that as 0R is
performance of the studied cases with an average block- increased, the blocking probability is reduced. For example,
ing probability of 3.3 · 10−1 for all traffic loads. The fact considering a traffic load of 0.3, a blocking probability of
that the blocking does not change with the traffic load sug- 1.36 · 10−4 , 3 · 10−6 , 4 · 10−6 and 1 · 10−6 was observed
gests that the main reason for blocked requests, in this case, for a value 0R equal to 0.25, 0.50, 0.75 and 1, respectively.
is the limited reach offered by the modulation formats under Two main reasons explain this behavior: Firstly, the increase
consideration. in the MAR (see Table 2) reduces the number of connection
NCI 5 exhibits a lower blocking than NCI 1, despite having requests blocked due to the inability to reach their destina-
the same optical reach. The lower blocking of NCI 5 is tion; secondly, the extended MAR for all modulation for-
achieved thanks to the use of 3R regenerators: some connec- mats allows the users to employ high order formats (that
tions with limited optical reach (that would have been blocked require a lower number of FSUs, as shown in Table 3) and
in NCI 1) can still be established in NCI 5 by regenerating thus, the blocking ratio is reduced due to increased spectrum
the signal at an intermediate node along the route. Besides, availability.
the better blocking performance of NCI 5 compared to Finally, we studied the source of blocking for all NCIs.
NCI 1, is still worse than the remaining transparent schemes. Results are shown in Fig. 6. Each network configuration

134362 VOLUME 9, 2021

A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

3 bars are shown, corresponding (from left to right) to traffic to NCI 2 with 0R = 0.50. The high energy consumption of
loads of 0.3, 0.5, and 0.8, respectively. The main cause of NCI 4 arises from the spans amplified purely by the Raman
blocking for NCI 1 is the limited MAR for all modulation amplifier. Note that this high energy consumption leads to
formats for all studied traffic loads. Only at a traffic load a better blocking performance compared to NCI 1. However,
of 0.8 capacity blocking is observed for NCI 1, with a value it offers a higher blocking probability than any NCI 2 studied,
of 1.65 · 10−2 . On the contrary, lack of available capac- as seen in Fig. 5.
ity is the only cause of blocking for NCI 2, NCI 3, and For all network configurations using DFRA (NCI 2-4),
NCI 4. Note that following the trend observed in Fig. 5, the high energy consumption is related to the Raman gain
the blocking ratio increases with the traffic load. For NCI 2-4, coefficient of a single-mode fiber. As stimulated Raman scat-
the MAR is enough to establish every connection request tering is a nonlinear effect, high power pumps are required to
independent of the traffic load. The blocking performance obtain signal amplification. Typically, Raman pump power
of NCI 5 is due to a mix of limited reach and capacity, but can be as high as 500 mW, approximately an order of magni-
reach blocking dominates the performance. As the number of tude higher than EDFA pump power.
available regenerators increases, a decrease of reach blocking Finally, the total energy consumed by NCI 5 depends on the
is observed for all traffic loads, as more connections can reach energy consumed by the EDFA amplifiers (same as NCI 1),
their destination using them. As reach blocking is reduced, the number of established connections using regenerators,
an increment in capacity blocking is observed in NCI 5, when and their bit rate.
more connections are established in the network. In this configuration, we used two different transceiver
The difference in performance between the transparent models to evaluate the energy consumption of regenerators,
(NCI 2-4) and translucent (NCI 5) solutions to improve the as described in Section III. The blue bars in Fig. 7 (b) rep-
transmission reach is because a regenerator can only serve a resent the energy consumption when the transceiver model
single user at a time to extend its transmission reach. This from [36]–[38] is considered, called Co-BVT herein. This
situation means that only a limited number of connections is the most common model used in previous work evaluat-
can be established using regenerators, and other requests that ing the energy consumption of regenerators and the values
cannot reach their destinations are blocked. On the other considered date from circa 2010. Additionally, the energy
hand, NCI 2-4 offers lower noise accumulation across the consumption of NCI 5 considering state-of-the-art values for
entire available spectrum, improving the quality of service the regenerator energy consumption, corresponding to CFP
for all network users. transceivers [41], [42] was studied, called DCO herein. The
energy consumption of this type of regenerator is shown in
B. ENERGY CONSUMPTION red bars in Fig. 7 (b). For a traffic load of 0.3, a low number
To evaluate the network’s energy consumption, we evaluate of connections use the available regenerators. On increas-
the simulation after 1000 time units for every NCI under ing the number of regenerators per node from 1 to 10,
investigation for traffic loads of 0.3, 0.5, and 0.8. We define no consumed energy addition is experienced, with a value
the duration of a single time unit as 1 second; thus, T = of 3.6 MJ and 11 MJ for the Co-BVT and DCO transceiver
1000 s and τ = 1 s were used to evaluate the total energy con- models, respectively. For higher traffic loads, however, differ-
sumption. As stated in Section III, only the energy consumed ent energy consumption is observed when more regenerators
by the optical amplifiers and regenerators was considered for are available. The consumed energy increases together with
this analysis, assuming that the energy consumption of all the traffic load and the number of available regenerators
other network elements in the different network configura- due to the higher number of connections using them. For
tions does not change. To evaluate the consumption of the a traffic load of 0.5, the consumed energy increased from
amplifiers using Eq. (5) NR was set as 2, conventionally used 3.9 MJ to 6.1 MJ, and from 10.4 MJ to 22 MJ when 1 and
to pump the C-band, and power conversion efficiencies for 10 regenerators were available, considering the DCO and
EDFA (ηEDFA ) and Raman (ηR ) of 5% and 3% were assumed Co-BVT transceiver models, respectively.
as in [35]. Two main conclusions can be drawn from the energy con-
Fig. 7 shows the total energy consumed by amplifiers and sumption using regenerators. Firstly, the significant differ-
regenerators for each network configuration, where Fig. 7 (a) ence observed between the two transceiver models, which
shows the transparent cases (NCI 1-4) and Fig. 7 (b) shows highlights the relevance of using state-of-the-art compo-
the translucent cases (NCI 5). For NCI 1-4, the energy nents, can greatly impact network operation costs. Secondly,
consumption is independent of the network’s traffic load, the energy consumption only depends on the network traffic
as the amplifiers were assumed to provide gain over the load in the translucent scenario, since in the transparent case,
entire available spectrum. NCI 1 exhibits the lowest energy no additional equipment is needed at different traffic loads.
consumption of 2.7 MJ due to the high power efficiency of
EDFA amplifiers. NCI 2 and 3, representing scenarios with
increased Raman gain ratio values, present a linear increase C. SPECTRAL AND ENERGY EFFICIENCY
in consumed energy with values ranging from 5.7 to 14.1 MJ. Fig. 8 (a) shows the spectral efficiency (SE) achieved by the
NCI 4 exhibits an energy consumption of 9.2 MJ, comparable different network configurations for a traffic load equal to 0.3.

VOLUME 9, 2021 134363

 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

FIGURE 6. Sources of blocking -reach blocking (RB) and capacity blocking (CB)- for every network configuration.

As shown in Fig. 8 (b), this behavior occurs because the

use of a higher Raman gain ratio allows for more efficient
modulation formats, as PM-16-QAM and PM-64-QAM, to be
used. For example, for NCI 1, only 8% of connections
were established using PM-16-QAM, and the remaining 92%
of them were established using PM-QPSK. Instead, in the
case of NCI 3 (0R = 1) 51%, 45% and 4% of the con-
nections used PM-QPSK, PM-16-QAM and PM-64-QAM,
respectively. These percentages explain the observed increase
in the spectral efficiency of the network in NCI 3 compared
to the NCI 1 shown in Fig. 8 (a).
Notice that the maximum achievable SE is 3.2 b/s/Hz,
which is lower than the SE achieved using QSPK over two
polarizations. This value can be explained considering the
requested bit rates used in the simulations, the available
modulation formats, and the constraints of EONs that an
established connection uses a minimum of 1 FSU. The sim-
ulated bit rate requests were 10, 40, 100, 400 and 1000 Gb/s.
In particular, bit rates of 10, 40 Gb/s require less than a
single FSU of optical bandwidth using the modulation for-
mats under consideration, which reduces the overall SE as
no other connections are established using the same FSU.
FIGURE 7. Network energy consumption considering T = 1000 s. Remark that, due to the uniform traffic distribution assumed
in this work, 10 and 40 Gb/s represent 40% of the simulated
connection requests.
Among the transparent network configurations (NCI 1-4), For the translucent configuration (NCI 5), only a slight
NCI 1 presents the lowest SE, equal to 2.65 b/s/Hz. Com- improvement compared to NCI 1 is observed. In this case,
pared to NCI 1, SE has increased in NCI 2, 3, and 4 thanks a SE of 2.675 b/s/Hz was obtained, 0.025 b/s/Hz higher than
to Raman amplification. The higher the value of the fac- NCI 1. That is, the use of regenerators to increase the reach
tor 0R , the higher the observed SE (with NCI 4 achieving of a connection does not lead to a more significant usage of
a spectral efficiency similar to NCI 2 with 0R = 0.25). more efficient modulation formats, as shown in Fig. 8 (b),

134364 VOLUME 9, 2021

A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

FIGURE 8. Spectral efficiency of the network for every NCI for a traffic FIGURE 9. Energy efficiency of the network for every NCI.
load of 0.3.

where a maximum of 17% of PM-16-QAM connections is overall energy consumption, and the number of established
observed for NCI 5. This behavior is explained considering connections will determine the energy efficiency. For a traffic
the limited number of regenerators per node, which translates load of 0.3, we observe a constant energy per bit of 0.267
into a maximum number of regenerated connections that and 0.666 J/b for the DCO and Co-BVT, respectively. As the
can be established simultaneously. Thus, only a fraction of traffic load increments, the required energy per bit is reduced
established connections could use modulation formats more as a greater throughput is established in the network.
efficiently than NCI 1. In particular, approximately 30 % of Noticeably, NCI 1 presents the best energy efficiency for
the established connections use regenerators, and only 30% of all traffic loads. However, it offers the highest blocking
those connections can improve their SE using PM-16-QAM probability. The use of regenerators (NCI 5) will strongly
due to the reach limitations of the studied formats. The depend on the deployed technology, with the DCO transceiver
network SE obtained for traffic loads of 0.5 and 0.8 was offering a consumed energy per bit similar to NCI 1 (0.27 J/b).
similar to that shown in Fig. 8 (a) for a traffic load of 0.3. The Co-BVT presents a consumed energy per bit comparable
In particular, the same trend was observed when changing to NCI 2 with 0R = 0.5, and the latter offers considerably
the NCI. lower blocking probabilities.
The energy efficiency of every network configuration is
shown in Fig. 9 considering traffic loads of 0.3, 0.5, and 0.8. D. FINAL CONSIDERATION
Fig. 9 (a) presents the energy efficiency for NCI 1-4, and The network configuration, in terms of the amplification and
Fig. 9 (b) for NCI 5. Similar to Fig.7 (a), for NCI 1-4 a regeneration scheme selected, impacts the service perfor-
higher energy per bit is required as the 0R increases from mance parameters directly in the service level agreement for
0 to 1 for all traffic loads. For every NCI, as the traffic load optical networks (O-SLA) [48] subscribed between a service
increases, the required energy per bit is reduced due to the provider and a subscriber, in particular the throughput. Since
greater throughput in the network. As the required energy for the network operator aims to reduce energy consumption
the transparent scenarios (NCI 1-4) is constant, the energy while meeting the specified values of the O-SLA, the best
efficiency varies only according to the number of established amplification and regeneration scheme will be determined by
connections. the quality of service required by connections. For example,
Unlike amplifiers, regenerators only consume energy when in a network serving high-throughput connections, NCI 3
they are operative. As seen in Fig. 7 (b), both the traffic would be the best choice. However, this is also the con-
load and the number of available regenerators increase the figuration with the highest energy consumption. Instead,

VOLUME 9, 2021 134365

 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

connections with meager throughput requirements could be All network configurations (NCI 1-5) assume that the ASE
served with the lowest energy consumption NCI 1. noise is an Additive White Gaussian Noise process and do
not consider wavelength-dependent noise and gain, which
V. CONCLUSION might lead to a different reach for every multiplexed chan-
We have studied the impact of the physical layer imple- nel. Other assumptions described in the manuscript include
mentation on the blocking probability and energy consump- ideal network components and ideal compensation of linear
tion of dynamic EONs, using two network topologies and impairments using digital signal processing. Despite all the
considering transparent and translucent network configura- mentioned limitations, a 2 dB margin for the amplifier gain in
tions. The transparent configurations were all-EDFA, hybrid each network configuration was included to reduce the impact
Raman-EDFA, all-DFRA, and alternating EDFA and DFRA of the limitations mentioned herein.
spans, termed NCI 1-4. The translucent configuration, termed In future work, we will extend our analysis to multiband
NCI 5, used all-EDFA amplification and a limited number and multicore transmission systems, two promising alterna-
of 3R regenerators in each node. tives to enlarge the capacity of optical networks. In such
For NSFNet and UKNet topologies, the blocking proba- systems, new physical impairments need to be considered
bility of NCI 1 was the highest of the studied configurations, (stimulated Raman scattering in multiband systems and inter-
followed by NCI 5 and 4. NCI 2 and 3 offered the lowest core cross-talk in multicore fibers) that might affect the
blocking probability, which decreased as the Raman gain network performance. We will also focus on including the
ratio (0R ) was increased. For NSFNet, the limited reach regenerator/HFA placement problem in the study.
dominated the blocking in NCI 1 and 5 (reach blocking) of
the available modulations formats, while for NCI 2, 3, and 4, REFERENCES
the blocking was driven by lack of capacity (capacity block-
[1] O. Gerstel, M. Jinno, A. Lord, and S. J. B. Yoo, ‘‘Elastic optical network-
ing). On the other hand, for UKNet the blocking in all NCI is ing: A new dawn for the optical layer?’’ IEEE Commun. Mag., vol. 50,
associated with capacity blocking; due to shorter routes there no. 2, pp. 12–20, Feb. 2012.
is no reach blocking, and more complex modulation for- [2] H. Waldman, ‘‘The impending optical network capacity crunch,’’ in Proc.
Int. Opt. Photon. Conf. (SBFoton IOPC), Oct. 2018, pp. 1–3.
mats are used. In general, as the blocking probability was
[3] Spectral Grids for WDM Applications: DWDM Frequency Grid, document
reduced, all studied NCI improved their spectral efficiency ITU-T G.694.1, Telecommunication Standardization Sector of ITU, Gine-
using more complex modulation formats. For the two topolo- bra, Switzerland, 2012.
gies, NCI 1 was shown to be the most efficient configuration [4] B. Chatterjee and E. Oki, Elastic Optical Networks: Fundamentals,
Design, Control, and Management. Boca Raton, FL, USA: CRC Press,
in terms of energy consumption, while the total consumed 2020.
energy increased linearly with the 0R for NCI 2 and 3. The [5] A. Fontinele, I. Santos, J. N. Neto, D. R. Campelo, and A. Soares, ‘‘An effi-
energy consumption of NCI 4 was shown to be similar to cient IA-RMLSA algorithm for transparent elastic optical networks,’’
NCI 2 with 0R = 0.75. The use of 3R regenerators in Comput. Netw., vol. 118, pp. 1–14, May 2017.
[6] F. S. Abkenar and A. G. Rahbar, ‘‘Study and analysis of routing and spec-
NCI 5 represented an increase in energy consumption relative trum allocation (RSA) and routing, modulation and spectrum allocation
to NCI 1. In the case of the UKNet, this behavior is observed (RMSA) algorithms in elastic optical networks (EONs),’’ Opt. Switching
for moderate to high traffic loads. For low traffic loads, 3R Netw., vol. 23, pp. 5–39, Jan. 2017.
[7] V. A. C. Vale and R. C. Almeida, ‘‘Power, routing, modulation level and
regenerators are not used at all. Furthermore, the technology spectrum assignment in all-optical and elastic networks,’’ Opt. Switching
used in the regenerators has a meaningful impact on the total Netw., vol. 32, pp. 14–24, Apr. 2019.
consumed energy and the energy efficiency of the network. [8] S. Araujo, A. Soares, A. Fontinele, D. R. Campelo, J. V. dos Reis, and
Finally, we identify NCI 2 with 0R = 0.25 as a promising E. Barbosa, ‘‘Circuit reallocation strategy aware of the physical layer
effects for elastic optical networks,’’ in Proc. IEEE Symp. Comput. Com-
alternative to reduce the blocking probability significantly mun. (ISCC), Jun. 2018, pp. 970–975.
in wide-area EONs, such as NSFNet, without a prohibitive [9] F. I. Calderón, A. Lozada, D. Bórquez-Paredes, R. Olivares, E. J. Davalos,
increase in energy consumption, with a factor of 2.1, and G. Saavedra, N. Jara, and A. Leiva, ‘‘BER-adaptive RMLSA algo-
rithm for wide-area flexible optical networks,’’ IEEE Access, vol. 8,
energy efficiency, with a factor of 1.4, compared to NCI 1, pp. 128018–128031, 2020.
as long as the O-SLA requirements are met. Notice that this [10] B. C. Chatterjee, N. Sarma, and E. Oki, ‘‘Routing and spectrum allocation
recommendation is not based on financial considerations. in elastic optical networks: A tutorial,’’ IEEE Commun. Surveys Tuts.,
We remark that the financial benefit-cost should inform the vol. 17, no. 3, pp. 1776–1800, Aug. 2015.
[11] B. C. Chatterjee, S. Ba, and E. Oki, ‘‘Fragmentation problems and man-
final decision made by a network operator of any alternative: agement approaches in elastic optical networks: A survey,’’ IEEE Commun.
an alternative will be feasible only if the financial benefit of Surveys Tuts., vol. 20, no. 1, pp. 183–210, 1st Quart., 2018.
decreasing the blocking ratio (due to increased revenue) com- [12] L. R. Costa, I. B. Brasileiro, and A. C. Drummond, ‘‘Low margin QoT-
pensates for the increased cost related to energy consumption. aware RMLSA with circuit invigoration in elastic optical networks,’’ in
Proc. IEEE Global Commun. Conf. (GLOBECOM), Dec. 2020, pp. 1–6.
The above-discussed results have been obtained under sim- [13] X. Luo, Y. Zhao, X. Chen, L. Wang, M. Zhang, J. Zhang, Y. Ji,
plified models that impose some limitations on the work. H. Wang, and T. Wang, ‘‘Manycast routing, modulation level and spectrum
Regarding the limitations of the physical layer model, Raman assignment over elastic optical networks,’’ Opt. Fiber Technol., vol. 36,
pp. 317–326, Jul. 2017.
and hybrid amplifiers assume ideal depolarized pump sources
[14] B. Li and Y.-C. Kim, ‘‘Efficient routing and spectrum allocation consider-
and lossless coupling components, which might result in ing QoT in elastic optical networks,’’ in Proc. 38th Int. Conf. Telecommun.
a lower equivalent noise figure than practical components. Signal Process. (TSP), Jul. 2015, pp. 109–112.

134366 VOLUME 9, 2021

A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

[15] T. Hashimoto, K.-I. Baba, and S. Simojo, ‘‘A study on routing, modu- [37] J. Zhang, Y. Zhao, X. Yu, J. Zhang, M. Song, Y. Ji, and B. Mukherjee,
lation level, and spectrum allocation algorithms for elastic optical path ‘‘Energy-efficient traffic grooming in sliceable-transponder-equipped IP-
networks,’’ in Proc. IEEE 3rd Int. Conf. Photon., Oct. 2012, pp. 395–399. over-elastic optical networks [Invited],’’ J. Opt. Commun. Netw., vol. 7,
[16] M. Klinkowski, K. Walkowiak, and M. Jaworski, ‘‘Off-line algorithms no. 1, pp. A142–A152, 2015.
for routing, modulation level, and spectrum assignment in elastic optical [38] G. A. Beletsioti, G. I. Papadimitriou, P. Nicopolitidis, E. Varvarigos,
networks,’’ in Proc. 13th Int. Conf. Transparent Opt. Netw., Jun. 2011, and S. Mavridopoulos, ‘‘A learning-automata-based congestion-aware
pp. 1–6. scheme for energy-efficient elastic optical networks,’’ IEEE Access, vol. 8,
[17] W. S. Pelouch, ‘‘Raman amplification: An enabling technology for long- pp. 101978–101992, 2020.
haul coherent transmission systems,’’ J. Lightw. Technol., vol. 34, no. 1, [39] J. Halder, T. Acharya, M. Chatterjee, and U. Bhattacharya, ‘‘E-S-RSM-
pp. 6–19, Jan. 1, 2016. RSA: A novel energy and spectrum efficient regenerator aware multipath
[18] A. Fallahpour, H. Beyranvand, S. A. Nezamalhosseini, and J. A. Salehi, based survivable RSA in offline EON,’’ IEEE Trans. Green Commun.
‘‘Energy efficient routing and spectrum assignment with regenerator place- Netw., vol. 5, no. 3, pp. 1451–1466, Sep. 2021.
ment in elastic optical networks,’’ J. Lightw. Technol., vol. 32, no. 10, [40] J. M. H. Elmirghani, T. Klein, K. Hinton, T. E. H. El-Gorashi, A. Q. Lawey,
pp. 2019–2027, May 2014. and X. Dong, ‘‘GreenTouch GreenMeter core network power consumption
[19] J. Pedro and N. Costa, ‘‘Optimized hybrid Raman/EDFA amplifier place- models and results,’’ in Proc. IEEE Online Conf. Green Commun. (Online-
ment for DWDM mesh networks,’’ J. Lightw. Technol., vol. 36, no. 9, GreenComm), Nov. 2014, pp. 1–8.
pp. 1552–1561, May 1, 2018. [41] H. Zhang, B. Zhu, S. Park, C. Doerr, M. Aydinlik, J. Geyer, T. Pfau,
[20] A. Ferrari, M. Cantono, A. Ahmad, and V. Curri, ‘‘Physical layer strategies G. Pendock, R. Aroca, F. Liu, and C. Rasmussen, ‘‘Real-time transmission
to save lightpath regenerators,’’ IEEE/OSA J. Opt. Commun. Netw., vol. 10, of 16 Tb/s over 1020 km using 200 Gb/s CFP2-DCO,’’ Opt. Exp., vol. 26,
no. 9, pp. 703–711, Sep. 2018. no. 6, p. 6943, 2018.
[21] M. Aibin and K. Walkowiak, ‘‘Adaptive modulation and regenerator-aware [42] J. Geyer, C. Rasmussen, B. Shah, T. Nielsen, and M. Givehchi, ‘‘Power
dynamic routing algorithm in elastic optical networks,’’ in Proc. IEEE Int. efficient coherent transceivers,’’ in Proc. 42nd Eur. Conf. Opt. Commun.
Conf. Commun. (ICC), Jun. 2015, pp. 5138–5143. (ECOC), 2016, pp. 1–3.
[22] M. Cantono, A. Ferrari, U. Waheed, A. Ahmad, S. M. H. Zaidi, A. Bianco, [43] P. J. Winzer, ‘‘High-spectral-efficiency optical modulation formats,’’
and V. Curri, ‘‘Networking benefit of hybrid fiber amplification for light- J. Lightw. Technol., vol. 30, no. 24, pp. 3824–3835, Dec. 15, 2012.
path regenerators saving,’’ in Proc. Opt. Fiber Commun. Conf., 2017, [44] R. S. Tucker, R. Parthiban, J. Baliga, K. Hinton, R. W. A. Ayre, and
Paper W4F.7. W. V. Sorin, ‘‘Evolution of WDM optical IP networks: A cost and energy
[23] E. F. da Silva, R. C. Almeida, H. A. Pereira, and D. A. R. Chaves, ‘‘Assess- perspective,’’ J. Lightw. Technol., vol. 27, no. 3, pp. 243–252, Feb. 1, 2009.
ment of novel regenerator assignment strategies in dynamic translu- [45] R. S. Tucker, ‘‘Green optical communications—Part I: Energy limitations
cent elastic optical networks,’’ Photonic Netw. Commun., vol. 39, no. 1, in transport,’’ IEEE J. Sel. Topics Quantum Electron., vol. 17, no. 2,
pp. 54–69, Feb. 2020. pp. 245–260, Mar. 2011.
[24] A. Ahmad, A. Bianco, H. Chouman, G. Marchetto, S. Tahir, and V. Curri, [46] D. Eppstein, ‘‘Finding the K shortest paths,’’ SIAM J. Comput., vol. 28,
‘‘Impact of fiber type and Raman pumping in NyWDM flexible-grid no. 2, pp. 652–673, 1999.
elastic optical networks,’’ in Proc. 18th Int. Conf. Transparent Opt. Netw. [47] Ì. Brasileiro, J. Valdemir, and A. Soares, ‘‘Regenerator assignment with
(ICTON), Jul. 2016, pp. 1–4. circuit invigorating,’’ Opt. Switching Netw., vol. 34, pp. 58–66, Nov. 2019.
[25] A. Ahmad, A. Bianco, V. Curri, G. Marchetto, and S. Tahir, ‘‘Raman [48] W. Fawaz, B. Daheb, O. Audouin, M. Du-Pond, and G. Pujolle, ‘‘Service
pumping as an energy efficient solution for NyWDM flexible-grid elastic level agreement and provisioning in optical networks,’’ IEEE Commun.
optical networks,’’ Int. J. Electr. Comput. Eng., vol. 7, no. 5, p. 2627, Mag., vol. 42, no. 1, pp. 36–43, Jan. 2004.
Oct. 2017.
[26] H.-C. Le, N. T. Dang, and N. Duc Nguyen, ‘‘Impact of optical regeneration
on dynamic elastic optical networks,’’ in Proc. Int. Conf. Adv. Technol. ASTRID LOZADA received the M.Sc. degree in
Commun. (ATC), Oct. 2017, pp. 11–15. electronic engineering from Universidad Técnica
[27] D. Borquez-Paredes, F. Calderon, N. Jara, A. Leiva, A. Lozada, Federico Santa María (UTFSM), Chile, in 2020,
R. Olivares, and G. Saavedra, ‘‘3R regeneration in elastic optical networks where she is currently pursuing the Ph.D. degree
and its impact on the network quality of service,’’ in Proc. 22nd Int. Conf. in electronic engineering. Her research interests
Transparent Opt. Netw. (ICTON), Jul. 2020, pp. 1–4.
include fiber optic communication systems and
[28] G. P. Agrawal, ‘‘Optical communication: Its history and recent progress,’’
nonlinear fiber effects.
in Optics in Our Time. Cham, Switzerland: Springer, 2016, pp. 177–199.
[29] D. Semrau, G. Saavedra, D. Lavery, R. I. Killey, and P. Bayvel, ‘‘A closed-
form expression to evaluate nonlinear interference in Raman-amplified
links,’’ J. Lightw. Technol., vol. 35, no. 19, pp. 4316–4328, Oct. 1, 2017.
[30] P. Poggiolini, G. Bosco, A. Carena, V. Curri, Y. Jiang, and F. Forghieri,
‘‘The GN-model of fiber non-linear propagation and its applications,’’ FELIPE CALDERÓN received the B.Sc. degree
J. Lightw. Technol., vol. 32, no. 4, pp. 694–721, Feb. 15, 2014. in electronic engineering from the Pontificia Uni-
[31] V. Curri, A. Carena, P. Poggiolini, G. Bosco, and F. Forghieri, ‘‘Extension versidad Católica de Valparaíso (PUCV), Chile,
and validation of the GN model for non-linear interference to uncom- in 2020, where he is currently pursuing the
pensated links using Raman amplification,’’ Opt. Exp., vol. 21, no. 3, M.Sc. degree in electrical engineering. His current
pp. 3308–3317, 2013. interests include optical networking and machine
[32] A. Mitra, D. Ives, A. Lord, P. Wright, and S. Kar, ‘‘Non-linear impairment learning.
modeling for flexgrid network and its application in offline network equip-
ment upgrade strategy,’’ in Proc. Int. Conf. Opt. Netw. Design Modeling
(ONDM), May 2015, pp. 57–62.
[33] D. Ives, P. Bayvel, and S. Savory, ‘‘Adapting transmitter power and
modulation format to improve optical network performance utilizing the
JOSÉ NÚÑEZ KASANEVA received the B.Sc.
Gaussian noise model of nonlinear impairments,’’ J. Lightw. Technol.,
vol. 32, no. 21, pp. 3485–3494, Nov. 1, 2014. degree in electrical engineering from the Universi-
[34] L. M. Zhang and F. R. Kschischang, ‘‘Staircase codes with 6% to 33% over- dad de Concepción (UDEC), Chile, in 2021, where
head,’’ J. Lightw. Technol., vol. 32, no. 10, pp. 1999–2002, May 15, 2014. he is currently pursuing the M.Sc. degree in electri-
[35] L. Lundberg, P. A. Andrekson, and M. Karlsson, ‘‘Power consumption cal engineering. His current interests include elas-
analysis of hybrid EDFA/Raman amplifiers in long-haul transmission sys- tic optical networks, optical network survivability,
tems,’’ J. Lightw. Technol., vol. 35, no. 11, pp. 2132–2142, Jun. 1, 2017. machine learning, and optimization.
[36] J. López, Y. Ye, V. López, F. Jiménez, R. Duque, and P. M. Krummrich,
‘‘On the energy efficiency of survivable optical transport networks with
flexible-grid,’’ in Proc. Eur. Conf. Exhib. Opt. Commun., 2012, p. P5-05.

VOLUME 9, 2021 134367

 A. Lozada et al.: Impact of Amplification and Regeneration Schemes on Blocking Performance and Energy Consumption

DANILO BÓRQUEZ-PAREDES received the NICOLÁS JARA (Member, IEEE) received the
B.Eng. degree in telematics engineering from Uni- B.Sc. and M.Sc. degrees in telematics engineering
versidad Técnica Federico Santa María (UTFSM), from Universidad Técnica Federico Santa María
Valparaiso, Chile, in 2012, and the Ph.D. (UTFSM), Chile, in 2010, and the Ph.D. degree
degree from Universidad Adolfo Ibañez, in from the Université de Rennes I, France, in 2017,
September 2018. He also received the professional and UTFSM, in 2018. He is currently an Assis-
title in telematics engineering from UTFSM. He is tant Professor with the Department of Electron-
currently a full-time Professor with the Faculty ics, UTFSM. His current research interests include
of Engineering and Sciences, Universidad Adolfo optical networks design, networks performability,
Ibañez. His research interests include the dynamic and simulation techniques.
allocation of resources in flexible optical networks, network virtualization,
graph theory, and optimization.

ARIEL LEIVA (Member, IEEE) received the B.Sc.

RICARDO OLIVARES received the B.Sc. degree
degree in electronic engineering and the M.Sc.
in electronic engineering from Universidad Téc-
degree in electrical engineering from the Pontif-
nica Federico Santa María (UTFSM), Chile,
icia Universidad Católica de Valparaíso (PUCV),
in 1983, and the M.Sc. and D.Sc. degrees in electri-
Chile, in 2003 and 2007, respectively, and the
cal engineering from the Pontificia Universidade
Ph.D. degree from Universidad Técnica Federico
Católica do Rio de Janeiro, Brazil, in 1994 and
Santa María, Valparaiso, Chile, in 2013. He is cur-
2001, respectively. He has been with the Depart-
rently a Lecturer with PUCV. His current interests
ment of Electronic Engineering, UTFSM, since
include fiber optic communication systems and
1986, where he is currently an Assistant Professor.
optical networking.
His current interests include RF measurements,
fiber optic communication systems, fiber optical sensors, and nonlinear fiber
optics.

GABRIEL SAAVEDRA (Member, IEEE) received

ALEJANDRA BEGHELLI received the B.Eng. the B.Eng. degree in telecommunication engineer-
degree in electronic engineering and the M.Sc. ing and the M.Sc. degree from the Universidad
degree from Universidad Técnica Federico Santa de Concepción, Chile, in 2013 and 2014, respec-
María, Chile, in 1996 and 2001, respectively, and tively, and the Ph.D. degree from University Col-
the Ph.D. degree from University College London lege London (UCL), London, U.K., in 2019. He is
(UCL), London, U.K. in 2006. She is currently a currently an Associate Professor with the Universi-
Lecturer with the Department of Electronic and dad de Concepción. His research interests include
Electrical Engineering, UCL. Her current research nonlinear fiber effects, nonlinear compensation
interests include dynamic resource allocation in methods, and digital signal processing for optical
optical networks, with an emphasis on the appli- communications.
cation of artificial intelligence techniques.

134368 VOLUME 9, 2021