Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW.

Survey of Photonic Switching

Architectures and Technologies in
Support of Spatially and Spectrally
Flexible Optical Networking [Invited]
Dan M. Marom, Paul D. Colbourne, Antonio D’Errico, Nicolas K. Fontaine, Yuichiro
Ikuma, Roberto Proietti, Liangjia Zong, José M. Rivas-Moscoso, and Ioannis Tomkos

Abstract—As traffic volumes carried by optical networks

continue to grow by tens of percent year over year, we are Index Terms—Optical fiber communication; Optical net-
working; Space division multiplexing; Wavelength division
rapidly approaching the capacity limit of the conventional
multiplexing; Wavelength-selective switches.
communication band within a single-mode fiber. New mea-
sures such as elastic optical networking, spectral extension
to multi-bands, and spatial expansion to additional fiber
overlays or new fiber types are all being considered as po-
tential solutions, whether near term or far. In this tutorial I. INTRODUCTION
paper, we survey the photonic switching hardware solu-
tions in support of evolving optical networking solutions
enabling capacity expansion based on the proposed ap-
proaches. We also suggest how reconfigurable add/drop
multiplexing nodes will evolve under these scenarios and
T he traffic carried by core optical networks as well as
the per-channel interface rates required by routers
are growing at a remarkable pace, ranging between 30%
gauge their properties and relative cost scalings. We iden-
tify that the switching technologies continue to evolve and
and 60% year-over-year [1,2]. Fiber-optic transmission
offer network operators the required flexibility in routing and optical networking advancements have so far satisfied
information channels in both the spectral and spatial do- these traffic requirements by delivering the burgeoning con-
mains. New wavelength-selective switch designs can now tent in a cost- and energy-efficient manner. However, we are
support greater resolution, increased functionality and approaching fundamental spectral efficiency limits of single-
packing density, as well as operation with multiple input
and output ports. Various switching constraints can be mode fibers and the growth capabilities of conventional
applied, such as routing of complete spatial superchannels, WDM networks operating on a fixed frequency grid are quite
in an effort to reduce the network cost and simplify the limited. While recent advancements and commercial avail-
routing protocols and managed pathway count. However, ability of multiple-bit-per-symbol modulation formats, in
such constraints also reduce the transport efficiency when
the network is only partially loaded, and may incur frag-
conjunction with coherent optical reception and polarization
mentation. System tradeoffs between switching granular- division multiplexing, obtain fiber link capacities up to
ity and implementation complexity and cost will have to be 20 Tb/s, further increases in capacity and spectral efficiency
carefully considered for future high-capacity SDM–WDM will be very challenging due to nonlinear distortions on ac-
optical networks. In this work, we present the first cost count of the optical Kerr effect [3,4]. New optical networking
comparisons, to our knowledge, of the different approaches
in an effort to quantify such tradeoffs. solutions need to be identified to resolve this situation. This
paper surveys recent technology innovations in the trans-
port layer driving additional network capacity increases and
assesses their projected implementation costs normalized to
Manuscript received July 22, 2016; revised October 21, 2016; accepted
October 22, 2016; published December 15, 2016 (Doc. ID 272038). the capacity gains so as to reveal their potential efficiencies.
D. M. Marom is with the Applied Physics Department, The Hebrew The short-term solution to increasing capacity is by
University, Jerusalem 91904, Israel (e-mail: [email protected]).
P. D. Colbourne is with Lumentum, Ottawa, Canada.
introducing hardware enabling elastic optical networking
A. D’Errico is with Ericsson Research, Ericsson Telecomunicazioni SpA, (EON) [5], which is intrinsically related to flexible-grid
Via Moruzzi 1, 56124 Pisa, Italy. channel allocation. EON can maximize the spectral effi-
N. K. Fontaine is with Nokia/Bell Laboratories, 791 Holmdel-Keyport ciency by adapting the provisioned bandwidth to the signal
Road, Holmdel, New Jersey 07733, USA.
bandwidth, with further efficiencies achievable by assem-
Y. Ikuma is with NTT Device Innovation Center, NTT Corporation,
Atsugi, Kanagawa, Japan. bling subchannels within spectral superchannels [6].
R. Proietti is with the Department of Electrical and Computer Spectral superchannels, defined as a set of contiguous car-
Engineering, University of California, Davis, Davis, California 95616, USA. riers (subchannels) switched as a single entity from a
L. Zong is with the Transmission Technology Research Department, source to a destination node throughout the network, pro-
Huawei, Shenzhen, China.
J. M. Rivas-Moscoso and I. Tomkos are with the Athens Information
vide the additional benefit that guard bands between sub-
Technology Center, Athens 15125, Greece. channels can be minimized or completely eliminated if
http://dx.doi.org/10.1364/JOCN.9.000001 spectrally compact multiplexing techniques such as Nyquist

1943-0620/17/010001-26 Journal © 2017 Optical Society of America

2 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

WDM or coherent optical (CO) OFDM are employed. in a single compact module. We first describe the function and
However, spectral superchannel transmission imposes re- principal elements of the conventional free-space WSS and
quirements on the network-switching elements to support then review the historical evolution of the ROADM node in
routing spectral slices of varying bandwidths [6]. Enhanced response to its expanding scope in terms of network manage-
functionality may be offered at the network nodes, such as ment of wavelength and direction assignments. We then ex-
the capability of all-optically adding/dropping subchannels plore recent research solutions enabling the realization of
to/from superchannels, leading to further requirements CDC-Flex ROADMs.
regarding ultra-fine resolution-filtering technologies [7].
The longer-term solutions will involve extending the fiber A. WSS Functionality
capacity beyond the C-band, with two possible routes: in-
creasing the telecom window or introducing parallelism.
A standard WSS system configuration is shown in
Multi-band (C-, L-, or S-bands) [8] and multi-fiber (fiber
Fig. 1(A), which consists of a linear array of input/output
overlays [9]) systems can still rely on elastic optical network-
single-mode optical fibers, followed by beam-conditioning
ing, but space division multiplexing (SDM) systems based on
optics (fiber-collimating lenses to reduce the beam diver-
new types of fibers with additional spatial conduits for infor-
gence of the optical fibers, polarization diversity optics for
mation transmission offer significant additional opportuni-
separating each beam into two co-polarized beams, and
ties for capacity enhancements [10]. Physical impairments
anamorphic prisms for beam expansion along one direction
introduced by SDM fibers, such as inter-core crosstalk or
to illuminate a wide grating section without increasing
mode coupling, and potential benefits of SDM, e.g., cost sav-
module height). Following the beam-conditioning optics,
ings related to the use of common lasers or joint digital sig-
parallel collimated elliptical beams are projected onto a dif-
nal processing (DSP) [10,11], may make it advisable to route
fraction grating. An input beam experiences angular dis-
all spatial channels as a single entity across the network. A persion upon diffraction from the grating, which is mapped
spatial superchannel is formed by grouping individual spatial to spatial dispersion by a Fourier lens. There are many var-
subchannels (on different fibers/cores/modes) at the same car- iants to the basic configuration of Fig. 1(A) in commercial
rier frequency and leverages on new SDM switching architec- WSSs, such as using a curved mirror in place of the lens or
tures employing switches adapted for operation across the employing a concave grating. These variants may offer
spatial domain [12,13]. A comparison of spectral and spatial greater compactness or reduce component count, but essen-
superchannel transmission in an SDM–WDM network can be tially achieve the same function, namely, to spatially dis-
found in [14,15]. Even though spectral allocation strategies perse and focus the wavelength channels composing the
result in better network performance, spatial allocation offers input beam at the lens’ back focal plane. Since all beams
the advantage that, by limiting flexibility to a certain extent, emerging from the input and output fibers (had light been
cost benefits can be derived from a reduction in the number of introduced from the latter) are parallel up to the Fourier
required switching elements without compromising the net- lens, following the Fourier lens they all focus to identical
work operation to an unacceptable degree [15,16]. positions on a wavelength basis. Thus, at the Fourier lens’
In this paper, we present a comprehensive view of the state back focal plane, the wavelength channels are dispersed
of the art of spectrally and spatially flexible reconfigurable op- and the input/output fiber beams are distinguished by their
tical add/drop multiplexer (ROADM) architectures and their unique incidence angle. At this plane, where the beams are
enabling technologies. In Section II, we present a classifica- dispersed and ports spatially superimposed, the switching
tion of spectral ROADM architectures and discuss the most action takes place by beam-steering the input beam toward
recent research proposals to achieve cost-effective colorless, a desired output fiber on a wavelength basis [18].
directionless, contentionless, and flexible-grid (CDC-Flex)
ROADMs. In Section III, we investigate ROADM solutions
incorporating sub-band switching capability. SDM switching
strategies and associated ROADM architectures are exam-
ined in Section IV. An evaluation of the different options to
extend the fiber capacity beyond the C-band (multi-band, fiber
overlays, and SDM) is discussed in Section V. Finally, in
Section VI, we present our conclusions and future outlook.

II. ACHIEVING CDC-FLEX ROADM FUNCTIONALITY

Directing wavelength channels throughout optical net-

works, whether routing at mesh network nodes toward a
channel’s ultimate destination or simply dropping the chan-
Fig. 1. Elements of a WSS. (A) Optical arrangement for dispers-
nel at an intermediate node and adding another in its place, ing WDM channels on the beam steering element. Technologies for
is performed all-optically by the ROADM switching hardware beam steering include (B) large MEMS tilting micromirrors,
[17]. The key element of the ROADM node is the wavelength- (C) fine MEMS tilting micromirrors, and (D) fine LCoS phase
selective switch (WSS), which performs the tasks of demulti- modulating pixels. The fine pixels and micromirrors allow flexible
plexing, switching on a wavelength basis, and re-multiplexing WDM channel allocation.
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 3

Glossary tilting them in groups, so that the channel center wave-

A/D Add/drop length and bandwidth can be adaptively defined and made
BV Bandwidth variable to match the signal bandwidth [20]. In place of customized
C/L/S-band Conventional/long/short-band MEMS micromirror arrays, ones modified from the image-
CapEx Capital expenditure projection industry can be substituted, with switching
CC Colorless, contentionless obtained by setting all the small, two-dimensional micro-
CD Colorless, directionless mirror pixels falling under the wavelength channel of
CDC Colorless, directionless, contentionless interest to switch to the desired port [21] [see Fig. 1(C)].
CO-OFDM Coherent optical OFDM Alternatively, another technology adapted from the display
DGD Differential group delay industry may be employed, based on a two-dimensional
DSP Digital signal processing liquid-crystal on silicon (LCoS) spatial light modulator
EON Elastic optical networking (SLM) [22] [see Fig. 1(D)]. Here, the liquid-crystal layer needs
FFT Fast Fourier transform to be modified from amplitude to phase modulation, and
FMF Few-mode fiber switching is performed by writing a linear phase ramp
GB Guard band (the equivalent to a mirror tilt). Since the phase-modulation
HSR High spectral resolution depth that can be inscribed by the LCoS SLM is limited to 2π,
LCoS Liquid crystal on silicon the phase ramp is applied modulo 2π and forms a blazed gra-
MCF Multi-core fiber ting. As such, its optical performance degrades with increas-
MCS Multicast switch ing steering/diffraction angle, giving rise to reduced efficiency
MEMS Micro-electro-mechanical system (or port-dependent loss) and spurious diffraction orders cre-
MIMO Multiple input, multiple output ating crosstalk to neighboring ports [23]. The main advan-
NA Numerical aperture tages of using two-dimensional SLM technologies adapted
OFDM Orthogonal frequency division multiplexing from the image-projection industry are the wide availability
OpEx Operational expenditure of devices (circumventing custom MEMS development ef-
OSNR Optical signal to noise ratio forts), the intrinsic support for flexible channel definition
OTN Optical transport network by software programming, and fine spectral channel boun-
OXC Optical cross-connect dary control (at the sub-pixel level) by preferentially allocat-
PL Photonic lantern ing edge pixels to the two boundary channels [24]. This
PLC Planar lightwave circuit allows a WSS to be adapted from a coarse to a fine WDM
ROADM Reconfigurable optical add/drop multiplexer grid, or to operate in a flexible-grid fashion in support of
SDM Space division multiplexing elastic optical networking [25].
SLM Spatial light modulator Finally, note that since all input and output fiber port
SMF Single-mode fiber beams are imaged in the wavelength-dispersed plane to
SPOC Spatial and planar optical circuit common positions, switching cannot be performed for more
TIDE Terabit interferometric drop-erase-add than one input beam to two different independent outputs.
ULI Ultrafast laser inscription Hence, the operation of the WSS is restricted to a single
WDM Wavelength division multiplexing input, multiple output case, denoted 1 × K WSS, allowing
WGR Waveguide grating router the input wavelengths to be distributed to the K output
WSS Wavelength-selective switch fibers; alternatively, the WSS can operate with multiple
WSXC Wavelength-selective cross-connect (K) inputs and a single output, denoted K × 1 WSS, in
which the connection from the desired input to the output
fiber is established independently for each wavelength
channel.
Early WSSs utilized micro-electro-mechanical system
(MEMS) tilting micromirrors for beam steering, one micro-
mirror per wavelength channel. Switching was performed B. ROADM Node Evolution
by redirecting the incident input beam on a wavelength
channel basis by individually tilting each micromirror, so ROADM node architecture has evolved in lockstep with
as to reflect from the micromirror in a direction commen- WSS technology over the past 15 years, as the two are
surate with the desired output fiber [19]. Upon reflection, intimately entwined. In Fig. 2, we chart the expanding
the beam back-traces the optical elements and then effi- capabilities of the ROADM node and its evolution toward
ciently couples to the selected output fiber. The width of CDC-Flex functionality.
the channelized MEMS micromirrors [see Fig. 1(B)] must
match the WDM channel plan of the optical network over 1) Blocker-Based Add/Drop Node: Early ROADM nodes
which it is to be deployed (e.g., 50 or 100 GHz), and the were placed in bi-directional optical transmission systems
optical spectrum must be well aligned to the device. The (long-haul line systems or metro-scale rings, where inter-
advantage of the MEMS micromirror approach is their mediate add/drop access to optical channels was required).
optical performance, as the mirror reflectivity is high, uni- These implementations were based on “wavelength block-
form, non-scattering, and independent of switching angle. ers,” as shown in Fig. 2(i), which are merely 1 × 1 WSS
A flexible-grid mode of operation can be achieved by em- (single input, single output) [26–28]. Their role is to clear
ploying MEMS tilting micromirrors on a finer pitch and the bandwidth occupied by the “drop” channel so that an
4 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

Fig. 2. Evolution of the network ROADM node architecture. (i) Blocker (1 × 1 WSS) based add-drop node and (ii) 1 × 2 WSS based add-
drop node for line systems with colored add/drop ports, (iii) colorless ROADM with 1 × K WSS, with K − 1 add/drop ports, (iv) degree
three-mesh network node with cross-connect functionality and colorless and contentionless (but direction-bound) add/drop, and (v) mesh
network node with colorless, directionless, and potentially contentionless (depending on D × L WSS capability) add/drop. The WSS can
provide flexible-grid attribute, depending on the employed switching technology.

“add” channel can be inserted in its place without suffer- of wavelength-tunable transceivers, the WSS output port
ing from deleterious crosstalk from the remnants of the count grew to a 1 × K design [with K
9 being a popular
“drop” signal. choice, as for a 40-channel WDM system it allowed 8-channel
(20%) wavelength access]. This allowed for “colorless” add/
2) 1 × 2 WSS-Based Add/Drop Node: Expanding the drop ports following the solution of Fig. 2(iii). Having color-
WSS to 1 × 2 functionality allows for better loss manage- less ports allows for efficient wavelength routing within the
ment within the ROADM node, as it eliminates a passive optical communication system; as long as a drop port is avail-
splitter [Fig. 2(ii)]. In both the 1 × 1 and 1 × 2 WSS cases, able and is terminated by a transceiver, any wavelength can
the “drop” channels appearing on the drop fiber are then be directed to it (i.e., the ROADM is “contentionless,” as the
demultiplexed (and “add” channels multiplexed) onto indi- WSS hardware imposes no wavelength contention).
vidual transceivers, each associated with a unique wave-
length (“color” port). Since tunable wavelength lasers 4) 1 × K WSS Colorless/Contentionless Mesh ROADM
could be had only for a significant premium at the time, Nodes: Adding WSS support of switching to multiple out-
optical transceivers were associated with a particular fixed puts allowed all-optical networking to come to fruition.
wavelength, and having color ports was not considered Optical networks typically follow a mesh architecture, hav-
a serious impediment. However, this implied that, for ing nodes with varying numbers of links to neighboring
unrestricted flexibility in choosing the “drop” wavelength nodes. The degree of the node denotes how many fiber links
channel during network operation, transceivers at all to neighboring nodes are attached to it (e.g., degree 3, 4,
wavelengths had to be deployed at each ROADM node, etc.). Wavelengths can be routed from any ingress of the
which was clearly not an economically viable solution. network node to any egress fiber, or channels can be locally
Hence, a smaller number of add/drop transceivers at prese- dropped, following the “Route-and-Select” arrangement of
lected wavelengths were deployed, thus limiting the opera- Fig. 2(iv) [29]. This introduces cross-connection capability
tional flexibility in routing channels to the desired ROADM at the ROADM node between the fiber links in addition to
node, which further constrains network management. add/drop support. Here, a 1 × K WSS at each ingress fiber
port distributes the incoming wavelength channels toward
3) 1 × K WSS-Based Colorless Add/Drop Node: In sup- egress fiber ports or to colorless drop ports. At the egress
port of greater operational flexibility, and with the advent fiber ports, a K × 1 WSS combines the routed and added
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 5

channels to the output WDM fiber. (There exists a detection, this should not be a detriment. While the MCS-
“Broadcast-and-Select” alternative arrangement that re- based solution provides the full CDC functionality using ru-
places the ingress 1 × K WSS with passive splitters, as a dimentary elements of splitters and selector switches, it is
lower-cost option [30].) The add/drop ports that are directly hampered by the high splitting losses inherent to the MCS
attached to the WSS at the ingress and egress fibers are solution. Therefore, an MCS architecture must be accompa-
associated with that direction. This can be considered as nied by an optical amplifier array to compensate for these
a limitation, as demonstrated next: Suppose each ingress losses, which adds to the cost of this solution, increases the
fiber has a finite number of drop transceivers deployed, power consumption due to the optical amplifiers, and degrades
and one direction’s drop count is fully occupied. If another the signal’s OSNR. These shortcomings have hampered the
wavelength channel is to be dropped from that fully occu- widespread acceptance of the MCS-based CDC-ROADM.
pied direction, the request cannot be fulfilled even though
ROADMs which employ passive demultiplexers in their
there are available idle transceivers at the node’s other in-
implementation (i.e., blocker and 1 × 2 WSS solutions) are
gress ports. Yet the wavelength channel cannot reach this
bound to the channel allocation plan of the demultiplexer.
resource, as it is bound to a direction. Hence, this architec-
ROADM architectures based solely on WSS inherit the
ture inherits the colorless and contentionless attributes
channel flexibility from the WSS attributes. Hence, if a
but the add/drop ports are direction-bound.
flex-grid WSS performs all the channel routing, then the
5) Colorless/Directionless/Contentionless ROADM Nodes: ROADM node can support elastic optical networking
To derive maximum benefit from node transceivers, it is [33]. CDC-Flex-ROADM capabilities generate considerable
desired that they be “directionless,” and hence able to serve interest in the optical communication industry, in support
drop (add) channels arriving from (directed to) any network of contentionless connection of any wavelength (and with
link connected to the node. This can be achieved by adding any bandwidth and modulation format) from any trans-
additional hardware that serves to aggregate the directions ceiver in any direction. This guarantees that the node
and distribute the add/drop channels to the available can evolve with the modulation formats being deployed
transceivers. One possibility is to introduce optical cross- and support elastic optical networking with bandwidth-
connect (OXC) functionality using a fiber space switch variable transceivers, making the solution future-proof.
(non-wavelength-selective) to the WSS-based ROADM CDC-Flex-ROADMs are further advantaged by imparting
node. The OXC may be placed within the WSS interconnec- the network with higher programmability, simplified man-
tivity [31], or only at the drop ports of all directions, and agement (less-restrictive routing constraints), and greater
establish connections to the transceivers [32]. Another pos- survivability of transceiver failure. However, the additional
sibility is to use a WSS operated with multiple (D) input hardware required for their implementation leads to in-
fiber ports, matching the direction count, and multiple creased node cost (i.e., CapEx increase), as well as possible
(L) output fiber ports for transceiver access (i.e., D × L extra complexity of intra-node fiber connections, node vol-
WSS), Fig. 2(v). However, as noted earlier, all input and ume, and sensitivity to failure of the direction aggregation
output fiber ports are imaged to the same position within hardware (possibly impacting all node channel drops or
a WSS, which implies that it cannot handle multiple inputs adds unless protection means are introduced with additional
and outputs simultaneously for independent switching. implementation complexity). Moreover, the MCS-enabled
Hence, the caveat in this implementation of a D × L CDC-Flex-ROADM requires additional amplifiers im-
WSS is that any particular optical wavelength can only ap- pacting cost and power dissipation (additional OpEx), while
pear on one input port at most. Thus, the node architecture the D × L WSS enabled Flex-ROADM offers only CD flexi-
is restricted to drop only a single occurrence of a particular bility, which constrains operations. Hence, the lack of an ef-
wavelength among all the node’s ingress links, which rep- fective CDC-Flex-ROADM solution that simultaneously
resents contention for the drop wavelength choice in net- achieves all desired attributes has hampered its deployment
work routing. Therefore, due to the inherent limitation by network operators, who are still seeking technological
of the conventional WSS, this node is classified as a color- advances in switching hardware to enable better CDC-
less, directionless (CD) ROADM, which falls short of Flex-ROADM solutions.
providing full CDC-ROADM functionality that system
operators desire.
A CDC-ROADM solution is possible, based on a multi-
C. Beyond the State of the Art on CDC-Flex-ROADM
cast switch (MCS) solution for the add and drop wave-
length channels. An MCS splits the drop signals and In order to address CDC-Flex-ROADM requirements
delivers a copy of all the drop channels to each attached without significantly increasing cost, footprint, and com-
transceiver. A multiple-input, single-output fiber switch plexity, technological upgrades of current devices will play
(non-wavelength-selective) determines which direction will a crucial role. New requirements need to be addressed in
be attached to the transceiver. A tunable wavelength filter future high-capacity flexible-grid optical nodes. Next-
may be required to isolate a single drop channel from the generation optical nodes will incorporate higher levels of
selected direction. However, for coherent detectors, this fil- flexibility in switching, bandwidth, and transmission rate.
ter may be discarded. The tunable filter at the drop ports The introduced flexibility will allow dynamic composition
may limit the channel bandwidth, hence impacting the sup- of network functions to cope with heterogeneous traffic
port for channel flexibility and elastic optical networking. demands. The CDC-Flex operation in the ROADM nodes
Since most elastic networking solutions utilize coherent gives the possibility of optimizing the resource utilization,
6 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

to reconfigure the network according to the variation of the

traffic pattern in a cost-effective way, while also supporting
re-routing functions in case of faults.
1) Ultrahigh-Port-Count WSS: Realization of ultrahigh-
port-count WSS can lead to an alternative CDC-Flex-
ROADM solution. High port counts may be challenging
to realize using a WSS design based on discrete elements.
An optical system that uses a guided-wave solution in com-
bination with free-space optics, termed spatial and planar
optical circuit (SPOC), has been demonstrated (see Fig. 3).
A SPOC-based WSS replaces some of the elements of the
front-end beam-conditioning optics with a waveguide-based
solution that serves as a spatial beam transformer. It applies
a waveguide grating router (WGR) of zero diffraction order
(non-dispersive) to each fiber port, thereby anamorphically
broadening each output radiating beam (Fig. 3, top), obviat-
ing the need for a microlens array and anamorphic prisms
and alignment procedures. The launched parallel beams are
collimated in the fast axis by a single cylindrical lens, fol-
lowed by the standard elements of the WSS (polarization
and dispersive optics, and an LCoS SLM) (Fig. 3, bottom).
A SPOC-based WSS was used to demonstrate a record
1 × 95 WSS configuration [34]. Similarly, an adiabatic wave- Fig. 4. CDC-Flex-ROADM implementation with high-port-count
guide broadening section may be implemented instead of the WSS. Each add/drop port incorporates a direction-selection switch.
zero-order WGR, as shown for a 1 × 40 WSS [35] and for a
1 × 93 WSS [36]. Both the zero-order WGR and the adiabati-
cally broadened waveguide essentially perform the same The number of add/drop transceivers supported by the
task of converting each individual I/O beam into a broad node is nearly equal to the WSS port count (D − 1 WSS
mode profile along the guided direction, hence eliminating ports are required for through-node traffic routing; the rest
the need for placing a microlens array and achieving very for add/drop port support). This architecture is advantaged
high beam fill factors (defined as beam size/port pitch) for by having very low loss budget to reach the transceiver, as
high port packing. well as having a per-transceiver aggregation means, thus
With ultrahigh-port-count WSS elements, the CDC- circumventing the common aggregator that can lead to an
Flex-ROADM can aggregate individual drop ports from ungracious failure mode. On the down side, there is an in-
each direction to a selector switch (non-wavelength depen- herent inefficiency to this solution where on average only
dent) placed in front of each deployed transceiver (Fig. 4). 1/D of the drop ports of any direction WSS are utilized [as
(D − 1)/D are allocated to drop ports from other directions].
This means that while the high-port-count WSS will com-
mand a cost premium, most of these extra ports will be
unutilized (considered the cost of implementing CDC func-
tionality). As these WSSs use LCoS SLMs, the solution is
an inherently flexible grid.

2) Multi-Input, Multi-Output WSS for Port Aggregation:

Another WSS alternative that is instrumental for achiev-
ing the full CDC-Flex-ROADM requirements is a conten-
tionless D × L WSS (having D input ports and L output
ports and used for add/drop port aggregation). A conten-
tionless D × L WSS can support multiple input ports
with the same wavelength channel appearing across
more than one of its input ports. As explained earlier
(Subsection II.A), all input and output fiber beams are im-
aged in the wavelength-dispersed plane to a common posi-
tion. This degeneracy has to be broken, which can be
achieved by having a unique position for the dispersed
spectrum for each of the D input (direction) ports, and
Fig. 3. High-port-count WSS utilizing a PLC front end for high- introducing the means to route each output (drop) port
density packing of fibers and implementing part of the anamorphic to a selected dispersed spectrum.
optics solution. Top: SPOC design. Bottom: Optical arrangement of
the high-port-count WSS with LCoS switching technology for One way of achieving a contentionless D × L WSS is to
wavelength flexibility. use a modified SPOC device, designed not only to broaden
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 7

a beam but also to prescribe a spatial tilt across it. This is functionality was designed and simulated. However, having
achieved by exciting the zero-order WGR with an off-center unique optical arrangements for the input and output beam
mode (see Fig. 5, top). An offset excitation of the waveguide delivery optics leads to a cumbersome arrangement.
array generates a linearly varying spatial phase, which is An alternative, simplified free-space arrangement was re-
preserved at the PLC facet. Here, D input waveguides ex- cently demonstrated [40], where MEMS tilting micromirrors
cite a single, central WGR of the SPOC through a slab lens were placed on every fiber port for beam-steering to the
at distinct positions, thereby serving to collimate all inputs unique positions on an LCoS SLM (where the fiber spectra
simultaneously while further instilling a unique tilt angle are dispersed), as shown in Fig. 6. This arrangement also
to each one due to the Fourier transform property of the allows for contentionless D × L WSS functionality, with
confocal slab lens. The subsequent free-space optical ar- the added benefits that the optical system is fully symmetric
rangement maps each tilt to a unique, offset position on (same optical elements in the path from fiber to spectral
the LCoS SLM (see Fig. 5, bottom). A wavelength channel plane and back, simplifying the optical design) and the
can then be routed from any of the D input dispersed sig- number of input fiber ports, D, can be software-reconfigured
nals to a desired output port (of the L available), by beam- according to the ROADM degree size, with the remaining
steering toward the port position on the SPOC device. fiber ports thereby available to support additional drop
There, the beam couples to the WGR associated with the ports. For example, contentionless 8 × 24 WSS functionality
desired output destination. Since the incidence angle onto is possible for degree 8 nodes (which occur with reduced fre-
the output-side WGR depends on the input port source, the quency). Should the node be of degree 4, the WSS can be
signal will emerge at one of the D waveguides attached to reprogrammed as 4 × 28, making optimal use of the fiber
the destination slab lens (the actual waveguide depends on port resource and availing extra drop ports.
the source input fiber). Hence, an active D × 1 switch is placed
afterward to select which of the D waveguides is connected to Contentionless D × L WSS serves to distribute dropped
the output fiber. This contentionless D × L WSS is also an in- channels to transceivers from any direction and likewise
herently flexible grid due to the LCoS switching engine, and aggregate added channels to egress fibers for CDC-Flex-
can be placed within the architecture of Fig. 2(v), enabling ROADM realization. Due to the lower losses of this solution
CDC-Flex-ROADM functionality. A SPOC-based contention- compared to the inherent losses of an MCS, the use of op-
less 8 × 24 WSS was demonstrated in [37]. tical amplifiers can be avoided. Note that optical amplifiers
can also be dropped in the MCS case if the node is con-
The contentionless D × L WSS can also be implemented strained to support a very small number of transceivers.
without resorting to a SPOC element for the beam condition- While D × L WSS tries to maximize the drop port count,
ing optics. As the beam-delivery optics (from the input fibers L, it is possible to have a low-port-count MCS, e.g., 6 × 6,
and dispersive optics that generate the separated dispersed which reduces the inherent losses by decreasing the split
spectra on the LCoS SLM) slightly differ from the beam- ratio. However, this solution reduces the scalability of the
reception optics (from the LCoS SLM to the output fibers network node. Fiber connectivity is also greatly reduced
including an active switching element), they require unique for a D × L WSS implementation, as dropped channels
arrangements that are placed side by side [38]. Its from each direction are still WDM multiplexed and only
scaling potential was studied in [39], where 8 × 128 WSS separated at the D × L WSS [compare fiber count in
Figs. 2(v) and 4].

Fig. 5. Contentionless D × L WSS utilizing a PLC front end. Fig. 6. Port-reconfigurable WSS array setup using free-space
Top: SPOC design variation. Bottom: Optical arrangement of the optics. MEMS micromirrors at the fiber beam path enable assign-
contentionless D × L WSS with LCoS switching technology for ments of the input/output ports for different realization of conten-
wavelength flexibility. tionless D × L WSS, matching the node size.
8 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

3) Wavelength-Selective Cross-Connect: A final variation

for CDC-Flex-ROADM implementation utilizes a wavelength-
selective cross-connect (WSXC). A WSXC was first demon-
strated by combining optical multiplexers/demultiplexers with
an optical cross-connect (fiber switch fabric) [41]. However,
the use of fixed demultiplexers subjects the switch to a chan-
nel grid. By switching dispersed light with an LCoS SLM,
the WSXC obtains channel flexibility. Since cross-connect
functionality requires two beam-steering events to complete
arbitrary connections from any input to any output, the op-
tical system is complicated by the fact that it must perform
an imaging operation between the two beam-steering
planes, as first demonstrated in [42] for a 5 × 5 WSXC, or
the spectral resolution of the channel definition would wor-
sen (as implemented in [43]). A new, greatly simplified reali-
zation of an 8 × 8 WSXC was demonstrated recently [44] (see
Fig. 7). Beam conditioning optics (collimators, cylindrical
lenses, anamorphic prisms, polarization diversity, and a dif- Fig. 8. Possible architecture of CDC-flex ROADM based on
fraction grating) applied to the input fibers disperses the 8 WSXC. Drop ports require tunable filters, which may be omitted
inputs onto the first LCoS SLM. On the output side, an iden- for coherent receivers.
tical arrangement disperses light from the output fibers to
the second LCoS SLM (the light is actually traversing the A suggestion for how to deploy such an 8 × 8 WSXC for a
elements in reverse). A special relay-imaging system be- CDC-Flex-ROADM functionality in a degree-4 node is
tween the first and second LCoS SLMs images the dispersed shown in Fig. 8 [45]. Four of the inputs and outputs handle
light in the wavelength-dispersed direction and Fourier- the ingress and egress node fibers. The remaining four in-
transforms it in the port direction (required for switching puts and outputs carry the traffic to be dropped and added.
with beam-steering technology). Such an 8 × 8 WSXC can Multiple replicas of the four drop fibers are created, neces-
perform any permutation on a wavelength basis between sitating ancillary optical amplification (similarly on the add
the input and output set without contention and with flex- side, where input sets are passively combined). Each copy
ible channel definition due to the use of LCoS technology. set of four fibers feeds a small space switch of dimension
4 × 4 for routing drop channels to an available transceiver.
The multiple copies made by passive splitters are required
Input block cylindrical lens input LCoS
to increase the number of supported add/drop transceivers.
grating Any dropped/added wavelength channel from any direction
can reach an available transceiver by selection of an avail-
collimation able wavelength slot on the WSXC and configuration of the
Relay output
lens small OXC. Since the dropped channels are wavelength mul-
system fiber array tiplexed, a tunable filter may be required in front of the
input
transceiver if direct detection is employed. Note that an
fiber array
collimation alternative 8 × 8 Flex-WSXC functionality was also recently
lens demonstrated utilizing a SPOC front-end design for spatial
beam transformation [46]. The salient features of a CDC-
grating
Flex-ROADM based on a single WSXC are the small number
of components required for its implementation and simple
output LCoS cylindrical lens Output block
fiber connections within the node, as the various cross-
polarization collimation
connections are carried out in free-space within the switch
collimation polarization input output
lens diversity
grating
LCoS LCoS
grating diversity lens confines. However, should the single WSXC fail, all traffic
through the node is interrupted. Additional architecture
variants for use of a WSXC in a CDC-Flex ROADM, includ-
ing 1  1 protection coverage, are introduced in [47]. The
Relay
system cost benefits and architectural simplicity of this solution
when protection is introduced are then reduced.

Input anamorphic cylindrical cylindrical anamorphic Output III. SPECTRALLY FLEXIBLE SUB-BAND ROADM
fiber array prisms lenses lenses prisms fiber array

Input block Output block Flex-grid ROADM solutions allow for adapting the
network-provisioned bandwidth to the wavelength
Fig. 7. Schematic of the optical prototype of the 8 × 8 flex WSXC. channel’s spectral support—the basis for EON. However,
Top: Free-space propagation in the dispersion direction. Bottom: due to the WSS spectral transfer function [19], which
Details of free-space propagation in the port direction. provides a “clear channel bandwidth” narrower than the
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 9

provisioned bandwidth, the provisioned bandwidth must al- optical splitters for broadcasting a dropped superchannel
ways be larger than the channel bandwidth if the channel is to multiple fixed subchannel receivers directly coupled to
to be routed in the network independently. The determining tunable transmitters, only part of which are turned on
factors in this over-provisioning are the WSS spectral roll- to create new superchannels composed of regenerated sub-
off bandwidth from the “pass” to “block” state, which is di- channels that can also be wavelength converted, thus fur-
rectly related to the WSS spectral resolution, as well as the ther dealing with defragmentation. However, the cost basis
number of ROADM nodes the signal will traverse, as each of these solutions, requiring at minimum banks of subchan-
passage through a WSS represents another channel-filter- nel coherent receivers and transmitters when only a single
ing event. Hence, when elastic optical channels are assigned channel add/drop might be required, is exorbitant. Instead,
to the network, they must include spectral guard bands the research community is seeking specialized ROADM
(GBs) between adjacent channels to allow for independent solutions which can perform the add/drop functionality
network routing. A GB is also required to take into account all-optically [7]. Such all-optical subchannel grooming
center wavelength accuracy, component aging, temperature capability has been proven to be able to improve network
dependence, manufacturing tolerances, etc. These GB form performance [50] and reduce the total equipment cost as
an inherent inefficiency, as their spectral support cannot be compared to optical networks based on end-to-end super-
utilized for carrying information. One way of increasing the channel transmission [51] and multi-layer networks with
spectral utilization is to combine several carriers into one built-in electronic aggregation [52]. The solutions for sup-
superchannel to be routed as a single entity, thus obviating porting sub-band ROADM differ depending on whether the
the need for inter-subchannel GBs. Consider the emerging subchannels composing the superchannel are spectrally
transmission of 400 Gb/s superchannels, which are typically overlapping or not.
composed of two closely spaced 200 Gb/s Nyquist-shaped
carriers (at 16 QAM over dual polarizations) [48], which
can be contained within 75 GHz of provisioned bandwidth. A. Subchannel Optical Add/Drop of Spectrally
In this case, channel filtering is negligible even if the Non-Overlapping Signals
400 Gb/s signal traverses multiple ROADM nodes (see
Fig. 9). Detection of this dual-carrier superchannel format With digital signal pre-processing, an optical transmit-
is accomplished with a power splitter followed by two ter can create modulation formats that are spectrally effi-
coherent detectors. cient, having nearly rectangular power spectrum (e.g.,
However, there are scenarios where we wish to enjoy the Nyquist-WDM and CO-OFDM). Such transmitters can
benefits of superchannel routing throughout the optical be used to create the multiple subchannels comprising
network (minimizing GB and thus maximizing spectral the superchannel. The subchannels can be set contiguously
utilization) as well as provide sub-carrier add/drop capabil- to create a continuous superchannel, when no intermediate
ity within the superchannel at certain network nodes add/drop support is required. However, when optical sub-
for operational flexibility. This can be done by deploying channel extraction from the superchannel is required then
a full OTN superchannel transceiver for adding and drop- minimal GB should be placed on either side of the subchan-
ping subchannel tributaries electronically and creating a nels to be dropped, along with fine filtering technology to
new superchannel in its place [49]. An alternative hybrid assist in selecting the subchannel. The size of the GB is
optical/electronic-based solution, SERANO [33], utilizes commensurate with the fine filter’s optical resolution.
An optical filter specially designed to filter arbitrary sub-
channels from a superchannel was first reported in [53].
The high spectral resolution (HSR) filter employed a wave-
guide grating router as the dispersive element instead of a
bulk diffraction grating, for customizing the resolution and
operational bandwidth. The WGR was designed with a free
spectral range of 200 GHz (i.e., operating bandwidth), and
its number of waveguide arms (250) was selected to achieve
1 GHz optical resolution. A superchannel introduced into
this WGR-based processor is angularly dispersed and its
frequency components spatially dispersed at a Fourier lens
back focal plane (Fig. 10), where a spatial light modulator
acts to pass or block arbitrary spectral components at fine
resolution for adaptive filtering of subchannels. The filter-
ing arrangement is very similar to that of a WSS (Fig. 1),
where the main difference is the spectral support (200 GHz
versus C-band) and optical resolution (1 versus ∼7.5 GHz),
Fig. 9. Dual-carrier 400 Gb/s superchannel with spectral support
within single WSS transfer function provisioning 75 GHz WSS
both of which are determined by the dispersive element. In
bandwidth (solid blue line) and a cascade of 10 WSSs (dashed blue order for the WGR to well resolve such fine spectral fea-
line). Each Nyquist-shaped carrier is modeled as a 30 GHz wide tures, its waveguide delays have to be very large, which
(25 GHz  20% overhead) raised cosine channel with 0.2 roll-off represents a challenge for maintaining the optical phases
factor. Channels’ carriers are separated by 31.25 GHz. at the WGR output facet due to fabrication tolerances. Two
10 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

selected for drop are directed toward a D × L WSS. A com-

plete superchannel can be dropped and detected by a full
coherent receiver, and another superchannel can be intro-
duced to the network in its place. Alternatively, at the lower
superchannel hierarchy level, we can access subchannels
following a blocker-based ROADM architecture [Fig. 2(i)].
The superchannel is split, with one copy destined to a single
(or multiple) subchannel coherent detector and the second
copy being filtered at fine resolution by the HSR filter, which
clears (or blocks) the bandwidth corresponding to the single
or multiple subchannels that have been dropped. New “add”
subchannels are introduced into the available spectral slots
via an optical coupler, reconstituting the superchannel as a
contiguous entity for reinjecting back to the optical transport
network. Consider the networking functionality for three
superchannels (red, green, and blue), each composed of seven
subchannels introduced to the ROADM node from the West
Fig. 10. Fine-resolution spectral filter (single input, single out- direction. The red superchannel is routed as a whole to the
put), utilizing a specially engineered WGR to span a typical super-
East direction. The blue superchannel drops locally its entire
channel bandwidth of 200 GHz and resolve fine spectral features of
1 GHz within. Spectral blocking regions are selected at 400 MHz contents and a new superchannel is added (cyan) and sent out
addressability with an SLM in the Fourier plane. the North direction. The green superchannel is also dropped
but then further processed to drop a single subchannel and
techniques have been developed for correcting the WGR introduce a new one (green striped) in its place after filtering
phases post-fabrication, one based on a secondary phase with the HSR filter.
SLM in an imaging configuration [53], and the other on Having the capability to filter subchannels at 1 GHz opti-
waveguide phase trimming with UV laser irradiation [54]. cal resolution allows for their extraction with only 2 GHz
Such a fine-resolution filter can be employed within wide GB set between subchannels. As the filter further sup-
a CDC-flex ROADM architecture following a dual-level ports spectral addressability of 400 MHz (the mapping of an
hierarchical approach, with the top level performing SLM pixel size to frequency offset), the subchannels’ band-
superchannel network routing using route-and-select topol- width can be defined with full flexibility. However, for opera-
ogy and the lower level handling subchannel access tional ease, it is simpler to define a finite spectral slot size by
(Fig. 11). At the top network routing hierarchy, a CDC-flex which spectrum is reserved, in this case 3.125 GHz following
ROADM aided by direction-side WSS performs superchan- the ITU trend of halving the grid size [55], which leads to
nel routing from ingress to egress ports, or superchannels nearly optimal spectral occupancy according to [56]. Figure 12
demonstrates a 200 GHz wide superchannel (spanning 64
spectral slots), provisioned over 218.75 GHz (70 spectral
slots), that incorporates 9 subchannels of varying widths

Fig. 12. Exemplary superchannel packing with subchannels of

varying bandwidth, when the HSR filter has 1 GHz resolution.
Subchannel spectral granularity of 3.125 GHz used (example
Fig. 11. CDC-flex-ROADM extended to support subchannel add/ shows 4, 5, …, 10, 11 slot assignments, total of 64 slots

drop from network level superchannels. A selected superchannel is 200 GHz assigned), with the net subchannel bandwidth in
split, with one copy incident on a subchannel coherent receiver and parentheses, with 2 GHz reserved for an inter-subchannel GB.
the other copy fine filtered to clear the bandwidth for an added WSS superchannel boundary in solid blue line, and cascade of
subchannel. ten WSSs in dashed blue line.
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 11

(4 to 11 slot assignments, demonstrating complete flexibility and subsequent electro-optic conversion. The performance
in subchannel bandwidth assignment according to the EON was assessed via simulations to identify its system sensi-
concept), with each subchannel having a clear bandwidth tivities, as proper erasure critically depends on destructive
2 GHz smaller than the spectral assignment since a GB optical field superposition. An alternative opto-electronic
is assumed about each subchannel (a worst-case scenario substitution technique based on frequency conversion in
which implies each subchannel will be dropped somewhere a nonlinear fiber was experimentally demonstrated for
along the superchannel path). Even with nine subchannels OFDM sub-carriers [63], with the detection and opposing
and their inter-subchannel GBs, the superchannel signal generation performed off-line.
provides 182 GHz of clear bandwidth for data payload at All-optical OFDM subcarrier extraction can be substi-
any modulation format. The unutilized bandwidth tuted for the opto-electronic approach for generating the
(218.75 − 182 ≅ 37 GHz, or 17%) is roughly equally divided cancelling subchannel of the interferometer, in a technique
between inter-subchannel GBs and superchannel edges called terabit interferometric drop-erase-add (TIDE) multi-
(three spectral slots on each side) required for supporting plexing for superchannels [64]. Starting with a copy of a
network routing (fiber networking level GB). In [57], simu- superchannel containing the subchannel to be dropped,
lation results of Nyquist-WDM transmission through a cas- an optical FFT is performed for filtering the OFDM subcar-
cade of ROADMs with ultra-selective A/D modules were rier [65], followed by temporal sampling with an optical gat-
presented, showing that the edge subchannels are more af- ing function and reshaping the sampled signal to an OFDM
fected by the two filtering stages, as was confirmed in experi- subcarrier with a subsequent optical FFT filter. The two
ments [58]. A loop experiment demonstrating the ability to paths are recombined with appropriate phase, time, power,
repeatedly drop and add the same subchannel from within a and polarization alignment so that the subchannel to be
superchannel was recently demonstrated [59]. dropped is interferometrically suppressed from the super-
An alternative approach for optical filtering non- channel, at which frequency a new subchannel can be added.
overlapping subchannels from a superchannel can be based An experimental demonstration of the all-optical TIDE
on a fixed, high-resolution, low free spectral range demul- architecture operating in real-time for BPSK and QPSK
tiplexer. High-resolution demultiplexers can be obtained by CO-OFDM superchannels was reported in [66].
combining finite and infinite impulse response devices (i.e.,
Whether all-optical or opto-electronic techniques are
chip scale interferometers assisted by high Q-factor micro-
used for optical add/drop of spectrally overlapping sub-
ring resonators). The use of a fixed device for separating a
channels from a superchannel, the customized hardware
superchannel to its constituent subchannels implies that
can be deployed in a two-tiered hierarchical ROADM archi-
the subchannels must reside on a fixed, predetermined
tecture, as shown in Fig. 11, replacing the HSR filter with
grid, thereby losing the flexibility trend of EON. In addi-
the interferometric arrangement.
tion, each subchannel experiences filtering operations, re-
gardless of whether the subchannel is required to be
isolated for add/drop operation. To its advantage, the filter- IV. ROADM FUNCTIONALITY FOR SDM–WDM
ing device can be fully passive, though in practice the high- NETWORKS
Q microrings need active thermal stabilization to keep
their resonance wavelength locked. A ring-assisted, sharp
Elastic optical networking and future advances such as
12.5/25 GHz interleaver was designed for separating
subchannel add/drop capability allow the full and efficient
12.5 GHz wide subchannels from a superchannel [60], and
utilization of the optical transmission bandwidth spanned
its cascadability in loop experiments was studied [61].
by the communication system. Once that communication
band is exhausted, other means for supporting additional
B. Subchannel Optical Add/Drop of Spectrally capacity must be introduced. One of the leading candidates
Overlapping Signals offering a significant capacity multiplier is SDM, introduced
via new fiber types offering additional conduits of informa-
tion channels such as spatial modes or cores. Expanding
Supporting add/drop operations with GBs between sub- fiber link channel counts to the spatial domain requires fur-
channels enables simple extraction with optical filters, but ther innovation and flexibility to route these channels at net-
the inserted GBs reduce the spectral utilization. Moreover, work nodes. We discuss SDM–WDM ROADM functionality
a superchannel can be constructed with overlapping carriers, and additional photonic-switching innovations in support of
as is the case in a CO-OFDM signal spanning the entire the spatial domain in this section.
superchannel. Under these circumstances, a different sub-
channel add/drop solution must be employed based on inter-
ferometric subtraction of the dropped subchannel. A. SDM–WDM Switching Strategies
Reference [62] reported on an add/drop module based on
an opto-electronic interferometer structure allowing the Previous ROADM solutions provided flexibility across
extraction of individual sub-carriers from Nyquist-WDM the wavelength domain, enabling the switching of wave-
and CO-OFDM superchannels, as well as the insertion of length channels (down to the subchannel granularity).
sub-carriers into existing superchannels, through intra- Employing both WDM and SDM within fiber links expands
dyne conversion of the optical signal to digital electronic the channel count M-fold, with M being the number of
baseband, followed by digital electronic signal processing spatial modes or cores. However, whereas wavelength
12 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

channels do not intermix in a linear medium, spatial chan- approaches can be adopted for transmission systems having
nels might. When spatial channels intermix, the informa- no spatial mode/core mixing. In addition, for fibers having no
tion can still be unraveled by employing multiple input, spatial mixing, the switching may offer the ability to switch a
multiple output (MIMO) signal processing across the wavelength channel from one spatial mode/core on an ingress
mixed channels. For MIMO processing to function, all fiber to a different one at the egress fiber. This SDM “lane
the mixed channels have to be present and jointly proc- change” can be used to circumvent some blocking situations,
essed. Hence, when mixing occurs, the mixed channels and can be recognized as the equivalent of “wavelength con-
must be kept as a group and cannot be independently version” in WDM networking, which has been extensively
switched to a distinct network destination. We distinguish investigated in the past for resolving blocking cases [68].
between three mixing situations influencing the switching However, whereas SDM lane changes can be performed with
strategies: (1) no spatial mixing, when cores are isolated or simple passive optical switches and are an attractive net-
several single-mode fibers are used to create a spatial working feature, wavelength conversion requires more com-
group, (2) mixing between all spatial modes, which may oc- plicated solutions that have not been adopted in commercial
cur in multimode fibers or in fibers with coupled cores, and WDM systems.
(3) mixing within subgroups, which may occur in few-mode,
multi-core fibers or irregularly spaced multi-core fibers.
With these mixing situations under consideration, we iden- B. SDM–WDM ROADM Architectures
tify four main SDM–WDM switching strategies, catego-
rized according to the switching granularity (i.e., the Implementing the different SDM–WDM ROADM nodes
smallest switched element) [67]: can be accomplished through many of the technologies that
have been developed for SMF, as well as some complemen-
1) Space Granularity: Switching is performed in the spa-
tary ones. Foremost of the latter is a spatial multiplexer
tial domain across the entire optical communication band
(demultiplexer), which is used for combining (separating)
(all WDM channels), forfeiting any wavelength flexibility.
M input (output) single-mode channels into (out of) an
The switching hardware is devoid of any dispersing means,
SDM fiber, ideally in a one-to-one mapping between mode/
thus relying on relatively simple space switches. This
core and individual SMF. Employing such SDM demulti-
switching strategy is only applicable to the no spatial mix-
plexers (demux) at the ingress ROADM ports separates
ing transmission case.
out the spatial channels to a discrete SMF set for sub-
2) Space–Wavelength Granularity: Switching can be sequent switching.
prescribed down to the single spatial and wavelength chan-
nel granularities in the fiber, exploiting the full switching 1) Space Granularity Architecture: Considering the
potential of an SDM–WDM transmission system. Due to its space granularity approach, fiber switches are then re-
offered finest granularity (resulting in greatest flexibility), quired to route the entire communication band on each sep-
its potential implementations are the most cumbersome. arated spatial channel to its destination. Two switching
As in the space granularity case, space–wavelength granu- scenarios can be considered: (1) routing from an ingress spa-
larity requires that the spatial modes/cores do not mix. tial channel (core or mode) is performed on the same spatial
channel on the egress fiber, i.e., without spatial lane change,
3) Wavelength Granularity: Contrary to space granular- and (2) unrestricted routing is performed permitting spatial
ity, which forfeited access to the spectral domain, wave- lane changes. Switching without lane changes can be real-
length granularity sacrifices the spatial domain in order ized by a bank of small OXCs, one per spatial channel
to simplify the switching hardware. Implementations rely [Fig. 13(A)]. The input/output port count of an individual
on modified WSSs in support of joint switching of all M spa- OXC (devoted to a spatial channel) is at minimum the num-
tial channels on a wavelength basis. This scenario is the ber of directions, D, i.e., D × D OXCs, and M such OXCs are
only one applicable when all the modes/cores intermix, required. (Additional OXC ports may be desired in support
as they must remain together as a group throughout the of local add/drop.) In support of lane changes, one large OXC
network. should handle all the node traffic [Fig. 13(B)]. In this case,
the OXC port count is D · M × D · M (again, at minimum).
4) Fractional Space, Full Wavelength Granularity: For
A significant risk associated with an architecture based on a
fibers where mixing is limited within spatial subgroups,
single large OXC is its failure impact. This can be circum-
this strategy adopts the wavelength granularity approach
vented by a second OXC placed in parallel for protection, but
to the subgroups, and each subgroup can be independently
this adversely impacts the cost, space requirement, and
switched on a wavelength basis. Attractive features of this
fiber-routing complexity. Without lane changes using a bank
strategy are the intermediate routing flexibility and mod-
of smaller OXCs, a single switch failure impacts only 1/M of
erate implementation complexity.
the node traffic.
Possible implementations for the four switching strategies
are shown in Fig. 13 (non-exhaustive) and detailed in the 2) Space–Wavelength Granularity Architecture: Space–
next subsection. While these switching strategies are geared wavelength granularity offers the finest switching granular-
for different mixing situations experienced in fiber transmis- ity at the expense of implementation complexity. Here,
sion, the more restrictive solutions can always be applied too, lane changes may be supported at the cost of additional
to the less restrictive ones. For example, the wavelength complexity. Access to the space–wavelength granularity is
granularity and fractional space–wavelength granularity achieved by first spatially demultiplexing the signal,
 Marom et al.

Fig. 13. Proposed implementations of SDM–WDM switching strategies for different switching granularities, drawn for four spatial modes and a degree-four node. (A) Space granularity
without SDM lane changes. (B) Space granularity with SDM lane changes. (C) Space–wavelength granularity without SDM lane changes. (D) Space–wavelength granularity with SDM
lane changes. (E) Wavelength granularity. (F) Fractional space and full-wavelength granularity without lane changes. (G) Fractional space and full-wavelength granularity with lane
changes. Cases (C)–(G) show the switching hardware for one direction only (West).
VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 13
14 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

followed by placing a WSS on each spatially separated chan- Fig. 13(F), or with lane changes, Fig. 13(G). The latter, while
nel for routing to other egress ports. If routing to the same indeed more cumbersome, is more tolerable if the subgroup
spatial channel on the other ports, then the number of WSS count, M∕P, is a small number.
output ports needs to be D, with routing to D − 1 other des-
tinations and at least one drop port [Fig. 13(C), with only one 5) Architecture Comparison: To better gauge the scaling
direction shown for simplicity]. If spatial lane change is de- of the SDM–WDM ROADM solutions just outlined, we
sired, then each ingress WSS should reach all other destina- demonstrate the implementation possibilities using rea-
tions and spatial channels, requiring the WSS to support sonable values for the node degree and spatial channel
M · D − 1  drop output ports [Fig. 13(D)]. The fiber wiring count in a hypothetical WDM-SDM network. Let us as-
can become unwieldly in this case. sume that D
4 to cover the most prevalent network node
An implementation alternative for the cross-connect degree size, and that the SDM channel count is M
12 (for
functionality provided by the route-and-select topology an order of magnitude capacity increase). We further as-
can be obtained by placement of a transparent optical sume that the 12 spatial channels can be separated out into
cross-connect switch, as has been envisioned early in the four groups of size P
3 (perhaps a four-core fiber, where
WDM era [69]. For space–wavelength granularity, each fi- each core supports three modes), though other options can
ber would need to be demultiplexed twice, per space and be considered. The hardware required for the implementa-
wavelength domains. However, such a solution would lose tion is tabulated in Table I.
the flexible bandwidth assignments and EON support. In It becomes immediately apparent that the technologies re-
addition, it would require an OXC per wavelength channel. quired to implement the network node architectures are
Hence, it cannot meet the modern networking require- available or have been demonstrated (as outlined in the next
ments, demands a large switching gear count, and presents section), and that many WSSs will be required in support of
a severe wiring problem. Thus, we do not foresee such a the SDM–WDM switching. We generalize the node-switching
solution as a viable alternative. requirements and summarize the attributes of each of the
SDM-switching strategies and enabling ROADM architec-
3) Wavelength Granularity Architecture: Implementation tures just described in Table II. Advantageous attributes
simplicity can be obtained by sacrificing flexibility, as has are shaded green, disadvantageous in red, and neutral in yel-
been the case for space granularity. In wavelength granu- low. While ideally it is desirable to switch at the smallest
larity, all spatial channels are routed as a whole per wave- granularity, this is not always possible (if modes are mixed),
length channel. This functionality can be provided by a and it comes at a cost premium. Hence, the architectural
single WSS modified to support joint switching over all choice is a multi-variable optimization problem that needs
SDM channels on a wavelength basis by employing spatial to take into account the physical-level impairments.
diversity [Fig. 13(E)]. In the particular case of a few mode
fiber (FMF), joint switching within the WSS can be imple- Based on this discussion, it should be clear that the
mented without requiring spatial diversity and its associ- choice of switching granularity directly affects the specific
ated space mux/demux, providing further implementation SDM–WDM ROADM architecture, routing algorithms, and
simplicity. Such joint-switching WSSs are described in the related network performance. For instance, for wavelength
next subsection discussing new technologies. granularity (strictly required for a MIMO-based SDM
system with mixing [70,71]), a spatial superchannel will oc-
4) Fractional Space–Wavelength Granularity Architecture: cupy all the M spatial modes for a given wavelength slot.
As a compromise between switching granularity size and While this constraint on the creation of spatial superchan-
implementation complexity and cost, the joint-switching nels will limit the flexibility to switch, add, and drop single
concept of spatial channels can be applied to subgroups of SDM channels, it simplifies the network control and provi-
the spatial channel count. Hence, if the M spatial channels sioning since, for instance, fragmentation will be limited to
are divided into subgroups of size P, then the number of the wavelength domain only. In the case of an SDM system
WSSs required to route subgroups is M∕P (which is obvi- not exhibiting mixing among its spatial channels, fractional-
ously bounded between the cases of wavelength granularity space, full-wavelength granularity or wavelength granular-
and space–wavelength granularity). This fractional space, ity switching can still be used so as to take advantage of
full wavelength granularity solution can be implemented its implementation simplicity and cost benefits. However,
without lane changes (or rather subgroup change), see fragmentation in both the spatial and wavelength domains

TABLE I
SWITCHING HARDWARE FOR A NETWORK NODE OF DEGREE 4 AND SPATIAL CHANNEL COUNT OF 12
Architecture Switched Granularity Switching HW
Space granularity with lane changes Entire commun. band per spatial mode Single 48 × 48 OXC
Space–wavelength w.o. lane changes Entire commun. band per spatial mode 12 units of 4 × 4 OXC
Space–wavelength with lane changes Wavelength channel per spatial mode 24 units of 1 × 37 WSSs per direction, 96 total
Space–wavelength w.o. lane changes Wavelength channel per spatial mode 24 units of 1 × 4 WSSs per direction, 96 total
Wavelength granularity All spatial modes per wavelength channel 8 12 × 1 × 4 WSSs (2 per direction)
Fractional space, with lane changes Group of three modes, per wavelength 32 units of 3 × 1 × 13 WSSs
Fractional space, w.o. lane changes Group of three modes, per wavelength 32 units of 3 × 1 × 4 WSSs
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 15

TABLE II
COMPARISON BETWEEN SDM–WDM SWITCHING STRATEGIES (ADAPTED FROM [65])

Space-wavelength Fractional space, full

Space granularity Wavelength granularity
granularity wavelength granularity
Minimum Entire optical
Single WDM channel on a Single WDM channel Single WDM channel over
switching communication band on a
single spatial mode. across all spatial modes. subset of spatial modes.
granularity single spatial mode.
No subgroup lane change:
No lane change: 2M WSSs No lane change: M OXC
2M/P high-port-count joint
per direction with D I/O ports with D×D port count (at
Two high-port-count joint switching WSS configured as
at minimum (to support at minimum, and additional
ports for local A/D). switching WSS per P×(1×D) WSS applied in
least one A/D port).
Realization With lane change: One large direction configured as an parallel to all spatial modes in a
With lane changes: 2M WSSs
M×(1×D) WSS applied to subgroup.
per direction with M·(D-1)+1 OXC with (D·M)×(D·M)
all spatial modes in With subgroup lane change:
I/O ports at minimum (to port count (at minimum,
parallel. 2M/P high-port-count joint
support at least one A/D and additional ports for local switching WSS configured as
port). A/D).
P×(1×[(D-1)·M/P+1]) WSS.

The complete optical Each spatial superchannel

Each mode/WDM channel
communication band is provisioned across all
independently provisioned Compromise solution using
routed across network. modes. Susceptible to
and routed. small SDM groups. More
Flexibility Coarse granularity. If WDM wavelength contention.
Spatial mode maintained. efficient when provisioning
channels need to be Add/drop bound to
Prone to wavelength low capacity demands.
extracted then WSS need to direction unless MCS
contention.
be introduced. technology expanded.

OXC supports foreseeable

Cost roughly independent
Switching node cost linearly mode and fiber counts, but is
of SDM count. Inefficient Cost scales as group count.
scales with capacity, not single point of failure.
for low capacity Groups can be turned on as
providing the price benefit Without lane changes (using
connections due to capacity grows, offering pay-
essential for SDM adoption. several smaller OXC),
Scaling minimum BW as-you-go alternative.
Integrated WSS array would failure impacts only 1/M of
provisioned across SDM. Maintaining small group sizes
help suppress the cost traffic. Pricing favorable but
Large SDM Rx/Tx are facilitates MIMO processing at
increase. when wavelength add/drop
integration and DSP Rx.
required, needs WSS
challenge.
modules.
3 dB per I/O fiber link being
10 dB per I/O fiber link. switched. If add/drop from
10 dB per I/O fiber link.
For MCF or FMF SDM fiber is extracted, 10 10 dB per I/O fiber link.
For MCF or FMF
Estimated transmission fiber, extra 4 dB dB excess loss for through. For MCF or FMF transmission
transmission fiber, extra 4
loss loss is induced by the spatial For MCF or FMF fiber, extra 4 dB loss is induced
dB loss is induced by the
MUX/DEMUX transmission fiber, extra 4 by the spatial MUX/DEMUX
spatial MUX/DEMUX
dB loss is induced by the
spatial MUX/DEMUX

might occur. It becomes then necessary to study new routing than achieving the same capacity increases by deploying
and defragmentation algorithms that take into account this additional SMF systems in parallel. This is only possible
new spatial dimension. While the discussion of the different via integration technologies incorporating the spatial
routing algorithms and strategies goes beyond the scope of degree of freedom.
this paper, the reader can refer to [14,16,72–75] for more
1) Spatial Multiplexer: As noted earlier, many of the
details.
ROADM designs make use of elements developed origi-
nally for SMF-based systems. When utilized for SDM
C. Technology Advances Supporting SDM–WDM systems, they must be preceded/followed by a spatial de-
multiplexer/multiplexer whose role is to convert between
Switching
the M cores or modes of the SDM fiber and M separate
SMFs. Ultrafast laser inscription (ULI) can be utilized to
The various SDM–WDM ROADM realizations which inscribe waveguides in three-dimensional space (under
are required for implementation of transparent optical programmatic control) inside a glass block, based on
networks with unprecedented capacity make use of ancil- permanent nonlinear densification of the glass at the focal
lary photonic innovations for multiplexing and switching region of the inscribing beam (Fig. 14, top). The written
(fiber amplifiers in support of SDM are beyond the scope waveguides can be made with mode sizes similar to
of this paper). It is imperative that these innovations in SMF for efficient edge coupling. 3D ULI waveguides can
support of the SDM capacity multiplier be more economical serve to interface between an SMF array and the cores
16 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

of a multi-core fiber (MCF), thereby establishing a one-to- Fig. 14, center). Fully selective PLs are hardest to fabricate
one mapping between cores and individual SMFs (some- and tend to have high losses and are limited to a small num-
times referred to as MCF fan-out or break-out). If the ber of modes, whereas completely scrambling PLs are sim-
MCF contains well-isolated cores (no coupling between pler to fabricate, have lower losses, and can scale to higher
them), then an SDM link composed of a multiplexer, numbers of modes. The mode scrambling is not perceived as
MCF, and demultiplexer will not exhibit mixing, enabling a limitation, as over long transmission distances (several
the space and space–wavelength granularity SDM–WDM hundreds of km), the modes become strongly mixed anyway
ROADM solutions. For mode-division multiplexed trans- and require MIMO processing to undo the mixing at the
mission, the ULI waveguides can be designed to excite spa- SDM receiver. The PL-induced mixing is taken out just as
tial modes by bringing them close to each other, such that well by the MIMO processing. Mode-selective PLs are useful
they begin to mutually couple, while adiabatically reducing in short-length propagation (<40 km), where mixing occurs
the waveguides’ size and pitch. As the individual wave- between fiber modes falling within the same mode group
guide modes expand (due to the decreasing core sizes), (having very similar propagation constants, delays, mode
the array structure starts to guide supermodes. It is the areas, and similar gains in amplifiers). Mode-selective
design goal to have the supermodes match the mode struc- and mode-group-selective PLs can be used to compensate
ture of the few-mode fiber, with the formed transition struc- for the differential group delay (DGD) of fiber spans or
ture called a photonic lantern (PL). When designed well, the modal gain differences in fiber amplifiers [76].
the PL can serve as a lossless mapping between M fiber PLs can also be fabricated using fibers, as an alternative
modes and M separate SMFs. to the 3D ULI approach. Multiple fibers are placed within a
While a PL can efficiently interface between FMF and the low-index capillary, and the structure is tapered until the
SMF array, they do not necessarily maintain a one-to-one light is guided by the fiber cladding and low-index capillary
mapping between modes and individual SMF. A PL can rather than by the individual fiber cores [77] (Fig. 14,
be designed with varying degrees of mode-selectivity, includ- bottom). The fiber PL is also an adiabatic device, has low
ing full selectivity (having a diagonal transfer matrix loss, can be spliced directly to the FMF, and can be scaled
between modes and individual SMF, or a one-to-one relation- to a large number of modes. When identical fibers are used,
ship), mode-group selectivity, and no mode selectivity (where the fiber PL has no mode-selectivity (i.e., a scrambling trans-
each SMF excites all modes with different efficiencies; see fer matrix). To add mode selectivity, fibers with different core
sizes can be used. Mode-group-selective devices have been
demonstrated with up to 15 spatial modes, enabling trans-
mission in fibers supporting 10 and 15 modes [78–80].
Drawbacks of the ULI PL compared to the fiber version
are higher insertion loss, difficulty in forming very large
multimode regions, and achieving a high NA.
2) Multiple WSS Packing: One of the detriments of
SDM–WDM ROADM designs in support of space–
wavelength granularity is their need for many WSS mod-
ules in their realization, resulting in an unfavorable cost
basis that scales with the mode count. Having multiple,
co-packaged WSSs within a single sub-system, sharing
the same optical, optoelectronic, electronic, and software
control elements can alleviate this largest disadvantage.
The SPOC-based processing front end within a free-space
WSS can be designed to support such functionality (see
Fig. 15). Here, the SPOC device implements an identical
spatial beam transformer at each zero-diffraction-order
WGR having multiple input/output waveguides at offset
positions to the slab lens. The offset positions lead to broad-
ened beams emerging from the WGR, with each offset po-
sition corresponding to a unique tilt angle. The subsequent
free-space optical arrangement maps each tilt to a unique
offset position on the LCoS SLM. The center WGR serves
as the input (shown with three inputs Ain , Bin , and Cin ),
which strike the LCoS SLM at unique positions (A, B, C).
The LCoS SLM directs the wavelength channel from
each independent WSS to its designated direction, and the
Fig. 14. Spatial mode demultiplexer between a multimode fiber
remaining WGRs serve the different output fiber ports
and SMF array based on a PL. Top: PL written by ultrafast laser (shown at the top output port position). The WGRs perform
inscription. Center: Transfer matrix characteristics between the function of angular multiplexing, allowing each spa-
modes and individual SMF for mode selective, mode-group selec- tially separated WSS to act independently, with its own in-
tive, and scrambling PL. Bottom: PL made by fiber tapering. put and output fiber ports that do not intermix. The spatial
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 17

Fig. 15. WSS multiplexing with a SPOC front end. Three 1 × 4

WSSs are integrated. The light paths of WSS_A, WSS_B, and
WSS_C are shown by solid, broken, and dotted lines, respectively.

overlap of the light paths among the WSSs makes the mod-
ule small, and the optical setup is as simple as that of a
conventional 1 × K WSS by sharing the common optical
components. This multiple WSS packing is thus nearly
identical to the SPOC-based D × L WSS (Fig. 5), but with-
Fig. 16. Top: WSS adapted by spatial diversity for joint-switching
out the selector switches on the output fibers [81]. The operation. SDM fibers (MCF shown, but other SDM solutions sup-
SPOC front end can be interfaced to SDM fibers using ported) are spatially demultiplexed and attached to the SMF ports
ULI waveguides in a glass block interface, as has been of the WSS. Bottom: Joint switching with space demux (using 3D
demonstrated for MCF integration [82]. waveguides) integrated within WSSs, for FMF WSSs supporting
three spatial modes.
3) Joint Switching WSS: Another approach for reducing
the SDM-ROADM hardware count and implementation cost FMF demux (three modes) and interfacing the SMF to a
is to employ WSSs supporting the joint-switching mode of commercial dual 1 × 20 WSS [83]. Alternatively, the mode
operation, required for the wavelength-switching granular- demux can be integrated within the WSS module, as dem-
ity approach. A joint-switching WSS accepts all the spatial onstrated in [84]. Spatial diversity was obtained from a 3D
channels and switches them as a group, on a wavelength- waveguide front end, which demultiplexes the modes from
channel basis, to desired output SDM destinations. As pre- three mode fibers and places them in linear array form in
viously noted, the design of a conventional WSS has all the the steering direction, but also reshuffles the beams such
fiber ports dispersed and imaged to the same positions at that the steering angle requirement for switching is reduced
the switching plane, where the input beam is directed to- (Fig. 16, bottom).
ward a desired output fiber on a wavelength basis (see
These joint-switching WSS solutions based on spatial
Subsection II.A). Since all the fibers are imaged to the same
diversity reduce the number of supported output SDM des-
position, the beam steering can simultaneously redirect
tinations by the spatial channel count, as exemplified by
multiple beams incident on it, with all beams experiencing
the 1 × 20 WSS turned into a 7 × 1 × 2 WSS. Thus instead
the same beam-steering shift. If the WSS input/output fibers
of having 20 output port options (in support of network
are arranged in a regularly spaced array, then the steering of
routing and add/drop support), functionality is reduced
a set of inputs is reimaged onto different sets of outputs.
to two output ports only (add and drop node). Two tech-
Associating the groups of fibers with SDM fiber interfaces
niques have been demonstrated to address this limitation:
allows the switching of all the spatial channels to their des-
a spatial diversity WSS that maps the fiber ports to two
tinations, gaining the SDM capacity multiplier without in-
dimensions and a WSS without spatial diversity that
creasing the WSS count (same WSS count as SMF
steers multimode beams and maintains port count.
implementations). This spatial diversity technique refor-
mats the channels of the SDM fiber to spatially separated Since a WSS images and disperses parallel beams at the
and parallel beams, enabling their joint switching with lens back focal plane where beam steering occurs, the spa-
the same hardware complexity as the SMF-based WSS. tial diversity can be extended beyond one-dimensional
The joint-switching concept was first demonstrated with realizations. By mapping the spatial channels to a 2D array
MCF interfaces with seven cores [12]. The cores were sepa- form while keeping the switching direction in the vertical
rated by an MCF breakout device and attached to the WSS. direction, switching action can be applied to rows of fibers
Two other sets of seven fibers were each attached to the out- of the array (Fig. 17). If SDM groups are associated with
put SDM fibers (Fig. 16, top). This enables switching the rows of the 2D array and an input SDM fiber is spatially
seven spatial channels to one of two output destinations, demultiplexed and fed to one row, then the entire SDM
for a 7 × 1 × 2 WSS functionality. This WSS requires 21 group can be jointly switched to output SDM fibers associ-
SMF at its interfaces, and indeed was based on a conven- ated with other rows. If an SDM group size is larger than
tional 1 × 20 WSS with fiber port reassignments. The same the column count, then multiple rows can be associated
joint-switching concept was subsequently demonstrated for with each SDM group. In [85], a 19 × 3 single-mode fiber
FMF, replacing the external MCF breakout devices with array (57 fibers) fed the WSS structure, which can support
18 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

Fig. 17. SDM WSS with spatial diversity spanning two dimen-
sions. Spatial channels are remapped to rows of the array, and
switching is performed in the vertical direction. Input row(s)
steered to desired output row(s). Fig. 18. FMF WSS with multimode beams propagated and
steered within the WSS.
3 × 1 × 18 switching functionality, demonstrating that an
SDM-WSS can be designed for joint switching of SDM impacting the spectral utilization of the communication
channels to many output destinations using spatial diver- band. Conversely, higher spectral resolution optics can
sity. Since the spatial demultiplexers are external to the be employed to achieve conforming transition bands. An al-
WSS, it can be mated to any SDM fiber solution. In ternative FMF–WSS implementation that addresses the
Ref. [85], the WSS was utilized in the first ROADM node channel passbands could be obtained if rectangular core fi-
realization within a heterogeneous network. Different ber (single-mode in one dimension, multimode in the other)
SDM fibers supporting six spatial channels were deployed were used for transmission instead of circular core fiber.
between four nodes of a line system (links with six SMFs A FMF–WSS was demonstrated with 10 fiber ports,
directly interfaced to the WSS, six-mode FMF with fiber PL where each I/O fiber supported three spatial modes
spatial demux, and six-mode MCF using a tapered fiber (LP01 and the degenerate LP11 modes), for 3 × 1 × 9
bundle spatial breakout). At the SDM ROADM node, the WSS functionality, and was used in a three-mode transmis-
spatial channels were separated and jointly switched with sion experiment over 300 km, where the WSS was passed
the 2D WSS (wavelength granularity approach as the up to ten times within a recirculating loop [13,86].
channels experienced mixing in fiber transmission).
SDM groups were mapped to two rows of the 19 × 3 fiber
array (since the 2D fiber array had three columns), with V. COST EVALUATION OF ROADM ARCHITECTURES FOR
the WSS then providing 6 × 1 × 8 functionality (one row TRANSMISSION OPTIONS BEYOND THE C-BAND
was discarded due to the odd row count). The WSS spectral
responses were nearly identical for each spatial channel We next attempt to compare the costs of different archi-
due to the spatial diversity solution. tecture options in support of capacity extending beyond the
Spatial diversity solutions require the WSS optics to single-mode communication band. This will be somewhat
handle many more free-space beams than switching ports, speculative because most of the components have not yet
due to the spatial multiplier. This strains the optical design been commercialized and their cost established. However,
of the WSS (performance and size). For FMF-based sys- we can make some comparisons with existing products.
tems, the fibers can be directly interfaced to the WSS fore- For example, a 1 × 5 joint-switching WSS for seven-core
going demultiplexing and spatial diversity. The beams fiber, or a 7 × 1 × 5 WSS (with a total of 42 ports), would
within the WSS become multimode, but otherwise traverse be about the same size and complexity as a conventional
the same optical elements as a conventional WSS. The 1 × 41 WSS, a product which does not yet exist. Since its port
number of free-space beams within the WSS equals the count is equal to that of a Twin 1 × 20 WSS, their cost should
number of switching ports (Fig. 18). However, the spatial be roughly equivalent. We can extrapolate to higher-port-
mode structure of the beams impacts the spectral channel count devices by comparing the relative costs to lower-
characteristics. At the beam-steering plane, where the port-count devices. However, we do not venture too far off
beams are spatially dispersed, flexible channel extent is de- in our extrapolation (i.e., we look at SDM increases of
fined by the LCoS SLM. Beam modes of optical frequencies up to factor 7), as it becomes increasingly speculative to
that strike the channel edges experience clipping, which estimate the cost of devices with port counts farther beyond
impacts each mode differently and disrupts mode orthogo- those of currently manufacturable devices. We consider only
nality (resulting in mode-dependent passbands and spec- the costs of the ROADM components, namely the WSSs,
trally dependent mode mixing at channel edges). This MCSs, and amplifiers associated with the MCSs. The
leads to larger spectral transition bandwidth between remaining network components (fiber, line amplifiers, trans-
the pass-band and the block-band. Hence, larger guard ceivers, etc.) are assumed to be cost neutral for different
bands are required between adjacent wavelength channels, architectures, which may not always be true due to
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 19

co-packaging cost benefits, but we have insufficient informa- against three-mode FMF) and larger (×7) capacity expan-
tion at this point to compare them; hence, they are excluded sion (for comparison against seven-core fiber).
from the analysis.
As a baseline, we choose a route-and-select CDC-Flex A. Single-Mode Fiber Overlays
ROADM node architecture using WSSs and MCSs in sup-
port of four network directions, as shown in Fig. 19. The
baseline system has a total of four Twin 1 × 10 WSSs The first capacity expansion option we consider is single-
(one Twin switch per direction), two Twin 4 × 16 MCSs mode fiber overlays with SDM lane changes. The overlay fi-
(two MCSs for add and two MCSs for drop, which can be bers are treated as additional node degrees except for the
used in protection mode, i.e., no single point of failure option to switch back to the same direction, which is not sup-
for add/drop channels, or in support of higher add/drop ported or needed. With no mode mixing present, this scheme
channel counts), and 16 amplifiers associated with the can support independent switching with space–wavelength
MCSs to compensate for their splitter/combiner losses. granularity for each overlay fiber. This could be considered
As protection schemes are beyond the scope of this paper, the default expansion option because it requires no new
we use the available add/drop ports to support a total of 32 technology development (at least for small numbers of over-
flexible channels. When the node capacity is scaled upward lay fibers) and is being implemented today for overloaded
in this study, we also scale the add/drop channel capacity in links. However, fiber overlays lead to increasing cost per
the same proportion. When comparing costs, we calculate bit because not only are more WSSs and MCSs required,
the relative cost per bit, so for example, if the capacity is but those components must be upgraded to support more
increased by a factor of 7, we divide the total ROADM cost ports. We consider only WSS options that do not exceed
by a factor of 7 to achieve a relative cost-per-bit basis. The the beam-steering capabilities of currently available tech-
cost model we employ is comparative, with the Twin 1 × 10 nology. For the MCS, its complexity scales as the product
WSS costing one unit, the relative cost of the Twin 4 × 16 of the number of common ports and A/D ports. This product
MCS module being 0.68, and the cost of a booster amplifier represents the worst-case number of MCS waveguide cross-
being 0.08. The total cost of the baseline single-mode sys- ings; each A/D port has a selector switch for connecting to
tem of Fig. 19 (four Twin 1 × 10 WSSs, two Twin 4 × 16 one of the common ports; the number of A/D ports deter-
MCSs, and 16 amplifiers) is normalized to one. In consid- mines the splitting ratio, and hence the inherent loss to each
ering device capability for higher port counts, we assume a common input port. The number of common ports must
50% price increase for a WSS (toward Twin 1 × 20 WSS) equal the degree of the node for unconstrained access; hence,
and 25% increase for an MCS (toward Twin 8 × 16 MCS). as more overlay fibers are added, the number of common
ports on each MCS increases. We assume that the complex-
We consider the following capacity expansion options:
ity of each MCS cannot increase without bound. Hence, we
A. single-mode fiber overlays, which can be implemented
limit the complexity by proportionally reducing the number
(i) with independent channel routing with SDM lane
of A/D ports when the common ports increase in an effort to
changes or (ii) without, in which case the cost per bit is
achieve a manufacturable product (using the 8 × 16 MCS as
essentially the same as the baseline since all elements
the state of the art scaling). This caps the cost and complex-
are simply replicated, or (iii) with joint-switching WSSs;
ity to that of existing 8 × 16 MCSs, but increases the number
B. few-mode fiber; C. multi-core fiber; and D. extension
of MCSs required and the number of WSS ports for connect-
to L-band and S-band. In each case, where appropriate,
ing to the MCS modules.
we consider a moderate (×3) capacity expansion (which
could also be achieved by spectral extension and compared Using these scaling assumptions, we arrive at a total
WSS cost of 4.4 times the baseline WSS cost for a 3× over-
lay (three fibers in each of the four directions, requiring by
12 units of Twin 1 × 20 WSSs), and 25.5 times the baseline
WSS cost for a 7× overlay (56 units of 1 × 80 WSSs), and a
total MCS cost of 5.7 times the baseline MCS cost for a
3× overlay (9 units of Twin 12 × 11 MCSs) and 31 times
the baseline MCS cost for a 7× overlay (49 units of Twin
28 × 5 MCSs). A total of 216 amplifiers are required for
the 3× overlay (9 MCSs ×12 common ports ×2 for add
and drop), but it is noted that for the 7× overlay case,
the number of A/D ports on each MCS is quite small
(and hence the splitting ratio within the MCS is low), so
we assume that amplifiers are not needed. Had amplifiers
been required, the 7× overlay configuration would have re-
quired more than 2000 amplifiers, which would have had a
detrimental impact on the overall cost. As an aside, con-
sider that a 28 × 16 MCS could be manufactured; the over-
all cost of the 7× overlay using this hypothetical MCS
Fig. 19. Baseline ROADM node architecture used for the cost variant would have comparable or possibly higher cost
analysis. than using the 28 × 5 MCS, as the cost benefits of reducing
20 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

the number of MCS modules and WSS port counts would be date fall short of present-day performance requirements,
offset by additional amplifier costs. In any case, it is appar- particularly in transient suppression, gain flatness, and
ent that the ROADM costs for fiber overlays increase faster power efficiency. We note only that the cost increase must
than the capacity increases. fall between a factor of 1 (same cost as conventional ampli-
In consideration of single-mode fibers without SDM lane fier) and a cost increase of 3 or 7 (which results if we amplify
changes, which is essentially M independent baseline net- the modes separately with three or seven conventional
works, this arrangement would enable capacity increase single-mode amplifiers; if the multimode amplifier cost is
with the same cost per bit as the baseline system. While higher than this we use instead multiple single-mode ampli-
this would result in lower cost compared to the aforemen- fiers). So we choose intermediate values and assume a cost
tioned system supporting SDM lane changes, the cost ad- increase of 2 times for three modes, and 3.5 times for
vantage can only be realized if every link of the network seven modes.
requires M times the capacity. At present, adding overlays
as node degrees is the preferred expansion option because
capacity growth is uneven throughout the network and the C. Multi-Core Fiber Systems
routing flexibility of SDM lane changes enables efficient
use of existing hardware. However, this could change in For MCF with joint switching, the architecture again
the future as the number of overlays increases. looks like that of Fig. 19, but with all components compatible
with multi-core fiber. The most practical method of making a
B. Few-Mode Fiber Systems WSS compatible with MCF is with spatial diversity, where
the multiple cores are separated and jointly switched (pos-
sibly with the remapping function to reduce the steering
For FMF, we assume the ROADM architecture looks the angle requirements). This makes the WSS realizable using
same as that shown in Fig. 19 (same number of components today’s WSS switching engine technology. The cost increase
with similar port counts) but with all components made com- for seven-core fiber is estimated to be a factor of 3.6, which
patible for FMF. Our task here is to estimate the cost of these includes a factor of two because the large port count pre-
FMF-compatible components. For a FMF WSS we choose the cludes making a Twin device, and a factor of 1.8 to account
spatial diversity approach (as shown in Fig. 17) rather than for the large port count. Note that the cost for an MCF-
propagating multimode beams (see Fig. 18), due to similar compatible WSS is estimated to be higher than the FMF-
spectral resolution requirements, facilitating the cost com- compatible WSS due to the stringent requirement to maintain
parisons to existing SMF WSSs. Here, we assume the sepa- core isolation, which leads to larger pitch fiber and collimator
rated modes are mapped to fibers that are grouped together
arrays and bigger overall implementation.
(without port reshuffling). These fiber ports can be spaced
more tightly than conventional WSS ports, as crosstalk be- For an MCS compatible with seven-core fiber, one could
tween modes of the same fiber is insignificant, thereby reduc- break out the modes and use seven conventional integrated
ing optics height and steering angle requirements and hence MCSs; however, this results in a cost increase of ×7.
reducing the cost of the WSS. For example, a 1 × 10 WSS for Instead, we assume that MCS can utilize free-space com-
three-mode FMF with compressed spacing between ports ponents designed to accommodate seven modes per mirror
from the same fiber would require about the same optics and beamsplitter, which results in a cost increase of only
height and beam steering requirements as a 1 × 20 WSS about 3.4, taking advantage of the assumption that we
for SMF, even though there are effectively 30 output ports. need only switching with wavelength granularity, in which
Thus, the WSS cost is estimated to be about the same as the case free-space optics can provide better cost scaling than
cost difference between a 1 × 10 WSS and a 1 × 20 WSS for integrated optics because the number of components does
SMF, a factor of about 1.5. Similarly, the WSS cost is esti- not increase with the number of cores. If lane (or core)-
mated to increase by a factor of 1.9 for seven modes (seven change capability is required with MCF, the architecture
modes used rather than the more common six to allow direct becomes similar to the overlay configuration, with compa-
comparison with seven-core fiber). rable cost.
FMF presents a problem for integrated MCS technology, For integrated MCF amplifiers, we assume a cost in-
because switches and splitters are difficult to implement crease between the limits of 1 and 7, where we use a cost
in PLC waveguides. We assume that free-space switches increase of 3.5 times for the seven-core fiber amplifier in
(MEMS-based) and free-space beamsplitters are used in- our cost estimates.
stead to realize the MCS functionality, and that this can The wavelength-granularity switching strategy (joint
be done for roughly the same cost as the integrated version switching in the space domain) can be applied to the single-
[87]. The sizes of the MEMS mirrors and beamsplitters mode fiber overlay architecture, with cost savings achieved
would need to be increased to accommodate the multimode by making use of the same WSS and MCS components we
beams, but otherwise the MCS design would be similar to a proposed for the MCF ROADM architecture with wavelength
single-mode version. We estimate the MCS cost to be in- granularity. The resulting costs are similar to those for the
creased by a factor of 1.3 for three modes and by a factor MCF ROADM architecture (which are much lower than the
of 1.7 for seven modes. costs for the case of overlay fibers employing ROADM with
Estimating the amplifier cost for multimode fiber is a space–wavelength granularity), except that the amplifier
challenge because multimode amplifier demonstrations to cost would be seven times for a 7× overlay (one amplifier
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 21

per MCF common port). Alternatively, switching with frac-

tional space, full wavelength granularity could be imple-
mented, which would have a cost intermediate between
those of switching with wavelength granularity and with
space–wavelength granularity.

D. Multi-Band Extended Systems

Finally, we consider extending the operating wavelength

range to the L and S bands as a means of increasing capac-
ity. The most straightforward approach would be to sepa-
rate the bands and have duplicate WSSs, MCSs, and Fig. 20. Comparison of ROADM normalized costs for considered
amplifiers for each band (presuming S-band amplifiers capacity expansion options.
become available), which increases the capacity by 2 or 3
times but at 2 or 3 times the ROADM cost. Two alternatives requires consideration of all costs, of which this is only a
may emerge: The first requires a redesign of the WSS to part. The actual SDM fiber may dominate the cost, and op-
support a larger communication band at the same resolv- erators may first choose the overlay solutions which utilize
ing power as today’s systems. This faces design challenges, an already available SMF plant buried in the ground.
as the WSS complexity scales with the number of resolv-
able points (or bandwidth support/optical resolution). As
a cost-saving alternative, we propose to preserve WSS com- VI. CONCLUSION AND OUTLOOK
plexity (i.e., the resolvable points) and accompany the ex-
tended bandwidth with a coarser spectral resolution by The persistent growth of traffic carried by optical net-
inserting a lower dispersion diffraction grating. Hence, this works, coupled with the fact that all the “easy” solutions
would enable extended band coverage without increasing to exploit the optical bandwidth available across the con-
the WSS cost. This does have the significant disadvantage ventional communication band have been exhausted, im-
that the channel width must increase to maintain spectral plies that new measures will have to be taken in the
efficiency, so it is unlikely to be adopted, but for the sake of years ahead to achieve additional network capacity. In this
presenting cost-saving options we consider it here. We as- survey paper we have outlined several measures, all intro-
sume that the WSS and MCS costs are unchanged (it is ducing innovations in optical switching systems to fully uti-
assumed that an MCS can be made to cover the extended lize the spectral domain and further extend into the spatial
frequency range), but we consider separate amplifiers for domain. EON will allow for the full exploitation of the
each band, so an amplifier cost scaling of 2 or 3 times C-band, with information-bearing signals having minimal
for two or three bands. It is worth pointing out that much GBs between wide superchannels. Technologies in support
of the apparent cost savings of this approach is simply due of subchannel addition and extraction within the super-
to the increase in size of the switchable data unit (the in- channel for signals with fine GBs between them or even
creased channel width), which results in fewer A/D ports spectrally overlapping have been identified. After fully ex-
needed for a given volume of dropped traffic. This cost- ploiting the conventional communication band, additional
saving technique is being applied to current networks, information can be encoded beyond its spectrum (extension
through the use of superchannels, with no change to the to the long and short bands) or to the spatial domain (SMF
ROADM infrastructure, provided it is capable of handling overlays or new SDM fibers). WSS solutions for all these
flexible channel widths. However, with lesser spectral scenarios have been discussed. The spatial domain prom-
resolving power, the guard bands between superchannels ises a larger capacity multiplier. However, network realiza-
would have to be increased, thereby negatively impacting tions with high-channel-count SDM solutions must provide
spectral utilization, i.e., capacity. the capacity benefits at small incremental costs (sublinear)
for the network to scale in a cost-effective manner in sup-
port of the projected capacity increases. We identify that
E. Cost Comparison some measure of loss of routing flexibility may need to
be incurred, whether in the frequency or space domains
Combining all of the calculations and cost projections on or in a combination of both, to obtain significant equipment
future devices, we arrive at the chart shown in Fig. 20, com- and cost reduction. The networking impact of such routing
paring the normalized ROADM cost per bit for the various restrictions continues to be a topic of debated research.
options. The main takeaways from this chart are the rap- As the switching solutions introduced in support of SDM–
idly increasing costs for the overlay approach with indepen- WDM information encoding can adapt to all fiber types (multi-
dent switching, and that capacity expansion options with ple SMF, FMF, MCF), we can safely conclude that the switches
joint switching (FMF, MCF, and overlays) result in lower alone will not serve as the determining factor in the identifi-
costs than the baseline. cation of the optimal SDM solution. Additional factors include
We emphasize that this analysis considers only the cost optical amplifiers (their cost, gain value and uniformity, and
of ROADM components. To make an architecture decision noise figure), receiver complexity (for cases where MIMO
22 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

processing is required), the cost of producing the fiber medium [4] R. Essiambre, G. Kramer, P. J. Winzer, G. J. Foschini, and
itself, and networking efficiency. However, the SDM–WDM B. Goebel, “Capacity limits of optical fiber networks,”
ROADM implementation complexity and cost may indeed J. Lightwave Technol., vol. 28, no. 4, pp. 662–701, 2010.
be influential. Taken together, we can imagine in one extreme [5] O. Gerstel, M. Jinno, A. Lord, and S. J. B. Yoo, “Elastic optical
case a multimode fiber supporting a large spatial channel networking: A new dawn for the optical layer?” IEEE
Commun. Mag., vol. 50, no. 2, pp. s12–s20, 2012.
count (at least an order of magnitude increase in capacity).
In this case, channels are prone to mixing and hence must [6] I. Tomkos, S. Azodolmolky, J. Solé-Pareta, D. Careglio, and E.
Palkopoulou, “A tutorial on the flexible optical networking
be jointly switched. The ROADM implementation is the sim-
paradigm: State-of-the-art, trends, and research challenges,”
plest of all SDM–WDM switching solutions, having the same
Proc. IEEE, vol. 102, no. 9, pp. 1317–1337, 2014.
WSS count per node as in SMF networks (but with modified
[7] D. Klonidis, S. Sygletos, D. M. Marom, S. Fabbri, A. Ellis,
WSSs), and will incorporate integrated SDM transceivers
E. Pincemin, C. Betoule, G. Thouenon, D. Hillerkuss, B.
and MIMO processing. If multimode optical fiber amplifiers Baeuerle, A. Josten, J. Leuthold, J. Zhao, S. Ben-Ezra, J. F.
are used, this solution appears to be the least expensive. Ferran, M. Angelou, G. Papastergiou, P. Zakynthinos, and
However, spatial channel utilization and fragmentation under I. Tomkos, “Enabling transparent technologies for the devel-
different network loads will lower the overall efficiency. At the opment of highly granular flexible optical cross-connects,” in
opposite extreme, the cases of MCF with well-isolated cores or Proc. 16th Int. Conf. Transparent Optical Networks (ICTON),
parallel SMF links will experience no mixing in transmission. 2014, paper We.D1.5.
The ROADM node complexity will be compounded when the [8] B. Shariati, P. S. Khodashenas, J. M. Rivas-Moscoso, S. Ben-
switching granularity is as fine as the individual wavelength Ezra, D. Klonidis, F. Jiménez, L. Velasco, and I. Tomkos,
and spatial channel due to many switching modules required “Investigation of mid-term network migration scenarios com-
for its implementation. However, the system can utilize to- paring multi-band and multi-fiber deployments,” presented
at the Optical Fiber Communication Conf. and Exhibition
day’s single-channel transceivers. This solution will likely
(OFC), Anaheim, CA, Mar. 2016, paper Th1E.1.
benefit the least from SDM integration, and hence the capac-
[9] F. Farjady, N. Antoniades, R. E. Wagner, and M. J. Yadlowsky,
ity increase will be accompanied by escalating cost, which may
“Value of fiber overlays in WDM metro networks,” IEEE
make this solution unattractive. Overall SDM system trade- Photon. Technol. Lett., vol. 15, no. 2, pp. 329–331, 2003.
offs in switching granularity need to be considered together
[10] K. Nakajima, P. Sillard, D. Richardson, M.-J. Li, R.-J.
with networking efficiency, cost, and technology maturation, Essiambre, and S. Matsuo, “Transmission media for an
which will likely take a few more years—until the traffic vol- SDM-based optical communication system,” IEEE Commun.
umes required for delivery cannot be well-sustained by the Mag., vol. 53, no. 2, pp. 44–51, 2015.
existing infrastructure. [11] M. D. Feuer, L. E. Nelson, X. Zhou, S. L. Woodward, R. Isaac,
The need to extend beyond optical networks operating on B. Zhu, T. F. Taunay, M. Fishteyn, J. M. Fini, and M. F. Yan,
the conventional communication band supported by SMF “Joint digital signal processing receivers for spatial super-
represents a great opportunity for optical technology innova- channels,” IEEE Photon. Technol. Lett., vol. 24, no. 21,
pp. 1957–1960, 2012.
tion, ranging from the fiber to ancillary components and sub-
[12] L. E. Nelson, M. D. Feuer, K. Abedin, X. Zhou, T. F. Taunay,
systems. In this paper, we have highlighted the switching
J. M. Fini, B. Zhu, R. Isaac, R. Harel, G. Cohen, and D. M.
solutions we foresee to address this challenge and discussed
Marom, “Spatial superchannel routing in a two-span
the merits and demerits of each proposed ROADM node im- ROADM system for space division multiplexing,” J.
plementation (see Table II for our comparative summary). Lightwave Technol., vol. 32, no. 4, pp. 783–789, 2014.
However, we anticipate there is room for additional innova- [13] D. M. Marom, J. Dunayevsky, D. Sinefeld, M. Blau, R. Ryf,
tion and ingenuity in this space, and encourage additional N. K. Fontaine, M. Montoliu, S. Randel, C. Liu, B. Ercan,
research in this developing and significant field. M. Esmaeelpour, S. Chandrasekhar, A. H. Gnauck, S. G.
Leon-Saval, J. Bland-Hawthorn, J. R. Salazar-Gil, Y. Sun,
L. Grüner-Nielsen, and R. Lingle, Jr., “Wavelength-selective
ACKNOWLEDGMENT switch with direct few mode fiber integration,” Opt. Express,
vol. 23, no. 5, pp. 5723–5737, 2015.
This work was supported in part by the European
Commission’s Seventh Framework Program under [14] D. Siracusa, F. Pederzolli, P. Khodashenas, J. M. Rivas-
Moscoso, D. Klonidis, E. Salvadori, and I. Tomkos, “Spectral
Grant Agreements 318415 (FOX-C project) and 619732
vs. spatial superchannel allocation in SDM networks under in-
(INSPACE project).
dependent and joint switching paradigms,” in Proc. European
Conf. Optical Communication (ECOC), 2015, paper Mo.4.6.2.
REFERENCES [15] P. S. Khodashenas, J. M. Rivas-Moscoso, D. Siracusa, F.
Pederzolli, B. Shariati, D. Klonidis, E. Salvadori, and I.
Tomkos, “Comparison of spectral and spatial super-channel
[1] P. J. Winzer, “Spatial multiplexing in fiber optics: The allocation schemes for SDM networks,” J. Lightwave
10× scaling of metro/core capacities,” Bell Labs Tech. J., Technol., vol. 34, no. 11, pp. 2710–2716, 2016.
vol. 19, pp. 22–30, 2014.
[16] B. Shariati, P. S. Khodashenas, J. M. Rivas-Moscoso, S. Ben-
[2] “Cisco Visual Networking Index: Forecast and methodology, Ezra, D. Klonidis, F. Jiménez, L. Velasco, and I. Tomkos,
2015–2020,” June 2016. “Evaluation of the impact of different SDM switching strate-
[3] A. D. Ellis, J. Zhao, and D. Cotter, “Approaching the nonlinear gies in a network planning scenario,” presented at the Optical
Shannon limit,” J. Lightwave Technol., vol. 28, no. 4, Fiber Communication Conf. and Exhibition, Anaheim, CA,
pp. 423–433, 2010. Mar. 2016, paper Tu2H.4.
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 23

[17] D. T. Nelson, C. R. Doerr, D. M. Marom, R. Ryf, and M. P. matrix switches,” in Proc. European Conf. Optical Communi-
Earnshaw, “Wavelength selective switching for optical band- cation (ECOC), 2010, paper Mo.2.D.2.
width management,” Bell Labs Tech. J., vol. 11, no. 2, [33] E. Hugues-Salas, G. Zervas, D. Simeonidou, E. Kosmatos, T.
pp. 105–128, 2006. Orphanoudakis, A. Stavdas, M. Bohn, A. Napoli, T. Rahman,
[18] P. Colbourne and B. Collings, “ROADM switching technolo- F. Cugini, N. Sambo, S. Frigerio, A. D’Errico, A. Pagano, E.
gies,” in Optical Fiber Communication Conf. (OFC), 2011, Riccardi, V. López, and J. P. Fernández-Palacios Giménez,
paper OTuD1. “Next generation optical nodes: The vision of the European
[19] D. M. Marom, D. T. Neilson, D. S. Greywall, C. S. Pai, N. R. research project IDEALIST,” IEEE Commun. Mag., vol. 53,
Basavanhally, V. A. Aksyuk, D. O. López, F. Pardo, M. E. no. 2, pp. 172–181, 2015.
Simon, Y. Low, P. Kolodner, and C. A. Bolle, “Wavelength- [34] K. Suzuki, Y. Ikuma, E. Hashimoto, K. Yamaguchi, M. Itoh,
selective 1 × K switches using free-space optics and MEMS and T. Takahashi, “Ultra-high port count wavelength selec-
micromirrors: Theory, design, and implementation,” J. tive switch employing waveguide-based I/O frontend,” in
Lightwave Technol., vol. 23, no. 4, pp. 1620–1630, 2005. Optical Fiber Communication Conf. (OFC), Los Angeles,
[20] R. Ryf, Y. Su, L. Moller, S. Chandrasekhar, X. Liu, D. T. CA, 2015.
Neilson, and C. R. Giles, “Wavelength blocking filter [35] M. Iwama, M. Takahashi, M. Kimura, Y. Uchida, J.
with flexible data rates and channel spacing,” IEEE J. Hasegawa, R. Kawahara, and N. Kagi, “LCOS-based flexible
Lightwave Technol., vol. 23, no. 1, pp. 54–61, 2005. grid 1 × 40 wavelength selective switch using planar light-
[21] T. A. Strasser and J. L. Wagener, “Wavelength-selective wave circuit as spot size converter,” in Optical Fiber
switches for ROADM applications,” IEEE J. Sel. Top. Communication Conf. (OFC), Los Angeles, CA, 2015.
Quantum Electron., vol. 16, no. 5, pp. 1150–1157, 2010. [36] M. Iwama, M. Takahashi, Y. Uchida, M. Kimura, R.
[22] G. Baxter, S. Frisken, D. Abakoumov, H. Zhou, I. Clarke, A. Kawahara, S.-I. Matsushita, and T. Mukaihara, “Low loss
Bartos, and S. Poole, “Highly programmable wavelength 1 × 93 wavelength selective switch using PLC-based spot size
selective switch based on liquid crystal on silicon switching converter,” in Proc. European Conf. Optical Communication
elements,” in Optical Fiber Communication Conf. (OFC), (ECOC), Valencia, Spain, 2015, paper Mo.4.2.2.
2006, paper OTuF2. [37] Y. Ikuma, K. Suzuki, N. Nemoto, E. Hashimoto, O. Moriwaki,
[23] D. Sinefeld and D. M. Marom, “Insertion loss and crosstalk and T. Takahashi, “8 × 24 wavelength selective switch for low-
analysis of a fiber switch based on a pixelized phase modula- loss transponder aggregator,” presented at the Optical Fiber
tor,” J. Lightwave Technol., vol. 29, pp. 69–77, 2011. Communication Conf. (OFC), Los Angeles, CA, Mar. 2015,
[24] S. J. Frisken, G. W. Baxter, H. Zhao, and D. Abakoumov, paper Th5A.4.
“Optical calibration system and method,” U.S. patent [38] P. Colbourne, “M × N wavelength selective optical switch,”
7457547 (Nov. 25, 2008). U.S. patent 20100172646 A1 (July 8, 2010).
[25] S. Frisken, G. Baxter, D. Abakoumov, H. Zhou, I. Clarke, and [39] Y. Yan, H. Zhao, L. Zong, and Z. Feng, “8 × 128 adWSS for
S. Poole, “Flexible and grid-less wavelength selective switch us- CDC ROADM,” in Int. Conf. Photonics in Switching (PS),
ing LCOS technology,” in Optical Fiber Communication Conf. Florence, Italy, Sept. 2015, paper WeI2.5.
(OFC), 2011, paper OTuM3. [40] L. Pascar, R. Karubi, B. Frenkel, and D. M. Marom, “Port-
[26] J.-K. Rhee, I. Tomkos, and M.-J. Li, “A broadcast-and-select reconfigurable, wavelength-selective switch array for colorless/
OADM optical network with dedicated optical-channel pro- directionless/contentionless optical add/drop multiplexing,” pre-
tection,” J. Lightwave Technol., vol. 21, no. 1, pp. 25–31, 2003. sented at Int. Conf. Photonics in Switching (PS), Florence, Italy,
[27] M. Vasilyev, I. Tomkos, M. Mehendale, J.-K. Rhee, A. Sept. 2015, paper PDP.2.
Kobyakov, M. Ajgaonkar, S. Tsuda, and M. Sharma, [41] R. Ryf, P. Bernasconi, P. Kolodner, J. Kim, J. P. Hickey, D. Carr,
“Transparent ultra-long-haul DWDM networks with “broad- F. Pardo, C. Bolle, R. Frahm, N. Basavanhally, C. Yoh, D.
cast-and-select” OADM/OXC architecture,” J. Lightwave Ramsey, R. George, J. Kraus, C. Lichtenwalner, R.
Technol., vol. 21, no. 11, pp. 2661–2672, 2003. Papazian, J. Gates, H. R. Shea, A. Gasparyan, V. Muratov,
[28] D. T. Neilson, H. Tang, D. S. Greywall, N. R. Basavanhally, L. J. E. Griffith, J. A. Prybyla, S. Goyal, C. D. White, M. T.
Ko, D. A. Ramsey, J. D. Weld, Y. L. Low, F. Pardo, D. O. Lopez, Lin, R. Ruel, C. Nijander, S. Amey, D. T. Neilson, D. J.
P. Busch, J. Prybyla, M. Haueis, C. S. Pai, R. Scotti, and R. Bishop, S. Pau, C. Nuzman, A. Weis, B. Kumar, D.
Ryf, “Channel equalization and blocking filter utilizing micro- Lieuwen, V. Aksyuk, D. S. Greywall, T. C. Lee, H. T. Soh,
electromechanical mirrors,” IEEE J. Sel. Top. Quantum W. M. Mansfield, S. Jin, W. Y. Lai, H. A. Huggins, D. L.
Electron., vol. 10, no. 3, pp. 563–569, 2004. Barr, R. A. Cirelli, G. R. Bogart, K. Teffeau, R. Vella, H.
[29] B. Collings, “New devices enabling software-defined optical Mavoori, A. Ramirez, N. A. Ciampa, F. P. Klemens, M. D.
networks,” IEEE Commun. Mag., vol. 51, no. 3, pp. 66–71, Morris, T. Boone, J. Q. Liu, J. M. Rosamilia, and C. R.
2013. Giies, “Scalable wavelength-selective crossconnect switch
[30] D. M. Marom, D. T. Neilson, J. Leuthold, M. A. Gibbons, and based on MEMS and planar waveguides,” in Proc.
C. R. Giles, “64 channel 4 × 4 wavelength-selective cross- European Conf. Optical Communication (ECOC), 2001.
connect for 40 Gb/s channel rates with 10 Tb/s throughput [42] N. K. Fontaine, R. Ryf, and D. T. Neilson, “N × M wavelength
capacity,” in Proc. European Conf. Optical Communication selective crossconnect with flexible passbands,” in Optical Fiber
(ECOC), 2003, pp. 816–817. Communication Conf. (OFC), 2012, paper PDP5B.2.
[31] J. M. Simmons and A. A. M. Saleh, “Wavelength-selective [43] F. Xiao and K. Alameh, “Opto-VLSI-based N × M wavelength
CDC ROADM designs using reduced-sized optical cross- selective switch,” Opt. Express, vol. 21, pp. 18160–18169, 2013.
connects,” IEEE Photon. Technol. Lett., vol. 27, no. 20, [44] L. Zong, H. Zhao, Z. Feng, and S. Cao, “Demonstration of ultra-
pp. 2174–2177, 2015. compact contentionless-ROADM based on flexible wavelength
[32] R. Jensen, A. Lord, and N. Parsons, “Colourless, directionless, router,” in Proc. European Conf. Optical Communication
contentionless ROADM architecture using low-loss optical (ECOC), 2014, paper P.4.4.
24 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

[45] L. Zong, H. Zhao, Z. Feng, and Y. Yan, “8 × 8 flexible wave- [58] M. Song, E. Pincemin, A. Josten, B. Baeuerle, D. Hillerkuss, J.
length cross-connect for CDC ROADM application,” IEEE Leuthold, R. Rudnick, D. M. Marom, S. Ben Ezra, J. F. Ferran,
Photon. Technol. Lett., vol. 27, no. 24, pp. 2603–2606, 2015. G. Thouenon, P. S. Khodashenas, J. M. Rivas-Moscoso, C.
[46] N. Nemoto, Y. Ikuma, K. Suzuki, O. Moriwaki, T. Watanabe, Betoule, D. Klonidis, and I. Tomkos, “Flexible optical cross-
M. Itoh, and T. Takahashi, “8 × 8 wavelength cross connect connects for high bit rate elastic photonic transport net-
with add/drop ports integrated in spatial and planar optical works,” J. Opt. Commun. Netw., vol. 8, pp. A126–A140, 2016.
circuit,” in Proc. European Conf. Optical Communication [59] M. Song, E. Pincemin, B. Baeuerle, A. Josten, D. Hillerkuss, J.
(ECOC), 2015, paper Tu.3.5.1. Leuthold, R. Rudnick, D. M. Marom, S. Ben-Ezra, J. F. Ferran,
[47] L. Zong, H. Zhao, Y. Yan, and Z. Feng, “Demonstration of S. Sygletos, A. Ellis, J. Zhao, G. Thouenon, C. Betoule, J. M.
quasi-contentionless flexible ROADM based on a multiport Rivas, D. Klonidis, and I. Tomkos, “Cascaded all-optical sub-
WXC,” J. Opt. Commun. Netw., vol. 8, pp. A141–A151, 2016. channel add/drop multiplexing from a 1-Tb/s super-channel
having 2-GHz guard-bands,” in Int. Conf. Photonics
[48] Y. Loussouarn, E. Pincemin, M. Song, S. Gauthier, Y. Chen,
Switching (PS), Niigata, Japan, July 2016.
and S. Zhong, “400 Gbps real-time coherent Nyquist-WDM
DP-16QAM transmission over legacy G.652 or G.655 fibre in- [60] L. Zhuang, C. Zhu, Y. Xie, M. Burla, C. G. H. Roeloffzen, M.
frastructure with 2 dB margins,” in Optical Fiber Hoekman, B. Corcoran, and A. J. Lowery, “Nyquist-filtering
Communication Conf. (OFC), Los Angeles, CA, 2015. (de)multiplexer using a ring resonator assisted inter-
ferometer circuit,” J. Lightwave Technol., vol. 34, no. 8,
[49] S. Melle, A. Deore, O. Turkcu, S. Ahuja, and S. J. Hand,
pp. 1732–1738, 2016.
“Comparing optical and OTN switching architectures in next-
[61] B. Corcoran, C. Zhu, J. Schröder, L. Zhuang, B. Foo, M. Burla,
gen 100 Gb/s networks,” presented at the Optical Fiber
W. P. Beeker, A. Leinse, C. G. H. Roeloffzen, and A. J. Lowery,
Communication Conf. and Exhibition, Anaheim, CA,
“Multipass performance of a chip-enhanced WSS for Nyquist-
Mar. 2013, paper NM3F-2.
WDM sub-band switching,” J. Lightwave Technol., vol. 34,
[50] P. S. Khodashenas, J. M. Rivas-Moscoso, D. Klonidis, G. no. 8, pp. 1824–1830, 2016.
Thouénon, C. Betoule, and I. Tomkos, “Impairment-aware re-
[62] P. J. Winzer, “An opto-electronic interferometer and its use in
source allocation over flexi-grid network with all-optical add/
subcarrier add/drop multiplexing,” J. Lightwave Technol.,
drop capability,” presented at the European Conf. Optical
vol. 31, no. 11, pp. 1775–1782, 2013.
Communication (ECOC), Valencia, Spain, Sept. 2015,
paper P.6.13. [63] T. Richter, C. Schmidt-Langhorst, R. Elschner, T. Kato, T.
Tamimura, S. Watanabe, and C. Schubert, “Coherent in-line
[51] P. S. Khodashenas, J. M. Rivas-Moscoso, D. Klonidis, I. Tomkos,
substitution of OFDM subcarriers using fiber-frequency con-
G. Thouénon, C. Betoule, and E. Pincemin, “Techno-economic
version and free-running lasers,” in Optical Fiber
analysis of flexi-grid networks with all-optical add/drop
Communication Conf. (OFC), San Francisco, CA, 2014.
capability,” presented at Int. Conf. Photonics in Switching
[64] S. Sygletos, S. J. Fabbri, E. Giacoumidis, M. Sorokina, D. M.
(PS), Florence, Italy, Sept. 2015, paper ThIII2.2.
Marom, M. F. C. Stephens, D. Klonidis, I. Tomkos, and A. D.
[52] G. Thouénon, C. Betoule, E. Pincemin, P. S. Khodashenas, J.
Ellis, “A novel architecture for all-optical add-drop multiplexing
M. Rivas-Moscoso, and I. Tomkos, “All-optical vs. electrical
of OFDM signals,” in Proc. European Conf. Optical
aggregations CAPEX comparisons in a fully-flexible multi- Communication (ECOC), Cannes, France, 2014, paper We.1.5.4.
layer transport network,” presented at the European Conf. on
[65] D. Hillerkuss, R. Schmogrow, T. Schellinger, M. Jordan, M.
Optical Communication (ECOC), Valencia, Spain, Sept. 2015,
Winter, G. Huber, T. Vallaitis, R. Bonk, P. Kleinow, F. Frey,
paper We.4.5.4.
M. Roeger, S. Koenig, A. Ludwig, A. Marculescu, J. Li, M.
[53] R. Rudnick, A. Tolmachev, D. Sinefeld, O. Golani, S. Ben-Ezra, Hoh, M. Dreschmann, J. Meyer, S. Ben-Ezra, N. Narkiss,
M. Nazarathy, and D. M. Marom, “Sub-banded/single-sub- B. Nebendahl, F. Parmigiani, P. Petropoulos, B. Resan, A.
carrier drop-demux and flexible spectral shaping with a fine Oehler, K. Weingarten, T. Ellermeyer, J. Lutz, M. Moeller,
resolution photonic processor,” in Proc. European Conf. M. Huebner, J. Becker, C. Koos, W. Freude, and J.
Optical Communication (ECOC), Cannes, France, 2014, Leuthold, “26 Tbit s-1 line-rate super-channel transmission
paper PD.4.1. utilizing all-optical fast Fourier transform processing,” Nat.
[54] N. Goldshtein, D. Sinefeld, O. Golani, R. Rudnick, L. Pascar, Photonics, vol. 5, no. 6, pp. 364–371, 2011.
R. Zektzer, and D. M. Marom, “Fine resolution photonic spec- [66] S. J. Fabbri, S. Sygletos, A. Perentos, E. Pincemin, K. Sugden,
tral processor using a waveguide grating router with perma- and A. D. Ellis, “Experimental implementation of an all-optical
nent phase trimming,” J. Lightwave Technol., vol. 34, no. 23, interferometric drop, add, and extract multiplexer for superchan-
pp. 379–385, 2016. nels,” J. Lightwave Technol., vol. 33, no. 7, pp. 1351–1357, 2015.
[55] “Spectral grids for WDM applications: DWDM frequency [67] D. M. Marom and M. Blau, “Switching solutions for WDM-
grid,” ITU-T Recommendation G.694.1, Feb. 2012. SDM optical networks,” IEEE Commun. Mag., vol. 53,
[56] P. S. Khodashenas, J. M. Rivas-Moscoso, B. Shariati, D. M. no. 2, pp. 60–68, 2015.
Marom, D. Klonidis, and I. Tomkos, “Investigation of spec- [68] R. Ramamurthy and B. Mukherjee, “Fixed-alternate routing
trum granularity for performance optimization of flexible and wavelength conversion in wavelength-routed optical net-
Nyquist-WDM-based optical networks,” J. Lightwave works,” IEEE/ACM Trans. Netw., vol. 10, no. 3, pp. 351–367,
Technol., vol. 33, no. 23, pp. 4767–4774, 2015. 2002.
[57] P. Torres-Ferrera, J. M. Rivas-Moscoso, D. Klonidis, D. M. [69] I. P. Kaminow, C. R. Doerr, C. Dragone, T. Koch, U. Koren,
Marom, R. Gutiérrez-Castrejón, and I. Tomkos, “Filtering ef- A. A. M. Saleh, A. J. Kirby, C. M. Ozveren, B. Schofield,
fects of cascaded flexgrid ROADMs with high spectral resolu- R. E. Thomas, R. A. Barry, D. M. Castagnozzi, V. W. S.
tion filters on the transmission of Nyquist and quasi-Nyquist Chan, B. R. Hemenway, D. Marquis, S. A. Parikh, M. L.
WDM superchannels,” in Proc. 13th Int. Conf. Optical Stevens, E. A. Swanson, S. G. Finn, and R. G. Gallager,
Communication Networks (ICOCN), Suzhou, China, 2014, “A wideband all-optical WDM network,” IEEE J. Sel. Areas
paper M12.6. Commun., vol. 14, no. 5, pp. 780–799, 1996.
Marom et al. VOL. 9, NO. 1/JANUARY 2017/J. OPT. COMMUN. NETW. 25

[70] P. J. Winzer, “Making spatial multiplexing a reality,” Nat. “Few-mode fiberwavelength selective switch with spatial-
Photonics, vol. 8, pp. 345–348, 2014. diversity and reduced-steering angle,” in Optical Fiber
[71] D. Richardson, J. Fini, and L. Nelson, “Space-division multiplex- Communication Conf. (OFC), 2014, paper Th4A.7.
ing in optical fibres,” Nat. Photonics, vol. 7, pp. 354–362, 2013. [85] N. K. Fontaine, T. Haramaty, R. Ryf, H. Chen, L. Miron, L.
[72] A. Muhammad, G. Zervas, D. Simeonidou, and R. Pascar, M. Blau, B. Frenkel, L. Wang, Y. Messaddeq, S.
Forchheimer, “Routing, spectrum and core allocation in flex- LaRochelle, R. J. Essiambre, Y. Jung, Q. Kang, J. K. Sahu,
grid SDM networks with multi-core fibers,” in Int. Conf. S. U. Alam, D. J. Richardson, and D. M. Marom,
Optical Network Design and Modeling (ONDM), 2014, “Heterogeneous space-division multiplexing and joint wave-
pp. 192–197. length switching demonstration,” in Optical Fiber
[73] R. Proietti, L. Liu, R. P. Scott, B. Guan, C. Qin, T. Su, F. Communication Conf. (OFC), 2015, paper Th5C.5.
Giannone, and S. J. B. Yoo, “3D elastic optical networking [86] R. Ryf, N. K. Fontaine, J. Dunayevsky, D. Sinefeld, M. Blau,
in the temporal, spectral, and spatial domains,” IEEE M. Montoliu, S. Randel, C. Liu, B. Ercan, M. Esmaeelpour, S.
Commun. Mag., vol. 53, no. 2 pp. 79–87, 2015. Chandrasekhar, A. H. Gnauck, S. G. Leon-Saval, J. Bland-
[74] D. Siracusa, F. Pederzolli, D. Klonidis, V. Lopez, and E. Hawthorn, J. R. Salazar-Gil, Y. Sun, L. Grüner-Nielsen, R.
Salvadori, “Resource allocation policies in SDM optical net- Lingle, Jr., and D. M. Marom, “Wavelength-selective switch
works (invited paper),” in Int. Conf. Optical Network for few-mode fiber transmission,” in Proc. European Conf.
Design and Modeling (ONDM), 2015, pp. 168–173. Optical Communication (ECOC), 2013, paper PD1.C.4.
[75] D. Klonidis, F. Cugini, O. Gerstel, M. Jinno, V. Lopez, E. [87] Y. Jung, S. Alam, and D. J. Richardson, “Compact few-mode
Palkopoulou, M. Sekiya, D. Siracusa, G. Thouenon, and C. fiber collimator and associated optical components for mode
Betoule, “Spectrally and spatially flexible optical network division multiplexed transmission,” in Optical Fiber
planning and operations,” IEEE Commun. Mag., vol. 53, Communication Conf. (OFC), 2016, paper W2A.40.
no. 2, pp. 69–78, 2015.
[76] B. Huang, N. K. Fontaine, R. Ryf, B. Guan, S. G. Leon-Saval,
R. Shubochkin, Y. Sun, R. Lingle, and G. Li, “All-fiber Dan M. Marom (S’99-M’00-SM’08) received the B.Sc. degree in
mode-group-selective photonic lantern using graded-index mechanical engineering and the M.Sc. degree in electrical engi-
multimode fibers,” Opt. Express, vol. 23, no. 1, pp. 224–234, neering, both from Tel-Aviv University, Israel, in 1989 and
1995, respectively, and was awarded a Ph.D. in electrical engineer-
2015.
ing from the University of California, San Diego (UCSD), in 2000.
[77] N. K. Fontaine, R. Ryf, J. Bland-Hawthorn, and S. G. Leon- From 2000 until 2005, he was a Member of the Technical Staff at
Saval, “Geometric requirements for photonic lanterns in the Advanced Photonics Research Department of Bell Laboratories,
space division multiplexing,” Opt. Express, vol. 20, no. 24, Lucent Technologies, where he invented and headed the research
pp. 27123–27132, 2012. and development effort of MEMS-based wavelength-selective
[78] S. G. Leon-Saval, N. K. Fontaine, J. R. Salazar-Gil, B. Ercan, switching solutions for optical networks. Since 2005, he has been
with the Applied Physics Department, The Hebrew University of
R. Ryf, and J. Bland-Hawthorn, “Mode-selective photonic lan-
Jerusalem, Israel, where he is now an Associate Professor leading
terns for space-division multiplexing,” Opt. Express, vol. 22, the Photonics Devices research group and pursuing his research in-
pp. 1036–1044, 2014. terests in creating photonic devices and sub-systems for switching
[79] A. M. Velazquez-Benitez, J. E. Antonio-Lopez, J. C. Alvarado- and manipulating optical signals in guided-wave and free-space op-
Zacarias, G. Lopez-Galmiche, P. Sillard, D. Van Ras, C. tics solutions using light modulating devices, nonlinear optics, and
Okonkwo, H. Chen, R. Ryf, N. K. Fontaine, and R. compound materials.
Amezcua-Correa, “Scaling the fabrication of higher order pho- Prof. Marom is a Senior Member of the IEEE Photonics Society
tonic lanterns using microstructured preforms,” in Proc. and a Fellow of The Optical Society (OSA). He was awarded the
IEEE Photonics Society Distinguished Lecturer Award for 2014
European Conf. Optical Communication (ECOC), 2015,
and extended to 2015. From 1996 through 2000, he was a
paper Tu.3.3.2. Fannie and John Hertz Foundation Graduate Fellow at UCSD,
[80] N. K. Fontaine, R. Ryf, H. Chen, A. V. Benitez, B. Guan, R. Scott, and was a Peter Brojde Scholar in 2006–2007. He currently serves
B. Ercan, S. J. B. Yoo, L. E. Grüner-Nielsen, Y. Sun, R. Lingle, E. as one of two Senior Editors for Photonics Technology Letters, han-
Antonio-Lopez, and R. Amezcua-Correa, “30 × 30 MIMO trans- dling all photonic-device-related submissions.
mission over 15 spatial modes,” in Optical Fiber
Communication Conf. (OFC), 2015, paper Th5C.1.
[81] Y. Ikuma, K. Suzuki, N. Nemoto, E. Hashimoto, O. Moriwaki, Paul D. Colbourne received the B.Sc. degree in mechanical en-
and T. Takahashi, “Low-loss transponder aggregator using gineering from Mount Allison University, Sackville, Nova Scotia,
spatial and planar optical circuit,” J. Lightwave Technol., Canada in 1983; the B.Eng. degree in engineering physics from
vol. 34, no. 1, pp. 67–72, 2016. the Technical University of Nova Scotia (now Dalhousie
University), Halifax, Nova Scotia, Canada in 1986; and the
[82] K. Suzuki, M. Nakajima, K. Yamaguchi, G. Takashi, Y. M.Eng. degree and Ph.D. degree in engineering physics from
Ikuma, K. Shikama, Y. Ishii, M. Itoh, M. Fukutoku, T. McMaster University, Hamilton, Ontario, Canada in 1988 and
Hashimoto, and Y. Miyamoto, “Wavelength selective switch 1993, respectively
for multi-core fiber based space division multiplexed network From 1993 to the present he has been at Lumentum (formerly
with core-by-core switching capability,” in Int. Conf. Photonics JDSU), designing components for fiber optic systems, including
Switching (PS), Niigata, Japan, July 2016. tunable filters, dispersion compensators, the Swept Wavelength
System, wavelength blockers, and, most recently, wavelength-
[83] J. Carpenter, S. G. Leon-Saval, J. R. Salazar-Gil, J. Bland-
selective switches. He holds more than 40 U.S. patents.
Hawthorn, G. Baxter, L. Stewart, S. Frisken, M. A. F. Dr. Colbourne is a member of The Optical Society (OSA) and
Roelens, B. J. Eggleton, and J. Schröder, “1 × 11 few-mode served on the Technical Committee for OFC 2006–2008.
fiber wavelength selective switch using photonic lanterns,”
Opt. Express, vol. 22, no. 3, pp. 2216–2221, 2014. Antonio D’Errico, received the Ph.D. degree cum laude in telecom-
[84] N. K. Fontaine, R. Ryf, C. Liu, B. Ercan, J. R. Salazar Gil, munication engineering from the Scuola Superiore Sant’Anna, Pisa,
S. G. Leon-Saval, J. Bland-Hawthorn, and D. T. Neilson, Italy, in 2008. Since 2009, he has been with Ericsson Research as a
26 J. OPT. COMMUN. NETW./VOL. 9, NO. 1/JANUARY 2017 Marom et al.

Senior Researcher in photonics and telecommunication systems. His includes high-speed optical transmission systems and all-optical
research interests also include advanced technological solutions for switching technology. He has authored over 40 papers published
optical networks. He is now working on photonic enabling technol- in peer-reviewed journals, international conferences, and patents.
ogies toward 5G Mobile Networks. He is author of more than
80 papers published in international journals, conference digests,
and patents. He is on the board of reviewers for a number of
international journals in the fields of optics and photonics. José Manuel Rivas-Moscoso obtained his Ph.D. in physics from
the University of Santiago de Compostela, Spain, in 2004. From
2004 to 2010, he worked as a Postdoctoral Researcher at the
Cambridge University Engineering Department, Cambridge,
Nicolas K. Fontaine obtained his Ph.D. degree in electrical UK, and the School of Telecommunication of the Technical
engineering at the University of California, Davis, in the Next University of Madrid, Spain.
Generation Network Systems Laboratory [http://sierra.ece. In 2011, he joined the Network Core Evolution Group at
ucdavis.edu] in 2010. In his dissertation, he studied how to gener- Telefónica I  D, Spain, where he worked on beyond-100G trans-
ate and measure the amplitude and phase of broadband optical port technologies. Since 2014, he has been working with the
waveforms in many narrowband spectral slices. Network and Optical Communications Group at AIT, Greece.
Since June 2011, he has been a Member of the Technical Staff at His research interests include elastic optical networks, flexible
Bell Laboratories in Crawford Hill, New Jersey, in the Advanced optical switching, and SDM.
Photonics Division. At Bell Labs, he develops devices for space-
division multiplexing in multi-core and few-mode fibers, builds
wavelength cross-connects and filtering devices, and investigates Ioannis Tomkos (B.Sc., M.Sc., Ph.D.) has been a Research Director
spectral slice coherent receivers for THz bandwidth waveform at the Athens Information Technology Center (AIT) since 2002,
measurement. He serves on the OFC Technical Program wherein he was/is involved in many (over 25) EU-funded research
Committee on fiber devices, the CLEO Technical Program projects with a consortium-wide leading role. He currently serves
Committee on lightwave communications, is an Associate Editor as a Consultant for executives of ICT companies and telecom regu-
for IEEE Photonics Journal, and is General Co-chair for The lators, Business Mentor for ICT start-up company founders, and
Optical Society (OSA) NETWORKS 2016 meeting. Adjunct Professor at the College of Optical Sciences of the
University of Arizona and at the Department of Electrical and
Computer Engineering of University of Cyprus Information. In
the past, he was an Adjunct Professor at Carnegie-Mellon
Yuichiro Ikuma (S’07-M’12) was born in Kanagawa, Japan, in University, (2002–2010), a Senior Scientist at Corning Inc., (1999–
1985. He received B.E., M.E., and Ph.D. degrees in electronics and 2002), and a Research Associate at the University of Athens,
electrical engineering from Keio University, Yokohama, Japan, in Greece (1995–1999).
2007, 2009, and 2012, respectively. From 2009 to 2012, he was a Dr. Tomkos is a widely recognized expert in the optical networking
Research Fellow of the Japan Society for the Promotion of Science. and broadband access networks community. Together with his
Since he joined NTT Photonics Laboratories in 2012, he has been colleagues and students, he has authored over 550 peer-reviewed
involved in the development of optical switches for ROADM systems. archival articles (with over 370 indexed in IEEE Xplore), including
He is currently with NTT Network Service Systems Laboratories, about 150 journal/magazine/book publications and over 400
Musashino, Japan. conference/workshop proceedings papers. For his research work,
Dr. Ikuma is a member of the IEEE Photonics Society and Institute he has received over 6000 citations and many distinctions and
of Electronics, Information and Communication Engineers (IEICE) awards. In 2007, he received the prestigious title of “Distinguished
of Japan. Lecturer” of the IEEE Communications Society, and was elected
“Fellow” of the Institute of Engineering and Technology–IET (2010)
and The Optical Society (OSA) (2012). He has served as the Chair of
Roberto Proietti’s biography was not available at the time of the International Optical Networking Technical Committee of the
publication. IEEE Communications Society (2007–2008), Chairman of the IFIP
working group on “Photonic Networking” (2008–2009), Chairman of
the OSA Technical Group on Optical Communications (2009–2010),
Liangjia Zong was born in Jiangxi, China, in 1984. He received
and Chairman of the IEEE Photonics Society Greek Chapter (2010).
the B.Sc. degree in optical information science and technology,
and the Ph.D. degree in physical electronics, from Huazhong He was also involved in several high-level IEEE and OSA committees
University of Science and Technology, Wuhan, China, in 2006 (most notably the OSA/IEEE Tyndall Award Evaluation/Nomination
and 2011, respectively. Committee) and in editorial boards of leading scientific journals (in-
He then joined Huawei as a Research Engineer in the cluding the Journal of Lightwave Technology and the Journal of
Transmission Technology Research Department. His research Optical Communications and Networking).