Content uploaded by Chuan Qin
Author content
All content in this area was uploaded by Chuan Qin on Aug 08, 2017
Content may be subject to copyright.
IEEE Communications Magazine • February 2015 79
0163-6804/15/$25.00 © 2015 IEEE
Roberto Proietti, Lei Liu,
Ryan P. Scott, Binbin
Guan, Chuan Qin, Tiehui
Su, and S. J. B. Yoo are
with the University of Cal-
ifornia, Davis.
Francesco Giannone is
with Altran Italia S.p.A.
INTRODUCTION
While the rapid deployment of wavelength-divi-
sion multiplexing (WDM) technologies sustained
the explosive and exponential network traffic
growths in the past decade, the continuing trend
of exponential traffic growth driven by data cen-
ters and emerging new services is demanding
deployment of more scalable and flexible net-
working technologies. The legacy WDM tech-
nologies can support traffic up to multiple
terabits per seconds on a single fiber, but this is
not sufficient to support future traffic demands
with peak link capacity beyond 10 Tb/s. It is
expected that commercial systems will need to
support link capacity as high as 100 Tb/s by 2018
[1]. The recent renaissance of coherent optical
communications with polarization-division multi-
plexing has enabled capacity of single fiber links
up to 20 Tb/s. Further increases in capacity and
spectral efficiency are very challenging because
of the nonlinear Shannon limit [2]. The high
optical power required for a high signal-to-noise
ratio (SNR) starts to degrade the transmitted
signal quality due to nonlinear optical effects in
the transmission link [2] even for moderate
(~500 km) transmission distances. Hence, trans-
mission capacities beyond 100 Tb/s must explore
a new and final frontier in optical communica-
tions — the space domain — by using space-
division multiplexing (SDM). SDM allows the
nonlinear Shannon capacity limit over a single-
mode fiber to be overcome, providing a pathway
toward petabit per second link capacities and
spectrum efficiencies beyond 10 b/s/Hz for prac-
tical transmission distances.
Several papers in the literature have already
discussed the advances and technological chal-
lenges in actual deployments of SDM. In addi-
tion to the straightforward bundled fiber
approach, where the SDM system is composed
of Nindependent single-mode fiber (SMF) sys-
tems, an SDM system can exploit few-mode
fibers (FMFs) [3], multi-core fibers (MCF), and
multi-(many)-mode fibers (MMFs) supporting
orbital angular momentum (OAM) [4, 5], or
other eigenstates. Each approach has benefits
and drawbacks related to some particular tech-
nological challenges. FMF systems strongly rely
on multiple-input multiple-output (MIMO) digi-
tal signal processing (DSP) due to strong mode
coupling, but can benefit from the availability of
FMF amplifiers [3]. MCF systems require com-
plex fiber fabrication and the use of Nseparated
amplification stages, but compared to FMF, the
crosstalk and coupling between cores can be
carefully controlled to levels that do not require
or strongly limit the use of MIMO DSP. OAM
states have relatively well defined azimuthal
orthogonality, leading to possibly simpler MIMO
DSP for propagation through ring-core FMFs or
MMFs compared to other multimode propaga-
tion methods.
Figure 1 shows some representations of the
three physical domains used to increase the
capacity in fiber optic communication systems.
In addition to the capacity increase, future net-
works must achieve flexible and agile utilization
of network resources with scalable network con-
trol and management (NC&M). Recently,
researchers proposed a new optical networking
ABSTRACT
Conventional elastic optical networking,
EON, uses elasticity in two domains, time and
frequency, to optimize utilization of optical net-
work resources in the presence of fluctuating
traffic demand and link quality. Currently, net-
working exploiting a third domain, space, is the
focus of significant research efforts since space-
division multiplexing, SDM, has the potential to
substantially improve future network capacity
and spectral efficiency. This article extends 2D-
EON to include elasticity in all three domains:
time, frequency, and space. We introduce
enabling technologies, architectures, and algo-
rithms for 3D-EONs. Based on sample network
topologies, we investigate algorithms for rout-
ing, spectrum, spatial mode, and modulation
format assignment — RSSMA. In particular, we
investigate fragmentation-aware RSSMA and
how the constraints in the formation of super-
channels in MIMO-based SDM systems can
impact the network performance in terms of
blocking probability.
SPATIALLY AND SPECTRALLY FLEXIBLE
ELASTIC OPTICAL NETWORKING
Roberto Proietti, Lei Liu, Ryan P. Scott, Binbin Guan, Chuan Qin, Tiehui Su, Francesco Giannone,
and S. J. B. Yoo
3D Elastic Optical Networking in the
Temporal, Spectral, and Spatial Domains
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 79
IEEE Communications Magazine • February 2015
80
technology called elastic optical networking
(EON) [6], in which each data link uses flexible
spectral bandwidths and variable modulation for-
mats. The flexibility in spectrum and bandwidth
allocation together with variable modulation for-
mats allow more optimized utilization of net-
work resources according to the given traffic
demand and the link conditions. Network
resources optimization spans the temporal, spec-
tral, and spatial domains, and efficient and effec-
tive algorithms need to be part of NC&M to
allocate the available resources optimally. For
instance, dynamic adaptation to varying traffic
demand can lead to stranded and fragmented
spectral resources in EON where there is no
fixed spectral grid. Recent studies investigated
spectral and temporal domain solutions for frag-
mentation-aware routing, spectrum, and modula-
tion format assignment (RSMA) [7] in
2D-EONs. New studies in the temporal, spectral,
and spatial domains for 3D-EONs are necessary.
At the physical layer, 2D-EON can employ
coherent optical orthogonal frequency-division
multiplexing (CO-OFDM), coherent optical
WDM (CO-WDM), Nyquist WDM [8], or
dynamic optical arbitrary waveform generation
(OAWG) and optical arbitrary waveform mea-
surement (OAWM) technologies [9], adopting
various modulation formats across the flexible
bandwidth depending on the reach. The 2D-
EONs have achieved ∼1.6×improvements com-
pared to the fixed-grid networks in capacity,
utilization, and availability in the network. To
achieve more than 10×improvements we must
introduce a new degree of freedom in the spatial
domain using SDM. The elasticity concepts dis-
cussed above can then also be extended to the
spatial domain. In [10] the authors proposed
spatial-spectral processing and coded modula-
tion to exploit elasticity in the spectral and spa-
tial domains. Reference [11] reports the first
demonstration of an elastic SDM system based
on MCFs capable of switching the traffic in the
spatial, spectral, and time domains. Reference
[12] discusses a new routing, spectrum and core
assignment (RSCA) for SDM systems based on
multicore fibers, taking into account the crosstalk
between cores. In 3D-EONs, the reliance on
electronic DSP (with or without MIMO) can
limit the total capacity.
This article discusses EON in the temporal,
spectral, and spatial domains (3D-EON) for
future optical fiber communication networks. In
the following sections, this article:
• Introduces the concept of elasticity in the
time, spectral, and spatial domains
• Presents enabling technologies for SDM
systems
• Discusses the routing, spectral, spatial
mode, and modulation format assignment
(RSSMA) algorithm, which extends RSMA
by exploiting the additional space domain
for resource allocation
ELASTICITY IN THE TEMPORAL,
SPECTRAL, AND SPATIA L DOMAINS
Figure 2 illustrates the concept of elasticity in
time, frequency, and space for 3D-EON. Let us
assume that we have an SDM system with four
spatial modes and that each node in the network
has a certain number of transceivers which can
tune in wavelength, line rate, and modulation
format. If the network needs to provision a 1
Tb/s connection between two nodes (e.g., from
Figure 1. Three physical domains (frequency, time-modulation format, and space) to support elastic
optical networking in the spectral, temporal, and spatial domains (3D-EON).
t
12.5 GBd
Time
Space
Frequency
Few-mode or multi-mode fiber
Orbital
angular
momentum
+z
l = 0
l = +1
l = 0
l = -1
-
t
Real part
16 QAM
Imag part
Symbol count
316
l = -1
l = 2
0
Multi-core fiber
2π
λ
λ
l = 1
λ
λ1λ2λ3λ4λ5λ6λ7λ8
25 GBd
100
32
10
3
1
Real part
QPSK
Imag part
0.0
0.3
0.6
0.9
1.2
1.5
1.8
2.1
2.4
2.7
The flexibility in
spectrum and
bandwidth allocation
together with
variable modulation
formats allow more
optimized utilization
of network resources
according to the
given traffic demand
and the link
conditions.
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 80
IEEE Communications Magazine • February 2015 81
node A to node D in Fig. 4a), the flexibility pro-
vided by the three domains (temporal, spectral,
and spatial) may allow this 1 Tb/s connection to
be established in different ways. At the begin-
ning, the network control plane could decide to
use the solution in Fig. 2 (left) by creating a
superchannel that spans across all four spatial
modes occupying five 12.5 GHz spectral slots on
each mode. Each spectral slot is filled in this case
with a 12.5 Gbaud, 16-quadrature amplitude
modulated (QAM) signal. If the network condi-
tion changes, for instance, the optical SNR
(OSNR) at the receivers decreases, the network
may be able to reconfigure the connection as
shown in Fig. 2 (center) by using only two spatial
modes, reducing the complexity of the modula-
tion format to quadrature phase shift keying
(QPSK) and occupying 20 spectral slots on each
mode. This reconfiguration exploits the elasticity
in all three domains (S+F+T). Figure 2 (right)
shows another example of elasticity in space, fre-
quency, and time. In this case the OSNR at the
receiver is high enough to provision the 1 Tb/s
connection by using one single spatial mode and
10 25-GHz spectral slots filled with 25 Gbaud 16-
QAM signals. In this case, the flexibility in the
three domains allows saving 50 percent of the
network resources in terms of number of
transceivers used.
While the scenario depicted in Fig. 2 is ideal,
an actual implementation of elasticity in the
time, frequency, and space domains could be
constrained by:
• Crosstalk among different spatial modes
• The need for MIMO processing to combat
crosstalk impairments and the associated
constraints on the formation of superchan-
nels
• MIMO DSP complexity
• Intermodal nonlinear impairments
• The SDM node architecture and its switch
and add/drop traffic granularity (trade-off
between flexibility and complexity)
• The number of TRXs available in the
transponders
All of the above aspects are active research
topics in spectral and spatial EON.
ENABLING TECHNOLOGIES FOR
SPAC E-DIVISION MULTIPLEXING
Although practical SDM technologies are still
under active development, we must understand
the current status of SDM technologies when
considering 3D-EON. Various technological
approaches to SDM fibers include MCFs, FMFs,
and MMFs. An MCF contains multiple cores
and operates in one of the two regimes, inde-
pendent cores with sufficiently large spacing to
limit the crosstalk or coupled cores to support
super modes. An FMF has a single core with a
dimension that restricts the number of modes
(e.g., less than a dozen distinct modes). An
MMF includes ring-core or “vortex” fibers that
support OAM modes with limited crosstalk. At
the research level, a few experiments have
already demonstrated long-distance (> 1000 km)
transmission of multiple spatial modes using
MCFs or FMFs (using either commercial single-
mode Erbium-doped fiber amplifiers or FMF
amplifiers [3]), while demonstrations of OAM-
based SDM fiber transmission systems are limit-
ed to a few kilometers so far. However, the
deployment of SDM systems can be conceivable
only if they provide benefit in terms of cost and
performance compared to their counterpart
using multiple stranded fibers. To this aim, as
discussed in [1, 3], optical amplification and inte-
gration are essential for practical implementa-
tion of SDM in actual systems.
A compact and integrated MUX/DEMUX
represents an important building block toward
the realization of compact and energy-efficient
transponders and reconfigurable optical
add/drop multiplexers. In the case of FMF, pho-
Figure 2. Three different possibilities to provision a 1 Tb/s connection exploiting flexibility in space, frequency, and time (S+F+T).
4 SDM modes 2 SDM modes 1 SDM mode
S+F+T
OSNR
decreases
Space domain
Time domain
Frequency domain
Real part
Imag part
Symbol count
316
λ1λ2λ3λ4λ20
100
32
10
3
1
Real part
S1
S0
12.5 GBd QPSK
(25 Gb/s)
25 GBd 16QAM
(100 Gb/s)
......
λ1λ2λ10
......
......
λ1
S1
S0
12.5 GBd 16QAM
(50 Gb/s)
λ2λ3λ4λ5
S2
S3
Imag part
0.0
0.3
0.6
0.9
1.2
1.5
1.8
2.1
2.4
2.7
S+F+T
OSNR
increases
Real part
Imag part
Symbol count
316
100
32
10
3
1
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 81
IEEE Communications Magazine • February 2015
82
tonic lanterns implemented from all-fiber or 3D-
waveguide technologies have shown promise as
mode-group-selective spatial multiplexers [13].
Alternatively, OAM multiplexers based on
hybrid 3D photonic integrated circuits (PICs),
when coupled with vortex fibers, may also pro-
vide limited crosstalk data paths through a single
fiber. As an example, we provide some details of
an OAM MUX/DEMUX device fabricated at
the University of California, Davis (UC Davis).
As Figure 3a illustrates, a light beam carrying
OAM exhibits a specific azimuthal phase varia-
tion determined by OAM state number, l, which
is an integer. OAM states with differing charge
numbers are orthogonal, and the sign of ldeter-
mines the handedness of the helical phase front.
The light beam can simultaneously support mul-
tiple OAM states (potentially an infinite num-
ber, subject to the Shannon limitation imposed
by the SNR). Therefore, OAM has the potential
to significantly improve the spectral efficiency or
photon efficiency of free-space and fiber optical
communications. Recent experimental results [5]
show successful terabit-per-second transmission
by multiple OAM modes propagating over a spe-
cially designed fiber without using computation-
ally intensive DSP MIMO algorithms.
Figures 3b and 3c show a device fabricated at
UC Davis performing OAM multiplexing and
demultiplexing based on PICs containing a free
propagation region (FPR). The FPR induces a
linear phase tilt on the input signals based on
their input waveguide position. After geometri-
cal transformation by the phase-matched wave -
guides, this phase tilt provides progressive phase
evolution in the azimuthal angle as required for
the OAM state generation (Figure 3b). The
superposition principle of waves indicates that
multiplexing of multiple OAM states is possible,
while the propagation of the waves in reverse
will achieve OAM state demultiplexing. The
device supports multiplexing and demultiplexing
of up to 15 OAM states, with both TE and TM
polarizations and relatively low loss performance
at 1.55 µm and crosstalk better than –15 dB. The
compact design and single-mode interface can
guarantee easy connection with other high-speed
optical components for future on-chip integra-
tion. The device was fabricated using 3D direct
laser inscribing technology [4] that has also been
used to fabricate photonic lantern mode-group
selective multiplexers for FMF [13].
RSSMA IN THE TEMPORAL,
SPECTRAL, AND SPATIA L DOMAINS
Although the details of particular SDM imple-
mentations are an active research topic, our dis-
cussion below on routing, spectrum, spatial
mode, and modulation format assignment
(RSSMA) is relatively agnostic to particular
SDM techniques, so the implications of the work
are general.
In a 3D-EON, an RSSMA algorithm is neces-
sary in order to route a connection request and
Figure 3. a) 3D illustration of OAM mode propagation for state numbers l= 0, +1, –1; b) integrated device conceptual configura-
tion for an OAM multiplexer/demultiplexer. The OAM multiplexer converts the linear phase front tilt to azimuthal varying
phase. Input OAM states at l= +2, …, –2 are each projected to an array of phase-matched waveguides to excite apertures with
azimuthally varying phase dependent on the OAM state number l. This way, multiple OAM states can be multiplexed spatially
(SDM); c) 3D illustration of the hybrid integrated device.
l = 0
(a)
(b) (c)
Inputs
Single-mode
apertures
Free-propagation
region
(FPR)
Azimuth
phase
2πl
2π
Phase-matched
waveguides
Linear phase
tilt
l = +2
l = +1
l = 0
l = -1
l = -2
0
-1
-2
1
0
1
2
Phase
ϕ
2
3
-1
0
1
2
-2
l = +1
0
-1
-2
1
0
1
2
2
3
-1
0
1
2
-2
l = +2
0
-1
-2
1
0
1
2
2
3
-1
0
1
2
-2
0
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 82
IEEE Communications Magazine • February 2015 83
assign resources for this request in the network.
Figure 4a shows a six-node 3D-EON network
with a centralized control plane. The control
plane runs RSSMA algorithms to maintain effi-
cient network level utilization. Each node can
have an impairment monitoring system that
transmits the information on the status of the
links to a centralized controller, which will use
this information for the RSSMA algorithm.
Figure 4b shows a system-level schematic of
the N-mode SDM links used in the 3D-EON
network of Figure 4a (i.e., the node A–node E
link). Mmulti-wavelength transmitters, imple-
menting elasticity in the spectral-temporal
domains, are spatially multiplexed. In addition to
the required DSP to implement CO-OFDM,
Nyquist WDM, or OAWG, it might also be nec-
essary to use MIMO DSP in the case of mode
mixing during transmission through the optical
link. An architecture on demand (AoD) pro-
grammable node [11] is placed between the
SDM MUX and the MTXs to guarantee recon-
figurability and switching granularity at the
wavelength level, and fully exploit the spectral
and spatial domain. The AoD consists of a large
port-count optical backplane (e.g., micro-electro-
Figure 4. a) Example of a six-node 3D-EON network with centralized control plane and performance monitoring (PM); b) node
A–node Elink system diagram; c) Fragmentation of D-E link status: (left) only spatial superchannels are used, leading to spec-
tral fragmentation only; (right) both spatial and spectral superchannel are used, leading to spectral and spatial fragmentation.
WDM TX/RX: multi-wavelength transmitters/receivers, implementing elasticity in the spectral-/temporal domains; AoD: archi-
tecture on demand; WSS: wavelength selective switch; EDFA: Erbium-doped fiber amplifier; MUX/DeMUX:
multiplexer/demultiplexer.
Spatial and spectral
SuperChannel
Spatial superchannel
ONLY
Fragmented D-E link
Wavelength (λ)
b
da a a
SDM
fiber
SDM
fiber
SDM
fiber
to Node D
Node B
DropAdd
Node E
(c)
AoD
SDM
MUX
WSS EDFA
EDFA
WSS
Centralized controller
(a)
bdc
bdc c
dca
dca
dc
b
Space
b
ba
PM
F
E
DB
C
A
PM
PM
PM
PM
PM
PM
AoD
(b)
Node A
SDM
MUX
WSS EDFA
EDFA
WSS
SDM
DeMUX
From Node D
WDM
TX1
WDM
TX2
WDM
TXM
WDM
RXM
WDM
RX2
WDM
RXM
DSP (with MIMO)
DSP (with MIMO)
SDM
DeMUX
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 83
IEEE Communications Magazine • February 2015
84
mechanical systems, MEMS), and several plug-in
modules such as flexible-grid wavelength selec-
tive switches (WSSs) and optical amplifiers.
After the SDM MUX, the SDM signal is
launched through an SDM-compatible fiber. At
node B, the different spatial modes are demulti-
plexed. A small amount of optical power is
tapped from each of the SDM DEMUX out-
puts to feed the impairment monitor system,
which provides information on the link quality
to the centralized controller. The NSDM
DEMUX outputs then connect to an AoD node
that serves as a reconfigurable optical add-drop
multiplexer (ROADM), allowing adding, drop-
ping and switching of different flexible channels
with granularity down to the wavelength level.
NAoD outputs connect to another SDM MUX
to launch the SDM signal into the next fiber
span and reach the next nodes all the way to
the final node, E. At node E, an SDM DEMUX
spatially demultiplexes the different spatial
mode. Each mode is then spectrally demulti-
plexed and fed into a bank of coherent receivers
with DSP.
THE RSSMA ALGORITHM AND
SIMULATION RESULTS
The amount of crosstalk among different spatial
modes and the need for MIMO processing will
not only impact the physical layer performance,
but also the way superchannels are formed and
routed. It has previously been pointed out [1, 3]
that in the case of an SDM system relying heavi-
ly on MIMO, it is necessary to create a super-
channel in the space domain with an end-to-end
signal that occupies all of the spatial modes for a
given wavelength slot (e.g., connections a, b, c,
and din Figure 4c on the left). It is evident that
this constraint on the creation of spatial super-
channels will limit the flexibility to switch, add,
and drop single SDM channels. However, at the
same time, it will simplify the network control
since fragmentation issues are confined to only
the wavelength domain (Fig. 4, left). Conversely,
in the case of an SDM system not relying on
MIMO (this is possible for certain MCF designs
and potentially for OAM systems), the creation
of super-channels and resource utilization would
Figure 5. a) Spectral fragmentation on the spectral-spatial-temporal dimensions; b) blocking probability comparison of different
algorithms in the 14-node NSFNET network; c) blocking probability comparison of different algorithms in the 28-node USBN
network.
Load (Erlangs)
(b)
500
0.0
Blocking probability
0.1
0.2
0.3
0.4
0.5
0.6
0 1000 1500 2000 2500 3000 3500
Load (Erlangs)
(c)
200
0.0
Blocking probability
0.1
0.2
0.3
0.4
0.5
400 600 800 1000 1200 1400 1600 1800
The fragmented spectrum
resources in link AB for mode m
E
DBDBCED
(a)
Frequency Space The spectrum misalignment between
neighboring links
......
Mode 0
Mode 1
Mode m
SP-FF-with-MIMO
SP-FF-without-MIMO
FA-RSSMA
FA-RSSMA-CA
SP-FF-with-MIMO
SP-FF-without-MIMO
FA-RSSMA
FA-RSSMA-CA
BA
EDBDBCEDBA
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 84
IEEE Communications Magazine • February 2015 85
be more flexible. In this case, however, fragmen-
tation in both the space and wavelength domains
might occur (Fig. 4c, right).
In conventional 2D-EONs, routing, spectral,
and modulation format assignment (RSMA) [7]
along with defragmentation in the spectral
domain [14] have been widely investigated. Con-
sidering that defragmentation operations are
complex, we investigated fragmentation-aware
and alignment-aware RSMA algorithms to mini-
mize the need for defragmentation. For 3D
EONs, we investigate routing, spectral, spatial
mode, and modulation format assignment
(RSSMA) with similar fragmentation awareness
and alignment awareness. Using an example net-
work topology of Fig. 4a, the dashed rectangle in
Fig. 5a indicates the fragmented spectrum
resources in link AB, which is the fragmentation
in the spectral dimension. In addition, consider-
ing that spectrum misalignment between neigh-
boring links (dashed rectangle in Fig. 5a) will
most likely increase the end-to-end blocking
probability, we propose two RSSMA algorithms.
The first is fragmentation-aware RSSMA (FA-
RSSMA), and the other is FA-RSSMA with con-
gestion avoidance (FA-RSSMA-CA). The
operational procedures for both algorithms are
different, but their objective is the same: any
new connection is set up to fragment the least
number of continuous spectral blocks on candi-
date links, while it fills up as many misaligned
spectral slots as possible on neighboring links for
a given spatial mode.
We have conducted simulations on sample
network topologies to compare the blocking
probability of the proposed fragmentation-
aware RSSMA algorithms with the commonly
used benchmark algorithm, that is, the shortest
path routing and first-fit spectrum (and spatial
mode) assignment (SP-FF) algorithm. For com-
parison purposes, we measured the performance
of SP-FF for the 3D EON deployed with an
SDM system relying on MIMO and without
MIMO (i.e., SP-FF-with-MIMO and SP-FF-
without-MIMO, respectively). We simulated
dynamic connection arrival and departure events
on a 14-node NSFNET network and a 24-node
U.S. backbone network (USBN), respectively.
Each fiber link has 400 spectral slots, and each
slot is set to be 12.5 GHz. In the simulation,
connection requests are randomly generated
between each source-destination pair according
to a Poisson process. The holding time of each
connection follows a negative exponential distri-
bution with an average of five time units. The
connection bandwidth is randomly selected in
the range of [1, 10] slots. The number of spatial
modes is 8. Figures 5b and 5c compare the
blocking probability of the different algorithms.
As expected, the use of MIMO has a significant
impact on the network blocking performance,
leading to the worst blocking probability caused
by an inflexible resource allocation due to the
constraint mentioned above on the creation of
spatial super-channels. The results in Fig. 5
(b–c) also show that both the FA-RSSMA and
FA-RSSMA-CA algorithms can greatly reduce
the blocking probability, compared to that of
the non-fragmentation-aware SP-FF algorithm.
The FA-RSSMA-CA outperforms FA-RSSMA,
but the difference is relatively small.
SUPERVISORY CHANNEL AND
PERFORMANCE MONITORING FOR 3D EON
In addition to the impairments already present
in a 2D-EON (optical noise accumulation, four-
wave mixing, self-phase modulation, etc.), the
spatial domain may introduce a significant
amount of in-band crosstalk due to the mode
coupling in the transmission media and crosstalk
in the spatial MUX/DEMUX elements distribut-
ed along the fiber links.
As we already experimentally demonstrated
Figure 6. a) Simulated correlation between the bit error rate of the data and supervisory signals as a function of the extinction
ratio (ER) parameter of the intensity modulator used to overmodulate the data signal (data signal is QPSK-modulated at 20
Gb/s, top, and 40 Gb/s, bottom); b) simulation results show strong correlation between the bit error rate of data and monitor sig-
nals as a function of the signal-to-crosstalk ratio for a 10 Gbaud QPSK signal with a 20 Mb/s monitor signal. The results suggest
that the supervisory channel technique can do performance monitoring for inband crosstalk in SDM systems.
Signal to crosstalk ratio [dB]
65
-10
-12
LOG10(BER)
-8
-6
-4
-2
0
7 8 9 10 11 12 13 14 15 16 17 18
Supervisory signal @ 40Mb/s
(a) (b)
8
-2
0
-12
-4
-6
-8
-10
6 10 12
E.R.=0.4
14 16 18
Supervisory signal @ 80Mb/s
OSNR (dB)
0
-15
LOG10(BER)
-5
-10
1210 14 16 18 20 22 24
Monitor
signal
Line signal
E.R.=0.8
E.R.=1.2
E.R.=1.6
E.R.=0.8
E.R.=0.4
E.R.=1.2
E.R.=1.6
Line signal@20Gb/s
Line signal@40Gb/s
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 85
IEEE Communications Magazine • February 2015
86
in [15], a real-time performance monitoring
method across the spread spectrum allows the
network to dynamically and adaptively adjust
the modulation format to maximize spectral
efficiency while maintaining the required qual-
ity of transmission (QoT) and bit error rate
(BER) performance even for signals that expe-
rience time-varying physical layer impairments.
The monitoring technique consists of a low-
speed supervisory channel modulating the
high-speed data with a low modulation index
(e.g., ~0.1). Since the data and the supervisory
channel signals follow that same path, the
BER of the data is strongly correlated with
that of the supervisory channel, as shown in
Fig. 6a. This correlation allows estimation of
the BER of the data at different modulation
formats, and the network control plane can
choose the modulation format that maximizes
the spectral bandwidth while meeting the
transmission requirements. The same spread-
spectrum supervisory channel can also be
applied to each spatial mode in a 3D-EON
network to monitor the QoT of each single
spatial mode.
The simulation results in Fig. 6 show how the
supervisory channel technique described above
can do performance monitoring in case of inband
crosstalk, which is a very critical impairment in
SDM systems due to mode coupling in multi-
modeMMFs or MCFs. Figure 6a shows the sim-
ulated correlation between the BER of the data
and supervisory signals, as a function of the
extinction ratio (ER) parameter of the intensity
modulator used to overmodulate the data signal.
The results are obtained for a data signal with
QPSK modulation format at 20 Gb/s (top) and
40 Gb/s (bottom). As expected, the ratio between
the line rate of the data and supervisory signals
remains constant (40 Mb/s for 20 Gb/s and 80
Mb/s for 40 Gb/s). More importantly, the simu-
lation results in Fig. 6b show the simulated cor-
relation between the BER of the data and
supervisory signals as a function of the signal to
inband crosstalk ratio in an SDM system for a 10
Gbaud QPSK signal with a 20 Mb/s supervisory
channel. The results show strong correlation
between the BERs of the data signal and super-
visory channel.
CONCLUSION
We discuss temporal, spectral, and spatial elastic
optical networking, and related physical layer
technologies and networking aspects. We intro-
duce the concept of elasticity in the time, fre-
quency, and space domains, and discuss recent
technological advances toward integrated pho-
tonic SDM multiplexers and demultiplexers. We
further discuss RSSMA algorithms, which take
advantage of the spatial dimension to reduce the
blocking probability. The study considers the
possible constraints on the allocation of spectral
and spatial slots given the use of MIMO process-
ing, which may be needed to combat in-band
crosstalk in SDM systems. Finally, we discuss the
application of a supervisory channel perfor-
mance monitoring technique in the new 3D-
EON to monitor in-band crosstalk resulting from
mode coupling in SDM systems.
ACKNOWLEDGMENTS
This work was supported in part by DOE under
grant DE-FC02-13ER26154, by NSF under
EECS grant 1028729, and by DARPA DSO
under grants HR0011-11-1-0005 and W911NF-
12-1-0311.
REFERENCES
[1] P. J. Winzer, “Making Spatial Multiplexing A Reality,”
Nature Photonics, vol. 8, 2014, pp. 345–48.
[2] R. Essiambre et al., “Capacity Limits of Optical Fiber Net-
works,” J. Lightwave Tech., vol. 28, 2010, pp. 662–701.
[3] D. Richardson, J. Fini, and L. Nelson, “Space-Division
Multiplexing in Optical Fibres,” Nature Photonics, vol.
7, 2013, pp. 354–62.
[4] B. Guan et al., “Free-Space Coherent Optical Communica-
tion with Orbital Angular, Momentum Multiplexing/Demul-
tiplexing Using a Hybrid 3D Photonic Integrated Circuit,”
Optics Express, vol. 22, 2014/01/13 2014, pp. 145–56.
[5] N. Bozinovic et al., “Terabit-Scale Orbital Angular
Momentum Mode Division Multiplexing in Fibers,” Sci-
ence, vol. 340, June 28, 2013 2013, pp. 1545–48.
[6] M. Jinno et al., “Spectrum-Efficient and Scalable Elastic
Optical Path Network: Architecture, Benefits, and
Enabling Technologies,” IEEE Commun. Mag., vol. 47,
2009, pp. 66–73.
[7] Y. Yin et al., “Spectral and Spatial 2D Fragmentation-Aware
Routing and Spectrum Assignment Algorithms in Elastic
Optical Networks [Invited],” J. Optical Commun. and Net-
working, vol. 5, 2013/10/01 2013, pp. A100–A106.
[8] G. Bosco et al., “On the Performance of Nyquist-WDM
Terabit Superchannels based on PM-BPSK, PM-QPSK,
PM-8QAM or PM-16QAM Subcarriers,” J. Lightwave
Technology, vol. 29, 2011, pp. 53–61.
[9] R. P. Scott et al., “Dynamic Optical Arbitrary Waveform
Generation and Measurement,” Optics Express, vol. 18,
Aug. 30 2010, pp. 18,655–70.
[10] M. Cvijetic, I. B. Djordjevic, and N. Cvijetic, “Dynamic
Multidimensional Optical Networking based on Spatial
and Spectral Processing,” Optics Express, vol. 20, Apr.
2012, pp. 9144–50.
[11] N. Amaya et al., “First Fully-Elastic Multi-Granular Net-
work with Space/Frequency/Time Switching Using
Multi-core Fibres and Programmable Optical Nodes,”
ECOC, Amsterdam, The Netherlands, 2012, Th.3.D.3.
[12] A. Muhammad et al., “Routing, Spectrum and Core
Allocation in Flexgrid SDM Networks With Multi-Core
Fibers,” 2014 Int’l. Conf. Optical Network Design and
Modeling, 2014, pp. 192–97.
[13] R. Ryf et al., “1705-km Transmission over Coupled-
Core Fibre Supporting 6 Spatial Modes,” ECOC, 2014,
pp. 1–3.
[14] R. Proietti et al., “Rapid and Complete Hitless Defrag-
mentation Method Using a Coherent RX LO with Fast
Wavelength Tracking in Elastic Optical Networks,”
Optics Express, vol. 20, Nov. 2012, pp. 26,958–68.
[15] D. J. Geisler et al., “Experimental Demonstration of
Flexible Bandwidth Networking with Real-Time Impair-
ment Awareness,” Optics Express, vol. 19, Dec. 2011,
pp. B736–B745.
BIOGRAPHIES
ROBERTO PROIETTI (rproietti@ucdavis.edu) received his M.S.
degree in telecommunications engineering from the Uni-
versity of Pisa, Italy, in 2004, and his Ph.D. in fiber optical
communications from Scuola Superiore Sant’Anna, Pisa,
Italy, in 2009. He is a project scientist at UC Davis. His
research interests include high-spectrum-efficiency coherent
transmission systems and elastic optical networking, access
optical networks and radio over fiber, and optical switch-
ing technologies and architectures for supercomputing and
data center networks.
LEI LIU [S’08, M’11] (liulei@ieee.org) received B.E. and Ph.D.
degrees from Beijing University of Posts and Telecommuni-
cations, China, in 2004 and 2009, respectively. From 2009
to 2012, he was a research engineer at KDDI R&D Labora-
tories Inc., Japan, where he was engaged in research and
development on control planes for optical networks. He is
now with UC Davis. He has coauthored more than 100
peer-reviewed papers in prestigious international journals
and conferences.
The operational pro-
cedures for both
algorithms are differ-
ent, but their objec-
tive is the same: any
new connection is
setup to fragment
the least number of
continuous spectral
blocks on candidate
links, while it fills up
as many misaligned
spectral slots as pos-
sible on neighboring
links, for a given
spatial mode.
PROIETTI_LAYOUT.qxp_Author Layout 1/30/15 1:49 PM Page 86
IEEE Communications Magazine • February 2015 87
RYAN P. SCOTT [M’03] (rpscott@ucdavis.edu) received his
Ph.D. degree in electrical and computer engineering from
UC Davis in 2009. He is currently an assistant project scien-
tist at UC Davis, and his present research interests include
optical arbitrary waveform generation and measurement,
optical comb generation, spatial division multiplexing, and
direct laser inscription of 3D optical waveguides.
BINBIN GUAN (rayguan@ucdavis.edu) received his B.S.
degree in optics engineering from Zhejiang University,
China, in 2008. He is currently working toward his Ph.D.
degree in electrical and computer engineering at UC Davis.
His current graduate research focuses on optical coherent
communications and optical signal processing.
CHUAN QIN (chaqin@ucdavis.edu) received his B.S. degree in
optical sciences and engineering from Zhejiang University,
Hangzhou, China, in 2011. He is currently working toward
his Ph.D. degree in electrical and computer engineering
from UC Davis. His research interests include high-speed
optical arbitrary waveform generation and measurement,
spatial division multiplexing, and coherent optical commu-
nications.
TIEHUI SU(tiesu@ucdavis.edu) received his B.S. degree in
electric engineering from Peking University, Beijing, China,
in 2008. He is currently working toward his Ph.D. degree in
electrical engineering at UC Davis. His research interests
include photonic integrated circuits design, orbital angular
momentum (OAM), and AWG devices design on various
material platforms.
FRANCESCO GIANNONE (frcgnn@gmail.com) received his M.S.
degree in telecommunications engineering from the Univer-
sity of Pisa in 2014. He was a visiting research scholar at UC
Davis from September 2013 to June 2014. He is currently
working as a consultant engineer at Altran Italia S.p.A.
S. J. BEN YOO [S’82, M’84, SM’97, F’07] received his B.S.
degree in electrical engineering with distinction, his M.S.
degree in electrical engineering, and his Ph.D. degree in
electrical engineering with a minor in physics, all from
Stanford University, California, in 1984, 1986, and 1991,
respectively. He currently serves as a professor of electrical
engineering at UC Davis. His research at UC Davis includes
optical switching devices, systems, and networking tech-
nologies for future computing and communications.
CALL FOR PAPERS
IEEE WIRELESS COMMUNICATIONS MAGAZINE
SOFTWARE DEFINED RADIO: 20 YEARS LATER
BACKGROUND
It has been two decades since the publication of the seminal tutorial paper "The Software Radio Architecture" in IEEE
Communications Magazine. The evolution of SDR systems had a profound impact in both military and multi-standard
commercial space. The intent of the Feature Topic is to capture key elements of the evolution of the enabling technolo-
gies as well as SDR solutions for flexible and reconfigurable wireless platforms. Topics should include, but are not limit-
ed to:
•Advanced radio frequency (RF) transceiver technology and flexible RF
•Analog-to digital converters and new converter architectures
•Processing technology: DSP, ASIP, FPGA, multi-core systems
•VLSI design methodology for reconfigurable platforms and HW/SW co-design
•Software architectures for SDR platforms
•SDR and spectrum sharing
•Regulatory issues including Dynamic Certification for SDR
•Cognitive radio: architectures, deployments, shared spectrum
•Commercial SDR platforms: base-stations, multi-standard terminals, cognitive radio
•Software-defined wireless networks and cross-layer design
SUBMISSIONS
Articles should be tutorial in nature and should be written in a style comprehensible to readers outside the specialty of
the field. All submissions will be reviewed based on technical merit, relevance and readability. Authors must follow the
IEEE Communications Magazine guidelines for preparation of the manuscript. Complete guidelines for prospective
authors can be found at http://www.comsoc.org/commag/paper-submission-guidelines. All articles to be considered for
publication must be submitted through IEEE Manuscript Central (http://mc.manuscriptcentral.com/commag-ieee). Sub-
mit manuscripts to the “September 2015/Software Defined Radio” category.
SCHEDULE FOR SUBMISSIONS
•Manuscript Submission Deadline: March 1, 2015
•Decision Notification: May 1, 2015
•Final Manuscript Deadline: July 1, 2015
•Publication Date: September 2015
GUEST EDITORS
Joe Mitola Preston Marshall Kwang-Cheng Che
Federated Wireless Google National Taiwan University
joseph.mitola@federatedwireless.com pres@google.com ckc@ntu.edu.tw
Markus Mueck Zoran Zvonar
Intel MediaTek
markus.dominik.mueck@intel.com zoran.zvonar@mediatek.com
PROIETTI_LAYOUT.qxp_Author Layout 2/2/15 10:56 AM Page 87
