1. ITU-T Kaleidoscope 2010 Beyond the Internet
ITU-T Kaleidoscope 2010 Beyond the Internet? - Innovations for future networks and services
Introducing Elasticity and Adaptation into the Optical Domain Toward More Efficient and Scalable Optical Transport Networks
M. Jinno, T. Ohara, Y. Sone, A. Hirano
O. Ishida, and M. Tomizawa
NTT Network Innovation Labs.
Pune, India, 13 – 15 December 2010

2. Outline Background: Growing anticipation
SE-conscious optical networking
Early initiatives by ITU-T
Elastic optical path network as a candidate to support future Internet and services
Adoption scenarios from rigid optical networks to elastic optical path network
Possible standardization study items and some solutions relevant to future ITU-T activities

3. Background (1): Successful Deployment of Optical Networks
Worldwide intensive R&D activities
Continuous initiative by ITU-T toward OTNs and ASONs
G.709 OTN augmentation to transport 100 GE traffic
100 M
1 G
10 G
100 G
1 T
10 T
100 T
1980
1990
2000
2010
2020
0.01
0.1 1 10
Year of commercialization in Japan
Per fiber capacity (b/s)
Spectral efficiency (b/s/Hz)
100 Gb/s x 80
(projected)
40 Gb/s x 40
10 Gb/s x 80
WDM
TDM
Optical networks have become widely spread and have assumed a role as mission critical infrastructures in our information society.
This is due to worldwide intensive R&D activities and continuous initiatives by ITU-T SG 15 toward optical transport networks (OTNs) and automatically switched optical networks (ASONs).
More recently, in close collaboration with the IEEE, the ITU-T has augmented its G.709 OTN standard to transport 100G Ethernet traffic over wide area networks.

4. Background (2): Slowing Down of SE Improvement
Fixed optical amplifier bandwidth (~ 5 THz)
Per fiber capacity increase has been accomplished through boosting SE (bit rate, wavelength, symbol per bit, state of polarization)
Bit loading higher than that for QPSK causes rapid increase in SNR penalty, and results in shorter optical reach
SE improvement for P2P is slowing down, meaning higher rate data need more spectrum
0.01
0.1 1
100
200
300
400
500 10
Bit rate per channel (Gb/s)
Relative optical reach with constant energy per bit (a.u.)
Spectral efficiency (b/s/Hz)
DP-QPSK
DP-16QAM
DP-64QAM
DP-256QAM
DP-1024QAM
QPSK
BPSK
600
@25 Gbaud
Optical amplifier bandwidth (~ 5 THz)
TDM
WDM
Multi-level mod.
At a fixed optical amplifier bandwidth, typically 5 THz, increases in the per fiber capacity have been achieved through boosting of the spectral efficiency by means of increasing the signal bit rate, wavelengths, symbols per bit, and states of polarization.
As a result, a cutting edge transport system employing dual-polarization and QPSK modulation will have a spectral efficiency reaching 2 b/s/Hz.
Unfortunately, it is well known that bit loading higher than that for QPSK to further increase the channel capacity causes a rapid increase in the SNR penalty.
Under the limited fiber launched power necessary to avoid excessive nonlinear signal distortions, this SNR penalty results in a shorter optical reach.
Due to this, despite the potential for more powerful forward error correction, FEC and a lower noise amplifier, we need to concede that the pace of improvement in spectral efficiency will be slowing down in the era beyond 100 Gb/s.
Multiplexing technology evolution
PDM

5. Background (3): Growing Concern of SE in Networking
Fiber capacity crunch concerns are driving optical networking toward a spectral-efficiency-conscious design philosophy
Right-sized optical bandwidth is adaptively allocated to an end-to-end optical path
Spectral-efficiency-conscious, adaptive networking approach has attracted growing interest
Ex. Elastic optical path network
2008.9
2010.9
2009.9
2009.3
2010.3
2011.3
ECOC2008
“Demonstration of novel spectrum-efficient elastic optical path network ….” (NTT)
Now, we should recognize that the spectrum resources of optical fibers are not limitless as previously thought but rather precious resources.
This idea is driving optical networking toward a spectral-efficiency-conscious design philosophy,
in which the right-sized optical bandwidth is adaptively allocated to an end-to-end optical path by “slicing off” the necessary spectral resources on a given route in the network.
For example, NTT has recently proposed and experimentally demonstrated the “Elastic optical path network”.
Since then, this spectral-efficiency conscious, adaptive optical networking approach has attracted growing interest and a number of relevant symposia and workshops have been held.
ECOC2009 Symposium
“Dynamic multi-layer mesh network”
OFC2010 WS
“How can we groom and multiplex data for ultra-high-speed transmission”
OECC2010 Symposium
“Future optical transport network”
ECOC2010
Symposium
“Towards 1000 Gb/s”
OFC2011 WS
“Spectrally/bit-rate flexible optical network”

6. Expected Early ITU-T Initiatives
Early ITU-T initiatives on studying possible extension of OTN and ASON standards are indispensable.
Greatly support rapid advance and adoption of spectrally-efficient and adaptive optical networks
Starting point regarding studying possible extension of OTN and ASON standards in terms of network efficiency
Clarify what should be inherited, what should be extended, and what should be created
The introduction of elasticity and adaptation will be a big leap forward from conventional rigid and fixed optical networks.
We, therefore, believe that early initiatives by the ITU-T will be indispensable in studying possible extension of the OTN and ASON standards.
This will greatly support the rapid advance and adoption of spectrally-efficient and adaptive optical networks.
In the remaining part of my presentation, I will briefly introduce the elastic optical path network and its possible adoption scenarios.
Then, as the starting point regarding studying the possible extension of OTN and ASON standards in terms of network efficiency,
I will clarify what should be inherited, what should be extended, and what should be created.

7. Elastic Optical Path Network
Spectrum-efficient transport of 100 Gb/s services and beyond through introduction of elasticity and adaptation into optical domain
Adaptive spectrum resource allocation according to
Physical conditions on route (path length, node hops)
Actual user traffic volume
SE-conscious adaptive signal modulation
SE-conscious elastic channel spacing
250 km
400 Gb/s
100 Gb/s
1,000 km
Fixed format, grid
Adaptive
modulation
QPSK
200 Gb/s
16QAM
Path length
Bit rate
Conventional
design
Elastic optical path network
The aim of the elastic optical path network is to provide spectrum-efficient transport of 100-Gb/s services and beyond through the introduction of elasticity and adaptation into the optical domain.
If based on the conventional design philosophy, every optical path is aligned on a fixed grid regardless of the path length, bit rate, and actual client traffic volume.
By taking advantage of spectral-efficiency-conscious adaptive signal modulation and elastic channel spacing, elastic optical path networks yield significant spectral-savings as shown in this figure.
For shorter optical paths, which suffer from less SNR degradation, we employ a more spectrally-efficient modulation format, such as 16QAM.
For client traffic that does not fill the entire capacity of a wavelength, the elastic optical path network provides right-sized intermediate bandwidth, such as 200 Gb/s.
Combined with elastic channel spacing, where the required minimum guard band is assigned between channels, elastic optical path networks accommodate a wide range of traffic in a spectrally-efficient manner.
Elastic channel
spacing

8. Enabling Hardware Technologies (1) Rate and Reach Flexible Transponder
Introduce coherent detection followed by DSP
Optimizing 3 parameters provides required data rate and optical reach while minimizing spectral width
(Symbol rate) x (Number of modulation levels) x （Number of sub-carriers）
Flexible reach
Change the number of bits per symbol with high-speed digital-to-analogue converter and IQ-modulator
Flexible rate
Optical OFDM is spectrally-overlapped orthogonal sub-carrier modulation scheme
Customize number of sub-carriers of OFDM
Flexible reach transmitter
100 G
400 G
Flexible rate/reach transmitter
100 G~
400 G
Using the next two slides, I will present enabling hardware technologies of the elastic optical path network.
The first technology is a rate and reach flexible optical transponder.
Introduction of coherent detection followed by DSP will yield a novel degree of freedom in designing transponders.
By optimizing 3 parameters, the symbol rate, the number of modulation levels, and the number of sub-carriers, we can provide the required data rate and optical reach while minimizing the spectral width.
For example, we can achieve a flexible reach transmitter by changing the number of bits per symbol with a high-speed digital-to-analog converter and IQ-modulator.
Optical OFDM is a spectrally-overlapped orthogonal sub-carrier modulation scheme.
We can achieve a flexible-rate transmitter by customizing the number of sub-carriers of the OFDM signal.

9. Enabling Hardware Technologies (2) Bandwidth Agnostic WXC
Introduce bandwidth-variable WSS based on, e.g., LCoS
Required minimum spectrum window (optical corridor) is open at every node along optical path
Required width of optical corridor is determined by factoring in signal spectral width and filter clipping effect accumulated along nodes. BV
WXC
WSS
transponder
Output
fiber
Input
Bandwidth agnostic WXC
Spatial light modulator
Bandwidth variable
wavelength selective switch (WSS)
Grating
Optical freq.
mittance
Trans-
400 Gb/s
400 Gb/s
40 Gb/s
40 Gb/s
100 Gb/s
The second technology is a bandwidth agnostic wavelength cross-connect.
Bandwidth agnostic wavelength cross-connects, WXCs can be achieved by using a continuously bandwidth-variable Wavelength selective switch (WSS) based on, for example, liquid crystal on silicon (LCoS) technology, as a building block.
In a bandwidth-variable WSS, the incoming optical signals with different optical bandwidths and center frequencies can be routed to any of the output fibers.
These technologies allow us to open the required minimum spectrum window at every node along the optical path.

10. Possible Adoption Scenarios
Step-by-step
Triggered by future
higher rate client signals
(e.g., 400 Gbps)
Earlier adoption
To facilitate
100 Gbps
ROADM design

11. Step-by Step Adoption Scenario: Higher Rate Client Triggered (e. g
Step-by Step Adoption Scenario: Higher Rate Client Triggered (e.g., 400 Gb/s)
Possible next Ethernet rate, 400 G, could appear around 2015.
Optical reach and SE are not independent parameters in 400 G era.
Balancing optical reach and SE in 400 G systems will most likely require elastic spectral allocation
Distance adaptive spectral allocation
High-SE multi-reach traffic accommodation
P2P
Network
Equally-spaced
Non-ITU-T grid
High-SE 400 G accommodation
P2P
1 G
10 G
100 G
1 T
1995
2000
2005
2010
2015
2020
Year of standardization
Bit rate (b/s) GE
10 GE
40 GE
100 GE
OTU1
OTU3
OTU2
OTU4
OTU5
(projected)
STM256
400 GE
STM64
Elastic
channel spacing
High-SE multi-rate traffic accommodation
This graph shows the standardization trend of Ethernet and OTN interfaces over time.
If we simply extrapolate this trend, we see that the possible next Ethernet rate of, say 400 Gb/s, will appear around 2015.
Since optical reach and spectral efficiency are not independent parameters in the 400G era,
in order to balance these parameters, 400 Gb/s point-to-point WDM systems will most likely require
an equally-spaced non ITU-T grid, or
elastic channel spacing if multiple rate traffic of 100 G and 400 G are accommodated.
When applied to ring or mesh networks, different path lengths between source and destination pairs will bring distance-adaptive spectral allocation for network-wide spectral savings.
Finally, dynamic spectral allocation to provide optical bandwidth-on-demand service and cost-effective high-availability transport service will be achieved through sophisticated operation based on the optical version of the link capacity adjustment scheme (LCAS) and bandwidth-squeezed highly-survivable restoration technologies.
Dynamic spectral allocation
Optical BoD, highly survivable restoration

12. Earlier Adoption Scenario: Large-Scale 100 Gb/s ROADM Design Facilitation
Even employing DP QPSK modulation, transmitting 100 Gbps signals over multiple hops of ROADMs on a 50 GHz grid is still tough task.
Distance adaptive spectrum allocation will facilitate 100 Gb/s ROADM design for longer paths
Significant spectral-saving when compared with the worst-case design on a 100 GHz grid.
112 Gb/s DP-QPSK
112Gb/s DP-16QAM 25 50 75
100 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Number of node hops
Allocated spectral width [GHz]
100 GHz grid
Distance
adaptive
Spectrum allocation maps
Network utilization efficiency 1 2 3 4 5 6 7
-45%
100 GHz grid
Distance
adaptive
Required total spectrum at most occupied link (THz)
Distance–adaptive spectrum allocation 12 1 11 2 3 5 4 6 7 8 9 10
The other scenario is an earlier adoption scenario to facilitate 100 Gb/s ROADM design.
Even employing a spectrally-efficient dual-polarization QPSK modulation format, transmitting 100 Gbps signals over multiple hops of ROADMs on a 50 GHz grid is still a tough task, especially for large scale networks.
One way to relax the system design while keeping a reasonable spectral efficiency would be to introduce a non-ITU-T grid or distance adaptive spectral allocation.
The middle figure shows the required spectral resources for 100 Gb/s signals as a function of the number of ROADM hops when non-ideal ROADM filtering characteristics and component frequency offsets are taken into account.
If we consider the worst-case design policy and the conventional frequency grid standard, we have to employ a 100 GHz grid to support a longer optical path with node hops of 10 or more.
The distance adaptive spectrum allocation with elastic channel spacing will alleviate 100 Gb/s ROADM design for longer paths, and result in significant spectral-savings when compared with the worst-case design on a 100 GHz grid.

13. Possible SG15 Study Items
OTN
NW Architecture
IF & Mapping
ASON
Protocol Neutral Spec.
Routing & Signaling
Physical Layer
Frequency Grid
Line-IF Application

14. OTN Network Architecture
G.872 “Architecture of optical transport networks” specifies functional architecture of OTN from network level viewpoint
Layered structure of Optical Channel (OCh), Optical Multiplex Section (OMS), and Optical Transmission Section (OTS)
Although data rate, modulation format, and spectral width of optical path in elastic optical path network may change, elastic optical path is naturally mapped into OCh
See no significant impact on current G.872
OMS
OTS
Mux
Demux Tx Rx 3R
ODUflex, ODUk
OTUflex, OTUk-xv
OCh
Bandwidth agnostic WXC
The first item I’d like to discuss is the Optical Transport Network, or OTN.
ITU-T Recommendation G.872 “Architecture of optical transport networks” specifies the functional architecture of OTN from a network level viewpoint.
G.872 defines an optical network layered structure that comprises an Optical Channel (OCh), Optical Multiplex Section (OMS), and Optical Transmission Section (OTS).
Although the data rate, modulation format, and spectral width of an optical path in an elastic optical path network may change according to the user demand and network conditions, an elastic optical path is naturally mapped into the OCh of the current OTN layered structure.
We, therefore, see no significant impact on the current G.872 when introducing the elastic optical path concept.

15. OTN Interfaces and Mapping: Current OTN
G.709 “Interfaces for the optical transport network (OTN)” specifies Interfaces and mappings of OTN
Conflicting operator requirements
Transport a wide variety of client signals while minimizing types of line-interfaces in order to reduce capital expenditures, which are dominated by line-interface costs.
LO/HO ODUs and ODUflex can address these conflicting requirements.
LO ODU offers versatility to accommodate various client signals and HO ODU offers simplicity in terms of physical interface.
100 Gb/s
ODU 4
OTU 4
ODU 0
OTU 3
The interfaces and mappings of OTN are specified in G.709 “Interfaces for the optical transport network (OTN).”
Originally the OTN specified client signal mapping into ODUk (k=1, 2, 3) [where k is one, two, or three], which have bit rates of approximately 2.5 Gb/s, 10 Gb/s, and 40 Gb/s, and their multiplexing to ODUk with a higher bit rate if necessary.
The multiplexed ODUk signal is then transported as an OTUk signal with an FEC code.
Although network operators should transport a wide variety of client signals, they must minimize the types of line-interfaces in order to reduce the capital expenditures, which are dominated by line-interface costs.
The concept of the Lower Order (LO)/Higher Order (HO) ODU and ODUflex can address these conflicting requirements.
The LO ODU [low order ODU] offers versatility to accommodate various client signals and the HO ODU [high order ODU] offers simplicity in terms of the physical interfaces.
ODU 3
10 Gb/s
ODU 2
OTU 2
OCh
ODUflex (L)
ODU 1
OTU 1
1 Gb/s
Client
signal
Map
Mux
Map
E/O
ODU (L)
ODU (H)
ODU
OTU
OCh

16. OTN Interfaces and Mapping: Possible Flexible OTU Extension
Rate-flexible OCh enables cost-effective transport of various client signals in fully optical domain w/o electrical multiplexing and grooming
Introduction of rate-flexible OTUs (OTUflex) and rate-flexible HO ODUs (HO ODUflex).
Rate-flexible
transponder
1 Tb/s
ODUflex (L)
ODUflex (H)
ODUflex
OTUflex
OTUflex
100 Gb/s
ODU 4
OTU 4
OCh
OTU 3
Once a rate-flexible OCh based on optical OFDM transponders and bandwidth-agnostic ROADMs/WXCs is introduced, cost-effective transport of various client signals will be enabled in the fully optical domain without intermediate electrical multiplexing and grooming processes.
As a natural step toward a rate-flexible OCh, we may need to introduce rate-flexible OTUs (OTUflex) as well as rate-flexible HO ODUs (HO ODUflex).
The OTUflex and HO ODUflex will be specified in the region of over Gb/s depending on the maturity of the device technology at the time.
ODU 3
10 Gb/s
ODU 2
OTU 2
ODU 1
OTU 1
Conventional
transponder
1 Gb/s
ODU 0
Client
signal
Map
Mux
Map
E/O
ODU (L)
ODU (H)
OTU
OCh

17. Physical Layer Specification (1): Possible Frequency Grid Extension
f=193.1 THz
f=193.2 THz
f=193.0 THz
100 GHz
50 GHz
25 GHz
12.5 GHz 1 2 3 4 5 6 7 8 -8 -7 -6 -5 -4 -3 -2 -1
Frequency grid (G.694.1)
G “Spectral grids for WDM applications: DWDM frequency grid”
Anchored to THz, and supports various channel spacings of 12.5 GHz, 25 GHz, 50 GHz, and 100 GHz
Explicitly allocate spectral resources to optical path
To quantize continuous spectrum into contiguous frequency slots with appropriate slot width. 1 3 4 5 6 7 8 2 -4 -3 -2 -8 -7 -6 -5 -1
Frequency slot (12.5 GHz width)
Next I’d like to discuss the physical aspects.
The current ITU-T frequency grid specified in G “Spectral grids for WDM applications: DWDM frequency grid” is anchored to THz, and supports various channel spacings of 12.5 GHz, 25 GHz, 50 GHz, and 100 GHz.
One way to explicitly allocate the spectral resources to an optical path may be to quantize the continuous spectrum into contiguous frequency slots with an appropriate slot width.
For example, the ITU-T frequency grid with a channel spacing of 12.5 GHz should be interpreted as a frequency slot on the grid with a slot width of 12.5 GHz, as shown in the figure to the lower right.
We can flexibly allocate the necessary spectral resources by assigning the required number of contiguous frequency slots. H L
50 GHz
125 GHz
37.5 GHz
Frequency slot allocation

18. Physical Layer Specification (2): Possible Intra-Domain Application Extension
Conventional systems:
Target distance and capacity are a fixed set of values
Elastic optical path network:
Line interfaces will have multi-reach functionality
Trade-off between optical reach and SE
Variable sets of parameters for target distance and capacity
(TD1, TC1)
Distance
Capacity
(TD2, TC2)
(TD3, TC3)
Elastic optical path network
(TD, TC)
Distance
Capacity
Conventional optical network
TD: Target distance
TC: Target capacity
BR: Bit rate
40.10G-20L652A(C)
Target Capacity
=40 x 10 Gb/s
Target distance
=20-span,
long-haul G.652.A-
fiber (C-band)
Recommendation G Longitudinally compatible intra-domain DWDM applications
Ex.
Considering the advanced functionalities that we are trying to achieve, it is natural to start with the intra-domain single vender longitudinal compatibility approach for physical layer specifications of elastic optical path networks, rather than the multi-vendor transversal compatibility approach.
Let us take line interface specifications for example.
In conventional systems, the target distance and capacity of line interfaces are a fixed set of values and defined as application codes.
For example, G defines “Longitudinally Compatible Intra-Domain DWDM Applications.”
This application indicates a target capacity of a 40-channel system with signals of the 10 G payload class, and a target distance of 20 long-haul spans of G.652A fiber.
If we recall that line interfaces of elastic optical path networks will have a multi-reach functionality and there will be a trade-off between the optical reach and spectral efficiency, there can be a variable set of parameters for the target distance and capacity for an application code of the line-interface.
This unique feature may bring additional degrees of freedom in defining the physical layer specifications and may result in a reduction in the capital expenditure.

19. ASON Control Plane
G. 805, G.7713, G.7714, and G.7715 provide network resource model, requirements, architecture, and protocol neutral specifications for automatically switched optical networks (ASONs),
Based on functional models for SDH (G.803) and OTN (G.872)
No significant impact on current ASON standards when introducing distributed control plane into elastic optical path networks
The third item is the Automatically Switched Optical Network, or ASON.
The ITU-T Recommendations on ASON provide a network resource model, requirements, architecture, and protocol neutral specifications for automatically switched optical networks with a distributed control plane.
We have already examined G.872 in the previous slide. As a result of preliminary investigation we consider that there will be no significant impact on the current ASON standards when introducing a distributed control plane into elastic optical path networks, although further studies are still necessary.

20. Possible Technology-Specific Extension of Routing and Signaling
Need discussion on extension of GMPLS protocols in IETF and OIF with ITU-T SG15
Define new parameters in signaling messages
Label request object
Upstream label object …
Explicit route object
Sender TSpec object
Label object
Record route object
Flow spec object
PATH message
RESV message
Switching type:
spectrum switching capable
Parameters in objects
Label:
(start slot, end slot)
As for the technology-specific aspects of routing and signaling in elastic optical path networks, we should discuss possible extension of GMPLS protocols in the IETF and the OIF in close cooperation with ITU-T SG 15.
We may define a new switching type “spectrum switching capable” and a new label “start slot number and end slot number” in signaling messages.
We will also introduce new parameters, the “symbol rate, number of sub-carriers, and modulation level” in the Sender TSpec and Flow Spec Objects.
Modulation format:
(symbol rate, no. of sub-carriers, modulation level)

21. Conclusions Elastic optical path network Possible adoption scenarios
Required minimum spectral resources are adaptively allocated
Possible adoption scenarios
Study items relevant to future standardization activities of ITU-T SG15
Possible extension of OTN, physical layer, and ASON standards in terms of network efficiency