Transcript A multi-channel effect

SCALING OPTICAL NETWORKS WITH ADVANCED PHOTONICS
Ove Parmlind
with acknowledgement to David Butler
1
© Nokia 2016
Nokia Public Use
AGENDA
1. How on earth did we get here?
2. An insatiable appetite………….
Tutorial
3. Our desires
4. What stops us
5. Squeezing out the last drop
6. Packing them in
Technology for Scalable Networks
7. The other piece
8. Tools to maintain performance
2
© Nokia 2016
Nokia Public Use
An Insatiable Appetite
[Stolen from P. J. Winzer, Bell Labs Tech. J., 2014]
20% TO 90% GROWTH PER YEAR, WIDELY DEPENDENT
ON MARKET SEGMENT, OPERATOR, AND GEOGRAPHY
3
© Nokia 2016
Nokia Public Use
Heart Attack Grill® Quadruple Bypass Burger®
AN INSATIABLE APPETITE
Global Internet Traffic Forecast
200
180
Exabytes per Month
160
140
120
100
80
60
23% CAGR 2014-2019
40
0.9dB/year
20
0
2013
© Nokia 2016
2015
2016
2017
2018
2019
2020
Cisco VNI Global IP Traffic Forecast, 2014-2019
Heart Attack Grill®
4
2014
Nokia Public Use
Demand
Supply
2014
2019
2025?
0.9 dB/year
Long Haul WDM Platform Capacity
WDM Capacity Tb/s per Fibre
100
0.8dB/year
0.25dB/year
3.1dB/year
10
Metro
60EB/Month*
168EB/Month*
Spectral Efficiency
of 4 bit/s/Hz
1
Spectral Efficiency
of 1 bit/s/Hz
0.1
0.6 dB/year
0.01
1990
1995
2000
2005
2010
2015
2020
2025
2030
2035
Long Haul
29EB/Month*
57EB/Month*
* Cisco VNI Global IP Traffic Forecast, 2014-2019
5
© Nokia 2016
Nokia Public Use
OUR DESIRES
Capacity
Reach
Power
Joshua Heller CC BY-SA 2.0
Low Cost
6
© Nokia 2016
Size
Nokia Public Use
Flexible
WHAT STOPS US
7
© Nokia 2016
Nokia Public Use
PULSE PROPAGATION
Receiver
Transmitter
A
i
1
2 A
2
i
 A 
2


A
A0
z
2
2
dT 2
© Robertson. Smithsonian Institution United States
8
© Nokia 2016
Nokia Public Use
PULSE PROPAGATION
A
i
1
2 A
2
i
 A 
2


A
A0
z
2
2
dT 2
Losses – predominantly in the fibre overcome with amplifiers generating noise
Chromatic Dispersion historically overcome with in line compensation
Nonlinear effects
Additionally time varying penalties, such as Polarization Mode Dispersion, need to be considered
9
© Nokia 2016
Nokia Public Use
NOISE
Generally dominated by Amplified Spontaneous
Emission
Added by each amplifier in the line
Noise power is proportional to the gain required
Noise is conventionally given in terms of Optical Signal to Noise Ratio (OSNR)
measured in 0.1nm (although one vendor uses 0.5nm!)
For a line system delivering a high OSNR is good
For a receiver tolerating a low OSNR is good
10
© Nokia 2016
Nokia Public Use
Bob Mellish CC BY-SA 3.0
CHROMATIC DISPERSION
Demonstration of error-free 25Gb/s duobinary transmission using a colourless reflective integrated modulator; Caroline P Lai et al.;
Optics Express Vol.21, Issue 1
11
© Nokia 2016
Nokia Public Use
Nonlinearity
JOHN KERR
Who is to blame?
Change in refractive index is proportional to
the SQUARE of the electric field
Public Domain
12
© Nokia 2016
Nokia Public Use
NONLINEARITY
Self-phase modulation (SPM)
Cross-phase modulation (XPM)
A single-channel effect
A multi-channel effect
2 dBm
17 dBm
Ps = 2 dBm
18 dBm
Pc = 13 dBm
20 dBm
Intensity distortion after single-channel propagation
in a 2x80km link (for different channel powers)
13
© Nokia 2016
Nokia Public Use
Intensity distortion after two-channel propagation in a
2x80km link
Nonlinearity
Four-wave mixing (FWM)
Stimulated Raman Scattering (SRS)
A multi-channel effect
A multi-channel effect
Generation of intermodulation frequency
components at f = fi + fj - fk
Energy transfer from lower-wavelength to higherwavelength channels
It generates an extra tilt, which is taken into account in
the link design
14
© Nokia 2016
Nokia Public Use
Power (0.5dB/Div)
Power (0.5 dB/Div)
Experimental spectrum recorded after 25km of G.653 fibre
with 3 channels at unequal spacing
2.3
dB
Fibre output (5.6 dBm/ch)
Fibre input
POLARIZATION MODE DISPERSION (PMD)
PMD is due to the asymmetry of the fiber strand. This can be caused by intrinsic geometric imperfections or
by the cabling putting stress around onto the core.
The birefringence in the optical fiber
slows down the X-polarized state that
sees the higher refractive index and
causes a differential group delay (DGD)
between the polarization states resulting
in pulse distortion
p
1.0E+00
1.0E–01
The probability of exceeding 3.2 times
the mean DGD is around 1E-05 which
corresponds to about 5 min/year
1.0E–02
1.0E–03
1.0E–04
1.0E–05
p
1.0E–06
1.0E–07
1.0E–08
1.0E–09
1.0E–10
0
< >
2 < >
3 < >

T1540440-00
DGD Probability Distribution Function
Average DGD (PMD)
15
© Nokia 2016
Nokia Public Use
“40Gb/s Networks and the PMD Challenge” ;richard.ednay & modesto.morais
POLARIZATION MODE DISPERSION (PMD)
16
© Nokia 2016
Nokia Public Use
PMD
Reach
10Gb/s 100Gb/s
(ps/√km)
(km)
(km)
0.2
0.5
1
2
5
2500
400
100
25
4
>10 000
>10 000
10 000
2500
400
NETWORK AUTOMATION
DESIGN, DEPLOYMENT & MANAGEMENT
• Network design and planning tools
• Integrated with Network Management
17
© Nokia 2016
Nokia Public Use
THE CHOICE OF FIBRE
Srleffler licensed under CC BY-SA 3.0
Typical
Zero
Dispersion Dispersio Effective
Fibre Type
@1550nm Wavelen (µm2)
(ps/nm(nm)
EX2000
18
© Nokia 2016
Nokia Public Use
20
1310
112
Excellent loss & non-linear tolerance.
G.652
17.1
1310
80
Very good non-linear performance
Teralight
6.5
1425
63
Probably the best performing G.655 fibre
TrueWave-RS
4.6
1446
56
Poor XPM in blue end primarily due to
LEAF
4.1
1511
73
Poor XPM in blue end primarily due to
TrueWave
3.5
1496
59
Poor XPM in blue end primarily due to
G.653
0.1
1549
48
Very poor non-linear performance
LS
-1.6
1571
56
Poor nonlinear performance particularly
SQUEEZING OUT THE LAST DROP
19
© Nokia 2016
Nokia Public Use
THE TOOLSET TO MEET NETWORK TRAFFIC GROWTH
FIVE PHYSICAL DIMENSIONS OF AN ELECTRO-MAGNETIC WAVE
Polarization
Space
Waveforms change in …
Time
•
Frequency
Amplitude & Phase
Same 5 physical dimensions across all communications technologies
(Wireless, DSL, Optics, …)
ORDER OF DEPLOYMENT MATTERS – HIGHLY APPLICATION SPECIFIC
Stolen from Peter Winzer
20
© Nokia 2016
Nokia Public Use
MANY MANY MOONS AGO……..
TIME AND AMPLITUDE
Polarization
Space
Time
Public Domain
21
© Nokia 2016
Nokia Public Use
Frequency
Amplitude & Phase
BEFORE “THE INTERNET” WAS CALLED “THE INTERNET”
ADD COLOUR
1993
1987
Polarization
Space
Time
1991
22
© Nokia 2016
Nokia Public Use
Frequency
Amplitude & Phase
BEFORE FACEBOOK
ADD PHASE
Polarization
Frequency
Space
Time
Amplitude & Phase
-1
0
1
Duobinary
Made 40Gb/s per carrier possible
23
© Nokia 2016
Nokia Public Use
0
1
1
1
0
0
1
0
DQPSK
& with Polarization Multiplexing............
Increasing Implementation Complexity
PDM-BPSK
PDM-QPSK
PDM16QAM
PDM64QAM
2
bits/symbol
4
bits/symbol
8
bits/symbol
12
bits/symbol
Spectral efficiency comes at a cost:
Higher signal-to-noise ratio (SNR)
requirements for higher-order modulation
24
© Nokia 2016
Nokia Public Use
45 GBAUD (90GS/S) OPERATION BECAME FEASIBLE @ 28NM CMOS
200Gb/s Options
40nm CMOS
32 GBaud
PDM-16QAM
28nm DAC with 45 GBaud and
Gaussian shaping
25
© Nokia 2016
Nokia Public Use
28nm CMOS
43 GBaud
“PDM-8QAM”
FLEXIBLE CHANNEL RATES
NEW FORMATS
• 64QAM offering maximum
capacity on short distances
NEW 100G HI-PERFORMANCE
SP-QPSK
• 8QAM well suited for long haul
200G applications
100G
QPSK
NEW 200G HI-PERFORMANCE
REAC
H
8QAM
• SP-QPSK for maximum 100G
reach in a single-carrier solution
200G
16QAM
NEW 400G HI-CAPACITY
EXAMPLE REACHES
SP-QPSK – 5000 km
64QAM
CAPACITY
26
© Nokia 2016
Nokia Public Use
QPSK
– 3000 km
“8QAM”
– 1500 km
16QAM
– 600 km
64QAM
– 150 km
Q² factor [dB]
WHY 200GB/S 8QAM?
Theory PDMQPSK
PDM-QPSK
(direct enc./dec.)
Theory PDM8QAM
PDM-8QAM
(cross)
PDM-8SP16QAM
~ 4.0 dB
10
12
14 16 18 20 22
OSNR [dB]/0.1nm
24
26
o PDM-8SP16QAM and PDM-8QAM have similarly tolerance to optical noise
27
© Nokia 2016
Nokia Public Use
-45
-55
-65
-75
1544.9
1545.3
wavelength [nm]
1545.7
PDM- 8SP16QAM (8 Set
Partitioned 16QAM )
Theory PDM16QAM
~ 3.25 dB
8
Power [dBm]
14
13
12
11
10
9
8
7
6
5
4
Native RRC 0.1
-35
PDM- Cross 8QAM
SP-QPSK
WHAT’S
THAT?
Polar. X
Polarization-based
Polar. X
Time-based
4D-SP-QPSK
4D-SP-QPSK
0
1
0
0
1
1
1
1
1
1
1
0
0
0
t
1
1
0
0
1
1
1
0
0
1
Q² factor [dB]
Q² factor [dB]
10
8
0.5 dB
6
8
43GBd 4D-SP-QPSK w/o ROADM
10
8
14
16
18
43GBd 4D-SP-QPSK w/ ROADM
43GBd 4D-SP-QPSK w/o ROADM
32.5GBd PDM-QPSK w/ ROADM
32.5GBd PDM-QPSK w/o ROADM
6
4
Theory - 43GBd PDM-QPSK
6
20
OSNR [dB]/0.1nm
© Nokia 2016
12
32.5GBd PDM-QPSK w/o ROADM
Theory - 32.5 GBd PDM-QPSK
4
SP-QPSK
50GHz spacing, SSMF testbed
(no DM)
14
10
Theory - 43 GBd 4D-SP-QPSK
12
No ROADM, SSMF testbed
(no DM)
12
6
43 GBd PDM-QPSK
10
Does not require common
phase recovery between
the two polarizations
43 GBd 4D-SP-QPSK
32.5 GBd PDM-QPSK
8
Requires common phase
recovery between the two
polarizations
Q² factor [dB]
14
12
Time Based
t
1
1
0
0
Polar. Y
Polar. Y
28
0
0
1
1
Polarization Based
8
10
12
14
# loops [x400 km]
16
18
20
4
6
8
10
12
14
# loops [x400 km]
16
18
• Back-to-back gain only about 0.5dB
• About 2dB gain with transmission
• Ideally used with super channels but still offers QPSK like performance through 50GHz filters
Nokia Public Use
20
64QAM
REALLY!
29
© Nokia 2016
Nokia Public Use
TITLE 20 PT TREBUCHET
SUBTITLE 16 PT TREBUCHET
RailPictures.net ©Phil Cotterill used with permission
Packing them in
30
© Nokia 2016
Nokia Public Use
FLEXIBLE GRID
SPECTRAL SHAPING AND SUPER CHANNELS
300GHz
50GHz
SPECTRAL SHAPING
-30
NRZ
-35
100G
RRC
-40
-45
-50
6x100G QPSK bundled
together from end to
end for maximum
transmission efficiency
-55
-60
37.5GHz
50GHz
-65
-70
1546.2
1546.3
1546.4
1546.5
1546.6
1546.7
1546.8
1546.9
1547
DSP-based spectral shaping at the Tx
100G
100G
225GHz
600G SUPERCHANNEL
SPECTRAL
SHAPING
31
© Nokia 2016
• Pulse-shaping functionalities on the 100G DSP allow transmitting narrower spectra
• Enables tighter channels spacing with negligible transmission penalty
• Combined with Flexgrid and Coherent filtering allows transporting “Superchannels”
Nokia Public Use
FLEXIBLE GRID COMES OF AGE
43GBAUD BASED SUPER CHANNELS
6 x 200Gb/s 16QAM in 300GHz
New Formats Now Practical Through
43GBaud
100Gb/s SPQPSK
200Gb/s “8QAM”
50GHz
2014
200G
≈600km Reach
250Gb/s 16QAM
50GHz
2015
400Gb/s 64QAM
≈1500km Reach
200G
200G
6 x200Gb/s “8QAM” in 312.5GHz
32
© Nokia 2016
Nokia Public Use
3rd coherent generation Coherent Chips
What is new?
64QAM
(Flexible FEC)
Flexible Modulation
16QAM
8QAM
QPSK
WSS Filter
shape
100Gb/s – 5000km
(>8Tb/s)
50Gb/s – 10,000km
(>8Tb/s)
BPSK
50GHz
Programmable Modulation optimizes spectral density to the
network
33
© Nokia 2016
Nokia Public Use
THE OTHER PIECE
WAVELENGTH ROUTING
34
© Nokia 2016
Nokia Public Use
WAVELENGTH ROUTING BENEFITS ANALOGY
FREEDOM TO EFFICIENTLY GET TRAFFIC WHERE YOU NEED IT
35
Must take assigned on-ramp
Choose the best on-ramp
Congestion, hard to divert
Efficient utilization, easy to divert
Slow new route utilization
Fast new route utilization
Running out of highway lanes
Future proof flexible lanes
© Nokia 2016
Nokia Public Use
FLEXIBLE GRID
COHERENT FILTERING
10G
100G
DEMUX
COHERENT RX
The wavelength is selected by
tuning the local laser in the RX
Direct Detection RX
COHERENT
FILTERING
36
© Nokia 2016
• Coherent detection allows to perform filtering in the electrical domain
• No need per-channel optical demux
• Enables tighter channels spacing and “Superchannels”
Nokia Public Use
FLEXIBLE GRID
ENABLING NARROWER AND FLEXIBLE CHANNEL SPACINGS
50 GHz
50GHz
37.5 GHz
75 GHz
WSS
37.5 GHz
FLEXIBLE
GRID WSS
100 GHz
50 GHz
FLEXGRID
WSS
37
© Nokia 2016
•
•
•
•
LCoS-based WSS enables flexible and granular allocation of bandwidth
12.5 GHz granularity
Enables advanced modulation formats for joint reach & spectral efficiency optimization
Enables “Superchannels”
Nokia Public Use
BARRIERS TO WAVELENGTH NETWORKING
Conventional ROADM architecture:
CDC-F architecture:
• Constrained
• Highly agile
• No truck rolls to route connectivity
• Efficient add/drop block
• Truck rolls to route connectivity
• Inefficient add/drop block
[1]
•
•
•
•
Ports are color specific
Ports are tied to a specific line
A/D blocks support a specific color only once
A/D blocks are separate for every direction
[1]
•
•
•
•
Ports are color independent (tunable)
Ports are routable to all lines
A/D blocks support a specific color when needed
A/D blocks are pooled across lines and efficiently used
[1] Based on “Flexible Architectures for Optical Transport Nodes and Networks”; S. Gringeri et al; IEEE Communications; July 2010
38
© Nokia 2016
Nokia Public Use
BENEFITS OF WAVELENGTH ROUTING
LOWER NETWORK COST OF OWNERSHIP WITH MORE SERVICE
AGILITY
35
%
Reduced
CAPEX
30
%
Reduced
CAPEX
$
New
services
39
© Nokia 2016
SCALE AT THE MOST EFFICIENT LAYER:
• capacity independent, in-service scale in all dimensions
• greener – over 30% less power to switch light
OPEX
costs
ROUTE WAVELENGTHS TO RECOVER NETWORK CAPACITY:
• Eases OEO scale
OPEX
• no on-site visits required to route wavelengths
costs
GENERATE NEW REVENUE OPPORTUNITIES:
• network agility to support SDN service goals
• automated control to support new services/OAM applications
Nokia Public Use
OPEX
costs
WAVELENGTH ROUTING TO RECOVER CAPACITY
EXTENDS NETWORK LIFESPAN
Wavelength Demands
A
A
A
A
B
B
B
C
C
C
D
E
E
F
BEFORE DEFRAGMENTATION
B
E
F
C
D
C
G
G
G
F
E
F
G
G
6
colors
Sites A-B Sites A-C Sites A-D Sites B-E Sites C-F Sites D-E Sites D-F Sites E-G Sites F-G
Site B
Site A
30% recovery of network capacity
Site C
4
colors
Site E
Site D
AFTER DEFRAGMENTATION
Site F
40
© Nokia 2016
Site G
Nokia Public Use
ADD AN OTN SWITCH
True Value of Flexible Networks; G. Wellbrock, TJ Xia; M3A.1 OFC2015
41
© Nokia 2016
Nokia Public Use
WHY BE FLEXIBLE?
42
© Nokia 2016
Nokia Public Use
NEW YORK – SAN JOSE
RFP DESIGN TO SUPPORT 88X100GB/S
The ALU RFP response was for a line design to support 88x100Gb/s channels and included two regenerators
39x100G
58x100G
53x100G
70x100G
66x100G
PDM-QPSK Regenerator Locations
RFP Demands per segment shown in BLUE
Congestion around
east coast routes
43
© Nokia 2016
Nokia Public Use
84x100G
ALTERNATIVE NEW YORK – SAN JOSE
DESIGN USING MIXED FORMATS
An alternative design using mixed modulation formats allows for relief of congestion on shorter east coast demands by using 200Gb/s
8QAM and greater reach on more lightly loaded segments by switching to 100Gb/s SPQPSK
PDM-SPQPSK
Regenerator Location
44
© Nokia 2016
Nokia Public Use
PDM-8QAM
MAINTAINING PERFORMANCE
Bert van Dijk CC BY-SA 2.0
45
© Nokia 2016
Nokia Public Use
OPTICAL TIME DOMAIN REFLECTOMETRY (OTDR)
An embedded OTDR card to assist with;
1.
Raman turn-up
 Provide point loss and reflection accuracy to within 1m for events
within 2km, and 2m accuracy out to 20km during Raman turn-up
 Give visibility of the first external connector (with shortest pulse width
setting)
 Provide an interactive “wizard” to simplify Raman turn-up
2.
Fiber cut detection and location
 It will monitor both ends of the fiber
 Provide location accuracy to within 10m for cuts no further than 60km
from either of the fiber ends, and for cuts that are further away (in
loss or distance), the distance resolution may degrade to 15m
 Fibre plant is aging and much has issues which require monitoring
3.
In-service loss distribution trending
 Store baseline at commissioning and flag significant changes
46
© Nokia 2016
Nokia Public Use
OTDR Results – 75km Dark fiber
OTDR Results – FIBER cut @ 50km
Event 2: Connector @ 75km
Event 1: Connector @ 50km
47
© Nokia 2016
Event 1 Signal Termination @
50km
Nokia Public Use
OPTICAL CHANNEL TRACE
Desired Configuration
Due to device failure, provisioning failure or incorrect fibering at the degree 4 node,
Service 1 and Service 2 are misdirected
Service 1
Service 2
A
Deployed Configuration: Intermediate Misconnection
B
A
B
C
C
D
D
The net result is misdirected traffic where one end is Service 1 and the other end is Service 2
48
© Nokia 2016
Nokia Public Use
Because the future isn’t what it used to be………..
Alien Wavelengths
An Alien Wavelength Interface / Demarcation is required
Alien wavelength power adjustment and power balancing needs to be achieved
Agnostic rate and format support to ensure compatibility with any of tomorrows formats
OSNR Monitoring of alien wavelengths
Automated commissioning and provisioning
Threshold alarms
ALU Line System
ITU Optic
Vendor x
49
© Nokia 2016
Nokia Public Use
ITU Optic
Vendor x
50
© Nokia 2016
Nokia Public Use