2020 07 17 Infinera Submarine Networks Today and Tomorrow

Submarine Networks –
Today and Tomorrow

Geoff Bennett
Director, Solutions and Technology
Geoff Bennett
Director, Solutions & Technology
Nottingham, United Kingdom
Today’s Rules
Have a Question?
Ask all questions in the Questions Tab
Experiencing Technical issues?

Use the Chat tab for all tech questions.
Poll Question
Poll Question
© 2020 Infinera. All rights reserved. Company Confidential.

5
From Telegraph to
Telephone to Data
Telegraph → Telephone Before 1956 calls were made over radio telephony
Transatlantic Cable Capacity (circuits)

4500
4k 4k
4000 LEO Telstar 1
July 10th 1962
3500
3000
GEO
2500
2000
Intelsat 1
1500 April 6th 1965
1000
845
500
36 48 138 138
0
TAT-1 TAT-2 TAT-3 TAT-4 TAT-5 TAT-6 TAT-7
1956 1959 1963 1965 1970 1976 1978
7
Copper vs Fiber Capacity Levels
Transatlantic Cable Capacity (circuits)
45000
40000
35000
30000
25000
20000
40k
15000
Copper
10000 Fiber
845 4k 4k
5000 36 48 138 138
0
TAT-1 TAT-2 TAT-3 TAT-4 TAT-5 TAT-6 TAT-7 TAT-8
1956 1959 1963 1965 1970 1976 1978 1988
8
Transatlantic Fiber Optic Cable Evolution
From TAT-10 onwards

cable capacity moves to
data rate, not
telephone channels…
…because, by 1992,
voice was just another
type of data
Source: Wikipedia

9
Today there are a lot of submarine cables! telegeography.com
Europe
They carry
>95% of
Mediterranean international
bandwidth
Asia

10
How much traffic is carried by Submarine Cables?
About 1.5 x 1015 bits per second Worldwide Growth
(1.5 petabits per second)
Growth is forecast at 39% ICPs dominate globally

CAGR from 2019-2026
Historically transatlantic and trans-
Source: Telegeography 2020 pacific, but spreading to all routes

11
Are we in another bubble?
Not even close…
Dotcom Bubble
Coherent
CapEx ($B)
Coherent
Post Bubble SLTE Upgrades ICP Surge

12
The anatomy of a
submarine cable
Laying the cable
Cable laying ship
Plough
Cable
Submarine cable
14
Amplifiers in Submarine Cables
25 year design life
+10kV -10kV
50-80km
15
Danger to Submarine Cables
• Sharks did used to attack galvanic

telegraph cables
• 1988 onwards – metal tape
screening introduced
• No more problems with sharks!
Sharks!
On average there are over 100

cable outages per year – but
you rarely hear about them

16
Open Cables: Key Trend in Submarine Cables
Before 2010 Circa 2009 Accelerated Innovation
Closed SLTE
Cable vendor gets 100%
transponder revenue
Closed cable
contracts expire
Coherent!
Cables Upgrades
2014→ Spectrum Sharing
New cables are • One network operator for the cable

designed to be “Open” – Possibly different operators per fiber pair
– Possibly different operators on a single fiber pair
Transponders
Vendor A
Vendor A Vendor B • Challenges
Wet Plant + .
.
. – How to “accept” the new cable is RFS?
Vendor Z – How to manage spectrum for multiple tenants
17
Open Cable Network Architecture
Vendor A Vendor B Infinera Cable Landing Station
Transponder(s) Transponder(s) Transponder(s)
ASE and/or
Idlers
Some or all transponders
will be located in PoP/DC
Power Management
ROADM Controller
PoP or  Wet plant
monitoring
Data Center
Backhaul
“Glass through” To the Wet Plant

18
Submarine Cable
Capacity Evolution
Poll Question
Submarine Cable Types
Ultra Long Haul
1 Dispersion Managed Cable (→2010) 2 Uncompensated (2014→)
+ve -ve +ve -ve +ve Positive Dispersion
Deployed up to 2010 Deployed 2014-2019
Examples: Hundreds of cable systems worldwide Examples: SeaBRAS-1, MONET, BRUSA, MAREA, AAE-1 etc.
3 Festoon/Unrepeatered 4 Space Division Multiplexing (2020→)

No Amp Chain
RFS 2020 onwards

Examples: Dozens of cable systems worldwide
Examples: Dunant – others in planning
Relatively Short Distance
22
Dispersion Managed Cables
1 Dispersion Managed Cable (→2010)

+ve -ve +ve -ve +ve
Deployed up to 2010
Examples: Hundreds of cable systems worldwide
These cables were designed Chromatic dispersion has to be

and deployed before coherent managed by alternating positive
technology was available and negative dispersion fibers
along the cable

23
Capacity Evolution over Dispersion Managed Cables
• Example: Transatlantic cable – Dispersion Managed – Transponder Evolution
– Four fiber pairs
– Design capacity: 800 Gb/s per fiber pair (80 waves at 10Gb/s per wave)
LC-PCS
60 Coherent
200G
Total Cable Capacity (Tb/s)
50 Coherent 48Tb/s
100G
Coherent 40Tb/s
40
32Tb/s
30 40G
Coherent
20 10G
IM-DD 12.8Tb/s
10
3.2Tb/s
0
2005 2010 2012 2016 2020

24
MAREA: A Coherent-Optimized Submarine Cable
The Cable System The Field Trial Results
6,644 km Production Gear – Hero Capacity
6,644 km One-Way Results (16QAM)
Bilbao 6.21 bit/s/Hz Spectral Efficiency
26.2 terabit/s capacity
Virginia Beach
13,210 km Loopback Results (8QAM)

4.46 bit/s/Hz Spectral Efficiency
▪ MAREA Cable System 18.6 terabit/s capacity
▪ Virginia Beach, USA to Bilbao, Spain Commercial Margin Capacity
▪ 6,644km
24.2 Tb/s
▪ Large Area, +ve dispersion cable

25
Where Next for
Increasing Cable
Capacity?
Submarine Amplifiers and Capacity
Amplifier Location
←C-Band EDFA→ • Submarine cables need amplifiers

←C-Band EDFA→
• Amplifiers must be powered
←C-Band EDFA→
←C-Band EDFA→
Let’s look at some key facts

+10-15 kV -10-15 kV
Amplifier power is Amplifiers are only Copper conductor Transponder

inserted at cable about 1.2% efficient has resistance of performance
end points about 0.75 Ω/km depends on high
OSNR – so high amp
power levels
27
Technology Option: Light C+L Bands
Amplifier Location
C+L doubles fiber
←C-Band EDFA→
←C-Band EDFA→ XTb/s capacity at the pair capacity, but has
←C-Band EDFA→ cost of 4 fiber pairs
←C-Band EDFA→ no effect on total
cable capacity
C Band only
Amplifier Location
←C-Band EDFA→
←L-Band EDFA→ XTb/s capacity at the Limited by ability to
←C-Band EDFA→ cost of 2 fiber pairs power the amp chain
←L-Band EDFA→
C+L Band

28
What is SDM? → How do we maximize capacity?
Amplifier Location
←C-Band EDFA→ All of these work to limit the

←C-Band EDFA→
←C-Band EDFA→
number of fiber pairs in the cable
←C-Band EDFA→
Maximize capacity Maximize capacity

per fiber pair
Cable design goals per cable
• Shorter amp spacing • Higher order modulation • Longer amp spacing • LC-PCS
• Higher amp power • Fewer fiber pairs • Lower amp power • More fiber pairs
• Examples: • Pump sharing • Example: Dunant
• MAREA, BRUSA, etc.
29
SDM Comparison: MAREA vs Dunant SDM Roadmap
Future transatlantic cable
40 fiber pairs @ 25Tb/s
Petabit scale cable
MAREA Dunant
State of the art
State of the art SDM cable
UNCOMPENSATED cable
• 6,600 km cable • 6,600 km cable

• 8 fiber pairs • 12 fiber pairs
• 24 Tb/s per fiber pair* • 25 Tb/s per fiber pair**
• Total capacity 192 Tb/s • Total capacity 300 Tb/s
*In service **Planned
30
Challenges for
Submarine
Transponder Design
Poll Question
Cable and Transponder Life Cycles
Submarine cable design lifetime Coherent optical engine technology cycle
25 Years 4 Years
The cable you lay this year may not be the “ideal cable” for future transponders
BUT… Regardless of cable type, each generation of transponder delivers more capacity

34
The only real solution…
You need a comprehensive toolkit

35
ICE6: 5th Generation Coherent Optical Engine
7nm DSP Analog ASIC PIC 20% 33%
Ultra High Baud Rate LC-PCS Subsea Modulations High Gain SD-FEC
(32-96 Gbaud) (ME-8QAM, etc)
• Compact DCO
• 1.6Tb/s Optical Engine
• 2λ x up to 800Gb/s
CHM6
Nyquist Subcarriers DBA Super-Gaussian Encryption
Gain Sharing
Groove (GX) C L
1567 1569
192.1
192.2
192.3
192.4
1529 1610
XTC Shared Wavelocker SD-FEC Gain Sharing Lightning Tolerance C+L Band
DRX

36
TERMINOLOGY: Nyquist Subcarriers: What and Why?
Carrier 1 Carrier 2 Carrier 1 Carrier 2 Carrier 1 Carrier 2 Carrier 1 Carrier 2
Subcarriers Subcarriers Subcarriers Subcarriers
Laser 1
Laser 2
Laser 1
Laser 2
Laser 1
Laser 2
Laser 1
Laser 2
1st Gen: No shaping 2nd Gen: Nyquist shaping ICE4: Nyquist subcarriers ICE6: Nyquist subcarriers
• Higher spectral efficiency • Higher spectral efficiency • Add subcarriers
• Driven by FlexGrid • Enhanced clock recovery • High Baud rate carrier
• Linear tolerance • Low Baud rate subcarriers
• Non-linear tolerance
37
Two options for increasing capacity*
*For both per-wavelength and total fiber capacity
QPSK 16QAM
Time Time
2 Bits/symbol 4 Bits/symbol Symbols per Second
Increase the modulation order Increase the Baud rate

(ie. bits per symbol) (ie. symbols per second)

38
Baud Rate Dependencies: Chromatic Dispersion
Chromatic Dispersion is a function of Square of the Baud Rate Compensating CD creates noise inside
(2xBaud Rate = 4xCD) the transceiver (lower modem SNR)
Coherent Transceiver
CD Compensation NOISE
Dispersion
Chromatic
Baud Rate
Uncompensated cables rely on high CD can be a big problem for

CD to mitigate nonlinear effects trans-oceanic (14,000km+)
39
Baud Rate Dependency: Nonlinear Effects
200Gb/s @ 16QAM 800Gb/s @ 64QAM
4 x 8 GBaud 8 x 12 GBaud
Most DWDM vendors
Single carrier
32 Gbaud → 66 Gbaud
A B
Distortion Penalty FWM SPM
Nyquist Sub-carrier
Optimized Performance
While maintaining economic
advantage of higher Baud rate
4 8 16 33 66
Baud Rate
(Gbaud)
40
Higher Order Modulation
X1 Spectral Efficiency X3
PM-QPSK PM-8QAM PM-16QAM PM-32QAM PM-64QAM
X30 Optical Reach

X1

41
Why do we lose so much reach at higher orders?
Maybe we could increase
PM-QPSK Let’s draw the symbols that a the distance between
receiver would see constellation points…
PM-QPSK PM-64QAM
PM-64QAM
The closest other The closest other symbols

symbols for QPSK are a for 64QAM are really close!
long way away But the further from the
constellation origin, the higher
the power of the symbol –
which triggers non-linear effects

42
The challenge with “hard” modulations
64QAM
1 Imagine if we could “smooth out” the sawtooth

Fiber Capacity
32QAM
16QAM Shannon Limit
8QAM
Imagine if we could “move the QPSK
curve” closer to the Shannon Limit
2 BPSK
Optical Reach
These are two aspects of Probabilistic Constellation Shaping

43
Probabilistic Constellation Shaping (PCS)
We know the outer
symbols are a
problem
More bits per symbol,
shorter reach
DM
Fewer bits per symbol,
longer reach
So we use a clever
Start with a Distribution Matcher to
64QAM manipulate the probabilities
of using certain symbols
constellation

44
Remember the “sawtooth” curve?
1
By adjusting the probability distribution
we can smooth out the sawtooth edges
2 The capacity-reach performance is
Fiber Capacity
moved closer to the Shannon Limit

More bits per symbol,
shorter reach
Shannon Limit
Fewer bits per symbol,

longer reach
Optical Reach

45
PCS and Subcarriers: Dynamic Bandwidth Allocation
Goal: Team to carry 800 lbs
DBA → 85 100
100 90 110 115 100
100 100 115 100
110 100
90 100
85 as far as they can in 1 hour
85 100
100 90 100
110 100
115 100
115 100
110 100
90 100
85
Impairments
ICE6 : 800Gb/s waves, 8x12GBd subcarriers
DBA Enables: Option 1: 100 lbs per person

Higher data rate over a given distance Weaker members slow the team down
Option 2: The strongest members carry a little
Go further at a given data rate
more than the weaker members
Result: The whole team goes further

46
But there is a theoretical limit
Current options revolve around the two different parts of the Shannon Equation
The “bandwidth term” The “log term”
C = B log2 ( 1 +
S
N
)
Bandwidth term
Capacity
This is why SDM is so
interesting for the near future
Log term
Capability
47
The Future of
Submarine Networks
A Situation Report
What do we know? What does that mean?
1: Maximize ROI from existing cables 2: More to extract from new cables
Submarine network
Amplifier Location
demand continues to grow Amplifier Location
←C-Band EDFA→ ←C-Band EDFA→

←L-Band EDFA→ ←C-Band EDFA→
←C-Band EDFA→
←C-Band EDFA→
←L-Band EDFA→ ←C-Band EDFA→
C = B log2
( 1 +
S
N ) 3: C+L delivers more capacity per 4: SDM points the way to a Petabit
PCS brings us close to fiber pair, not per cable transatlantic cable
the Shannon Limit
49
What does a future transponder look like?
Imagine a future Petabit

transatlantic cable
Somehow you
40 fiber pairs must deal with…
25 Tb/s per FP 2,500

400 Gb/s per  …transponders*
*At each end

50
What can we do to help?
C = B log2 ( 1 +
S
N
)
There may be a Shannon …but not on power
Limit on fiber capacity… or volume
Imagine a 400G QSFP28 module

that can span the Atlantic

51
Large Scale Photonic Integration
Instead of one wavelength per
transponder and then muxing them…
Transponders

..
.
  Mux or ROADM
Implement many wavelengths on

a single chip and mux them


..
.


52
Questions?
Enter your question in chat
or
Click “Raise Your Hand” to ask your question live.
Thank You
For Joining!

2020 07 17 Infinera Submarine Networks Today and Tomorrow

Uploaded by

Copyright:

Available Formats

2020 07 17 Infinera Submarine Networks Today and Tomorrow

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2020 07 17 Infinera Submarine Networks Today and Tomorrow

Uploaded by

Copyright:

Available Formats

Submarine Networks –

Today and Tomorrow

Experiencing Technical issues?

© 2020 Infinera. All rights reserved. Company Confidential.

Transatlantic Cable Capacity (circuits)

From TAT-10 onwards

© 2020 Infinera. All rights reserved. Company Confidential.

© 2020 Infinera. All rights reserved. Company Confidential.

Growth is forecast at 39% ICPs dominate globally

© 2020 Infinera. All rights reserved. Company Confidential.

Not even close…

© 2020 Infinera. All rights reserved. Company Confidential.

Cable laying ship

25 year design life

• Sharks did used to attack galvanic

On average there are over 100

© 2020 Infinera. All rights reserved. Company Confidential.

New cables are • One network operator for the cable

© 2020 Infinera. All rights reserved. Company Confidential.

Deployed up to 2010 Deployed 2014-2019

3 Festoon/Unrepeatered 4 Space Division Multiplexing (2020→)

RFS 2020 onwards

1 Dispersion Managed Cable (→2010)

Examples: Hundreds of cable systems worldwide

These cables were designed Chromatic dispersion has to be

© 2020 Infinera. All rights reserved. Company Confidential.

© 2020 Infinera. All rights reserved. Company Confidential.

13,210 km Loopback Results (8QAM)

© 2020 Infinera. All rights reserved. Company Confidential.

←C-Band EDFA→ • Submarine cables need amplifiers

Let’s look at some key facts

Amplifier power is Amplifiers are only Copper conductor Transponder

© 2020 Infinera. All rights reserved. Company Confidential.

←C-Band EDFA→ All of these work to limit the

Maximize capacity Maximize capacity

• 6,600 km cable • 6,600 km cable

Submarine cable design lifetime Coherent optical engine technology cycle

© 2020 Infinera. All rights reserved. Company Confidential.

You need a comprehensive toolkit

© 2020 Infinera. All rights reserved. Company Confidential.

*For both per-wavelength and total fiber capacity

Increase the modulation order Increase the Baud rate

© 2020 Infinera. All rights reserved. Company Confidential.

Uncompensated cables rely on high CD can be a big problem for

Most DWDM vendors

Distortion Penalty FWM SPM

PM-QPSK PM-8QAM PM-16QAM PM-32QAM PM-64QAM

X30 Optical Reach

© 2020 Infinera. All rights reserved. Company Confidential.

The closest other The closest other symbols

© 2020 Infinera. All rights reserved. Company Confidential.

1 Imagine if we could “smooth out” the sawtooth

16QAM Shannon Limit

These are two aspects of Probabilistic Constellation Shaping

© 2020 Infinera. All rights reserved. Company Confidential.

moved closer to the Shannon Limit

Fewer bits per symbol,

© 2020 Infinera. All rights reserved. Company Confidential.

ICE6 : 800Gb/s waves, 8x12GBd subcarriers

DBA Enables: Option 1: 100 lbs per person