Upload
nhat-minh
View
216
Download
0
Embed Size (px)
Citation preview
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
1/59
The All Optical NetworkWhy we need it how and when we get it
David Payne BTCo-inventor of TPON and Principal Consultant, Optical Networks
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
2/59
David Payne BT 2007
Thanks to the optical team:
Alan Hill
Alan McGuireAlbert Rafel
Andrew Lord
Dave McCartneyDerek Nesset
Ed Sikora
Ian hope
Ivan Boyd
John Wright
Justin Kang
Kristan Farrow
Martin Wade
Paul TomlinsonPaul Wright
Peter Chidgey
Peter HealeyPhil Barker
Ruben Gorena
Russell Davey
Shamil Appathurai
Steve Hornung
Tim Gilfedder
Yu-rong Zhou
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
3/59
David Payne BT 2007
The electro-magnetic spectrum
The optical fibre window1.87x1014 to 2.38x1014
(1600nm to 1260nm)
Why optical networks? the physics perspective
Higher carrier frequencies
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
4/59
David Payne BT 2007
Channel Capacity Limits
1.0E-04
1.0E-02
1.0E+00
1.0E+02
1.0E+04
1.0E+06
1.0E+08
1.0E+10
1.0E+12
1.0E+14
1.0E+16
1.0E+18
1.0E+20
1
.00E+05
1
.00E+06
1
.00E+07
1
.00E+08
1
.00E+09
1
.00E+10
1
.00E+11
1
.00E+12
1
.00E+13
1
.00E+14
1
.00E+15
1
.00E+16
1
.00E+17
1
.00E+18
1
.00E+19
1
.00E+20
Frequency (Hz)
RelativeCh
annelCapacity
Why optical networks? the physics perspective
10 24
10 22
10 20
10 18
10 16
10 14
10 12
1010
10 8
10 6
10 4
10 2
1
Capacity per Hz
Cumulative capacity
Channel capacity bits/s
Fibre spectrum
The optical fibre window50 THz
~1000Tb/s
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
5/59
David Payne BT 2007
It will not be worth going beyond ~3nm wavelength (~1017Hz) due tofundamental quantum noise limitations.
Current generation optical communications systems are within twoorders of magnitude of that limit in information capacity terms.
But we have no technology currently being developed that exceeds thecapability of optical fibre or that could exploit the higher frequencies.
In the past we have had a fundamentally better technology (usuallymore expensive) for core networks than the access network.
With fibre ubiquitously deployed in access, metro and core networksthis is no longer the case.
Therefore end users of a ubiquitous fibre network can neverhave the full bandwidth of a fibre dedicated to them!
This has consequences for the choice of network architecture.
Why optical networks? the physics perspective
Summary 1
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
6/59
David Payne BT 2007
Why optical networks? the commercial perspective
Meeting customer demands for new high bandwidth
services
New revenue generation
Reducing operational costs
Meeting competition threats
Staying locally or internationally competitive Attract inward investment
Drivers
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
7/59
David Payne BT 2007
Bandwidth demand New services including:
video, hdtv etc.
Large file transfer eg. hi-res image & video content, photographs etc.
Mass market network storage services
Distributed servers
Thin client computing
Etc.
End users need access to high speed pipes
To get low delay, fast file transfer & rapid response time
But they dont need the bandwidth all of the time.
Therefore redistribute the unused bandwidth DBA for all services!
Why optical networks? the commercial perspective
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
8/59
David Payne BT 2007
Impatience Index
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 5 10 15 20 25 30 35
time to download (Secs)
%Customersthatwait
DialU
p(circa2000)
Early
Bband(circa2002-3)
Bband(2
006)
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
9/59
David Payne BT 2007
Time to transmit files
200ms
1 sec
10 sec
1min
1 hour
8 hours
1 day
10 days
100 days
1 10 100 1,000 10,000 100,000
File size (Mbytes)
Timeto
transmit
Modem
56kb/s
ADSLdo
wn2Mb/s
Postal Service
ADSLUp0
.5Mb/s
FTTC(V
DSLUp)3M
B/s
FTTP100Mb
/scust
port
FTTP1
Gb/scustpo
rts
Instantaneous
Customers accept delay
Customers get impatient
Delay a major barrier
Physical transport faster!
}}}10 min
DVD HDTVCD
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
10/59
David Payne BT 2007
StreamingServices
2008 2017 2008 2017
Internet Surfing 60% 100% 40% 70%
IP HDTV HR 0% 90% 70% 80%
IP HDTV HR 0% 80% 50% 60%
IP HDTV LR 10% 30% 30% 40%
IP HDTV LR 10% 12% 10% 20%
IP SDTV 90% 50% 70% 80%
IP SDTV 80% 25% 50% 60%
IP SDTV 40% 20% 30% 40%
IP SDTV 20% 10% 10% 20%Video comms Cam
corder 2% 15% 20% 30%
Web Cam comms10% 15% 20% 30%
Voice 100% 100% 20% 20%
Voice 75% 75% 15% 15%
Voice 50% 50% 10% 10%Thin client
computing 0% 30% 10% 50%
Thin client
computing 0% 20% 10% 50%
e-mail 0% 80% 10% 20%
e-mail 0% 60% 10% 15%
Photos 30% 70% 5% 10%
Photos 20% 30% 5% 10%
Video clips 20% 60% 15% 25%
Video clips 20% 40% 15% 25%
Music 30% 40% 5% 15%
Music 20% 30% 5% 15%
Documents 10% 20% 5% 10%
Documents 5% 15% 5% 10%
Software 5% 10% 3% 8%
Software 2% 5% 2% 5%Web site uploads 3% 10% 1% 5%
Web site uploads 0% 1% 1% 5%
Peer to peer file
sharing 5% 40% 40% 50%
Peer to peer file
sharing 3% 20% 30% 40%
Storage services5% 30% 1% 5%
Storage services3% 15% 1% 2%
Table 1 Example service scenario 1
Probability of takeup Probability of using service inbusy hour
File Transfers
Example Service Scenario
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
11/59
David Payne BT 2007
Bandwidth Usage ResultsFrom example service scenario 2008
This chart shows the distribution ofthe average FTTP user bandwidth forthe service scenario parameters foryear 2008.
The mean is 2.0 Mb/s Note the
significant probability that someusers do not use any services in thebusy hour.
This chart shows the distribution ofthe maximum user bandwidth if theuser had simultaneously used all theservices they had signed up for.
This is a dumb way of defining userbandwidth and would push the meanto 29.9 Mb/s 15 times the requiredlevel if statistical multiplexing is takeninto account.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
12/59
David Payne BT 2007
Bandwidth Usage ResultsFrom example service scenario 2017
This chart shows the distribution of theaverage user bandwidth for the servicescenario parameters for year 2017.
The mean traffic has grown from 2.0 Mb/s7.5 Mb/s.
The probability of users using no serviceshas reduced significantly but is still largeenough to remain the mode of thedistribution.
This chart shows the distribution of themaximum user bandwidth if the user hadsimultaneously used all the services theyhad signed up for.
This is the dumb way of defining userbandwidth and would push the mean to82 Mb/s, 11 times the required level ifstatistical multiplexing is taken intoaccount.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
13/59
David Payne BT 2007
End user bandwidth growth
Effective end user average bandwidth
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
0 2 4 6 8 10
Years
Mb
/s
Peak user bandwidthAverage peak user bandwidthAverage user bandwidth
Mb/s
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
14/59
David Payne BT 2007
Streaming mode
Used bandwidthprofile
End useraccess pipes
End user 1
End user 2
End user 3
End user 5
End user 4
Access network bandwidth Pipe(eg. GPON DS bandwidth)
CPs
Streaming mode
user 1
user 2
user 4
user 3
user 1user 5
Spare bandwidth pool =access pipe size used bandwidth
Spare bandwidth not usedin streaming mode
Streaming mode
Used bandwidthprofile
End useraccess pipes
End user 1
End user 2
End user 3
End user 5
End user 4
Access network bandwidth Pipe(eg. GPON DS bandwidth)
CPs
Streaming mode
user 1
user 2
user 4
user 3
user 1user 5
Spare bandwidth pool =access pipe size used bandwidth
Spare bandwidth not usedin streaming mode
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
15/59
David Payne BT 2007
End useraccess pipes
End user 1
End user 2
End user 3
End user 5
End user 4
Access network bandwidth Pipe(eg. GPON DS bandwidth)
CPs
Burst-mode
Streaming mode
Used bandwidthprofile
All spare bandwidth isused to deliver files as fast
as possible
Burst-modeUsed bandwidth
profile
Spare bandwidth
shared betweenactive users
End useraccess pipes
End user 1
End user 2
End user 3
End user 5
End user 4
Access network bandwidth Pipe(eg. GPON DS bandwidth)
CPs
Burst-mode
Streaming mode
Used bandwidthprofile
All spare bandwidth isused to deliver files as fast
as possible
Burst-modeUsed bandwidth
profile
Spare bandwidth
shared betweenactive users
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
16/59
David Payne BT 2007
512 Customers, 10G PON Bandwidth, 1000M Access Port, No Broadcast
0.00
100,000,000.00
200,000,000.00
300,000,000.00
400,000,000.00
500,000,000.00
600,000,000.00
700,000,000.00
Teleph
one
Vide
opho
ne
SDVideo
HDVideo
Basic
Servic
es
Premium
Servic
es
Teleph
one
Vide
opho
ne
SDVideo
HDVideo
Basic
Servic
es
Premium
Servic
es
AverageSessionBa
ndwidth(bits/s)
10%
15%
20%
30%
40%50%
60%
70%
80%
90%
Burst Mode ScenarioStreamed Scenario
Improvements in End User experience
from burst-mode operation
700
600
500
400
300
200
100
0AverageSessionbandwidthMb
/s
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
17/59
David Payne BT 2007
Why optical networks? - commercial &
services perspective
There are a number of powerful drivers for all optical networkingparticularly Fibre to the Premises (FTTP) networks:
Attracting inward investment
Operational cost savings
New services for end users
New revenues into the industry
Improved end user experience click and its there!
Users need high speed access + higher ave. bandwidth
Summary 2
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
18/59
David Payne BT 2007
Barriers High upfront capital investment
Civil infrastructure build costs
Also time taken to build
Increased opex costs if legacy not replaced Uncertain take-up rates
Uncertain revenues from new broadband services
Financial barriers
So why arent optical networks everywhere?
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
19/59
David Payne BT 2007
Todays optical access network options
Point to Point Solutions
Point to point fibre
WDM PON
Active Star
Passive Optical Network:
BPON
GEPON
GPON
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
20/59
David Payne BT 2007
Options for optical access
Big
Business
CustomerLocal Exchange
FTTP
Customers
Cabinet
Non-FTTP
Customers
backhaul/metro
network
Cabinet
Point to point fibre.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
21/59
David Payne BT 2007
Options for optical access
Big
Business
CustomerLocal Exchange
FTTP
Customers
Non-FTTP
Customers
backhaul/metro
network
Point to point fibre active cabinet.
Cabinet
Active star
Cabinet
Active star
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
22/59
David Payne BT 2007
Cabinet
Options for optical access
Big
Business
Customer
StreetMSAN
Non-FTTP
Customers
copper
backhaul/metro
network
FTTP
Customers
WDM PON.
Local Exchange
WDM
Splitter
WDM
Multiplexer
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
23/59
David Payne BT 2007
Pt-Pt optical access network options
Point to Point Solutions: Point to point fibre to exchange
+ High bandwidth per customer
+ Good security
- Large fibre count cables- two opto-electronic modules per customer
- Opex and capex issues
Active Star
+ Shares feeder fibre
- Active street electronics opex issues
WDM PON Logical point to point
+ WDM enables some fibre sharing
- Requires advanced WDM optics
- Nails wavelengths to customers very inefficient use of optical spectrum
No bandwidth sharing Separates access and metro networks
Keeps traditional architectures and cost structure
Keeps electronic nodes - either in exchanges, in street nodes or both
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
24/59
David Payne BT 2007
Cabinet
CabinetMSAN
Options for optical access
Passive Optical Network (PON)
Big
Business
CustomerLocal Exchange
FTTP
Customers
Non-FTTP
Customers
backhaul/metro
network2.5G
b/s
1.25/2.5Gb/
s
copper
~32 way
split
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
25/59
David Payne BT 2007
PON options
Passive Optical Networks:
BPON
GEPON
GPON
A-GPON
LR-PON
Features
+ Infrastructure sharing
+ Passive infrastructure no active street electronics
+ Enables dynamic bandwidth sharing
+ Lower cost - particularly extended/long reach systems+ PONs can be protected economically
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
26/59
David Payne BT 2007
Options for optical accessSummary 3: Claims and Pros & Cons of PON v Pt-Pt
Common Claim: Pt to Pt is more future proof than PON
Customers must share fibre bandwidth by multiplexing bandwidth from access fibres into core fibres.
Therefore customer bandwidth is always less than the fibre bandwidth!
All the fibre bandwidth can never be allocated to individual customers
No fundamental difference between the End User bandwidth upgrade potential for Pt to Pt or PON
architectures
PON is a more flexible architecture - can be considered to be the most future proof architecture.
PONs can support legacy protocols (TDM, ATM etc.) if required on a common transmission protocol.
PONs can support multi-casting/broadcasting natively (only one copy of a channel require for N customers).
Customers can access many tens of channels simultaneously without increasing access and backhaul bandwidthrequirements.
WDM can be used to up-grade PON capacity and user bandwidth to the limits of the fibre capacity.
The question becomes:
What is the simplest, lowest cost and lowest power consumption multiplexing technique?
Pt-Pt uses an electronic multiplexer with a port per customer on the access side. The equipment requirespower and is housed in a building or street cabinet
PON systems use passive optical multiplexing which consumes no power and can fit into external plant.
In practice both Point to Point and Passive Optical Networks will co-exist inthe network for different market situations. However PON will be the massmarket solution.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
27/59
David Payne BT 2007
Economic IssuesFTTP drives a bandwidth demand paradigm shift
New broadband services in particular HDTV/IPTVenabled by FTTP will massively increase demand onthe metro/core network
DSL has been successful at enabling good userexperience for many non-video Internet applications
But it will struggle to provide a future basket of services(e.g. HDTV, 3D applications, real time video, peer topeer file sharing including video etc.)
So currently were at the onset of a worldwide accessfibre program
What would be the impact of an access bandwidthexplosion driven by FTTP on the end to end network?
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
28/59
David Payne BT 2007
Bandwidth Growth
The Margin Challenge
0
10
20
30
40
50
60
70
80
90
100
2000 2001 2002 2003 2004 2005 2006
Revenues
Relative
Relativegrowth
Costs
0
10
20
30
40
50
60
70
80
90
100
0
10
20
30
40
50
60
70
80
90
100
2000 2001 2002 2003 2004 2005 2006
Revenues
Relative
Relativegrowth
Incremental Costs
0
10
20
30
40
50
60
70
80
90
100
2000 2001 2002 2003 2004 2005 2006
Costs
Relative
Relative
growth
Bandwidth
0
10
20
30
40
50
60
70
80
90
100
2000 2001 2002 2003 2004 2005 2006
Relative
Relative
growth
Bandwidth
1 2 3 4 5 6 7
1 2 3 4 5 6 7
But costs rise faster
Margins are eroded
Greater bandwidths- New services- Maintain/grow
revenues
years
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
29/59
David Payne BT 2007
The economics of current FTTP solutions do notlook attractive for mass market deployment.
Cost of bandwidth growth enabled by FTTP can
easily exceed revenue growth.
Can new network architectures enable FTTP andfuture bandwidth growth to become economicallyviable?
Must consider end to end network solutions:Access, Metro & Core together.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
30/59
David Payne BT 2007
BTs 21st Century Network
Inner Core PoPs
Outer Core PoPs
Metro PoPs
Access PoPs with WDM
Access PoPs
Electronic conversion whencrossing layer
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
31/59
David Payne BT 2007
Doubling the bandwidth more than doubles the electronics
Electronics cost reductions are not fast enough
Bandwidth cant become cheap enough whilst maintaining profit No long-term incentive to provide high bandwidth services
Too many exchange buildings and to much equipment
CapEx and OpEx too high
Environmentally unfriendly with large power usage
Network optimised around existing / legacy / known traffic / services
Not configured for content distribution or highly dynamic bandwidth
The architecture has to changeThe architecture has to change
Will this architecture scale economically with
FTTP?
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
32/59
David Payne BT 2007
Integrated access and backhaul network
Cabinet
Big Business
Customer
FTTP
Customers
Non-FTTP
Customers
copper
backhaul/metro
Long reach access.
10Gb/
s
2.5/10Gb/s
100km
CabinetMSAN
All ONUs have fixed band-pass blocking filterswhich only select the initial operating wavelength and
block all others.
When WDM is added at a later stage the new
ONUs will have a band pass filter at one of the
additional wavelengths
EDFA technology constrains operation
to C band wavelength range
~500 to 1000 way split
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
33/59
David Payne BT 2007
Integrated access and backhaul add circuit switchedoptical core
Cabinet
Big
Business
Customer
FTTPCustomers
Non-FTTP
Customers
copper
backhaul/metro
Long reach access.
10Gb/s
2.5/10Gb/s
100km
CabinetMSAN
All ONUs have fixed band-pass blocking
filters which only select the initial operating
wavelength and block all others.When WDM is added at a later stage the
new ONUs will have a band pass filter at
one of the additional wavelengths
EDFA technology constrains
operation to C band wavelength
range
~1000 way split
Intelligent
Photonic Inner
core Network
Metronodes
Opticalswitches
Intelligent
Photonic Inner
core Network
Metronodes
Opticalswitches
I t t d d b kh l ith h t i
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
34/59
David Payne BT 2007
Integrated access and backhaul with photonic coreLonger Term Vision Hybrid WDM/TDM + Flexible wavelength
assignment
Network Reduced to ~100 exchanges
IntelligentPhotonic Inner
core Network
Metronodes
Optical
switches
Cabinet
Big
Business
Customer
FTTP
Customers
CabinetCabinet
Big
Business
Customer
FTTP
Customers
Tunable & self install ONU (for residential customers)
backhaul/metro
Long reach access.
10Gb/s
10Gb/s
copper
CabinetMSAN
CabinetMSAN
Power splitter (notWDM splitter) to
enable any or
combination ofs
to any customer
New amplifier technology (e.g. quantumdot) enables operation across all the
fibre spectrum
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
35/59
David Payne BT 2007
Optics-centric architecture
Long reachWDM/TDM
PONaccess
~100
nodeall-
opticalcore
Access nodes replaced
by passive optics +optical amplifiers
Inner / outer core
hierarchy scrapped
No backhaul / metronetwork
All nodes both access
and core
E l i i f BT k
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
36/59
David Payne BT 2007
Evolution scenario of BT network
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
37/59
David Payne BT 2007
Operational & Environmental Benefits
21C MSAN
20 racks
50-100 kw
Long Reach PON
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
38/59
David Payne BT 2007
Amplified GPON
Adding amplifiers to GPON can be an interim solution for LR-PON
ONU
ONU
ONU
ONU
4x32-waySplit
4
X4
60km (100km)
128-way total split
Local ExchangeOLT W
Tx
Rx
OLT STx
Rx
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
39/59
David Payne BT 2007
Status of Amplified GPON
Extender box standardisation active topic in FSAN
Standard expected early 2008.
First development prototype of commercial amplifiedGPON system (Motorola) now in BTs Labs.
Blocking filter standard for WDM upgrade standardagreed.
Could see standards compliant systems commercially
available inside two years.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
40/59
David Payne BT 2007
Status of NGA Long Reach PON (LR-PON)
Principles demonstrated 4 years ago by BT
Active topic in NGA working group in FSAN
System components actively being developed incollaborative projects.
First demonstration prototype from major supplier(Nokia-Siemens) in BT Labs September 2007.
Evolution path from GPON to LR-PON determined.
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
41/59
David Payne BT 2007
Nokia-Siemens LR-PON Concept
512
subscribers Node
AccessExchange Access
ExchangeAccess
Exchange
AccessExchange
512 subscribers512 subscribers
512 subscribers
10Gbit/s
2.5Gbit/s
OLT
100km
(10G ready)
up to 2048 end userssharing backhaul fibre
E i i
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
42/59
David Payne BT 2007
Economic comparisons
Cash Flow
-1.00
-0.50
0.00
0.50
1.00
1.50
2.00
0 1 2 3 4 5 6 7 8 9 10 11 12
Years
B
illions
Option 3: Std GPON + 21cn GE B'haul
Option 0: Pt-Pt fibre GE B'haul
Option 3: Std GPON + 21cn GE B'haul
Option 4: Amplified GPON
Option 5: LR-PON
Option 8: VDSL Cab + 21cn GE B'haul
Comparison of time to breakeven v
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
43/59
David Payne BT 2007
Sustained Cust B'wdth v Time to Breakeven
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
0.1 1 10 100
B'width Mb/s
Timeto+vec
ashflow(years)
Std GPON
A-GPON
LR-PONVDSLCab+21cn
Comparison of time to breakeven v
sustained end user bandwidth
Assumes no correlation of revenue with bandwidth + self install ONT
VDSLCabupstream
bandwidthexceeded
VDSLCabdownstreambandwidthexceeded
Time to breakeven v PON split size
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
44/59
David Payne BT 2007
Time to breakeven v split
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
10 100 1000
PON split size
Timet
obreakeven
LR-PON - -
A-GPON
Std GPON
Time to breakeven v PON split size
End user sustained bandwidth ~10 Mb/s
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
45/59
David Payne BT 2007
Optical access & metro network
Summary 4 Todays optical access and separate metro/backhaul
network solutions not economic for mass deployment. Cost of bandwidth growth would exceed any revenue growth.
Need to reduce amount of electronic equipment andnumber of network nodes.
Achieved by combining access and metro/backhaul.into one network
Long Reach PON proposed as possible solution.
Amplified GPON as interim solution. LR-PON economics look promising, FTTP can be competitive with
FTTCab particularly as end user bandwidths increase.
Impact of FTTP & long reach access on core
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
46/59
David Payne BT 2007
Impact of FTTP & long-reach access on core
Some orders of magnitude:
~10 million customers, 100km long-reach access, 10 Mb/ssustained rate busy hour
~100 core nodes
Total sustained traffic = 100 Tb/s
Traffic per node = 1 Tb/s
If equally distributed amongst core nodes
Average traffic between each set of node pairs = 10 Gb/s Assuming little turn-around traffic, no protection, no bandwidth contention, etc.
Assign 1 between all node pairs
No need for grooming / intermediate electronics.
Long-reach access core Metro/core-node structure
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
47/59
David Payne BT 2007
Long-reach access core Metro/core-node structure
Metro/core node
Ave 100,000 customers
100km
10km
500 per PON
WDM backhaul
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
OLT
Averag
e200OLTs
Average
100lambdas
To othernodes
O
ptical
switching
Duct routes out ofexchange
Aggregation
/routing
Effect on Locations for Head-Ends (OLTs)
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
48/59
David Payne BT 2007
Effect on Locations for Head Ends (OLTs)
UK With All Exchanges UK with~100 21cn nodes
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
49/59
David Payne BT 2007
Numbers of nodes bypassedTraffic Paths
0%
2%
4%
6%
8%
10%
12%
14%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Physical Hops
Assumes 1 betweenall node pairs
Significant amount ofnode bypass
Therefore need cheap
large scale node bypassoptics to make thisarchitecture cost in
Potential future UK all-optical core network
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
50/59
David Payne BT 2007
All-optical meshed core
Light-paths set up betweenall core nodes
Optical path bypassesintermediate nodes using noelectronics
Potential future UK all optical core network
Flat versus hierarchical core
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
51/59
David Payne BT 2007
Flat versus hierarchical core
Full logical mesh between all routers to create a flat core
Assumes routers/switches can handle connectivity to ~100 nodes
Optical switching to allow for flexibility, meshed restoration, etc.
Optical interfaces much cheaper than router interfaces
Router/switches smaller do not need to handle through traffic
Hierarchical core allows for intermediate electronic grooming forlimited traffic levels
inner core
outer core
Router
OXC?
Router
OXC?
Router
OXC?
Router
OXC?
flat core
Router
OXC
Router
OXC
Router
OXC
Router
OXC
Router
OXC
Router
OXC?
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
52/59
David Payne BT 2007
Flat vs hierarchical coreNetwork Hierarchy Comparsion
0
10000
20000
30000
40000
50000
60000
0 20 40 60 80 100 120 140 160 180 200
Network Traffic (Tb/s)
LineCards
Two-Tier Flat
2 tier modelassumes 20 innercore nodes
Flat architecturepreferred whenwavelength fill >40%
Network node dimensioning
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
53/59
David Payne BT 2007
Network node dimensioning
Nodal Capacity
0
500
1000
1500
2000
2500
3000
3500
Nodes
Total Wavelength Ports Utilised Wavelength Ports
Assumption of long reach access + flat all-optical core leads to huge coretransport + very large optical switches
Optical switch capacity up to 40 fibre pairs, assuming 80 channel DWDM
Up to 20 fibre pairs on busy links
Possible new optical core node architecture
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
54/59
David Payne BT 2007
p
Fibre switch
Wavelength switch
Electronic
(IP / Ethernet)
To other nodesTo other nodes
Scaleable to many fibre pairs Fibre switch enables fibrebypass of node. eg. NxN MEMs
Some fibres diverted towavelength switch for localnode wavelength add/drop
and also possible wavelengthgroomingTo access side
of node
Wavelength switch could use
WSS technology
Cascading nodes leads to requirement for ultra low loss
How do we provide such huge optical
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
55/59
David Payne BT 2007
How do we provide such huge optical
bandwidths? How do we optimise split between Electrical switching,
Wavelength Division Multiplexing (WDM) and Fibre DivisionMultiplexing (FiDM)?
Low fibre count solution - 40Gb/s, 100Gb/s, C-L-Sbands, highspectral efficiency
Eggs in one basket
Multiple fibres Cheaper simpler 10G transmission
More lambdas more optical switching flexibility
Multiple line systems = more line equipment
More resilient?
Many optical technologies presented at conferences OFC, ECOCetc over last 10 years now need to be considered.
Economics and resilience key to deciding best approach
Optical Core
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
56/59
David Payne BT 2007
Summary 5 Fibre access enables unprecedented bandwidth growth.
Challenging economics needs new architectures
the bandwidth growth will also pressurise the core
All-optical flat architecture with no separate metro networkprovides economic scalability.
In the UK this equates to:
~ 100 core nodes with over 10Tb/s on some links
Up to 1500 10G wavelengths bypassing nodes
Large optical switches at fibre and wavelength level will be needed
High bit rates / multiple wavebands (C,L,S) will all be required
Economics and resilience at the fibre, cable and duct level willdetermine optimal design
Technology and evolution timeline
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
57/59
David Payne BT 2007
gy
2008 2012
GPON
PoweredCabinets
amplifiedGPON(60 km)
10 Gbit/sLR-PON(100+ km)
Non-
Greenfieldaccess
Backhaul
Greenfield
access
Flexible LR-PON+10Gbit/s
+ scaleprotocol to1024 split
+WDMinbackhaul
WDMLR-PON
+colourlessONUs
+tunableoptics
Ethernet
GP
ON
WDM
SDH
PIEMAN
Overall Summary
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
58/59
David Payne BT 2007
Overall Summary
We have the opportunity with the all optical networkof radically changing the architecture to:
Greatly reduce the capital expenditure required for a true broadbandnetwork
Massively reduce operational costs.
Produce an environmentally friendly network
>90% reductions in energy consumption. Revolutionise the customer experience.
Click and its there!
Provide a reusable and continuously upgradeable physical
infrastructure for the Next Generation 21st Century Network withtrue broadband capability!
7/31/2019 All Optical Networks Presentation - David Payne - 27.09.07
59/59
David Payne BT 2007
Thank you