All Optical Networks Presentation - David Payne - 27.09.07

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


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


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


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


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


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.


Summary 1


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


7/59

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


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)


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


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


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.


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.


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


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



End user 1

End user 2

End user 3

End user 5

End user 4


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


15/59

David Payne BT 2007


End user 1

End user 2

End user 3

End user 5

End user 4


CPs

Burst-mode

Streaming mode


All spare bandwidth isused to deliver files as fast

as possible

Burst-modeUsed bandwidth

profile

Spare bandwidth

shared betweenactive users


End user 1

End user 2

End user 3

End user 5

End user 4


CPs

Burst-mode

Streaming mode


All spare bandwidth isused to deliver files as fast

as possible

Burst-modeUsed bandwidth

profile

Spare bandwidth

shared betweenactive users


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


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


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?


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


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.


21/59

David Payne BT 2007


Big

Business


FTTP

Customers

Non-FTTP

Customers

backhaul/metro

network

Point to point fibre active cabinet.

Cabinet

Active star

Cabinet

Active star


22/59

David Payne BT 2007

Cabinet


Big

Business

Customer

StreetMSAN

Non-FTTP

Customers

copper

backhaul/metro

network

FTTP

Customers

WDM PON.

Local Exchange

WDM

Splitter

WDM

Multiplexer


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


24/59

David Payne BT 2007

Cabinet

CabinetMSAN


Passive Optical Network (PON)

Big

Business


FTTP

Customers

Non-FTTP

Customers

backhaul/metro

network2.5G

b/s

1.25/2.5Gb/

s

copper

~32 way

split


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


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.


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?


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


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.


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


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?


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


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


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


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


36/59

David Payne BT 2007

Evolution scenario of BT network


37/59

David Payne BT 2007

Operational & Environmental Benefits

21C MSAN

20 racks

50-100 kw

Long Reach PON


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


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.


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.


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


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


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


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


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


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


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)


48/59

David Payne BT 2007

Effect on Locations for Head Ends (OLTs)

UK With All Exchanges UK with~100 21cn nodes


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


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


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?


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


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


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


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


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


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


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!


59/59

David Payne BT 2007

Thank you

Date post:	05-Apr-2018
Category:	Documents
Upload:	nhat-minh
View:	216 times
Download:	0 times