Revolutionary Communications
Transcript of Revolutionary Communications
RevolutionaryRevolutionaryCommunicationsCommunications
Will StewartChief Scientist, Marconi
Communications Traffic Forecast (1996)-2005
0
2000
4000
6000
8000
10000
12000
14000
CAGR1996-2005
Internet 95.8%
Voice over IP30%
Data Traffic30%
Circuit Switched 12.1%
Ter
abyt
es/D
ay
Source : Renaissance Analysis
1994 1996 1998 2000 2002 20042000
5000
10000
20000
50000
100000.
200000.
MegaMinutes pa
Year
Mobile
Fixed
UK Mobile (voice) growth, OfTEL
How much data?
About 40,000 movies have been released in the US - at 4GigaBytes/movie (DVD) thisis about 1015 bits = 1000 TeraBit ~ 1 MovieBrain
A 10 Terabit (1013)/sec optical linkwill transmit this in about 2 minutes!
1 MB/minute MP3 file
Service Provider RevenuesService Provider Revenues
$$bnbn
YearYear
9993 94 95 96 97 0098 0201 0403
Switched Voice RevenuesSwitched Voice Revenues
Data & IP Revenues Data & IP Revenues
Source: DataquestSource: Dataquest
Service Provider Revenues/2Service Provider Revenues/2
$$bnbn
YearYear
9993 94 95 96 97 0098 0201 0403
Switched Voice RevenuesSwitched Voice Revenues
Data & IP Revenues Data & IP Revenues
Source: DataquestSource: Dataquest
New Data Service RevenuesNew Data Service Revenues
IP Revenues IP Revenues
Switched Voice RevenuesSwitched Voice Revenues
Source: RHK
New Public Network Architecture
Internet
CorporateData
PSTN MobileCable
Intelligent OpticalTransport & Switching
BroadbandAccess
NPNEnhanced Service &
Connection “servers”
Key Issues
• Transport capacity -> Photonics• Switching -> shift functions to the optical layer (transparency)• Mobility -> Mobile, IP
• Control moving to the edge• IP quality of service -> MPLS/MPlamS• Integration with legacy systems
NPN Applications
DEMAND FOR:
Application Class
Sp
ee
d/u
se
r
Lo
w L
ate
nc
y
Lo
w D
elay
Po
ten
tia
l u
se
r b
as
e
AG
R
Ne
two
rk i
nfo
Se
cu
rity
Ne
two
rk C
on
tro
l
Telephone 64k L H Billions 12% Low M NASP, remote support, (Luminate, Xdrive, Medical) ? H H 100's k 109% HVideo/TV (BBC..) 10-250M L no 100's M LRadio (Live 365, realplayer) 64k-? L L 100's M LInternet 50k-10M H M 10'sM 96% LGames 50k-10M H M 100's M 100% MOnline Auctions (Ebay, Dovebid) 64k-10M H H He-commerce ? H M 100'sM VHExotic Caching (Akamai, ValueClick) H H H H MGames support (Powerplay) H H 100'sM H M HGrid Computing/Thin client (Sychron) ?1-10M M-H M ? H M MFreecalls 64k L H Billions H M LPositional data/WAP 10-300k H M 100's M 14%? H L L
Deployment Trend of SONET/SDHand WDM
SONET/SDH
WDM
Access/ Edge
Largestnetworks
Fibre
Core
Mediumsized
networks
“Metro”
Wave CharacteristicsWave Characteristics
Wavelength -> size of antenna
Frequency - > ability to carry information
velocity
750 600 500 430
Frequency,Frequency, TeraHertz TeraHertz
0.4 µm
Wavelength, µmWavelength, µm
0.55 µm 0.7 µm
The Electromagnetic Spectrum
100 MHz
Atmospheric absorptionFog
Fibre
10 GHz
2.7°K microwave background
hot bodyRadio 4
Sunlight
1PHz10 THz
1000
100
10
1
0.1
dB/km attenuation
-2 -1 0 1 2 3-2 -1 0 1 2 3
Electromagnetic Information
10 MHz 1PHz
Atmospheric absorptionFog
Fibre
10 GHz
Log scales
Linear Frequency scale
100100 THz THz
1985 1987.5 1990 1992.5 1995 1997.5 2000Year
50
100
150
200
250
300
1985 1990 1995 2000 2005
Year
1
10
100
1000
10000
Total Rate, GBitês
Factorêyr = 1.72209
Optical Fibre Capacities
Growth rate ~ *1.7 per year
1
0.1
0.01
Ter
abit/
sec/
fibr
e
Number of channels
Channel Rate
1985 1990 1995 2000 2005Year
10
100
1000
Channel Rate, GBitês
Factorêyr = 1.60061
Spectral Efficiency Bits/sec/Hz
1992 1994 1996 1998 2000 2002 2004Year
0.02
0.05
0.1
0.2
0.5
BitsêHz
Factorêyr = 1.58227
Rate v range
100 200 500 1000 2000 5000 10000Range, km
200
500
1000
2000
5000
10000
Total Rate, GBitês
Log Gradient = -0.285356
Moore's Law
year1970 1975 1980 1985 1990 1995 2000
1 0 3
1 0 4
1 0 5
1 0 6
1 0 7
1 0 8T
rans
isto
rs p
er c
hip
after G Moore, Intel
Growth rate ~ *1.4 per year
New Optical Networks
The growth in opticalfibre capacity (60-80%pa) is outstrippingThe (Moore’s law)growth in processorspeeds (~25% pa andfalling)*
*classic x2/18 months includes increasing chip sizes!
1975 1980 1985 1990 1995 2000 2005 2010Year
0.1
10
1000
TotalRate, GBitês
Optical fibre
Silicon
Sources: fibre: reported data major conferences Silicon: International roadmap 1999
New Optical Networks
This led in the mid-90’sto the rise of WDM:-
1975 1980 1985 1990 1995 2000 2005 2010
Year25
50
75100
125
150
175
200
Source: fibre: reported data major conferences
Increasing channels/fibre
2000 2002 2004 2006 2008 2010
10
20
50
100
200
500
performance êcost
Disk Drive (Philips)
Single optical Fibre(conference data)
Silicon Performance (International Roadmap)
DRAM (Philips)
Trends Performance v cost for Fibre and Silicon
Dense WDM for switching
Switch
Switch
Transport Transport
But now the processorsin switches cannot copewith the fibre datadelivery rate and someof the the switchingfunctions are movinginto the optical regime:-
Dense WDM for routing
MPLS(Lambda)
Convergence - IP / MPLS / WDM
Fibre
SwitchingSwitching
SignallingSignalling
Call controlCall control
ATM
IP
ServicesVoice Data
SDH
MPLS(Label)
WDM
Off-loading Switched Data Trafficonto the Managed Photonic Layer
IP IP
IP cloud
IP
Direct
Transit andrestoration
traffic: 60-70%
Reducingnetwork costs
Transport cloudWDM
Switched glass &air cheaper thansilicon & copper!
New Optical Networks
• Control MPLS• Lasers• Detectors
• Electrical XCs• Optical ADMs
• SDH networks
New Technologies:-
• MPLambdaS• tunable lasers• tunable detectors• wavelength translators• Optical XCs -> wavelength-selective OXCs• Optical ADMs
• Static WDM overlayed SDH ->Dynamically reconfigurable SDH/WDM
VCSEL
Modulator
Tunable laser
Silicon Technology
Frequency, THz
Scal
e, µ
mScale v speed for Electronics & Scale v speed for Electronics & PhotonicsPhotonics
Er band
Yariv
GaAs Modulator RF Response
-9
-6
-3
0
3
0 10 20 30 40 50 60
Frequency (GHz)
Op
tica
l Mo
du
lati
on
(d
B)
GaAs Modulator Modelling
GaAs Modulator Measured
LiNb03 Modulator
0 15 20 45 60 75
Gbit/s data transmission
GaAs/AlGaAs Mach-Zender Modulator
RFConductive n-doped layer
S.I. Substrate
Fiber Grating Holographic Fabrication Technique
Performance of actual filters
Wavelength (nm)
-40
-30
-20
-10
01545.32 1546.12 1546.92 1547.72 1548.52 1549.32 1550.12
Tra
nsm
issi
on
(d
B)
Fiber Grating (100 GHz)
Bulk Diffraction Grating (100GHz)
Dielectric filter
3D MEMS OXC
• 256 ports > 1,000 ports
Fibre arrays
Mirror arrays
SG-DBR Tunable Laser Source
Op
tica
l Ou
tpu
t P
ow
er (
mW
in f
ibre
)
Module 9831L Tuning Comb; Superimposed Spectra
-60
-55
-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
1525 1530 1535 1540 1545 1550 1555 1560 1565 1570 1575
Wavelength (nm)
Laser StructureLaser Structure
Principle of Optical Cross-connect Basedon Tuneable Laser and Passive Router
TPTPTP
Transponder
Tuneable laser
TPTPTP
Waveguide Grating Router/ NxN AWG Router
Routes inputsto outputs
according toinput port
number andinput
wavelength
#1 #1
#8 #8
Passive, reliable,piece of glass!
Minimum IC structure dimensions
(as reported at the IEEE SSC conference)
1985 1990 1995 2000 2005
0.05
0.1
0.5
1
5
10
0.05
0.1
0.5
1
5
Min
imum
dim
ensi
on, µ
m
YearAfter Alec Broers
Forecast average
Range for optical microstructures
Zoom MovieZoom Movie
All actual pictures of electronic devices(movie by Richard Beanland)
pbg integration technology
High confinement implies dense circuits without crosstalk -- needed to match electronic densities and switch performance
Hole Array (Glasgow)
Self-assembled PyramidArray (Imperial College)