Documente Academic
Documente Profesional
Documente Cultură
by E.Desurvire
ALCATEL
One day there will be a telephone in every major city in the USA
A.G.Bell, c.1880
Outline
Technology
> capacity X distance (Moore) > fiber bandwidth > channel capacity (Shannon)
Conclusion
N PO B HF C FTTH
VD SL
N GPO
last mile
FITL
NY Broadway 1890
Source: http://www.uh.edu/engines/epi1640.htm
Optical communications in 2025 4
NY Broadway 1910
NY Broadway 2025 ?
From today to 2025, the main continental traffic routes to remain unchanged But P2P traffic intensity to increase orders of magnitude Optical networks to become more geographically pervasive with proliferation of local L2-L3 DXC/OXC/IProuters ..
1 Pbit/s
voice
8%/Y
Gbit/s
1 Tbit/s
115%/Y
1 Gbit/s
1998 2000 2002 2004 2006 2008 2010 2012
Source Corning International Traffic Model / LIGHTWAVE July 2000, see http://www.corning.com/docs/opticalfiber/r3461.pdf
Optical communications in 2025 6
So far, the installed capacity (eg. Atlantic) provides sufficient (x2) margin to satisfy needs up to 2007 With current submarine-cable technology, a single 5Tbit/s protected-traffic system can fully handle such a need But the situation will be very different a few years from now.. !!
2004
2005
2006
2007
120
Atlantic IP traffic and installed capacity (Tbit/s)
80 60 40 20 0 2005
115% 25%
20 %
100
50%
The scenarios of 50-115% internet traffic growth are catastrophic in a 3.5-7 years horizon ! A 25%-only Internet traffic growth should make it >>> But is it a scenario we can believe for the next 20 years ?
2010
2015
2020
2025
504.5 Gbit/s
181.4Gbit/s 66.3Gbit/s
Capacity (Tbit/s)
1999
2000
2001
2002
2003
2004
2005
2006
Capacity (Tbit/s)
70 60 50 40 30 20 10 0 2005
38-fold
increase from 2005 traffic
2010
2015
2020
2025
2004
2025
0.5 Tbit/s
95 Tbit/s
0.18 Tbit/s
47 Tbit/s
0.066 Tbit/s
25 Tbit/s
World total
80 60 40 20 0 1995
US
9%
recovery 17%
1997
1999
2001
2003
2005
2007
bubble pops
* and 1997-2007 data:
KMI Research, 2002
fiber glut
100
0 1995
2000 2005
2010 2015
2020 2025
23%/Y
Yields 70M home passed in 2010, or 17-23M subscribers New installed fiber : 245,000 km
(3.5m/home)****
60 50 40 30 20 10 0 2002
Japan
2004
2006
2008
2010
2012
Sources: *Yankee Group, sept. 2004; **FTTH Council, 2004; *** Press/announcements; **** =38,8000 miles /18 million homes
Optical communications in 2025 16
Homes passed : 1.52 Billion (19%) 555 Million (7%) 285 Million (3%)
(% of 2025 world population*)
1,520
1000
23%/Y
100
New fiber installed (max): 5.32 Mkm Total 2025 traffic (max):
(100Mbit/s @50%)
10
15%/Y 10%/Y
555 285
19-25 Pbit/s
? 10 every 4 years
PETA
Raman FEC (>1999) DM, C+L WDM EDFA coherent 1.5m DSF 1.3m SMF 0.8 m MMF 82 86 90 94 98 02
TERA GIGA
108
MEGA
78 06
year
(1/2)
C? D ? 5.178 ? 2
year
2005
Y ? 2000 1.053
Pbit .km/ s
2013 2017 2021 2025
2009
139 0.01
1936 0.19
26,951 2.69
375,055 37.5
1,E+06
disruptive technology
100Tbit/s
incremental technology
C? D PbitKm /s ? 10 ? 2
3.75 years = 45months
Y 3.75
1,E-08 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025
Year
Optical communications in 2025 20
100 Tbit/s over 10,000 km 1 Pbit/s over 1,000 km 10 Pbit/s over 100 km
Tbit/s =
Key question: Can 2005 Lightwave Communications experts confirm right now wether such performance is theoretically possible ?
Fiber bandwidth
Attenuation coefficient (dB/km) 400nm=54.5 THz Raman 0.4 10x more dB loss per 100km OH peak 200nm=23.5 THz Raman minimal attenuation = minimal ASE U/XL ? (nm) 1200
25nm (1295-1320) 80nm (1340-1420) E.Desurvire et al., J.Wiley & Sons, Inc., 2004
Optical communications in 2025 22
bending loss
0.3
EDFA E S C L 1600
40nm (1575-1615) 60nm (1420-1480) 30nm (1530-1560)
0.2
1300
1400
1500
1700
60nm (1630-1690)
superband B
(dB/km)
0.4 0.3 0.2 200nm=23.5 THz
superband A
? (nm)
1200 1300 1400 1500 1600 1700
Assumption: Two fiber pairs of priority traffic (2X usable fiber capacity) + 2:2 protection
Phase I, 2015: superband A (about 25THz): 50 Tbit/s RZ Phase II, 2025: superband B (about 50 THz): 100 Tbit/s RZ Superband capacity upgrade: M-ary formats (M=4,8,16..)
10 bit/s/Hz
0,8
QAM
M= 128 64 32 16 8
Effect of modulation format QAM and PSK M-aries make possible to reach ISDs much greater than 1 bit/s/Hz But lots of signal levels ! (e.g. >256 for 10 bit/s/Hz) And coherent receivers required, making implementation complex and demanding very high fiber linearity (AM, PM) High ISD requires high SNR
0,6 0,4
ink th it n ve t t e bou n a do
4
PSK
1 bit/s/Hz
IM-DD
FS K
BER = 1 0 - 3 to 10 -11
16
18
20
22
24
26
CONCLUSION (1/2)
Traffic growth over the next 20 years phenomenal compared to the past 30 years (+115%/Y world average) But serious problem of installed capacity exhaustion in 3.5-7 years ? Ultra-conservative scenario: - 20% installed capacity growth for undersea links - 25% Internet growth > yields rock-bottom need for 50-100Tbit/s per fiber pair in 2025 > Implementation needs superbands concept phase I/II (yet in two priority-traffic fiber pairs) > 10/20-y challenge: achieve 50-100THz commercial systems
CONCLUSION (2/2)
Optical (fiber) bandwidth is not infinite Optical Moores law is now meaningless and misleading 20-y objectives can only be reached though tech-driven research and there is an urgent need to get started
Disclaimer :
This presentation being essentially speculative, conclusions may not be used towards any business purposes
purely linear
2 3
14 12 10
SNRmax ?
SNR
2 ? Pth ? 3 ? 2 PN ?
model 1
? ? ? ?
The capacity of the optical channel cannot be indefinitely increased with signal power since SNR is bounded by a maximum limit which is function of the nonlinearity threshold of the channel
80
90
100