OPTICAL NETWORK SCALING - University of Rochester...• Optical transport has run out of physical...

COPYRIGHT © 2013 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

OPTICAL NETWORK SCALING SPATIAL AND SPECTRAL SUPERCHANNELS Peter J. Winzer Optical Transmission Systems and Networks Research Bell Labs, Alcatel-Lucent, USA


ACKNOWLEDGMENT Bell Labs S.Chandrasekhar A.R.Chraplyvy C.R.Doerr R.-J.Essiambre N.K.Fontaine G.J.Foschini A.H.Gnauck H.Kogelnik G.Kramer X.Liu S.Randel G.Raybon R.Ryf M.Salsi R.W.Tkach C.Xie

Optical Fiber Solutions D.DiGiovanni J.Fini L.Gruner-Nielsen R.V.Jensen X.Jiang R.Lingle A.McCurdy D.Peckham T.F.Taunay S.Yi B.Zhu

Sumitomo Electric T.Hayashi M.Hirano Y.Yamamoto T.Sasaki T.Taru

AND MANY OTHERS


OPTICAL NETWORKS SCALING CHALLENGES IN A NUTSHELL

1986 1990 1994 1998 2002 2006

10

100

1

10

100

Seri

al In

terf

ace

Rate

s an

d W

DM

Cap

acit

ies

Gb/

s Tb

/s

2010 2014 2018 1

2022

• How can we scale to Petabit capacities ? • WDM systems available up to ~20 Tb/s, saturating at ~50 Tb/s

[P. J. Winzer, IEEE Comm. Mag. 26-30 (2010)] ?

3

?

• How do we get to Terabit channels ? • Interface rates available up to 200 Gb/s per carrier


SCALING LINE INTERFACE RATES

Space

Polarization Frequency

Time Quadrature

Physical dimensions

1986 1990 1994 1998 2002 2006

10

100

1000

Seria

l int

erfa

ce r

ates

[G

b/s]

2010 2014 2018 1

2022

107-Gbaud on/off keying

• 100-Gbaud not widely supported by electronics • Intolerant to CD (~R2) and PMD (~R) • Doesn’t scale WDM spectral efficiency

[Winzer et al., ECOC 2005]

CD: Chromatic dispersion PMD: Polarization-mode dispersion

4



Space


Time Quadrature

Physical dimensions

1986 1990 1994 1998 2002 2006 2010 2014 2018 2022

56-Gbaud DQPSK

2x parallel (I/Q)

Seria

l int

erfa

ce r

ates

[G

b/s]

• 100-Gbaud not widely supported by electronics • Intolerant to CD (~R2) and PMD (~R) • Doesn’t scale WDM spectral efficiency Exploit phase dimension (2x parallelization)

[Winzer et al., ECOC 2006]

10

100

1

1000

5



Space


Time Quadrature

Physical dimensions

1986 1990 1994 1998 2002 2006 2010 2014 2018 2022

2x parallel (I/Q)

2x parallel (pol.)

[C.R.S.Fludger et al., OFC 2007]

Seria

l int

erfa

ce r

ates

[G

b/s]

Polarization multiplexing lowered symbol rates to ~10 Gbaud (40G) and ~25 Gbaud (100G) Can use fast (CMOS) A/D converters Digital coherent (intradyne) detection

10

100

1

1000

6


P M


Space


Time Quadrature

Physical dimensions

1986 1990 1994 1998 2002 2006

10

100

2010 2014 2018 1

2022

107 Gbaud QPSK

1000

Seria

l int

erfa

ce r

ates

[G

b/s]

107 Gbaud 16-QAM Current state of research:

• 856-Gb/s PDM-16-QAM (107-Gbaud)

[G. Raybon et al., ECOC (2013)]

7


DIGITAL COHERENT DETECTION 2x2 MULTIPLE-INPUT MULTIPLE-OUTPUT (MIMO)

2 x 2 MIMO DSP engine

PBS

x

y Polarization rotation Signal 1

Signal 2

Axxx + Axyy

Ayxx + Ayyy

PBS Fiber

8

MIMO DSP: Multiple-input multiple-output digital signal processing

• Digital polarization de-rotation

• Digital frequency & phase locking

• Digital impairment compensation


[Doerr et al., OFC 2009]

Monolithic Silicon Coherent Receiver

MODERN OPTICAL TRANSPONDERS LOTS OF DIGITAL PROCESSING

ADC ADC

ADC ADC

DAC DAC DAC DAC

Receive DSP

Transmit DSP

Client Interfaces

PDM I/Q-MOD

TX Laser

PDM Coherent Receiver

LO Laser

FEC Enc.

FEC Dec.

9

FEC: Forward error correction PDM: Polarization-division multiplexing DSP: Digital signal processing DAC: Digital-to-analog converter ADC: Analog-to-digital converter

t

f

F


CONSTELLATION SIZE VERSUS SYMBOL RATE

1 2 3 4 5 6 7 8 9

100

200

300

Bits per symbol

Line

rat

e [G

b/s]

16-QAM

[Winzer, J. Lightwave Technol. 30, 3824 (2012)]

PDM 512-QAM 3 GBaud (27 Gb/s) [Okamoto et al., ECOC’10]

PDM 256-QAM 4 GBaud (32 Gb/s) [Nakazawa et al., OFC’10]

PDM 64-QAM 21 GBaud (128 Gb/s) [Gnauck et al., OFC’11]

PDM 16-QAM 80 GBaud (320 Gb/s) [Raybon et al., PTL’12]

PDM QPSK 107 GBaud (224 Gb/s) [Raybon et al., PTL’12]

ADC & DAC resolution Bandw idth

10

[R. H. Walden, JSAC (1999), and Proc. CSIC (2008)] [A. Khilo et al., Opt. Ex. (2012)]


TERABIT INTERFACES AND BEYOND OPTICAL SUPERCHANNELS

Space


Time Quadrature

Physical dimensions

1 Tb/s (2 subcarriers) 16-QAM 5.2 bit/s/Hz

3200 km transmission [Raybon et al., IPC’12]

263 GHz

1.5 Tb/s (8 subcarriers) 16-QAM 5.7 bit/s/Hz

5600 km transmission [Liu et al., ECOC’12]

1.2 Tb/s (24 subcarriers) QPSK 3 bit/s/Hz

7200 km transm. [Chandrasekhar et al., ECOC’09]

300 GHz

Electronics complex ity Optics complex ity

200 GHz

1 Tb/s (4 subcarriers) 16-QAM 5.0 bit/s/Hz

2400 km transmission [Renaudier et al., OFC’12]

200 GHz

11


TERABIT INTERFACES AND BEYOND OPTICAL SUPERCHANNELS

Space


Time Quadrature

Physical dimensions

Electronics complex ity Optics complex ity

Mod.

f

Mod.

Mod.

Mod.

Laser f

+ Laser

Laser

Laser f

f

f

• Cohesive spectral entity • Parallelization requires integration

DWDM Superchannel

12


SUPERCHANNEL TRANSPONDER INTEGRATION SPECTRAL SUPERCHANNEL

DAC TX DSP

PDM IQ-MOD

DAC DAC DAC DAC DAC DAC DAC DAC DAC DAC DAC

ADC ADC ADC ADC ADC ADC ADC ADC ADC ADC ADC ADC

PDM IQ-MOD

PDM IQ-MOD

+

PDM IQ-MOD

PDM IQ-MOD

PDM IQ-MOD

TX DSP

TX DSP

RX DSP

RX DSP

RX DSP

Client Interface

RX DSP

TX DSP

[H. Yamazaki et al., Opt. Ex. 19, B69 (2011)]

10 x 14.25 GBd PDM-QPSK Transmitter

[F. A. Kish et al., JSTQE 17, 1470 (2011)]

13

Laser

Laser

Laser

Laser

Laser

Laser


SUBCARRIER ADD/DROP IN SUPERCHANNELS AN OPTO-ELECTRONIC INTERFEROMETER

[P. J. Winzer, J. Lightwave Technol. 1775 (2013)] (earlier patents by Taylor, Morita, Winzer)

10 dB 10 dB 10 dB

RS RS RS

Frequency Frequency Frequency

Spe

ctru

m [d

B]

Optical input Optical output

Optical processing

Optical input Optical output

Digital electrical processing

14


DSP

SUBCARRIER ADD/DROP IN SUPERCHANNELS AN OPTO-ELECTRONIC INTERFEROMETER

[P. J. Winzer, J. Lightwave Technol. 1775 (2013)] (earlier patents by Taylor, Morita, Winzer)

10 dB 10 dB 10 dB

RS RS RS

Frequency Frequency Frequency

Spe

ctru

m [d

B]

fi

Sin(f)

f

Sout(f)

fi f

90-d

eg.

hybr

id

Laser

A/D

I

Q

fL≈ fi

Subc

arrie

r m

odul

atio

n

Add data input

B si(t)

Drop data output

si(t)

Subc

arrie

r de

mod

ulat

ion

A Subcarrier extraction

si(t)

fi f

π/2

D/A

I/Q modulator

I Q

15


OPTICAL NETWORKS SCALING CHALLENGES IN A NUTSHELL

1986 1990 1994 1998 2002 2006

10

100

1

10

100

Seri

al In

terf

ace

Rate

s an

d W

DM

Cap

acit

ies

Gb/

s Tb

/s

2010 2014 2018 1

2022

• How can we scale to Petabit capacities ? • WDM systems available up to ~20 Tb/s, saturating at ~50 Tb/s

?

16

?

• How do we get to Terabit channels ? • Interface rates available up to 200 Gb/s per carrier

[P. J. Winzer, IEEE Comm. Mag. 26-30 (2010)]


THE (LINEAR) SHANNON LIMIT SPECTRAL EFFICIENCY VS. SIGNAL-TO-NOISE RATIO

• For every bit of higher spectral efficiency, ~3 dB more SNR are needed [C. E. Shannon, BLTJ (1948)]

0 5 10 15 20 25

1

10

Required SNR per bit [dB]

Spec

tral

eff

icie

ncy

per

pola

rizat

ion

[b/s

/Hz]

3.7 dB 8.8 dB

2x

2x

16-QAM turns out to be a good compromise

16

64

4

256

17

SE = log2 (1 + SNR)


[Karlsson & Agrell, ECOC’10]

[Cai et al., ECOC’10]

ISI

MAP

APPROACHING SHANNON IN OPTICS 4D, CODED MODULATION, SHAPING, MAP DETECTION

0 5 10 15 20 25

1

10

Required SNR per bit [dB]

16-QAM

64-QAM

4-QAM

256-QAM Constellation shaping

Over-filtering & MAP Coded modulation

[Liu et al., OFC’12]

Spec

tral

eff

icie

ncy

[b/s

/Hz]

ISI: Inter-symbol interference; MAP: Maximum a posteriori

[P. J. Winzer, JLT, 2012]

18


5

10

15

2 100 1,000 10,000

Transmission distance [km]

Spec

tral

eff

icie

ncy

[b/s

/Hz]

<2x

THE NONLINEAR SHANNON LIMIT COMMERCIAL WDM PRODUCT NEEDS

30 … 60% traffic growth per year

Commercial WDM needs ca. 2016 … 2018

Current WDM products

19

[R.-J. Essiambre et al., J. Lightwave Technol. 28(4), 662-701 (2010)] [P. J. Winzer, Phot. Technol. Lett. 23, 851 (2011)]


BETTER FIBER: NOT A LONG-TERM SOLUTION

Capacity is fairly insensitive to (heroic!) improvements of fiber loss, nonlinearity, or dispersion Improving single-mode fiber is not a long-term solution

[R.J.Essiambre and R.W.Tkach, Proc. IEEE, 2012]

20


PHYSICAL DIMENSIONS IN USE TOMORROW

Polarization

Time Quadrature

Physical dimensions Frequency

Space

21

Q Q Q Q Q Q

Courtesy: X. Liu

t

f

F

O E S C L

1300 0

1400 1500 1600

0.3

0.6

0.9

1.2

Wavelength [nm]

Loss

[dB

]

1360 1260 1460 1530 1565

1625

~5x

Integration

Integration

~100x


RELIABLE MODE DIVISION MULTIPLEXING ESSENTIAL INGREDIENTS

PBS

x

y Polarization rotation Signal 1

Signal 2

Axxx + Axyy

Ayxx + Ayyy

PBS Fiber

22

1. MIMO processing of all modes (both polarizations)

2. Coherent detection of any complete orthogonal set of modes (both orthogonal polarizations, any basis)

3. Selective excitation of any complete orthogonal set of modes (both orthogonal polarizations, any basis)

[Winzer and Foschini, Proc. OFC, OThO5 (2011)]


HOW TO PERFORM SELECTIVE EXCITATION UNITARY TRANSFORM INTO (ANY COMPLETE) BASIS

23

SMF port 2

SMF port 1

SMF port 0

Phase Plates

Beam Splitters

f1 f2

Lenses Mirror

FMF

[R. Ryf, et.al., JLT 30(4), 2012]

LP01 X-pol LP11a X-pol LP11b X-pol

Inte

nsit

y Ph

ase

LP01 Y-pol LP11a Y-pol LP11b Y-pol

[H. Bülow et al., ECOC2011 (Tu.5.B); N. Fontaine et al. ECOC 2012 (Th.2.D.6)]

[R. Ryf et al., PTL, 1973 (2012)]


INTEGRATION FOR ECONOMIC SUSTAINABILITY

TX RX

TX RX

TX RX

Deploying M parallel paths is better than using multiple regenerators But: M systems cost M times as much & consume M times the energy Cost/bit (or energy/bit) remains constant

Integration is key to scale capacity in parallel systems

24


TRANSPONDER INTEGRATION

TX RX

TX RX

TX RX

25


SUPERCHANNEL TRANSPONDER INTEGRATION SPECTRAL SUPERCHANNEL SPATIAL SUPERCHANNEL

DAC TX DSP

PDM IQ-MOD



PDM IQ-MOD

PDM IQ-MOD

+

PDM IQ-MOD

PDM IQ-MOD

PDM IQ-MOD

TX DSP

TX DSP

RX DSP

RX DSP

RX DSP

Client Interface

RX DSP

TX DSP

DAC TX DSP

PDM IQ-MOD



PDM IQ-MOD

PDM IQ-MOD

PDM IQ-MOD

PDM IQ-MOD

PDM IQ-MOD

TX DSP

TX DSP

RX DSP

RX DSP

RX DSP

Client Interface

RX DSP

Laser

Laser

TX DSP

26

Laser

Laser

Laser

Laser

Laser

Laser

[S. Randel, OFC 2013, Tutorial OW4F.1]


TRANSPONDER INTEGRATION MULTI-CORE FIBER INTERFACING

[C. R. Doerr et al., Photon. Technol. Lett. 23(9), 597-599 (2011)]

27


OPTICAL AMPLIFIER INTEGRATION

TX RX

TX RX

TX RX

28

TX


OPTICAL AMPLIFIER INTEGRATION SOME EXAMPLES

[A. Chavez-Pirson et al., OAA, TuC3 (2003)]

Multimode-pumped Er doped

SOA array

[M. H. Hu et al., OAA, OWB3 (2006)]

[K. S. Abedin et al., Opt. Ex. 19, 16715 (2011)]

Core- and cladding-pumped 7-core

[H. Takahashi et al., ECOC, Th.3.C.3 (2012)] [K. S. Abedin et al., Opt. Ex. 20, 20191 (2012)]

29

Few-mode doped & Raman [Y. Yung et al., ECOC 2011] [E. Ip et al., ECOC 2011] [R. Ryf et al., ECOC 2011] [M. Salsi et al., ECOC 2012]


SDM WAVEGUIDE INTEGRATION

TX RX

TX RX

TX RX

30

Integration Integration


SDM RESEARCH HAS STARTED WITH FIBERS MULTI-CORE AND FEW-MODE

[K.Imamura et al., ECOC 2011]

[T. Hayashi et al., ECOC 2011]

[B.Zhu et al., ECOC 2011] [Sakaguchi et al., OFC 2012]

[H. Takara et al., ECOC 2012]

31

[C. Cia et al., Sum. Top. 2012]

[R. Ryf et al., OFC 2012] [M. N. Petrovich et al, ECOC 2012]


VERY IMPRESSIVE TRANSMISSION RESULTS BEYOND THE SINGLE-MODE SHANNON LIMIT

Transmission distance [km] 1,000 10,000 100

10 Aggr

egat

e sp

ectr

al e

ffic

ienc

y [b

/s/H

z]

30

40

50

60

10

20

[Liu, ECOC 2011]

[Sakaguchi, OFC 2012] [Gnauck, ECOC 2012]

[Chandrasekhar, ECOC 2011]

70

80

90 [Takara, ECOC 2012]

[Ryf, OFC 2013]

[Takahashi, ECOC 2012]

32


ROADM SDM

ROADM TX

TX

TX

RX

RX

RX

USE OF SPATIAL PATHS FOR NETWORKING ? THE ANSWER LIES IN THE PHYSICAL LAYER

• The benefits of SDM come through integration • Integration typically comes at the cost of impairments

• … such as crosstalk

33

MIMO DSP: Multiple-input multiple-output digital signal processing

MIMO DSP

• Crosstalk can be accommodated by MIMO processing • MIMO requires detection of all cross-talking channels Forces the use of spatial superchannels


CONCLUSIONS • Optical transport has run out of physical dimensions

• Need parallelism in frequency or space Superchannels

• Parallelism in frequency • Spectral superchannels to reach Terabit/s interface rates • Multi-band systems as a stop-gap solution to capacity crunch

• Parallelism in space

• Spatial superchannels to reach Petabit/s fiber capacities

• Integration is essential to reduce cost per bit • Transponders, optical amplifiers, ROADMs, connectors/splices, fiber • Integration eventually leads to crosstalk and the need for MIMO

• A smooth upgrade path from existing infrastructure is essential

• Start with parallel deployed fiber strands • Mixed fiber and mixed technologies • Leverage existing components at telecom wavelengths

34

OPTICAL NETWORK SCALING - University of Rochester...• Optical transport has run out of physical...

Documents

Transcript of OPTICAL NETWORK SCALING - University of Rochester...• Optical transport has run out of physical...