Download (PDF, 5.99MB)

IEEE Talk(April 24, 2015)

Generation, Detection, and Transmissionof Terabit/s and Higher “Channel” Rates

Using Multicarrier Solutionsg

S. ChandrasekharB ll L b t i Al t l L tBell Laboratories, Alcatel-Lucent

Holmdel, New Jersey, USA

All Rights Reserved, Alcatel-Lucent1S. Chandrasekhar, IEEE Talk, April 2015

AcknowledgementsI would like to thank many colleagues at Bell Labs for collaboration and discussionI would like to thank many colleagues at Bell Labs for collaboration and discussion. Among them are:

Xiang Liu (now at Futurewei)P. J. WinzerS. RandelG. RaybonA AdamieckiA. AdamieckiN. FontaineR. RyfD. NeilsonP DP. DongA. R. ChraplyvyR. W. Tkach


Outline

Modern Optical Transmission Systems

Superchannel ConceptSuperchannel Concept

Implementation

S ft D fi d T i iSoftware-Defined Transmission

Enabling Technologies

Flexible-Grid WDM

Novel Superchannel Techniques

Conclusion


Modern Optical TransmissionModern Optical TransmissionModern Optical Transmission Modern Optical Transmission SystemsSystems


Optical networks: backbone of the internet

Line interface

Reconfigurable opticaladd/drop multiplexer(ROADM)

Client interface(e.g., 4 x 25 Gbit/s)

(e.g., 100 Gbit/s)Router

Optical network

WDM system( 80 100 Gbit/ ) S t l ffi i (SE) i [b/ /H ]

Tx Rx

~ 5 THz bandwidth ~ 100 km of fiber

(e.g., 80 x 100 Gbit/s)

• Increase per-wavelength interface rate• Increase aggregate per fiber capacity

Spectral efficiency (SE) in [b/s/Hz]: SE = per-channel bit rate / WDM spacing


• Increase aggregate per-fiber capacity• Increase network flexibility

Evolution of optical transport networks1980 1990 2000 2010 2020

2.5 Gb/s 10 Gb/s 40 Gb/s 100 Gb/s Single-span

Channel rate:System type: Multi-span (EDFAs) Optically routed networks SDON?

1 Tb/s

10

100

paci

ties

s

1

10

d W

DM

Cap

Tb/s

2.5 dB/yr

100

ce R

ates

an (78%/yr)

[P. J. Winzer,

10

Seria

l Int

erfa

Gb/

s

1IEEE

IEEEITU-T

IEEEITU-T

[ ,IEEE Com. Mag., July 2010 ]


1986 1990 1994 1998 2002 2006

S

2010 2014 20181

2022

Dimensions for modulation and multiplexing[P. J. Winzer, S. Chandrasekhar and X. Liu, “Modulation Formats and Receiver Concepts for Optical Transmission Systems,”OFC’14 Short Course SC105]OFC 14, Short Course SC105]

http://www.occfiber.com/

SpaceSpacePolarization

Wavelength or FrequencySpaceSpace Wavelength or Frequency

Time Quadrature

Physical dimensions

Current R&D explores all these dimensionsCurrent R&D explores all these dimensions


Current R&D explores all these dimensionsCurrent R&D explores all these dimensions

Modulation format choices

Quadrature

M ltil lIntensity Phase

Memoryless

ASK/PSKQAM

Multilevel

Memoryless With memoryBinary MultilevelCorrelative codingBinary Pseudo-MultilevelMultilevel

OOK

C-NRZDST

CRZACRZ

M-ASK CSRZVSB-CSRZAP formats DB

PSBTPASSCAPSAMI

(DCS)

(D)PSK (D)QPSK

Chirp-free ChirpedPartial response

NRZ RZ NRZ RZNRZ RZ NRZ RZ

VSBSSB

DST ACRZ

Im{E}

I/Q modulationI/Q modulation

E : Optical field

Re{E}


E : Optical field

“Si l i i t it b t h it f ll bi th i bilit f“Si l i i t it b t h it f ll bi th i bilit f

DSP – Innovatively tailored for optical transport

“Signal processing is at its best when it successfully combines the unique ability of mathematics to generalize with both the insight and prior information gained from

the underlying physics of the problem at hand” [S. Haykin, “Signal processing: Where physics and mathematics meet,” IEEE Signal Process. Mag, 2001]

“Signal processing is at its best when it successfully combines the unique ability of mathematics to generalize with both the insight and prior information gained from

the underlying physics of the problem at hand” [S. Haykin, “Signal processing: Where physics and mathematics meet,” IEEE Signal Process. Mag, 2001]

uenc

ym

atio

n

cisi

on

ADC1

dem

ux

hase

mat

ion

com

p.

ecov

ery

Freq

Est

i m

DecADC2

and

Pol

-d Ph

Est

im

CD

c

Clo

ck R

eADC3

ADC4 D c

omp.

a

com

p.

quen

cym

atio

n

cisi

on

hase

mat

ion

CADC4

PM

D

CD

Freq

Est

im

DecPh

Est

imAll Rights Reserved, Alcatel-Lucent9S. Chandrasekhar, IEEE Talk, April 2015

12 TeraOperations/sec (TOPs) for 40Gb/s, and ~30 TOPs for 100Gb/s (~7,000 PCs)

Moore’s Law is still marching on! Dr. Kevin Kahn, Intel Senior Fellow


22 Superchannel ConceptSuperchannel Concept22 Superchannel ConceptSuperchannel Concept


In the context of optical transport networks the term “superchannel” was

The term “Superchannel”In the context of optical transport networks, the term superchannel wasfirst used in 2009* to refer to multiple single-carrier-modulated signals that areseamlessly multiplexed under the coherent optical orthogonal frequency-division multiplexing (CO-OFDM) conditions.

*: [S. Chandrasekhar, X. Liu, B. Zhu, and D. W. Peckham, “Transmission of a 1.2-Tb/s 24-carrier no-guard-intervalcoherent OFDM superchannel over 7200-km of ultra-large-area fiber,” ECOC’09, PD2.6]

The superchannel concept was later generalized to any collection of opticalsignals that are

modulated and multiplexed together with high SE at a commonoriginating site,transmitted and routed together over a common optical link andtransmitted and routed together over a common optical link, andreceived at a common destination site.

For signals that are not seamlessly bundled together, i.e., there is a spectralb t i l th t “ h l” i l ft dgap between signals, the term “super-channel” is also often used.

Much work has been done since then. ITU has recently standardized the useof superchannels in “flexible grid WDM” [ITU-T G.694.1]


p g [ ]

IEEE Xplore search on “Superchannel”Search by “superchannel”Search by “superchannel” Search by “super channel”Search by “super channel”Search by superchannelSearch by superchannel Search by super-channelSearch by super-channel

First record in 2009.215 Google citations as of 4/21/2015

First record in 2009.215 Google citations as of 4/21/20154/21/2015.4/21/2015.


Benefits of superchannel

Higher “per-channel” data rate Avoiding electronic bottlenecks and easing wavelength management

Higher spectral efficiency (SE)Due to better spectrum utilization, especially with flexible-grid WDM

Better leveraging large-scale integration of photonic integrated circuits (PICs) and application specific integrated circuits (ASICs)

Increased efficiency in digital signal processing (DSP)

Supporting software-defined optical transmission and potentially future software-defined optical network (SDON)

As transmitter DSP (required for spectral shaping) is already in place


Channel (or carrier) Spacing = Δf; Symbol Rate = B

Conventional WDM classifications Channel (or carrier) Spacing = Δf; Symbol Rate = B

Δf/B >5.0: legacy WDM and coarse WDM10G over 100-GHz

1.2 < Δf/B ≤ 5.0: common Dense WDM (DWDM)10Gbaud/28Gbaud over 50-GHz; 40Gbaud over 100-GHz;14.3Gbaud over 25-GHz.

1.0 < Δf/B ≤ 1.2: “Quasi-Nyquist-WDM” and “Guard-banded-OFDM”32Gbaud PDM-QPSK on 37.5GHz

Δf/B = 1.0: “Nyquist-WDM”Δf/B 1.0: Nyquist WDM28Gbaud PDM-QPSK on 28GHzCoherent WDM and NGI-CO-OFDM are special cases of Nyquist-WDM thatadditionally satisfy the OFDM conditions

Δf/B <1.0: “Super-Nyquist-WDM” (or “Faster-than-Nyquist”)30Gbaud PDM-QPSK on 25GHz

The last three WDM classes are commonly supported by superchannels


The last three WDM classes are commonly supported by superchannels

Electrical OFDM

OFDM: where modulation and multiplexing mergeOFDM: where modulation and multiplexing merge

f

56 Gb/s net rate (65 Gb/s line rate)32-QAM per subcarrier[Takahashi et al., OFC’09]All subcarriers modulated at oncef

f

Mod.Laser

Optical OFDMMod.

ff M d

f

All subcarriers modulated individuallyParallel optical hardware

Comb

f

f

f

f

Mod.

Mod.

Mod.

60 GHz

65 GHz300 GHz

448 Gb/s (10 subcarrier) 16-QAM5 bit/s/Hz 2000-km transmission

[X Li t l OFC’10]

606 Gb/s (10 subcarrier) 32-QAM7 bit/s/Hz 1600-km transmission

[X Li t l ECOC’10]

1.2 Tb/s (24 subcarrier) QPSK3.74 bit/s/Hz 7200-km transmission[S Ch d kh t l ECOC’09]


[X. Liu et al., OFC’10] [X. Liu et al., ECOC’10][S. Chandrasekhar et al., ECOC’09]

[S. Chandrasekhar and X. Liu, Opt. Ex., p.21350, 2009]

The OFDM condition

Frequency-locked carriersPhase locking is not a necessary conditionPhase locking is not a necessary condition

Carrier Spacing = Symbol Rate

Symbols in the carriers time-alignedSynchronous at de-multiplexing

S ffi i t t itt b d idthSufficient transmitter bandwidth

Sufficient sampling speedOversampling along with appropriate anti-aliasing filters in the DSPOversampling along with appropriate anti aliasing filters in the DSP

See also:“Coherent WDM” [A Ellis et al PTL p 504 2005]


Coherent WDM [A. Ellis et al., PTL, p.504, 2005]“2-Carrier NGI-CO-OFDM“ [E. Yamada et al., OFC’08, PDP8, 2008]

Multiplexing with a guard band (GB)[S. Chandrasekhar and X. Liu, “Advances in Tb/s Superchannels”, Optical Fiber Telecommunications VI, Edited by Ivan P Kaminov Tingye Li and Alan Willner Academic Press 2013]

⎟⎞

⎜⎛ ⋅⎟

⎞⎜⎛

∑∑∞∞

22 FFTNGBGB

To avoid the stringent OFDM condition, signals can be multiplexed with a GB. The amount of the worst-case crosstalk imposed on the edge subcarriers can be expressed as

Edited by Ivan P. Kaminov, Tingye Li, and Alan Willner, Academic Press, 2013]

⎟⎟⎠

⎞⎜⎜⎝

⎛+=⎟⎟

⎠

⎞⎜⎜⎝

⎛+

Δ= ∑∑

=

−

=

−

0

2

0

2 )]([log10)]([log10)(n OFDM

FFT

n SC

nB

NGBn

fGBdBX ππ

where ΔfSC, NFFT, and BOFDM are the OFDM subcarrier spacing, FFT/IFFT size used in OFDM, and the spectral bandwidth of the OFDM signal, respectively.

-16

-14

-12

-10

ubca

rrier

s (d

B)

NFFT=128

NFFT=256

567

dB]

QPSK 16-QAM 64-QAM

[P. J. Winzer, Proc. ECOC (2011)]

24

-22

-20

-18

alk

on th

e ed

ge s

u

1234

NR

pen

alty

[d

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18 0.2-28

-26

-24

GB/BOFDM

Cro

ssta

10 15 20 25 30 35 40 45

01

5Crosstalk [dB]

OS


A GB-to-Baud ratio of ~0.1 gives a good tradeoff between SE and simplicity

ImplementationImplementationImplementationImplementationGenerationGeneration


( ) OFDM b d ltib d (b) N i t WDM i ( ) N i t WDM i

Generation methods

OF

OF

(a) OFDM-based multiband transmitter

(b) Nyquist-WDM usingoptical filtering

(c) Nyquist-WDM usingdigital filtering

SB-TX

Multicarrier SB-TX

bine

r

DMUX SB-TX

SB-TX

bine

rLaser

Laser

SB-TX with DF

SB-TX with DF

bine

rLaser

Laser

OF

…Multicarriergenerator

SB-TX

…

Com

bSB-TX

…

Com

b

Laser

…

SB-TX with DF

…

Com

b

Laser

…(frequency-

locked carriers)

DMUX: wavelength de-multiplexer; SB-TX: single-band transmitter; OF: optical filter; DF: digital filter.

MultibandTX type

Frequency locking?

OF needed?(band spacing)

SB-TXtype used*

TX-BW ADCspeedyp g ( p g) yp p

(a) OFDM-based

Yes No(B)

SB-(a), SB-(b) >~2B* >~4B*SB-(c) ~B ~1.5B

(b) Nyquist-WDM w/ OF

No Yes (~1.1B*)

SB-(a), SB-(b) ~B ~2B( )

(c) Nyquist-WDM w/ DF

No No(~1.1B*)

SB-(b), SB-(c) ~B 1.5B~2B

* SB-(a): Single-carrier PDM-QAM w/o DAC; SB-(b): Single-carrier PDM-QAM w/ DAC;SB-(c): CO-OFDM with PDM-QAM subcarriers


SB (c): CO OFDM with PDM QAM subcarriers.

OFDM-based seamless multiplexing[S. Chandrasekhar and X. Liu et al., ECOC’09, PD2.6] [X. Liu and S. Chandrasekhar et al, OFC 2011]

1.12-Tb/s 24-carriersNGI-CO-OFDM PDM-QPSK

0

485-Gb/s 10-carrier RGI-CO-OFDM PDM-16QAM

-20

-10

(dBm

)

-40

-30

1.2-Tb/s Superchannel

PO

WE

R (

-501554 1555 1556 1557

1.2 Tb/s Superchannel24 Frequency-locked Carriers

WAVELENGTH (nm)( )

12.5-GHz spaced carriers12.5-Gbaud PDM-QPSK per carrier

Intra-channel SE = 3.74 b/s/Hz7200 km over ULAF

6.5-GHz spaced carriers/bands48.5-Gb/s PDM-16QAM per carrier

Intra-channel SE = 7.0 b/s/Hz4800 km over ULAF


7200-km over ULAF 4800-km over ULAF

Guard-banded multiplexing[X. Liu and S. Chandrasekhar et al., ECOC’12, post-deadline paper Th.3.C.5]

λ1λ3λ5λ7

EDFA EDFA

Cou

pler

ler

PDM I/Q modulator

I Q QIEDFA

Modulation: (1) Reduced-guard-interval (RGI)

xI

λ2λ4λ6λ8

EDFA EDFA

Cou

pler C

oupl

xQ yI yQ

xI xQ yQyI

PDM I/Q modulator

Delay CO-OFDM with PDM-16QAM(2) Single-carrier PDM-16QAM with

Nyquist filteringSymbol rate: ~32 Gbaud.

DAC DAC DAC DACTransmitter-side

offline DSP

50~64 GSa/s Guard band to baud ratio: ~10%.

1.6-Gb/s quasi-Nyqusit-WDM superchannel

-10

0

wer

(dB

)-10

-5

0

wer

(dB)

Even signalsOdd signalsEntire superchannel

1.5-Gb/s RGI-CO-OFDM superchannel

-30

-20

Rel

ativ

e po

w

OSA resolution: 4 GHz

35

-30

-25

-20

-15

Rel

ativ

e P

ow

OSA resolution: 0.02 nm


-150 -75 0 75 150Frequency offset (GHz)

1549.5 1550 1550.5 1551 1551.5 1552-35

Wavelength (nm)

[G. Raybon et al, Proc. IEEE Photonics Conference, Sept. 2012, post-deadline paper PD1.2.]

1-Tb/s 2-carrier quasi-Nyquist-WDM superchannel

80-Gb/s Quadrature

(Q)

Delay

PRBS Gen40 Gb/s

2:1Mux

DFB 50 GHZ

Channel

200 ps

80 Gb/s

π/2EDFA

ECL

DFB

DFB

DFB

DFB

OF

50 G

Sub-carrier 280-Gb/s In-phase (I)

215-1

PRBS Gen40 Gb/s

2:1Mux

3 dBDFB 50 GHZ

-20

-10

0

Wavelength, nm1548 1557

Sub-carrier 1Sub ca e

-50

-40

-30

f0 f2f1Very high speed electronics at transmitter; 160-GSa/s 66-GHz ADC at receiverOptical equalizer to shape modulation format and overcome bandwidth limitationHighest achieved single carrier bit-rate is 640Gb/s, 80 Gbaud PDM-16QAM

Very high speed electronics at transmitter; 160-GSa/s 66-GHz ADC at receiverOptical equalizer to shape modulation format and overcome bandwidth limitationHighest achieved single carrier bit-rate is 640Gb/s, 80 Gbaud PDM-16QAM


Successful transmission over 3200-km with a SE of 5.0-b/s/Hz Successful transmission over 3200-km with a SE of 5.0-b/s/Hz

Super-Nyquist-WDM -1Super-Nyquist-WDM (SNW): Modulation symbol rate < channel spacingSuper-Nyquist-WDM (SNW): Modulation symbol rate < channel spacing

[J. -X. Cai et al., OFC2009, PDPB10 (2009)]

Typically, SNW offers higher SE at the expense of (i) additional OSNR penalty and (ii) increased DSP complexity for removing the ISI caused by SNW.With PDM QPSK and modest SNW low complexity SNW receiver may be realized:With PDM-QPSK and modest SNW, low-complexity SNW receiver may be realized: an interesting approach when a slight SE increase is desired.

[J. Li, M. Sjödin, M. Karlsson, and P. A. Andrekson, Opt. Express 20, 10271-10282 (2012) ]

11 carriers spaced at 25-GHz28-Gbaud PDM-QPSK (112G per carrier)

TX-side: optical pre-filtering RX-side: duobinary-shaping & MLSD


( p )Intra-channel SE ~ 4.19 b/s/Hz

y p gAdditional OSNR penalty: ~1.8 dB @BER=10-3

Super-Nyquist-WDM -2140 7 Tbit/s 7 326 km transmission of 7x201 channel 25 GHz spaced140 7 Tbit/s 7 326 km transmission of 7x201 channel 25 GHz spaced140.7-Tbit/s, 7,326-km transmission of 7x201-channel 25-GHz-spaced Super-Nyquist-WDM 100-Gbit/s optical signals using a 7-core fiber, achieving a record capacity-distance product of 1.03 Exabit/s⋅km,

was recently demonstrated [K. Igarashi et al., ECOC’13, post-deadline paper PD3.E.3.]

140.7-Tbit/s, 7,326-km transmission of 7x201-channel 25-GHz-spaced Super-Nyquist-WDM 100-Gbit/s optical signals using a 7-core fiber, achieving a record capacity-distance product of 1.03 Exabit/s⋅km,

was recently demonstrated [K. Igarashi et al., ECOC’13, post-deadline paper PD3.E.3.]

(a): Concept of Super-Nyquist-WDM transmission (30Gbaud over 25-GHz: SE=4 b/s/Hz after removing 20% overhead for SD-FEC); (b): Impulse response; (c): Power spectrum of


duobinary-pulse shaping; (d): Duobinary trellis for two-level amplitude modulation.

ImplementationImplementationImplementationImplementationDetectionDetection


Banded detection of superchannelsOptical Filter-free detection (In ROADM jargon- Colorless Drop)

Example: N-band superchannel and M detection bands (M<N)

Optical Filter free detection (In ROADM jargon Colorless Drop)Receiver hardware complexity reduced by detection of more than 1 band

Each detected band uncorrelated

Dispersion compensation t OFDM ditimust ensure OFDM conditions

satisfied among carriers ofeach band


Oversampling Ratio for multiband detectionBanded detection: X. Liu et al., Proc. OFC’10, OWO2

Oversampling ratio (OSR) sampling rate/net detected baudOversampling ratio (OSR) = sampling rate/net-detected-baud

fOLO=f0

Δf li =4xB Δfsampling=4xB

fOLO=(f0+f1)/2 fOLO=f0

Δfsampling=4xBΔfsampling 4xB

f0 f1=f0+B f0 f1f-1 f2 f0 f1f-1f-2 f2

OSR=4 OSR=2 OSR=1.33


485-Gb/s PDM-16QAM with 10 bands detected with 80-GSa/sOSR = (80/485)*8 = 1.31

728-Gb/s PDM-64QAM with 10 bands detected with 80-GSa/sOSR = (80/728)*12 = 1.32

Various superchannel demonstrations

263 GHz300 GHz

200 GHz 200 GHz

1 Tb/s (2 subcarriers) 16-QAM5 2 bit/s/Hz

1.5 Tb/s (8 subcarriers) 16-QAM5 7 bit/s/Hz

1.2 Tb/s (24 subcarriers) QPSK3 74 bit/s/Hz

1 Tb/s (4 subcarriers) 16-QAM5 0 bit/s/Hz5.2 bit/s/Hz

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

5.7 bit/s/Hz5600 km transmission

[Liu et al., ECOC’12]


[Chandrasekhar et al., ECOC’09]

ADC&DAC b d idth & l tiADC&DAC b d idth & l ti O ti l ll liO ti l ll li


[Renaudier et al., OFC’12]

ADC&DAC bandwidth & resolutionADC&DAC bandwidth & resolution Optical parallelismOptical parallelism

Few carriersHigh symbol rate

Large # of carriersLower symbol rate

The optimal choice depends on implementation platforms and applicationsThe optimal choice depends on implementation platforms and applications

g y y


Modulation Format Superchannel Composition ISE* Reach ISEDP

Recent Tb/s-class superchannel demonstrationsModulation Format Superchannel

data rateComposition ISE

(b/s/Hz)Reach(km)

ISEDP(km⋅b/s/Hz)

Seamless CO-OFDM (with frequency-locked optical carriers)

PDM-QPSK 1200 Gb/s 24× 50Gb/s 3.74 7200 26928

PDM-16QAM 1500 Gb/s 15× 100Gb/s 7.00 1200 8400

RGI OFDM 16QAM 485 Gb/ 10 48 5Gb/ 6 20 4800 29760RGI-OFDM-16QAM 485 Gb/s 10× 48.5Gb/s 6.20 4800 29760

Guard-banded CO-OFDM (with frequency-unlocked carriers)

OFDM-16QAM 1864 Gb/s 8× 233Gb/s 5.75 5600 32000

Quasi-Nyquist-WDM (with frequency-unlocked carriers)

PDM-64QAM 538 Gb/s 8× 67Gb/s 8.96 1200 10752

PDM-16QAM 1280 Gb/s 2× 640Gb/s 5.00 3200 16000

Super-Nyquist-WDM (with frequency-locked carriers)

PDM QPSK 1232 Gb/s 11× 112Gb/s 4 19 >640 >2560


PDM-QPSK 1232 Gb/s 11× 112Gb/s 4.19 >640 >2560*ISE: intra-channel spectral efficiency (net channel data rate over channel bandwidth)

44 SoftwareSoftware--defineddefined44 SoftwareSoftware defined defined transmissiontransmission


Adaptive modulation for Tb/s-class superchannels[S. Chandrasekhar and X. Liu, “Advances in Tb/s Superchannels”, Optical Fiber Telecommunications VI, Edited by Ivan P Kaminov Tingye Li and Alan Willner Academic Press 2013]

PDM-16-QAM PDM-32-QAM PDM-64-QAM PDM-256-IPM

Edited by Ivan P. Kaminov, Tingye Li, and Alan Willner, Academic Press, 2013]

PDM-16-QAM PDM-32-QAM PDM-64-QAM PDM-256-IPM(8 bits/symbol) (10 bits/symbol) (12 bits/symbol) ( 16 bits/symbol)

The transmitter DSP and DAC, required to perform spectral shaping for superchannel formation, naturally allows for adaptive modulation for

varying data rates (depending on demand/reach)


Adaptive modulation with fine granularityCorrelated information between I and Q phases Time domain interleaving of formats

Adaptive 4-D modulationAdaptive 4-D modulation Adaptive hybrid-QAMAdaptive hybrid-QAM

Correlated information between I- and Q-phases Time-domain interleaving of formats

[X. Zhou, L. E. Nelson, and P. Magill, “Rate-Adaptable Optics for Next Generation Long-Haul Transport Networks ”

[J. K. Fischer, S. Alreesh, R. Elschner, F. Frey, M. Nölle, and C Schubert “Bandwidth-Variable Transceivers Based Optics for Next Generation Long Haul Transport Networks,

IEEE Commun. Mag. 51, 41-49, March 2013] and C. Schubert, Bandwidth Variable Transceivers Based on 4D Modulation Formats for Future Flexible Networks,” ECOC’13, invited paper Tu.3.C.1]

QPSK QPSK16-QAMSymbol duration


Also see: [D. Coelho and N.Hanik, ECOC’11, Mo.2.B.4 (2011)]; [M. Sjödin,et al., Opt. Express 20, 8356-(2012)]; [J. Renaudier et al., OFC’13, OTu3B.1].

[G. Bosco, V. Curri, A. Carena, P. Poggiolini, and F. Forghieri, “On the performance of Nyquist-WDM terabit superchannels based on PM BPSK PM QPSK PM 8QAM or PM 16QAM subcarriers ” JLT 29 53 61 (2011)]

Tradeoff between capacity and reachon PM-BPSK, PM-QPSK, PM-8QAM or PM-16QAM subcarriers, JLT 29, 53–61 (2011)]

Maximum reach with BER≤4⋅10-3 versus capacity in the C-band (bottom axis) and spectral efficiency (top axis) for Nyqusit-WDM superchannel with 27 75-Gbaud PM-BPSK (crosses) PM-QPSK (diamonds) PM-


axis) for Nyqusit-WDM superchannel with 27.75-Gbaud PM-BPSK (crosses), PM-QPSK (diamonds), PM-8QAM (squares) and PM-16QAM (circles). Solid lines refer to SSMF and dashed lines refer to NZDSF.

Joint optimization of modulation and FEC coding[J Renaudier et al., “1-Tb/s Transceiver Spanning Over Just Three 50-GHz Frequency Slots for Long-Haul Systems ,” ECOC’13 t d dli PD2 D 5]ECOC’13, post-deadline paper PD2.D.5]

FIXED SYMBOL RATE

Superchannel total data rate fixed. Number of carriers fixed. Therefore spectral extent is fixed.Vary constellation size for same symbol rate. Use extra symbol rate for FEC overheadFEC threshold changes, therefore reach changes.

Demonstrated a 1Tb/s transceiver, occupying only three 50-GHz slots and leveraging a joint optimization procedure of modulation format and error correction codingDemonstrated a 1Tb/s transceiver, occupying only three 50-GHz slots and leveraging a joint optimization procedure of modulation format and error correction coding


a jo t opt at o p ocedu e o odu at o o at a d e o co ect o cod gIt enabled 51-Tb/s transmission over 2000 km, at 6.7-bit/s/Hz spectral efficiency a jo t opt at o p ocedu e o odu at o o at a d e o co ect o cod gIt enabled 51-Tb/s transmission over 2000 km, at 6.7-bit/s/Hz spectral efficiency

Alternative superchannel formation – fixed information rate[Q. Zhuge, M. Morsy-Osman, M. Chagnon, X. Xu, M. Qiu, and D. V. Plant, “Demonstration of energy efficient and format-

transparent digital signal processing for Tb/s flexible transceiver ” in Proc ACP'13 post deadline paper AF2E 7 (2013)]transparent digital signal processing for Tb/s flexible transceiver,” in Proc. ACP'13, post-deadline paper AF2E.7 (2013)]

10 carriers100 Gb/s (net) per carrier1 Tb/s (net) per superchannel

10 carriers100 Gb/s (net) per carrier1 Tb/s (net) per superchannel1 Tb/s (net) per superchannelVarying baud with format:

PDM-QPSK: 30 GbaudPDM-8QAM: 20 GbaudPDM-16QAM: 15 Gbaud

1 Tb/s (net) per superchannelVarying baud with format:

PDM-QPSK: 30 GbaudPDM-8QAM: 20 GbaudPDM-16QAM: 15 GbaudPDM 16QAM: 15 GbaudPDM 16QAM: 15 Gbaud

Formationmethods

Total data rate *

Spectralefficiency*

TX/RX cost/bit*

Reach Client interface

(1) Fixed log2(M/4) log2(M/4) 1/log2(M/4) Lower as M **

Load baud increases** changes

(2) Fixed data rate

1 log2(M/4) 1 Slightly betterthan (1)

Load unchanged

* Normalized to the case of PDM-QPSK. M is the number of the constellation points of M-QAM.


Q p Q**: Larger M can be combined with higher overhead FEC to achieve longer reach for a given SE.

[J Renaudier et al., ECOC’13, post-deadline paper PD2.D.5]

5555 Enabling TechnologiesEnabling Technologies


DSP gate counts

Major DSP ASIC milestonesg

100

1000

llion

s)

NEL 2nd gen

Ciena WaveLogic 3

7

8

9

bit

ADC resolution vs. sample rate

ALU 56GS/s

10

100

TE C

OU

NT

(Mi

~70% per year

4

5

6

Num

ber o

f

StatedENOB

12004 2006 2008 2010 2012 2014

GA

T

2

3

0 10 20 30 40 50 60Sample rate [GS/s]

ENOB

70M+ gates70M gates


Nortel electronic pre-EDC 10G Tx (2005)20GS/s DAC

Nortel 40Gb/s PDM-QPSK (2007)20GS/s ADC/DSP

Alcatel-Lucent 112Gb/s (2010)56GS/s ADC/DSP

200-Gb/s transponder - pulse shaping, equalization, soft decoding[http://www3.alcatel-lucent.com/400g-pse/ ; http://www.convergedigest.com/2012/09/alcatel-lucent-cites-deployment-of-400g.html; http://www nationmultimedia com/technology/Alcatel Lucent tackles the capacity crunch caused 30177739 html]

DACDAC

TX

200 PDMFEC

http://www.nationmultimedia.com/technology/Alcatel-Lucent-tackles-the-capacity-crunch-caused--30177739.html]

ADC

DACDACDSP

0G C

lient Inte

I/Q-MODTX Laser

PDM LO Laser

Enc.

ADCADC

ADCRXDSP

erfaces

PDMCoherentReceiver

LO LaserFECDec.

Alcatel-Lucent’s 400G Photonic Service Engine

0

-10

ower

(dB

)

Digitally pre-filtered for supporting a SE

gIndustry’s first 400G coherent optical chip Frequency (GHz)

-20Po

-20 0 20


of up to 5.3 b/s/Hz (400G over 75 GHz)

400G dual-carrier ASIC and optics[http://www3.alcatel-lucent.com/400g-pse/ ; http://www.convergedigest.com/2012/09/alcatel-lucent-cites-deployment-of-400g.html; http://www nationmultimedia com/technology/Alcatel Lucent tackles the capacity crunch caused 30177739 html]

DACDAC20 Dual-carrier optics integration

http://www.nationmultimedia.com/technology/Alcatel-Lucent-tackles-the-capacity-crunch-caused--30177739.html]

ADC

DACDACDACTX

DSP

00GD

ata Inp

PDMI/Q-MOD

TX Laser

FECEnc. • Dual-PDM Coherent RX

• Dual PDM I/Q-Mod

with Alcatel-Lucent’s 400G

ADCADC

ADCADC

RXDSP

ut/Output

PDMCoherentReceiver

LO LaserFECDec.

with Alcatel Lucent s 400G

coherent ASIC

DACDACDAC

TXDSP

200GD

a

FECEnc.

PDMI/Q-MOD

TX Laser

ADCADCADC

DAC

RXDSP

ata Input/Outp

FECDec.

TX Laser

PDMCoherentR i

LO Laser


ADCADC

put Receiver

500-Gb/s photonic integrated circuits (PIC)[J. Rahn et al., OFC/NFOEC’12, post-deadline paper PDP5D.5 (by Infinera)]

The PIC-based transmitter (TX) has 10 integrated lasers, 40 modulatorsrunning at 14.3 GBaud, and an optical Array Waveguide Grating (AWG).The PIC-based transmitter (TX) has 10 integrated lasers, 40 modulatorsrunning at 14.3 GBaud, and an optical Array Waveguide Grating (AWG).The PIC-based receiver (RX) integrates the AWG, per-carrier DFBs,optical hybrids, and photodiodes.Modes of operation

500Gb/s: 10x 25-GHz-spaced 14.3-Gbaud PDM-QPSK, SE=2 b/s/Hz

The PIC-based receiver (RX) integrates the AWG, per-carrier DFBs,optical hybrids, and photodiodes.Modes of operation

500Gb/s: 10x 25-GHz-spaced 14.3-Gbaud PDM-QPSK, SE=2 b/s/Hz


250Gb/s: 10x 25-GHz-spaced 14.3-Gbaud TCM-PDM-QPSK, SE=1 b/s/Hz250Gb/s: 10x 25-GHz-spaced 14.3-Gbaud TCM-PDM-QPSK, SE=1 b/s/Hz

Silicon-photonics based coherent PIC[P. Dong, X. Liu, S. Chandrasekhar, L. L. Buhl, R. Aroca, Y. Baeyens, and Y.-K. Chen, OFC’13, post-deadline paper PDP5C.6]

224-Gb/s PDM-16-QAM Modulator and Receiver based on Silicon Photonic Integrated Circuits

Compact transmitterCompact transmitterpCompact receiverScalablePotentially

pCompact receiverScalablePotentially integratable w/ DAC/driver/ADC/DSPFurther performance improvement and V

integratable w/ DAC/driver/ADC/DSPFurther performance improvement and Vimprovement and Vπ

reduction neededimprovement and Vπ

reduction needed


66 FlexibleFlexible--grid WDMgrid WDM66 FlexibleFlexible grid WDMgrid WDM


Agile networking via superchannels[Excerption from a presentation made by X. Liu, P. J. Winzer, and R. W. Tkach to DARPA on December 16, 2008]Al [S Ch d kh d X Li “T bit h l f hi h t l ffi i t i i ” ECOC’10 i it d T 3C 5]Also see: [S. Chandrasekhar and X. Liu, “Terabit superchannels for high spectral efficiency transmission,” ECOC’10, invited Tu3C.5]

Rate, Format, and Reach Agile Optical GroomingCurrent DWDM System

10Gb/s

stem

Rea

ch 10Gb/s40Gb/sDPSK

40Gb/sDQPSK

100Gb/sDQPSK

100Gb/sDP-QPSK

Current DWDM System3000km

Reduced guardband

Increased capacity

frequency50 GHz

Sys

Flexible-Band* OFDM** System

Increased capacity

4x10Gb/s

em R

each

1x40Gb/s

1x100Gb/s

10x10Gb/s

2x40Gb/s

1 Terabit/s 16QAM**-OFDM over 200-GHz

1 Terabit/s QPSK-OFDM over 400-GHz

w/ 2nd layer banding by coherent-WDM

frequency50 GHz

Syst

e

*: 1st layer banding by DSP in OFDM; 2nd layer banding by coherent-WDM**: PDM-QPSK assumed unless otherwise indicated

OFDM over 200 GHz


DARPA meeting, 12/16/2008

The flexible-grid DWDM standard[ITU-T G.694.1, “Spectral grids for WDM applications: DWDM frequency grid”, approved in June, 2012;http://www.itu.int/rec/T-REC-G.694.1-200206-S/en]

Details about the new ITU standard:Bandwidth granularity: 12.5 GHzCenter frequency granularity: 6.25 GHzq y g yThe allowed frequency slots have a nominal central frequency (in THz) defined by: 193.1 + n × 0.00625, where n is an integer

Any combination of frequency slots is allowed as long as no two frequency slots overlapThe flexible WDM architecture is supported by the recent availability of flexible-grid WSSs


pp y y g

Flexible-grid superchannel examples

2-Carrier-32.5Gbaud 400-Gb/s superchannelson a 87.5-GHz grid, supporting

20 Tb/s over the typical 4 4-THz C-band

5-Carrier-32.5Gbaud 1-Tb/s superchannelson a 200-GHz grid, supporting

22 Tb/s over the typical 4 4-THz C-band20 Tb/s over the typical 4.4-THz C-band

0

10

“S-Channel” 1 “S-Channel” 2 “S-Channel” 30

10

“S-Channel” 1 “S-Channel” 2 “S-Channel” 3

-5

0

5“S-Channel” 3“S-Channel” 2“S-Channel” 1

-5

0

5“S-Channel” 3“S-Channel” 2“S-Channel” 1

22 Tb/s over the typical 4.4-THz C-band

er (d

B)

er (d

B)

-20

-10

-20

-10

-25

-20

-15

-10

-25

-20

-15

-10

aliz

ed p

owe

aliz

ed p

owe

-40

-30

-175.0 -87.5 0 87.5 175.0

400G in 87.5-GHzSE = 4.6 b/s/Hz

-40

-30

-175.0 -87.5 0 87.5 175.0

400G in 87.5-GHzSE = 4.6 b/s/Hz

-40

-35

-30

-400 -300 -200 -100 0 100 200 300 400

1T in 200-GHzSE = 5 b/s/Hz

-40

-35

-30

-400 -300 -200 -100 0 100 200 300 400

1T in 200-GHzSE = 5 b/s/HzN

orm

a

Nor

ma

FREQUENCY (GHz)FREQUENCY (GHz) FREQUENCY OFFSET (GHz)FREQUENCY OFFSET (GHz)Frequency offset (GHz) Frequency offset (GHz)

The frequency spacing between adjacent superchannels is made to be sufficiently The frequency spacing between adjacent superchannels is made to be sufficiently


large to allow multiple passes through flexible-bandwidth-ROADMs large to allow multiple passes through flexible-bandwidth-ROADMs

Evolution of ROADM subsystems[S. Frisken, S. B. Poole, and G. W. Baxter, "Wavelength-Selective Reconfiguration in Transparent Agile Optical Networks,”P di f th IEEE l 100 5 1056 1064 M 2012]Proceedings of the IEEE , vol.100, no.5, pp.1056-1064, May 2012]

* CDC-f: Colorless, Directionless, Contentionless, and FlexgridWSS: Wavelength-Selective Switch; PLC: Planar Lightwave Circuit; ROADM: Reconfigurable Optical Add/Drop Multiplexer; AWG: Arrayed Waveguide Grating


AWG: Arrayed Waveguide Grating.

Flexible-grid-WSS[S. Frisken, S. B. Poole, and G. W. Baxter, Proceedings of the IEEE , vol.100, no.5, pp.1056-1064, May 2012]

stal

on

CoS

)Li

quid

cry

ssi

licon

(LC

The input is dispersed across a switch element, and the channel plan for switching wavelengths is defined by software and created by writing the appropriate phase image to the LCoS.


Measured pass-band spectra of a WSS employing a flexible grid with varying bandwidth.

Software-defined transmission in flexible-grid WDM[ S. Gringeri, N. Bitar, and T. J. Xia, "Extending software defined network principles to include optical transport,“C i ti M i IEEE l 51 3 32 40 M h 2013]Communications Magazine, IEEE , vol.51, no.3, pp.32,40, March 2013]


Illustration of a software-defined transceiver having several adjustable features

Capacity gain from superchannel on flexible grid[ P. Palacharla at OFC 2013 Workshop] [Courtesy of Drs. P. Palacharla, M. Sekiya, and X. Wang of Fujitsu Labs, USA]

Simulation Model Quasi-static traffic @ 400 Gb/s: demands with randomly chosen S-D pairsReference: Current network with fixed modulation (PDM-QPSK), inverse mux of 4x100GCase 1: Super-channel transmission with distance-adaptive modulation (4-carrier PDM-QPSK, 2-carrier PDM-16QAM) with fixed 50-GHz gridCase 2: Super-channel transmission with distance-adaptive modulation with flexible grid (ITU-T G 694 1 12 5-GHz granularity)

Capacity gains of 40~160%

(ITU T G.694.1, 12.5 GHz granularity)Capacity Gain (%) = (Capacity1,2 – Capacityref)/Capacityref

Capacity gains of 40 160% are achieved using flexible grid capability

Much lower capacity gains are obtained with fixed 50-GHz grid capability


77 Novel SuperchannelNovel Superchannel77 Novel Superchannel Novel Superchannel TechniquesTechniques


Multi-Tb/s superchannel with large-scale integration[X. Liu, S. Chandrasekhar, and P. J. Winzer, “Digital Signal Processing Techniques Enabling Multi-Tb/s SuperchannelT i i ” IEEE Si l P i M i 2014]Transmission”, IEEE Signal Processing Magazine, 2014]

OLOI1x’

Q1x’

ADCR Polarization-

PDDACI1x

Q1x PDMMulti-Tb/s superchannel

transmissionS1

Q1x

I1y’

Q1y’

ADC

ADC

ADC

R1Polarization-

DiversityOptical Hybrid

PD

PD

PD

y

DAC

DAC

DAC

y

1x

I1y

Q1y

λ1PDMI/Q

Modulator

FiberLink

S litt I

DS

P

AS

IC A

rray

DS

P

AS

IC A

rray

I CombinerWDMMUX

WDMDMUX

SplitterSM

OLO

IMx’

QMx’

IMy’

QMy’

ADC

ADC

ADC

ADCRM

Polarization-DiversityOptical Hybrid

PD

PD

PD

PD

DAC

DAC

DAC

DAC

IMx

QMx

IMx

λ1PDMI/Q

Modulator

Illustration of a multi-Tb/s superchannel transponder embedded in a WDM system (DSP: digital signal processor, preferably processing the entire superchannel as a whole)

OLO ADCPDDACQMy


( g g p p y p g p )

Joint DSPAs the constituents of a superchannel share the same optical transmission path theAs the constituents of a superchannel share the same optical transmission path, themonitoring of some transmission properties of the superchannel can be simplified

Carrier frequency recovery may be simplified[M. D. Feuer et al., “Joint digital signal processing receivers for spatial superchannels,” PTL 24, 1957-1960, 2012.]

Li t lk th h l tit t ld b t dLinear crosstalk among the superchannel constituents could be compensated[C. Liu et al., “Super receiver design for superchannel coherent optical systems,” Proc. SPIE 8284, Next-GenerationOptical Communication: Components, Sub-Systems, and Systems, 828405, 2012.]

Nonlinear crosstalk among the superchannel constituents could be partially compensated.[E Ip et al "Interchannel nonlinearity compensation for 3λ× 114 Gb/s DP 8QAM using three synchronized sampling[E. Ip et al., Interchannel nonlinearity compensation for 3λ× 114-Gb/s DP-8QAM using three synchronized samplingscopes," OFC’12, paper OM3A.6 (2012)][N. K. Fontaine, X. Liu, S. Chandrasekhar et al., "Fiber nonlinearity compensation by digital backpropagation of anentire 1.2-Tb/s superchannel using a full-field spectrally-sliced receiver," ECOC’13, paper Mo.3.D.5 (2013)][C. Xia, X. Liu, S. Chandrasekhar, N. K. Fontaine, L. Zhu and G. Li, "Multi-channel Nonlinearity Compensation of128-Gb/s PDM-QPSK Signals in Dispersion-Managed Transmission Using Dispersion-Folded Digital BackwardQ g p g g p gPropagation", OFC’14, paper Tu3A.5 (2014)]

The DSP efficiency of EDC may be improved.[K.-P. Ho, “Subband equaliser for chromatic dispersion of optical fibre,” Electronics Lett. 45, 1224-1226, 2009][A. Tolmachev and M. Nazarathy, “Filter-bank based efficient transmission of Reduced-Guard-Interval OFDM,” OpticsExpress, vol. 19, no. 26, pp. 370–384, 2011]

FEC gain could be increased through wavelength diversity by averaging PDL, OSNR, andnonlinear effects.

[J. Rahn et al., “Transmission improvement through dual-carrier FEC gain sharing,” OFC’13, paper OW1E.5, 2013][H Zh l “30 58 Tb/ i i 7 230 k i PDM h lf 4D 16QAM d d d l i i h 6 1


[H. Zhang et al., “30.58 Tb/s transmission over 7,230 km using PDM half 4D-16QAM coded modulation with 6.1b/s/Hz spectral efficiency” OFC’13, paper OTu2B.3, 2013.]

[N. K. Fontaine, X. Liu, S. Chandrasekhar et al., "Fiber nonlinearity compensation by digital backpropagation of an entire1 2 Tb/s superchannel using a full field spectrally sliced receiver " ECOC’13 paper Mo 3 D 5 (2013)]

Full-field spectrally sliced receiver1.2-Tb/s superchannel using a full-field spectrally-sliced receiver, ECOC 13, paper Mo.3.D.5 (2013)]

A 1.2-Tb/s 5-carrier quasi-Nyquist-WDM superchannel (PDM-16QAM) detected by 5 receivers each at 40 GSa/s, achieving both time and phase alignments.

A 1.2-Tb/s 5-carrier quasi-Nyquist-WDM superchannel (PDM-16QAM) detected by 5 receivers each at 40 GSa/s, achieving both time and phase alignments.


(This type of receiver can be used for SDM-superchannel as well)(This type of receiver can be used for SDM-superchannel as well)

[N. K. Fontaine, X. Liu, S. Chandrasekhar et al., ECOC’13, paper Mo.3.D.5 (2013)]

Joint fiber nonlinearity compensation

Transmission distance: 960 km in dispersion-uncompensated 80-km SSMF spans

(a) Nonlinearity compensation performance at 6-dBm superchannel launch power;(b) Nonlinearity compensation performance at 8-dBm superchannel launch power;(b) Nonlinearity compensation performance at 8 dBm superchannel launch power;(c) Q2 factor vs. signal launch power for EDC only, single-carrier DBP, and entire

superchannel DBP (showing the benefit of the joint DSP); and(d) Recovered constellation diagrams at 6-dBm launch power.


Tradeoff between performance and DSP complexity needs to be madeTradeoff between performance and DSP complexity needs to be made

Conclusion

To enable sustainable communication capacity growth, the cost per

information bit needs to be continuously reduced.

Superchannels, with their improved spectral efficiency and natural

compatibility with large-scale photonic/electronic integration, are

expected to be well suited to meet this demand.

Equipped with digital signal processing at both the transmitters and theEquipped with digital signal processing at both the transmitters and the

receivers, superchannel-based transmission may play a key role in the

future evolution of optical networks to support the ever-increasing

demand for internet traffic and cloud services and enable software-

defined optical networking.


Thank you all!

Backup Slides


SUPER-NYQUIST WDM – 3BUT EVEN HERE, SHANNON IS THE LIMIT,

10] 10

64-QAM256-QAM

y [b

/s/H

z]

ISI

16-QAM

4-QAMOver-filtering, Super-Nyquist

effic

ienc

y

[Cai et al., ECOC’10]

MAP

1

4 QAM

Spec

tral

e

[P J Wi JLT 2012]

0 5 10 15 20 25Required SNR per bit [dB]

S [P. J. Winzer, JLT, 2012]

COPYRIGHT © 2014 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

58

SPECTRAL EFFICIENCY – AND ITS PRICE IN SNR10z]

1

al e

ffic

ienc

yza

tion

[b/s

/H

~1 dB/year

10

Hz] 256

SE = log2 (1 + SNR)

0.01

0.1

Spec

tra

per

pola

riz

1990 1994 1998 2002 2006 2010

~1 dB/year

eff

icie

ncy

atio

n [b

/s/H

2x

2x16

64

41990 1994 1998 2002 2006Year

2010x-pol

y-pol

x-pol

y-pol 1

Spec

tral

per

pola

riza 2x4

Laser & filter stabilityModulation

PDMDetection

0 5 10 15 20 25Required SNR per bit [dB]

p

3.7 dB 8.8 dB

20 years ago: Device physics set engineeringlimits on spectral efficiency

T d I f ti th t f d t l

Simple on-off modulation,direct detectionHi h d d l i

PDM

COPYRIGHT © 2014 ALCATEL-LUCENT. ALL RIGHTS RESERVED.

59

Today: Information theory sets fundamentallimits on spectral efficiency

Higher-order modulation,coherent detection

Download (PDF, 5.99MB)

Documents

Transcript of Download (PDF, 5.99MB)