Digital Coherent Transmission: A Paradigm Shift of Optical Transmission Technology

conference & convention enabling the next generation of networks & services Digital Coherent Transmission: A Paradigm Shift of Optical Transmission Technology Shoichiro Oda, Toshiki Tanaka, and Takeshi Hoshida Fujitsu Laboratories Ltd. 1

Outline Introduction Merit of digital coherent transmission systems Digital coherent transceiver and its key components Digital signal processing algorithms Nonlinear mitigation algorithms Dispersion map optimization Summary 2

Introduction 1980: Proposal of coherent (heterodyne) receiver T. Okoshi and K. Kikuchi, Electron. Lett., 28th Feb. 1980, vol. 16, no. 5, pp. 179-181. F. Favre and D. Le Guen, Electron. Lett., 28th Aug. 1980, vol. 16, no. 18, pp. 709-710. During 1980s: Many R&Ds on coherent systems. 1991: Proposal of digital coherent (intradyne) receiver F. Derr, Electron. Lett., 7th Nov. 1991, vol. 27, no. 23, pp. 2177-2179. Coherent (intradyne) detection 3 ADC + DSP R&Ds on coherent systems were postponed because of advents of erbium doped fiber amplifier (EDFA) on 1989 and WDM technology on early 1990s.

Introduction 2003~: Reactivating R&Ds on digital coherent technology Demand of increasing capacity while keeping the same reach and total bandwidth as current system Demand of mitigation of large distortion due to CD and PMD in high bit rate system. Progress of CMOS technology, catching up with bit rate in optical communications. Relaxation of laser requirement because of increase of bit rate in optical communications. 2003: Experimental demonstration of digital coherent detection with digital equalization of chromatic dispersion M.G. Taylor, ECOC 2003, Paper We4.P.111. 2004: Proposal of feed-forward digital carrier recovery R. Noe, OECC 2004, Paper 16C2-5. 2005: Experimental demonstration of demodulation of 40Gb/s polarization multiplexed QPSK after 200 km transmission S. Tsukamoto et al., OFC 2005, PDP29. QPSK: Quadrature phase shift keying CD: Chromatic dispersion PMD: Polarization mode dispersion 4

Introduction Key results in long-haul and high-capacity transmission experiments presented in OFC and ECOC post-deadline session Conference Paper No. Affiliation Ch No. x Bitrate/ch Distance Features ECOC2009 PD2.5 Bell Labs, Alcatel- Lucent 155 ch x 112 Gb/s 7200 km Digital coherent, 112Petabit/s km, transatlantic distance OFC2010 PDPB7 NTT 432 ch x 171 Gb/s 240 km Digital coherent, 69.1 Tb/s of total capacity OFC2010 PDPB8 Bell Labs, Alcatel- Lucent 10 ch x 224 Gb/s 1200 km Digital coherent, 4 bit/s/hz, >200Gb/s/ch OFC2010 PDPB10 Tyco 96 ch x 112 Gb/s 10,608 km Digital coherent, 3 bit/s/hz, transpacific distance Digital coherent technology is essential for long-haul and high-capacity transmission system. 5

Features of digital coherent receiver Exemplified digital coherent receiver Signal Local laser (LO) Polarization diversity 90 degree hybrid AD convertors (4ch) Digital signal processing Features of digital coherent receiver Higher OSNR tolerance Full-access to optical field of signal 7 OSNR: Optical signal to noise ratio

OSNR tolerance OSNR tolerance of coherent detection 10-2 ~2.2 db improvement: roughly translates to 50% reach extension 10-3 Bit error ratio 10-4 10-5 10-6 10-7 Coherent detection Direct detection 10-8 10-9 10-10 10-11 43.2 Gb/s, RZ-DQPSK 8 9 10 11 12 13 14 15 16 17 18 19 20 OSNR [db] RZ-DQPSK: Return to zero differential quadrature phase shift keying 8

Full-access to optical field Signal E h = A h exp( j h ) E v = A v exp( j v ) Local laser (LO) Polarization diversity 90 degree hybrid A h cos h A h sin h A v cos v A v sin v AD convertors (4ch) DSP Recovering signal information, amplitude, phase, polarization Full-access to optical field of the signal Digital equalization: CD, PMD, nonlinearity Digitizing signal Handling in DSP Multi-level modulation formats (high spectral efficiency): polarization multiplexing, QAM, OFDM, and etc. CD: Chromatic dispersion, PMD: Polarization mode dispersion QAM: Quadrature amplitude modulation, OFDM: Orthogonal frequency divition multiplexing 9

Merit of digital coherent transmission systems Conventional Many optical components Tx Tx VDC VDC Fiber VDC VDC receiver receiver Tx VDC Insufficient reach VDC receiver VDC: Variable dispersion compensator Digital coherent Less optical components: reduced CAPEX and OPEX Multi-level modulation: high capacity Tx Tx Fiber Digital coherent Rx Digital coherent Rx Tx Extended reach Digital coherent Rx Longer reach, higher capacity, and lower cost 10

Digital coherent transceiver and its key components Transmitter H-pol. I H-pol. Q Tunable Laser V-pol. I V-pol. Q PBC Integrated transmitter Receiver Tunable Laser Polarization diversity 90 degree hybrid Integrated receiver ADC + DSP 12

Standardization (1) Transmitter integrated photonics Integrated polarization multiplexed quadrature modulated transmitters Implementation agreement approved on March 2010 in OIF Functionality shown in yellow area Electrical interface Mechanical specification Opto-electronic properties Etc. OIF-PMQ-TX-01.0 OIF: Optical internetworking forum 13

Standardization (2) Receiver integrated photonics Integrated dual polarization intradyne coherent receivers Implementation agreement Functionality (shown in yellow area) High-speed electrical interface (between the receiver and ADC) Low-speed electrical interface Mechanical specification Etc. (To be approved) Signal Local oscillator PBS BS 90 deg optical hybrid 90 deg optical hybrid X I X Q Y I Y Q ADC ADC ADC ADC 14

Digital coherent receiver Digital coherent receiver Receiver front-end (FE) Digital signal processing Polarization diversity 90 degree hybrid AD convertors (4ch) FE imperfection compensation Semi-fixed equalizer Adaptive equalizer Laser frequency Offset compensation Carrier phase recovery Decision Mux / Output interface Local laser (LO) Intradyne detection Digital equalization Front-end imperfection compensation We discuss the detection of DP-QPSK signal. 16

Coherent detection Frequency Control of local laser Receiver bandwidth Electrical spectrum f Heterodyne Homodyne Intradyne sig f LO Automatic frequency control Optical PLL Free-running commercial DWDM source 2x ~ symbol rate ~ symbol rate ~ symbol rate f IF BW 0 f f f f f BW sig 0 LO f sig LO Fluctuation by laser frequency drift 0 f We focus on intradyne detection. 18

Phase recovery algorithms Received signal: E n Aexp[ j( ( n) ( n) nt )] s e QPSK symbol phase Carrier phase offset e Carrier frequency offset Signal LO frequency Random phase deviation much slower than symbol rate (less than ~10MHz) Phase rotation due to laser frequency mismatch (5 GHz max.) Symbol phase rotates due to carrier phase and frequency offset. 19

Frequency offset compensation Phase rotator e E n Aexp[ j( ( n) ( n) nt )] s e Frequency offset estimator E' n Aexp[ j( s ( n) e( n))] Examples of frequency offset estimation algorithms Mth power based estimator A. Leven et al., PTL 19, pp.366-368, 2007. I+jQ Z -1 ( )* ( ) m N arg( )/m T Pre-decision-based angle differential estimator (PADE) Li et al., OFC08, OWT4. I+jQ arg( ) Data decision Z -1 Z -1 abs( )>? No Loop filter T 20

Carrier phase recovery Mth power carrier phase recovery algorithm A. J. Viterbi et al., Trans. Inform. Theory, IT-29, pp. 543-551, 1983. D. S. Ly-Gagnon et al., JLT, 24, pp. 12-21, 2006. n e E' Aexp[ j( ( n) ( n))] s e Carrier phase estimator m 1 ) arg( m ( ) 1 N m e arg m n 1 E'( n) Phase rotator E r n Aexp[ j ( n)] s Symbol phase is recovered by digital cancellation of carrier phase and frequency offset. 21

Semi-fixed equalizer Semi-fixed equalizer is responsible for coarse compensation of large dispersion. Semi-fixed equalizer is implemented by temporal domain filter (FIR filter) and frequency domain filter. Time domain filter (FIR filter) Input T/2 T/2 T/2 Frequency domain filter k(1) C(1) C(M-2) C(M-1) C(M) Input FFT k( M 1) IFFT Output Output k(m ) 23

Complexity comparison between time and frequency domain filter 24

Adaptive equalizer Adaptive equalizer is responsible for Polarization demultiplexing Equalization of distorted waveform due to PMD, residual CD, etc. Butterfly structure FIR filters H pol. input FIR xx H pol. output FIR xy V pol. input FIR yx FIR yy Adaptation algorithm V pol. output Adaptation algorithms LMS (least mean square) algorithm: Data aided equalization CMA (constant modulus algorithm): Blind equalization 25

PMD compensation 43Gb/s DP-QPSK transmitter Polarization controller DGD emulator Optical power P = Px + Py Power splitting ratio = Px /(Px +Py ) P Digital coherent receiver OSNR=13dB 4x7-tap Butterfly FIR filter Q (db) 11.80 11.80-12.00 11.60 11.60-11.80 11.40 11.40-11.60 11.6 11.20 11.20-11.40 11.4 11.00 11.00-11.20 11.2 10.80 10.80-11.00 11.0 1121 125 87.5 10.00 10.20 10.40 10.60 10.80 11.00 11.20 11.40 11.60 11.80 12.00 100 62.5 8.00 8.20 8.40 8.60 8.80 9.00 9.20 9.40 9.60 9.80 75 0.10 0.20 0.40 12.5 37.550 50 0 25 0.3 0.5 0 050 26 DGD (ps) DGD: Differential group delay

Digital coherent receiver Digital coherent receiver Receiver front-end (FE) Digital signal processing Polarization diversity 90 degree hybrid AD convertors (4ch) FE imperfection compensation Semi-fixed equalizer Adaptive equalizer Laser frequency Offset compensation Carrier phase recovey Decision Mux / Output interface Local laser (LO) Intradyne detection Distortion equalization Front-end imperfection compensation 27

Receiver front-end imperfection Signal quality may be impaired due to receiver front-end imperfection. Phase error of 90 degree optical hybrid Bandwidth imbalance Signal Local laser (LO) Polarization diversity 90 degree hybrid AD convertors (4ch) Digital signal processing Skew Amplitude imbalance 28

Skew tolerance Q-penalty [db] 2 1 112 Gb/s DP-NRZ-QPSK OSNR 18dB CD 21,000 ps/nm 3ps skew LO w/o. skew Receiver front-end 1 2 3 4 Skew= 1-2 I Q I Q ADC (4ch) DSP 0.2 0-40 -20 0 20 40 Skew [ps] Improvement of skew tolerance is required. 29

Digital skew compensation DSP T. Tanimura et al., ECOC2009, 7.3.2. LO Receiver front-end ADC (4ch) Digital skew compensator Equalizer Phase recovery Decision I (H pol.) Q (H pol.) I (V pol.) Q (V pol.) input Z -1 Z -1 C 0 C 1 C 2 output Digital skew compensator with 3 tap FIR. 30

Improvement of skew tolerance Q-penalty [db] 2 1 with Skew compensator without Skew compensator 7x improvement 112 Gb/s DP-NRZ-QPSK OSNR=18 db CD=21,000 ps/nm 0.2 0-40 -20 0 20 40 Skew [ps] Requirement of skew can be relaxed by digital skew compensator. 31

Mitigation of nonlinear impairments OSNR improvement by increasing optical power Nonlinear limit OSNR limit Performance degradation due to nonlinear impairments Q-value Optical power Algorithms for mitigating the both intra and inter-channel non-linear impairments are expected to extend the transmission reach. 33

Digital backpropagation nonlinear equalizer (DBP-NLE) S. Oda et al., OFC2009, OThR6. T. Tanimura et al., ECOC2009, 9.4.5. Tx Tx LO Receiver front-end ADC (4ch) DSP Digital backpropagation Adaptive Equalizer Phase recovery Decision distorted signal input 1st stage LE NLE N-th stage LE NLE equalized signal output H-pol. V-pol. 2 x x x 2 x x exp(-j ) + + x exp(-j ) H-pol. V-pol. 34

Experimental demonstration Launched power: +5.5 dbm/ch. 112Gb/s Symbol-aligned DP-NRZ-QPSK, symbol-interleaved DP-RZ-QPSK ch. 10 1546.12nm ch. 1 ch. 20 112 Gb/s DP-QPSK 112 Gb/s DP-QPSK PC PC CPL PC SW SW ASE CPL LO CPL DEMUX Optical hybrid PD PD PD PD DSO 50 GSa/s 60 km SMF 60 km 60 km 60 km 60 km ROADM #1 ROADM #5 ROADM #4 ROADM #3 ROADM #2 EDFA WSS EDFA No optical dispersion compensation WSS: Wavelength-selective switch, DSO: Digital storage oscilloscope 35

Q-improvement vs. spans Q-improvement [db] 2.5 2.0 1.5 1.0 0.5 Symbol-aligned NRZ Symbol-interleaved RZ OSNR=20 db 0.0 0 5 10 15 20 Number of Spans > 2dB Q-improvement Q-improvement by DBP-NLE exceeded 2dB regardless of pulse formats. 36

Nonlinear polarization crosstalk Polarization crosstalk due to XPM L. Li et al., OFC2010, Paper OWE3. V t XPM V t Polarization crosstalk H H H-pol: S h V-pol: S v Polarization crosstalk monitor R h R v S h W hv=(r h S h )/S v Averaging S v W vh =(R v S v )/S h Averaging 37 W hv W vh H-pol: R h = S h + W hv S v V-pol: R v = S v + W vh S h Monitored autocorrelation of W hv Auto correlation of W hv (a.u.) 0.02 0.01 0-50 -30-10 10 30 50 Symbol Period

Nonlinear polarization crosstalk canceller (NPCC) DSP L. Li et al., OFC2010, Paper OWE3. Signal LO Receiver front-end ADC (4ch) Equalizer Phase recovery NPCC Decision R h Delay Delay R v + - W hv S h W hv=(r h S h )/S v Averaging R h W vh =(R v S v )/S h Averaging R v S v Delay Delay 38 R h + W vh - R v

Experimental demonstration 40 x 112 Gb/s DP-QPSK 100 GHz channel spacing Launched power: +2.5 dbm/ch. ch. 1 ch. 19 ch. 2 ch. 40 Tx 112 Gb/s DP-QPSK 112 Gb/s DP-QPSK PC PC CPL PC SW ASE LO CPL DEMUX Optical hybrid PD PD PD PD DSO 50 GSa/s SW CPL Node #5 60 km Node #4 SMF 60 km 60 km 60 km 60 km Node #3 Node #2 39 EDFA Node #1 DCF WSS EDFA 105% in-line dispersion compensation

Q-improvement vs. distance Q Improvement (db). 1.5 1 0.5 0 OSNR=18dB NRZ aligned 300 600 900 1200 1500 1800 Distance (km) With DBP-NLE, Q-improvement by NPCC further improved besides the Q-improvement by the DBP-NLC itself. 0.9 db Q-improvement 40

Dispersion map optimization Digital coherent Rx with large chromatic dispersion tolerance possibly can remove in-line dispersion compensation (DC) and pre-dcf Optimizing amount of in-line DC ratio and pre-dcf System model Launched power: 0dBm/ch. 50 GHz channel spacing 100Gb/s DP-QPSK Tx Node Digital Coherent Rx 100Gb/s DP-QPSK Tx Pre- DCF SMF 60 km WSS DCF x25 Digital Coherent Rx 42

Results Q penalty (db) 7 6 5 4 3 2 1 0 In-line DC ratio 98% 0% 20% 40% 60% 80% -25000-20000 -15000-10000 -5000 0 5000 10000 15000 Pre-DCF [ps/nm] 0% in-line DC ratio with optimum pre-dcf is best performance. Pre-DCF improves performance regardless of in-line DC ratio. The improvement by pre-dcf increases with in-line DC ratio. 43

80ch x 112Gb/s DP-QPSK, 3000km-long DCF-less transmission experiment No optical dispersion compensation ch. 1 ch. 79 ch. 2 ch. 80 112 Gb/s DP-QPSK 112 Gb/s DP-QPSK SW ASE LO Optical hybrid PD PD PD PD 50 GSa/s DSO SW SSMF 100 km SSMF 100 km SSMF 100 km SSMF 100 km SSMF 100 km SSMF 100 km Repeater #6 Repeater #5 Repeater #4 Repeater #3 Repeater #2 Repeater #1 After 3000km Q= 8.8dB at OSNR=15dB Q H-pol. I 44 Q V-pol. I Please visit Fujitsu booth for further information.

Summary Basics and recent progress of digital coherent technology have been reviewed. Features of digital coherent receiver: Higher OSNR tolerance, Full-access to optical field Digital signal processing algorithms: Phase recovery, Digital equalization Nonlinear mitigation algorithms: DBP-NLE, NPCC Dispersion map optimization: DCF-less transmission system Digital coherent technology is indispensable for long-haul and high-capacity transmission systems. 45

Acknowledgements The authors thank the contributions of the following colleagues Jens C. Rasmussen Hisao Nakashima Takahito Tanimura Zhenning Tao Lei Li Liang Liu Izumi Yokota Akira Sugiyama Hiroshi Nakamoto Hiroyuki Irie 46

2010 conference & convention enabling the next generation of networks & services The 7th International Conference & Convention on Undersea Telecommunications Pacifico Convention Plaza Yokohama & InterContinental The Grand Yokohama 11 ~ 14 May 2010 www.suboptic.org 48