Single- versus Dual-Carrier Transmission for Installed Submarine Cable Upgrades

Transcription

1 Single- versus Dual-Carrier Transmission for Installed Submarine Cable Upgrades L. Molle, M. Nölle, C. Schubert (Fraunhofer Institute for Telecommunications, HHI) W. Wong, S. Webb, J. Schwartz (Xtera Communications) Heinrich-Hertz-Institute Berlin

2 Presenter Profile Lutz Molle is a research associate at the Fraunhofer Institute for Telecommunications, Heinrich- Hertz-Institute Berlin (HHI) since 2005, where he is with the dept. Photonic Networks. He has been working on long-haul system design and upgrades of submarine systems. Lutz.Molle@hhi.fraunhofer.de

3 Introduction Fact: The ever-increasing capacity demand on back-bone networks Two ways for capacity increase on submarine infrastructure: New-builds Upgrades New & improved fibre, no DCF Fixed wet plant Larger bandwidth (C+L-band) Fixed bandwidth ( C-band) Coherent technology Coherent technology Advanced modulation (mqam) Lower-order modulation ( QPSK) 10Tb/s per fibre for >9000 km 3 Tb/s per fibre for > 9000 km More challenging, but Very long planning & lead time Quicker set up Very expensive Much lower cost This talk is on upgrades of 2 nd generation submarine links

4 What We ll Discuss Today Upgrade Aspects Transmission Link and 100G Test Equipment Single-carrier transmission on 3600-km link Single- & Dual-carrier transmission on 5600-km link Summary

5 Upgrade Paths for Legacy Links 2 nd generation systems were designed for 10G channels in 50GHz grid 10-fold capacity: replace old 10G channels by 100G-QPSK really? Possible problems (some) Q-margin low for 100G-QPSK at typ. delivered OSNR of 14 16dB Possible solutions (some) Stronger FEC Fewer channels (limited by nonlinearities) < 100Gb/s bitrate per channel Nonlinear impairments (e.g. SPM, phase flips, XpolM, ) Robust modulation (e.g. BPSK) Optimization of transmit pulse DSP (robust CPE, pilot symbols, MAP/MLSE decoding, DBP, ) This talk: Dual-carrier BPSK, RZ shaping, polarization interleaving

6 Investigated Wet Plant Links Two segments of 2 nd -gen. deployed submarine link, optical loop-back. NZDSF link, dispersion compensation by hybrid DCF/NZDSF spans Far End Terminal Station #1 Link #1 Link # km Near End Terminal Station km Far End Terminal Station #2 Link Parameter Link #1 (loopback) Link #2 (loopback) Length 3600 km 5800 km No. of Repeaters Repeater Spacing 75 km avg. 70 km avg. Repeater Output Repeater Bandwidth Repeater Noise Fig. 11 dbm 15 nm 5 db Link #2 Dispersion Map avg. dispersion slope ~0.07 ps/km/nm² THz ( nm) THz ( nm) THz ( nm)

7 Channel spacing [GHz] Expected Linear Performance Maximum reach due to noise (ASE) accumulation along link: Maximum system length (in km) for 70km repeater spacing (BW EDFA = 14.0nm, P EDFA = 11.0dBm, NF EDFA = 5.0dB, fiber = 0.21dB/km) Calculate required OSNR at FEC limit for used modulation format and bit rate 2. Add sufficient OSNR margin 3. Select proper channel spacing 4. Read maximum reach No. of Channels e.g. 120Gb/s-DP-QPSK: Required OSNR* = 11.2 db Additional margin = 3.0 db Required link-osnr = 14.2 db Chanel spacing = 50 GHz Required Rx-OSNR [db in 0.1nm] Reach is noise limited to ~7000 km with load of G-QPSK channels * req. OSNR for pre-(sd-)fec-ber =

8 20 l odd ILV 20 l even 100G Transmitter Test Equipment l1 RZ DP-IQ DGD EDFA 120Gb/s test channel: Dual-carrier BPSK or single-carrier QPSK DATA TDC EDFA 3dB to line l2 RZ DP-IQ DGD EDFA 3dB 100G QPSK 100G QPSK 100G Loading Channels blue mid red RZ-BPSK Pol-ILV 100G channel as dual-carrier DP-BPSK or single-carrier DP-QPSK Flexible RZ pulse carving, optional polarisation interleaving (pol-ilv) Insertion of test channel in loading comb in blue, mid or red band

9 Pol. Diversity 90 opt. Hybrid 100G Receiver Setup + Offline DSP EDFA from line 3dB RX 2 VOA EDFA opt. BPF OSNR EDFA LO Laser BD1 BD2 BD3 BD4 I X Q X I Y Q Y ADC: 4 80GS/s (8bit) PC for offline DSP Receiver DSP (Offline) Resampling Correction of optical frontend For dual-carrier : sub-carrier pre-filtering 14dB OSNR 120Gb/s DP-QPSK 60Gb/s DP-BPSK (1l) CD compensation LO offset correction Blind equalizer (CMA, DFE) Carrier recovery (V&V) Polarization-diversity coherent receiver High-bandwidth (36 GHz) real-time scope for ADC Offline-DSP for signal recovery and BER counting Differential decoding BER

10 Q (db) 100G Transmission on Link #1 (3600 km) Influence of RZ pulse shape for 100G single-carrier DP-QPSK Sweep over RZ pulse-width* Plain NRZ as ref. RZ-50 gains ~2 dbq Weak improvement with wider RZ-spectra (~0.5 dbq) Further tests: PC-RZ pulse RZ-50 PC-RZ 7.5 NRZ rms spectral width (pm) 500 * By adjustment of RZ-modulator bias. Only for MID-band channel, with Pol-ILV & const. channel power (RX-OSNR=19dB)

11 Q / db 100G Transmission on Link #1 (3600 km) Influence of Polarisation Interleaving for 100G single-carrier QPSK Sweep over channel power* RZ-50 w/o pol-ilv as reference Pol-ILV gains ~1 dbq PC-RZ gains ~0.8 dbq on top Increase of optimum Q value directly related to increase of non-linear limit * Only for RED-band channel B2B FEC Pol-ILV RZ-50 RZ-50 Pol-ILV PC-RZ OSNR (db in 0.1nm)

12 Q / db 100G Transmission on Link #2 (5600 km) Blue, mid and red band for 100G single- and dual-carrier format 100G single-carrier QPSK: about 2dB margin for red & blue band, <1dB@mid Blue and red band performance similar 100G dual-carrier BPSK: about db higher optimum Q Simultaneous dual-carrier detection resulted in ~1.4dB lower optimum Q due to electrical bandwidth limitations Simultaneous detection still desirable for cost savings * 120-Gb/s Single-carrier PC-RZ-QPSK ** 120-Gb/s Dual-carrier PC-RZ-BPSK B2B FEC Dual Carrier** Single Carrier* OSNR / db in 0.1nm

13 Summary Field test of 100G single- vs. dual-carrier on NZDSF link Mid band most critical (as expected): ~1.5dB penalty 3600-km transmission of QPSK improved by optimised RZ pulse-shaping & Polarisation Interleaving 5400-km transmission of QPSK is nonlinearly limited 5400-km transmission of dual-carrier BPSK has about 2.5dB higher optimum Q (despite same linear performance) dual-carrier BPSK more robust against nonlinearities, but requires almost doubled hardware effort and results in lower spectral efficiency

14 Thank you for your attention!