Enabling Technologies and Challenges in Coherent Transport Networks David Dahan, Ph.D. ECI Telecom Ltd.

Drivers and Impact for Optical Networking Ecosystem Other IP Traffic IP Video Traffic 90% Video and more Video…. Internet streaming 2013 - IP at 5x 2008 levels with 90% Video Shifting to the Cloud… Enterprise and personal IT are moving to the cloud computing Service Providers become “All Play” providers… Exponential Traffic Growth Dynamic Traffic Networks Slow Revenue Growth Technology challenges Need more optical channel capacity : 100G/400G/1T Reduce cost /bit/switch/transport Business challenges Improve service provisioning, time and resource utilization : SDT/ROADM/SDN

Optical Network Evolution History and roadmap 2015- 2012 2011 2008 400G/1T Superchannel Bandwidth on Demand N:M ROADM configuration Gridless ROADM 400G/1T transceivers Fully automated network 40G/100G Coherent 100G Coherent networks support DCFless networks Colorless / directionless / contentionless WSON GMPLS-based 40G 80 Channels IP over WDM Mesh topology ASON GMPLS-based ODU basednetworks 10G/40G channels, ready for 100G coherent Plug and play 2006 10G 40 Channels SDH / Sonet / EoS services support Ring topology East/west protection Reconfigurable OADM WDM over OTN 10G channels SDH / Sonet Networks Increase capacity Point-to-point CWDM/DWDM Up to 40 channels 2.5/10G channels 2.5G CWDM, DWDM Continued demand for bandwidth from all applications

From Direct Detection to Coherent Detection Up to 10G (SE = 0.2 b/s/Hz) 40G/100G/200G coherent solution (SE > 2 b/s/Hz) Intradyne Coherent detection Phase and polarization diverse receiver Frequency Locked Lasers (<+/- 2 GHz) Digital Signal Processing at TX/RX TX RX 40G non coherent solution (SE = 0.8 b/s/Hz)

Current 100G Coherent Transceiver architecture Modulation format : DP-QPSK (Symbol Rate is ¼ Bit Rate : 2bit/s symbol x 2pol) Integrated Coherent receiver Integrated PDM QPSK MZM LiNBO3 Modulator 40 nm CMOS ASIC with 4 (8 bit resolution) x63 Gsamples/s ADC * Nelson et al, “A Robust Real-Time 100G Transceiver With Soft-Decision Forward Error Correction” J. OPT. COMMUN. NETW, vol 4, no. 11,2012

Current 100G Coherent Transceiver architecture Ix Qx Qy Iy DSP block ADCs Coh. Rx S LO 90° Hybrid & Detector j Frequency & phase recovery Slicing Clock recovery &Interpolation Resampling Soft Symbol estimation SD- FEC decoder 120G OTU4 112G D>60000 ps/nm PMD>30 ps 1 bit +reliability bit info 6-8 bits Gen Type Code FEC Over head Pre-FEC BER TH. For post FEC<10-15 Coding gain [dB] 1st HD BCH (Bose-Chaudhuri-Hocquenghem) and RS (Reed-Solomon) codes 7% ~10-4 6-7 dB 2nd Concatenation of RS codes, Viterbi convolutional codes and BCH codes (CBCH) ~1x10-3 (EFEC) 8.5-10 dB 3rd SD BCT (Block Turbo code) or Turbo Product Code (TPC) and LDPC (Low Density Parity Check) codes 15%-20% ~2x10-2 10.5-11.5 dB

100G submarine Field trial over 4600 km The 100G trial was carried out over Bezeq International’s live operational submarine fiber, in conjunction with the TeraSanta Consortium : demonstration of advanced capabilities of ECI 100G transmission system and technologies in compensating for non-linear channel impairments and chromatic dispersion utilizing advanced SD-FEC algorithms.  DMUX 100G Apollo Platform 100G MUX

Flexgrid tunable laser Next Generation of Coherent Transceiver : : Software Defined Transceiver (SDT) Si Photonics IC with Electronic and Optical functionality 28 or 20 nm CMOS ASIC with DAC/ADC and DSP capabilities in both TX/RX Power reduction Higher computational strength Adapt modulation format/Symbol rate Technol. Gate ADC (8bits) DAC (8bits) GA 28 nm 150-200M 110-130 GS/s 1.7W 0.7W 2013 20 nm 200-250M 1.2W 0.5W 2014 Client Data Rate 100G/150G/200G/400G/1T FEC overhead 0%-30% Modulation format BPSK/QPSK/ 8-QAM/16QAM TX DSP Pulse Shaping Optical Carrier Flexgrid tunable laser (C/L band)

Raised Cosine FIR filter New DSP features Nyquist spectral shaping at TX : increases of the spectral efficiency by reducing the channel bandwidth to ~ symbol rate Raised Cosine FIR filter

New DSP features Self diagnostic monitoring features : Accumulated Chromatic Dispersion monitor PMD monitor OSNR monitor ESNR monitor Still missing : Efficient nonlinear compensation technique Current state of the art techniques based on digital back propagation or Volterra Series are too complex for real time ASIC implementation Nonlinear optical impairments are the ultimate limitations in optical network

Transmission Technology options for 400 Gb/s Modulation Gbit/s OSNR min [dB] DP-QPSK 120 12.5 DP-16QAM 240 18.5 480 21.5 DP-256QAM >30 f 4x120G 4 bands with DP-QPSK (30Gbaud) No spectral efficiency improvement over 100G Suitable for long haul (>2000 km) Symbol Rate Limitations of DACs/ADC and electronics 90 Gbaud 60 Gbaud 1x480G 1 bands with DP-16 QAM (60 Gbaud) High spectral efficiency Reach Limited to Metro (~700 km) 30 Gbaud 2 bands with DP-16 QAM (30 Gbaud) High spectral efficiency Reach Metro /Long Haul distances f 2x240G 1 bands with DP-256 QAM (30 Gbaud) Extremely high spectral efficiency Reach Limited (~100 km) f 1x480G QPSK 8-QAM 16-QAM 32-QAM 64-QAM 256-QAM Constellation size 1 Reach Limited <<100km 2 3 4 Subcarriers/band

Hybrid Raman Amplifiers Improving transmission reach Complex Coherent modulation formats like 200G DP-16QAM require for 6-8 dB OSNR improvement with respect with current 100G DP-QPSK modulation format The use of hybrid Raman-EDFA amplification schemes is required to improve the received OSNR or mitigate the nonlinear penalties by lowering the launched power into the fiber : can improve the transmission reach by 100% Non linear impairments Low OSNR With Hybrid Raman –EDFA amplification Non linear impairments Low OSNR Non linear impairments Low OSNR

Improving spectral efficiency beyond 100G Superchannels Improving spectral efficiency beyond 100G Future services of 400Gb/s and 1T will be packed into super channels, in order to provide optimum flexibility and reach performance tradeoffs : 400G : 2 channels spaced by 37.5 GHz 1T : 5 channels spaced by 37.5 GHz For optimized spectral efficiency, Super channels use Nyquist spectral shaping and Flexgrid WSS ROADMs

Flexgrid Networks To increase spectral efficiency, we move from a fixed channel grid (50GHz/100GHz) to flexible channel grid management : 6.25 GHz grid 12.5 GHz bandwidth granularity The channel spectral slot is adapted on a per channel basis using : 10G/ 40G on 25 GHz slot 100G and 200G on 37.5 GHz slot 400G on 75 GHz slot 1T on 187.5 GHz slot 400G 1T 100G 50 GHz f Fixed 50GHz grid 10G 40G 50 GHz 400G 1T 100G f Flex grid 40G 10G Increase by 25 % the available useable fiber bandwidth

Flex Grid Technology enablers Very stable tunable lasers compatible with 6.25 GHz grid resolution Flexgrid ROADMs : First generation of WSS allocated a channel on a single MEM based pixel Flexible WSS based on LCoS technology use a flexible matrix based wavelength switching platform with megapixel matrices allowing programmable channel bandwidth * EXFO Webinar : “400G Technologies: the new challenges that lie ahead”,04/02/2014 http://www.exfo.com/library/multimedia/webinars/400g-technologies-challenges

Optical Network Node with Full Flexibility Network node capabilities are enhanced with new features allowing full flexibility : Flexgrid : any channel/ superchannel can be directed towards any other node Colorless Directionless Contentionless Flexgrid WSS

Optical Network Node with Full Flexibility Network node capabilities are enhanced with new features allowing full flexibility : Flexgrid Colorless : any wavelength can be added or dropped at any port Directionless Contentionless

Optical Network Node with Full Flexibility Network node capabilities are enhanced with new features allowing full flexibility : Flexgrid Colorless Directionless : any wavelength can be directed at any direction an reach a given port Contentionless

Optical Network Node with Full Flexibility Network node capabilities are enhanced with new features allowing full flexibility : Flexgrid Colorless Directionless Contentionless : Multiple channels of the same wavelength can be dropped or added by a single module

Optimum management of the optical spectrum resources Optimized routing and resource allocation algorithms for flexible optical networking Conventional Routing and Wavelength Assignment (RWA) algorithms can be used only for rigid grid networking New paradigms based on Routing and Spectral allocation Assignment (RSA) algorithms should be developed for flexible grid networking Physical Impairment awareness and optimal combination of Software Defined Transceiver parameters (modulation format/symbol rate, FEC overhead) will be required Need to find optimum strategies for spectrum defragmentation Rigid grid network Flex grid network

Software Defined Networking : Why ? Flexible Multi-layer Networking Bandwidth hungry services (video, mobile data, cloud services) lead to new traffic characteristics : Rapidly changing traffic patterns High Pic to average traffic ratio Large Data chunk transfers Asymmetric traffic between nodes SDN will turn the networks into programmable virtualized resource for better efficiency and automation

Software Defined Networking Flexible Multi-layer Networking Application requirements Dynamic connectivity Bandwidth QoS Resiliency User I/Fs Network Apps Open APIs SND Control Plane Hardware Abstraction & Virtualization SDN Control Plane Aware of Application requirement Optimized resource and configuration OpenFlow Multi layer Network Elements Ethernet switch/MPLS router OTN switch ROADM, SDT Fiber switch Multi layer network elements

Conclusion The future optical transport networking will provide better Capacity : coherent modulation formats, superchannel, better SE Flexibility : software defined transceivers, flexible grid, flexible CDC ROADMs nodes Resource utilization : impairment aware- RSA algorithms, SDN The future optical transport networking needs to provide : Efficient nonlinear optical impairment compensation techniques Strategies for pro-active and re-active spectrum defragmentation and fragmentation awareness in service expansion and contraction policies Energy efficient strategies Capex and Opex reductions