1. The Dawn of the Terabit Age 100G and 1T Transport Architectures Geoff Bennett: Director, Solutions & Technology

2.  Internet demand “set to grow by a factor of 1,000 over the next 20 years” 2© 2011 Infinera Corporation Confidential & Proprietary “Technology and Architecture to Enable the Explosive Growth of the Internet”* *Adel A. M. Saleh and Jane M. Simmons; IEEE Communications Magazine, January 2011 Let’s look at one way we can enhance spectral efficiency in practical transport networks

3. Private Line SONET transport Transport Network (OTN/WDM) Wholesale Bandwidth SAN Core / Backbone Network Understanding The Transport Layer IP Core Routers IP Peering IP-VPNL2 VPN Internet Access IPTVVoIP Service Routers

4.  A key factor in scaling the network is the product of the data rate of a single wavelength, and the number of wavelengths on a fiber In other words, the total fiber capacity  How has the optical transmission industry responded to the challenge to scale total fiber capacity? 4© 2011 Infinera Corporation Confidential & Proprietary Total Fiber Capacity

5. Progress in Optical Communications Spectral Efficiency (bit/sec/Hz) 0.1 1 10 Voice-band Modems (V34) Microwave Radio Wireless (per site) Optical Fiber 1980199020002010 10G on 25 GHz 40G coherent DSP Electronic PMD comp. 100G on 50GHz benefiting from advances in DSP and VLSI Electronic CD compensation 5© 2011 Infinera Corporation.

6. What’s changed so far Since the advent of DWDM… now Phase Modulation Coherent Detection ITU Frequency Grid Intensity Modulation Direct Detection ITU Frequency Grid 6© 2011 Infinera Corporation.

7. What Comes Next For Terabit Transport? Since the advent of DWDM… …so what has to change Phase Modulation Coherent Detection ITU Frequency Grid Intensity Modulation Direct Detection ITU Frequency Grid Quadrature Amplitude Modulation (QAM) Coherent Wave Combining and Separation Grid-less FlexChannels 7© 2011 Infinera Corporation.

8. Advanced Modulation Formats Pol-Mux QPSK Pol-Mux 8-QAM Pol-Mux 16-QAM IM-DD PM- BPSK 1.68121624 C-Band Capacity (Tb/s) 0 0.2 0.4 0.6 0.8 1 1.2 Capacity * Reach Product 8© 2011 Infinera Corporation Confidential & Proprietary

9.  Phase-based modulation is generally more tolerant of fiber impairments  We can put more bits per symbol Magnitude of fiber impairments scale with square of symbol rate  Phase-based modulation allows us to use coherent detection Advanced digital signal processing is then possible  The result is that 100G waves have the same, or even better reach than 10G IM-DD 9© 2011 Infinera Corporation Confidential & Proprietary Advanced Modulation Benefits

10. 10© 2011 Infinera Corporation Confidential & Proprietary But we have a problem… Practical advanced modulation and coherent detection only get us to 100G waves How do we get to “Terabit waves”? Do we even want “Terabit waves”? We need a Terabit unit of capacity that can be managed in “one chunk”, and has the maximum spectral efficiency

11. 11© 2011 Infinera Corporation Confidential & Proprietary The Problem With Single Carrier Terabit Coherent detection requires very high speed Analog to Digital Converters This is a simplified coherent detector Can we actually build them?

12. © 2011 Infinera Corporation Confidential & Proprietary12 1Tb/s Single Carrier: The A/D Converter Problem 124681012 1 2 3 4 5 6 7 8 9 10 OSNR Penalty (dB) Number of bits per symbol PM-BPSKPM-QPSK PM-8QAM PM-16QAM PM-32QAM PM-64QAM By 2014 commercial ADCs are expected to operate at ~64GBaud 640GBaud320GBaud 210GBaud 160GBaud 128GBaud 105GBaud

13.  For the foreseeable future, Terabit units of Transport Capacity will probably not be built using a single wavelength  Given that a multi-wavelength approach is the practical option… 13© 2011 Infinera Corporation Confidential & Proprietary Interim Conclusion What technologies are needed to make practical Terabit “super-channels” possible?

14. Since the advent of DWDM… …let’s look at that evolution Quadrature Amplitude Modulation Coherent Wave Separation Grid-less FlexChannels On-Off Keyed Modulation Direct Detection ITU Frequency Grid What Comes Next For Terabit Transport? 14© 2011 Infinera Corporation.

15. wavelength demux DWDM Direct Detection PD Spacing on the fiber needed between waves: “Guard Bands” Spatially separate the channels using a wavelength demux 15© 2011 Infinera Corporation. …11010110… Direct conversion of photons into electrons that “look like” bits Little or no requirement for complex detector c. 1995 – 2010 OC-1 to 10G

16. wavelength demux Spatially separate the channels using a wavelength demux DWDM Coherent Detection PD Spacing on the fiber needed between waves: “Guard Bands” ADC DSP PD LO Use a local oscillator to choose the “color” we want to “detect” to match the demux port color 16© 2011 Infinera Corporation. c. 2010 - 40G – 100G

17.  We use coherent detection to achieve greater reach at higher data rates (eg. 100G)  BUT…we are still tied to a grid-based, wavelength demux architecture  This is the current “state of the art” for 40G and 100G long haul optical transmission We can run 40G and 100G on relatively poor fiber, and over even greater distances than 10G IM-DD 17© 2011 Infinera Corporation Confidential & Proprietary Wavelength Demux with Coherent Detection

18. What Comes Next For Terabit Transport? Since the advent of DWDM… …so what has to change Quadrature Amplitude Modulation Coherent Wave Separation Grid-less FlexChannels On-Off Keyed Modulation Direct Detection ITU Frequency Grid 18© 2011 Infinera Corporation.

19. How 1Tb/s Might Look… Conventional WDM Channels  Guard bands to allow for individual wavelength demux Conventional Per-Channel WDM Filtering 1Tb/s 19© 2011 Infinera Corporation. Q:Is there a wavelength separation technology that does not require these guard bands? A:Yes, it’s called coherent wave separation

20. LO … © 2011 Infinera Corporation Confidential & Proprietary20 Coherent Wave Separation 1Tb/s Multi-Carrier FlexChannel Separate the Polarizations Separate the Wavelengths LO Separate the phase states All this is performed in a complex optical circuit The result is the ability to do this Start with super-channels

21. How 1Tb/s Might Look… FlexChannels  Closely-packed groups of wavelengths  Fewer guard-bands  25% increase in useable amplifier spectrum 1Tb/s Multi-Carrier FlexChannel 21© 2011 Infinera Corporation. LO … Separate the Polarizations Separate the Wavelengths LO Separate the phase states Start with super-channels

22. © 2011 Infinera Corporation Confidential & Proprietary22 Large Scale Photonic Integration Is The Key To Economic Terabit Networking Let’s take a look at that complex optical receiver circuit…

23. wavelength demux Spatially separate the FlexChannels using a wavelength demux Coherent wave combining and separation PD LO Block of similar-colored waves: a “Flex-Channel” We now need multiple local oscillators to separate the waves ADC DSP PD

24. ADC PD Optical Mixer ADC PD ADC PD...... … Local Oscillators Wavelength Phase state Polarization DSP From Waves to Super-Channels  Instead of demuxing individual wavelengths, we group waves together, into “super-channels”  Super-channels are optically separated, using guard bands  When a super-channel falls on a coherent detector, the individual carriers are separated using multiple LOs Guard Band Super-Channel

25. Coherent Wave Separation of Super-Channels “Super-channel demux”

26. 0 0.2 0.4 0.6 0.8 1 1.2 024681012 Capacity * Reach Product Bits / Dual-Pol-Symbol Which Modulation Technique? PM QPSK 0 1 2 3 4 5 6 Spectral efficiency [b/s/Hz] C-Band Capacity (Tb/s) 1.68121624 The result of FlexChannels: more capacity… …at the same reach… …for a given modulation type 26© 2011 Infinera Corporation.

27. 27© 2011 Infinera Corporation Confidential & Proprietary This is a PM-QPSK Transmitter PBS LD X Polarizations Y Polarizations To switch it to PM-BPSK mode, we would “switch off” half the circuit If you were building this board using discrete optical components, there would be a strong temptation to economize, and build either a PM- QPSK or a PM-BPSK board A PIC-based design costs the same – no matter what mode of operation you use – so one board for all modulation types

28. Infinera 100G Transmission Differentiators FlexCoherent™ Modulation PM-BPSK PM-QPSK 8 QAM REACHCAPACITY 16 QAM  FlexCoherent DSP ASIC 40nm CMOS technology for low power consumption Enhanced compensation of optical impairments: Chromatic Dispersion, PMD  FlexCoherent = “Dial-A-Reach” Per-wavelength selection of the optimum coherent modulation 28© 2011 Infinera Corporation.

29. FlexChannels Increase Total Fiber Capacity More complex modulation → more capacity per fiber PM-QPSK 8-QAM 16-QAM 1Tb/s 12 Tb/s 18 Tb/s 25 Tb/s 29© 2011 Infinera Corporation.

30. Reach, Spectral Efficiency, and Co-Existence 1Tb/s PM-8QAM FlexChannel 1Tb/s PM-8QAM FlexChannel 1Tb/s PM-16QAM FlexChannel 1Tb/s PM-16QAM FlexChannel 10x100G PM-QPSK 1Tb/s PM-QPSK FlexChannel 1Tb/s PM-QPSK FlexChannel or A A E E B B C C D D 30© 2011 Infinera Corporation.

31.  Pure “Terabit waves” are unlikely to become practical in the near to mid term technology horizon Super-channels are more likely Flexible modulation is essential  Super-channels will be made up of multiple, closely packed waves that do not align on the current ITU grid Large scale PICs are essential for practical super-channels  A programmable optical circuit will transform super- channels into FlexChannels for maximum flexibility and minimum spares management issues PIC-based optics are essential for FlexCoherent modulation  Photonic Integration is the key to enabling the Terabit Age 31© 2011 Infinera Corporation Confidential & Proprietary Summary

32. Thank You! gbennett@infinera.com © 2011 Infinera Corporation Confidential & Proprietary32