Awaited Emerging Optical Components for All-Optical Ultra-Dense WDM-Networks

Transcription

1 Optical Networking in the Layered Internet Model Awaited Emerging Optical Components for All-Optical Ultra-Dense WDM-Networks Bo Willén, KTH Problems Applications Keep contact Network access End Users FTP, Telnet TCP / IP Ethernet Physical Media Layer 6 Layer 5 Layer 3-4 Layer 2 Signal definition Layer 1 The wire Layer 0 Content Multiplexing Techniques Introduction Key Technologies Wavelength selection Optical device limitations Conclusions - Wishes 1

2 System Capacity and Network Traffic TDM vs. WDM Gb/s %/yr 25%/yr Year TDM Research WDM Research TDM Commercial WDM Commercial Total Traffic inow 09 TLI R. Tkach (Alcatel-Lucent), APOC 08 5 Ultra-Dense 25 GHz Number of Channels Example: 40 Gb/s Mb/s 100 Gb/s on the research level. 5 GHz channel separation 1000 channels (reported but difficult) Potential still unexplored Maximum aggregate rate (N B) max = 14 Tb/s HECTO Overview Optical Networking 2x50Gb/s electrical input signals Transmitter Access Network Multiplexer & Driver amplifier Developed dby FhG / IAF Packaging by FhG / IAF DFB-TWEAM Developed by KTH & Syntune Packaging by U2T 100Gb/s electrical interfaces 100 Gb/s optical signal ADM WDM Metro Network Passive Optical Network (1 Gb/s 30 Mb/s upstream) 2x50Gb/s electrical output signals Clock and Data Recovery & Demultiplexer Developed by FhG / IAF Packaging by FhG / IAF Photodetector & Preamplifier Developed by FhG / HHI Packaging by U2T Receiver Gb Ethernet WDM network 32 2

3 Optical Key Technologies Narrowband / Tunable Lasers Narrowband / Tunable Receivers Optical Splitters / Combiners Reconfigurable Optical Add-Drop Multiplexors Wideband Optical Amplifiers Reconfigurable Wavelength Routers All-Optical Wavelength Converters Routing and wavelength assignment problem Available Bandwidth in Practice Optical Fibre 50 THz Absorption 20 THz EDFA C-band Gain flattening Tunable laser 5 THz GHz spacing DMUX 100 GHz spacing R-OADM Architectures Static Optical Crossconnect Flexible Banded Band Band Drop Add Wavelength-Selective Switching N x N Switch Matrix VOA s DEMUX MUX DEMUX MUX Drop Channels Add Channels Drop Channels Add Channels Broadcast and Select Drop Module Add Module Coupler Coupler Wavelength Blocker DEMUX MUX Bus-Like & Minimal Switching 2x1 Switches VOA s Drop Module Coupler DEMUX DEMUX MUX Drop Channels inow 09 TLI Add Channels Drop Channels Add Channels Add Module Source: W. Lin (ANDevices) Reconfigurable Couplers 3

4 Mach-Zehnder Interferometer Acousto-Optic Tunable Filter Non-linear element varying the time delay. Multi-wavelength! MOEMS R - Optical Switch Technologies Type Size Loss Crosstalk PDL Switching Wavelengths Bulk Mechanical 8 x ms 1.3 and D MEMS 32 x ms 1.3 and D MEMS 1000 x ms 1.3 and 1.55 Thermo-optic 8 x low 3 ms Bubble-based 32 x ms Liquid crystal 2 x ms Polymer 8 x low 2 ms 1.3 and 1.55 LiNbO 3 4 x ps SOA 4 x low 1 ns 4

5 Wavelength Selection Techniques Key Parameters Multiplexers and Filters Passband width Passband flatness Crosstalk suppression Temperature coefficient Insertion loss Polarisation dependence Cost ( CRAP ) Fabry-Perot Filters Fabry-Perot Filters d c Δf FSR = 2d R Mirrors Δλ FSR F π 1 λ 2 = 2d R R 5

6 Thin-Film Filters Thin-Film Filters Cavity Cavity Dielectric reflectors Glass substrate A resonant multi-cavity thin-film filter. Thin-Film Filter MUX / DMUX Bragg Gratings Short period gratings ~ 0.5 µm Fibre gratings Waveguide gratings Long period gratings ~ 100 µm Fibre gratings 6

7 Bragg Mirror Bragg Mirror Bragg phase-matching condition: Bragg wavelength: 2πηeff β1 = β0 = λ 0 2π β0 β1 = Λ λ 0 η eff = 2Λ Efficiency: η = tanh 2 (κl) Fractional bandwidth: Δλ λ 1 B N B Bragg Mirror OADM Transmission Gratings Bulk, fibre, waveguide Pitch = a θ d θ i Grating equation m λ sin( θd ) = sin( θi ) a θ depends on wavelength d Imag ing plane 7

8 Reflection Gratings Reflection Grating MUX / DMUX Stimax Grating α = blaze angel Concave mirror Semiconductor Waveguides Mach-Zehnder Interferometer 8

9 Cascaded Mach-Zehnders Mach Zehnder Interferometer with Add/Drop Arrayed Waveguide MUX / DMUX AWG C FSR = λη 0 mη g FSR N max = int λ 0 40 output channels 0.8 nm channel spacing (100 GHz) 30 GHz -1dB C-band 9

10 High Channel Count AWG Cross Connect Multistage Banding λ, λ λ2, λ3, 1 4 λ, λ λ2, λ3, 4 4 Band 1 Band 1 Band 2 Band 32 Band Guard space Band Channel Multiplexer / Demultiplexer AWG, Polarisation Compensation Table 3.1 Penalty-free polarisation compensation of Si02/Si arrayed waveguide grating wavelength multiplexers using stress release grooves E. Wildermuth, Ch. Nadler, M. Lanker, W. Hunziker and H. Melchior ELECTRONICS LETTERS 20th August 1998 Vol. 34 No

11 Optical Network Elements Optical Elements: Drawbacks Any problems?? Physical size ( footprint ) 100s to 1000s of wavelengths in length, order of wavelength in transverse dimension Compare electronics with FET gate lengths <10nm, but interconnects important for photonics as well as electronics Functionality, photonics lacks: RAM-type memory Signal processing capability (regeneration, clock-recovery) Cost Too expensive (too much handcraft) Wideband filters > 5 GHz But there are ideas how to resolve this! Optical Isolator: MO Kerr Effect Optical Router cross-connect 8 fibre ports 32 wavelengths 640 Gb/s 2.5 Gb/s per channel Source:19th Annual Meeting IEEE LEOS, pp , October

12 Electronics: Devices THE OPPORTUNITY OF SILICON PHOTONICS Enormous CMOS manufacturing infrastructure, process learning, and capacity: Cheap Manufacturing of Optical Subsystems Draw from continued investment in Moore s law Potential to integrate multiple optical devices Micromachining could provide smart packaging Potential ti to converge computing & communications: High-Speed Interconnect Technology Smaller dimensions are required! inow 09 TLI To benefit from the above, optical wafers must run alongside existing CMOS products. Source: H. Rong (Intel) Silicon Photonics 10 x 10GE Silicon PIC Tx & Rx Multiplexing Chips Example of limited scalability: Optical transparency Electrical Input 10 x 10 GE 10 x Laser Driver Monitoring Photodiodes MUX 10 x CWDM DFB Laser Arrays 100 Gb Ethernet DEMUX 10xPhotodiodes 10 x TIA Electrical Output 10 x 10 GE Compensate for loss Optical amplifiers Linear - analog Noise accumulation No retiming, pulse shaping etc Limited distance Power out Tx Multiplexer (Planar Echelle Grating) Rx De-multiplexer P 1 Digital electronics optical bistability bilit 0 Power out 1R log BER Power in 2R, 3R Optical amplifiers inow 09 TLI Source: A. Martin (Kotura) 0 Ex (1-p)^N Power in 12

13 Regeneration in Optical Networks Optical Bistability The signal has to be regenerated after passage of a certain number of nodes and transmission length. Intensity dependent refractive index (Kerr Effect) in a FP-cavity resonator where resonance depends on optical path length Reamplification Reshaping Retiming Resetting Wavelength η eff (P) Optical logic is required! Optical Logic: AND Gate Optical Logic: AND Gate A B output A B output Field distribution of Optical AND gate Field distribution of Optical AND gate 13

14 Optical Logic: AND Gate Optical Logic: AND Gate A B output A B output Field distribution of Optical AND gate Field distribution of Optical AND gate Wavelength Conversion Buffering A CW-source is mixed with the optical signal pulses through nonlinear effects such as Cross-gain modulation (XGM) Cross-phase modulation (XPM) Cross-absorption modulation (XAM) 4-wave mixing Header processing Port contention Fibre delay line Deflection routing Preferably in a Mach-Zehnder type setup: Optical memories are required! 14

15 Optical Memories 5 GHz Bandwidth Micro-Ring Resonator Photonic Crystal FSR = λ η L eff 2 Electromagnetically Induced Transparency 2 GHz Promising Technologies Cost - Developed fabrication technologies CMOS Size - New materials Memory -Electromagnetically Induced Transparency Logic - Nonlinear optics - Photonic crystals Bandwidth - Micro-Ring resonators - Photonic crystals Speculative Conclusions & Requests Today s dark fibres will be employed within 5 years 5 THz available bandwidth: GHz GHz Mb/s 32 (32 5 GHz) R-OADM will be the workhorse of all-optical networks: AWGs with R-couplers There are ideas how to solve basic limitations of optical devices, but DMUXing of <5 GHz channels required for WDM-penetration of the Access Network: AWG-like Photonic Crystals? Wavelength conversion of a complete band incorporating ~32 channels? 15