Optical Networks: Evolving from Fat Pipes to Cognitive Network Infrastructures High Performance Networks Group Dimitra Simeonidou: dsimeo@essex.ac.uk Contributors: Dr Reza Nejabati, Dr Georgios Zervas, Dr Siamak Azodolmolky, Dr Yixuan Qin, Dr Eduard Escalona and Mr Norberto Amaya
Main Consideration for Future Intelligent Networks Adapt to Users Requirements Rapid deployment of services Privacy, user rights Protect digital assets Respond to Applications Diversity From sensory data of Kb/s to gigantic data transactions of Tb/s Communications schemes and protocols optimised to application and terminal Seamless NW services over wireless and wired access Linked Supercomputers Future Internet: Network Centric IT and content delivery services Scalable to data and number of terminals Total traffic in Peta-bytes or Exa-bytes Billions of tiny/thin terminals (tags and sensors) Very high volume terminals (HPC, Data centres, scientific facilities) Evolve network with minimum replacement of infrastructure Very large number of transactions and sessions Page 2 Reduce environmental Impact Low power consumption (technology) Green network services (end-to-end) Evolve network with minimum replacement of infrastructure
Research Space Data Centers Supercomputers Content Servers Scientific Facilities Core/Metro Networks Fixed Access Network Wireless and Mobile Access Network Page 3 Home/Office Network
Intelligent Optical networks: Aware Adaptive Cognitive Optical networks that: Sense and gather all or part of their environment s (e.g. network load, loss, delay) information are considered aware systems Autonomously modify its operating parameters to be considered adaptive (e.g. topology, control protocol, aggregation- scheduling, modulation format, hardware configuration) Learn through their operation are considered cognitive Page 4
Enabling Technologies Data Plane: Flexible, Elastic Optical Layer Architectures on Demand Control Plane Targeted extensions for dynamic and data plane-aware network services Optical Network Infrastructure Virtualisation, Slicing and Isolation Software/Hardware Defined Network Programmability For infrastructure and service adaptation Optical Network Cognition Page 5
Data Plane: Flexible, Elastic Optical Layer Page 6
Research Motivation: Challenges Emerging applications with a wide range of transport requirement Healthcare/Medical High-speed data 400G, 1Tb/s User generated networked content Multimedia Production Residential Scientific Future applications with unknown requirements Flexible and efficient optical networks to support existing, emerging and future applications Page 7
Proposed Solution: Elastic Resource Allocation Flexible allocation of resources in time and frequency in order to: Accommodate applications with arbitrary requirements Video conference/virtual Presence High-speed data transmission 400G, 1T Page 8Education/Remote Learning Gaming
Research Motivation: Problem Current optical network solutions are inflexible: node architectures with rigid allocation of BW resources Wavelength High-speed IN-1 Current Optical Node Architectures with Rigid BW Resource Allocation OUT-1 Waveband IN-2 IN-3 Circuit Switch OUT-2 OUT-3 Packet Switch Sub-Wavelength IN-N OUT-N Page 9
MUX DEMUX Technologies towards Gridless/elastic operation: Spectrum Selective Switch DEMUX WSS Band DEMUX SSS 1 2 3 4 5 6 7 8 9 1 1 2 3 1 2 2 10G 40G 40G HIGH SPEED λ 6 10G 40G 40G HIGH SPEED λ 3 7 4 2 1 1 1 1 1 1 1 1 1 2 λ 1 2 λ 10G 40G 40G HIGH SPEED λ 3 10G 40G 40G HIGH SPEED λ 3 4 SSS: Spectrum Selective Switch (e.g. Finisar WaveShaper) 4 Page 10
Technologies used for Implementation Frequency allocation using WaveShaper Min bandwidth 10 GHz Max bandwidth 5 THz Min step 1 GHz Time allocation using PLZT (Polarized Lead Zirconium Titanate) switch Switching time 10 ns Variable frame size and variable slot size Minimum slot size 0.97 μs Slot-size step 1 μs Page 11
Splitter Splitter DEMUX Band- DEMUX Split DEMUX DEMUX Architecture on Demand Aimed to develop an optical node that can adapt its architecture according to the traffic profile and supports elastic allocation of resources High-speed A E Wavelength Elastic Add Sub-Waveband B F Drop Drop C G EDFA EDFA Sub-Wavelength SSS Band-MUX EDFA EDFA PLZT switch Page 12
3D-MEMS Switch for AoD Backplane Up to 320x320 per shelf Up to 960x960 per rack Switching time <20ms IL < 3dB Xtalk < -65dB 470 mw per connection Page 13
Splitter Flexible OXC Configuration Backplane implemented with 96x96 3D-MEMS Asymmetric configuration per port Flexibility to implement and test several switch architectures on-the-fly Switching time 20ms 1 A EDFA PLZT switch EDFA Elastic Add 2 3 4 SSS E C G Page 14
Architecture on Demand Benefits Support for arbitrary switching-granularity Dynamic architecture reconfiguration Fast fault recovery Faulty module replacement Alternative architecture Modular design that can grow with requirements Easy to upgrade with new modules Wavelength conversion Regeneration Other signal processing Page 15
Multi-granular Elastic Testbed Page 16
Gridless & Time-Shared Optical Node Demonstration Page 17
Research Exploitation: London Olympics 2012 Implement and demonstrate experimental services for real-time 4K and 8K multi-view video formats Remote location Sporting event AoD UHDTV Screen UHDTV Cameras (Multi-view) Elastic Transport Network Page 18 Audience
SHV Olympics Demo Consortium Network Infrastructure Content Provider Technology Development Future Networks & Visualisation Facilities Networked Media Technologies Page 19
Control Plane: Extensions for impairment and resource awareness and service optimisation Page 20
Control Plane Services Control Plane Services Fast provisioning of dynamic BoD Multi-technology support Multi-granularity support Traffic engineering Automatic resilience (Protection and restoration) Routing and Wavelength Assignment (RWA) Circuit oriented provisioning Current Control Plane standards OIF UNI IETF GMPLS (RSVP-TE, OSPF-TE, LMP) IETF PCE Page 21
Control Plane Extensions to Support Advanced Networked Media Services Extensions for FlexiGrid and Gridless optical infrastructures Support for multi-granularity including sub-wavelength technologies Optimised RWA PLI constraints Energy efficiency Extensions for media and IT aware networks IT/Cloud support, coding/transcoding, storage, displays Seamless Network and non-network resource provisioning Support for virtual infrastructure control and management Page 22
Control Plane Extensions Advanced IT + NET services ROUTING Extensions for PLI advertisement Service Layer PHYSICAL LAYER IMPAIRMENTS (PLI) PCE RWA PLI aware path computation Node Controllers NCP NET Resource configuration CCI PLI information Page 23
Control Plane Extensions Advanced IT + NET services Service Layer ROUTING Extensions for Energy metrics ENERGY AWARE CONTROL PLANE PCE RWA Energy aware path computation Node Controllers NCP NET Resource configuration CCI Energy information Page 24
Control Plane Extensions Service Layer SUB-WAVELENGTH SUPPORT Advanced IT + NET services ROUTING Extensions for time slots advertisement PCE NCP SIGNALLING Extensions for time slot labels Node Controllers NET Resource configuration Sub-wavelength network Page 25
Control Plane Extensions Advanced IT + NET services Service Layer ROUTING Extensions for IT resource advertisement MEDIA + IT AWARE CONTROL PLANE PCE RWA IT + NET path computation NCP SIGNALLING Extensions for IT resource requests Node Controllers NET Resource configuration IT Resource configuration Page 26
Optical Layer Virtualisation: Application/service specific infrastructure adaptation Page 27
Optical Network Virtualization Optical Network Virtualization Creation of logical instances of optical network resources Behaviour is the same of their corresponding physical optical network resources. Several virtual optical network resources can be connected together into logically isolated Virtual Optical Network Infrastructures Simultaneously coexisting over shared Physical Optical Network Infrastructures Each virtual network can be managed by an independent administrative entity Virtualization of optical network resources is often achieved by partition and/or aggregation Optical node virtualization Optical link virtualization Page 28
Optical Network Virtualization Virtual Infrastructure Virtual Infrastructure Virtual Resources Page 29
Architectural Consideration for Virtualization Virtual Infrastructure Control & Management Virtual Infrastructure Composition Resource Virtualization [Partitioning &/OR Aggregation] Resource Abstraction and Description Page 30
Optical Resource Virtualization Optical Resource Slicing Optical Resource Aggregation Page 31
Effect of Physical layer impairments on virtual infrastructure Support for isolation and coexistence of multiple virtual infrastructures, sharing a common ANALOGUE physical infrastructure Virtual infrastructures: physical layer constraints aware New virtual infrastructure setup may affect existing virtual infrastructure Nonlinear Impairments Linear XPM,FWM,SRS SBS Optical Inband Optical Out of Band Coherent InCoherent Page 32
Optical Network Virtualization GEYSERS Project Virtual Infrastructur e Operator Virtual Infrastructure Provider Physical Infrastructure Provider Page 33
Optical Network virtualization: An OpenFlow approach ISP A Client Private Line High-end Client Controller C Client Controller C Controller C OpenFlow Protocol FLOWVISOR Unified virtualization OpenFlow Protocol Isolated Client Network Slices C P C C C P C P P Single Physical Infrastructure of Packet & Circuit Switched networks Page 34
Software Hardware Defined Networking: Service adaptation Page 35
Software/Hardware Defined Adaptable/Programmable Network (SHDAN) Definition A framework provides software/hardware definition ability and flexibility enabling an adaptable and or programmable network environments Network architectures and elements can be dynamically reconstructed and/or optimised on demand Update key software and hardware modules on the fly driven by: Service/application s request, e.g. networked media, cloud computing Service provider s request, e.g. service updates New technology, e.g. Elastic/gridless devices New algorithms, e.g. RWAs PLI and energy aware Open research environments: e.g. optical network virtualisation Page 36
How SHDAN Fulfil Adaptability How to completely redefine the network architecture to open up infrastructure to new applications HW Definition HW Definition APP APP APP HW Definition SHDAN Enabled Network Operating System HW Definition HW Definition APP APP APP SW/HW Instruction Operating Set System Specialized Packet Forwarding Hardware HW Definition HW Definition APP APP APP SW/HW Instruction OperatingSet System HW Definition HW Definition APP APP APP SW/HW Operating Instruction Set System Specialized Packet Forwarding Hardware HW Definition HW Definition APP APP APP SW/HW Instruction OperatingSet System Specialized Packet Forwarding Hardware HW Definition HW Definition APP APP APP Specialized Packet Forwarding Hardware SW/HW Instruction Operating Set System Specialized Packet Forwarding Hardware Page 37
How SHDAN fulfil adaptability More circuit switch HW Definition APP HW Definition SHDAN Enabled Network Operating System Network Operating System OPENFLOW Protocol APP More packet switch Networking Applications Unified Control Plane Packet Switch Circuit Switch Packet & Circuit Switch Underlying Data Plane Switching Page 38
Programmable Hardware Network Control Capabilities NEoC 1) Routing, 2) Signalling 1) Networking planning with physical layer constraints computation 2) RWA calculation Topology and node emulation on Chip 1) 40-100 Gb/s Tx/Rx 2) Scheduling 3) Line speed processing time GMPLS PCE Sub-wavelength Switching Page 39
Cognitive Optical Networks Page 40
Cognitive optical network (COGNITION) architecture Requirements Layer Application Layer Service Plane Control Plane MAC Layer Physical Layer 15/06/2011 Page 41
Cognitive optical node architecture On the top level, cognitive service plane, cognition can create a self-optimized BW virtualization (buffer self-management and re-configuration) to suit service requirements under node conditions (buffer availability and structure). On control plane layer, distributed signalling and routing protocols can also be re-configurable, selfmanaged and self-healing by modifying protocol interactions, state machines and routines to reserve resources and disseminate availability information. At the MAC layer the transport module can support a re-configurable MAC protocol by supporting adaptive bandwidth allocation e.g. under different characteristics of network load, number and type of applications transferred, type of available PHYs. The O/E transponder consists of a cognitive PHY able to re-configure, adapt and re-purpose internal functions. Bit-rate, symbol rate, modulation format, wavelength(s) used, power level, continuous/burst mode operation can be automatically adjusted to respond to environmental stimuli. Cognitive Service Plane: BW virtualization Service-to-transport cognitive translation Cognitive Control Plane: Routing and signaling protocols and mechanisms Etc Cognitive MAC: Re-configurable MAC Adaptive bandwidth allocation, aggregation Etc Cognitive PHY Adaptation, re-configuration, repurposing Bit-rate, symbol rate Modulation format Wavelength(s) Power control etc Cognitive Transponder Cognitive Transponder Cognitive Service Plane: Switch virtualization Service-to-switching cognitive translation Cognitive Control Plane: Routing and signaling protocols and mechanisms Cognitive Control & Reconfiguration: Architecture selfreconfiguration Observation Self-control Cognitive Multi-Granular OXC Page 42
Thank you Page 43