Janet Aurora Evolution of a dark-fibre facility David Salmon – Janet(UK) Research results from: Bristol, Cambridge, Southampton, UCL, and

Evolution of a facility… Important research results The struggle to sustain the infrastructure Success and the forward look Background

Aurora – establishment and early years

– Early discussions within the e-Science programme on access to dark- fibre to support photonics and optical systems R&D – Included within the UKLight programme, but low priority – Funding via JISC Support for Research committee – UKLight proceeds – layer2 point-to-point services – ethernet/SDH 2006 – UKLight steering committee permits dark-fibre to proceed & funding agreed – Procurement and preferred supplier chosen late – Fibre and route selection, contract with supplier signed June – Cambridge, Essex, UCL – Infrastructure accepted December 2007 Aurora Timeline…beginnings

2008 – Researchers install equipment on network, unfunded, use whatever is to hand in labs and existing projects. – NO explicit funding strand to support projects & exploitation. – Phase 2 fibres bring in Aston and Southampton 2009 – Research builds and significant results published – Begin to make case to UK Engineering and Physical Sciences Research Council (EPSRC) for long-term requirement for a dark-fibre R&D facility. – EPSRC rates facility highly, but uncertainties over longer-term fibre infrastructure funding hold-up progress. Aurora Timeline – Establish & grow infrastructure

JANET Aurora Intermediate equipment co-location point University / JANET access point Fibre spans International Lightpath access to other NRENs via JANET & GEANT JANET Lightpath access to other UK Locations University of Cambridge University of Essex UCL University of Southampton Telehouse London 550km of fibre-pairs VM Brentford PoP VM Camberley PoP VM Crawley Court PoP VM Chelmsford PoP VM Enfield PoP VM Ipswich PoP VM Bury St Edmonds PoP 31km 52km 58km 43km 30km 55km 57km 54km 50km 72km

Highlights of early research results

OFC’09 PD: Field Trial of WDM-OTDM Trans-multiplexing Two experiments performed to show functionality and performance in above scenario: First experiment: between A and D Second experiment: between C and F Dark fibre network Network scenario WDM transmission (A); OTDM grooming (B); OTDM transmission (C); tributary separation (D); WDM bridge (E); and re-aggregation (F) A B C D E F 43 Gb/s 129 Gb/s

1. DPSK regeneration using Phase-sensitive amplification (You probably have these already)

Network configuration University of Southampton Telehouse London 40 km 60 km 40 km

Results: 2 roundtrips Fig A: Use of mid-point regenerator reduces the power penalty by a factor of two (e.g., at BER=10 -6 from 3 dB to 1.5 dB and at BER=10 -8 from 5 dB to 2 dB. Fig. A: Regenerator reduces error floor by about 1 order of magnitude, meaning it is preventing 90% of errors to occur. Fig. B: When the regenerator is used, about 2 dB less power can be launched through the link for identical BER performance.

3. Transfer of broadband frequency combs (collaboration with NPL)

ORC – NPL collaboration: Transfer of broadband frequency comb over 86km of the AURORA network Experiment explored the potential for dissemination of accurately referenced comb-based microwave frequency combs to remote laboratories or industry over installed fibre links Part of a joint PhD; combined NPL frequency metrology expertise with ORC photonics technology expertise Experiment required a large (unregulated) optical bandwidth: Aurora dark fibre network ideal test bed

Advanced Coherent Optical Communication Systems at UCL

15 Motivation Signals in optical communication systems may be weak or have low signal-to-noise ratio –Achieving the highest possible sensitivity or noise rejection remains a challenge Synchronous coherent demodulation offers the best theoretical performance –Also enables demodulation of complex modulation formats (e.g. PSK, QPSK, etc.), and frequency selectivity without optical filtering But it requires optical carrier recovery Carrier Recovery E(t) = E 0.x(t).sin{  o t+  (t)} sin{  o t} x(t) or  (t) We have been investigating a carrier recovery technique called the Optical Injection Phase Lock Loop

Coherent and Direct Detection of IM Intensity modulation (or amplitude shift keying) has a strong residual carrier component that the OIPLL receiver can lock to Demodulation of ASK data is straightforward and gives improved performance compared to direct detection at low optical signal-to-noise ratio (OSNR) 16 Intensity Modulation Gb/s, PRBS 1 nm filter M. J. Fice, A. Chiuchiarelli, E. Ciaramella, and A. J. Seeds, “Homodyne coherent optical receiver using an optical injection phase-lock loop,” J. Lightwave Technology, vol. 29, 2011, p. 1152–1164.

TX: 100Gb/s+ range of modulation formats RX: DD/coherent +DSP Optical transmission test-bed for the study of: - nonlinear signal distortion/pulse propagation - mitigation of transmission impairments - variety of dispersion maps - effects of cascaded optical add-drop multiplexers - device performance - transmission over installed fibre using Aurora dark fibre network in the loop - straight line experiment (1200km) in progress - UCL Optical Networks Terabit Test-bed

Aurora – the battle to sustain the facility

– e-Science programme closes and funding finishes – JISC service extension funds fibre for one more year to four sites – Drop link to Aston – too costly (long route) – Late 2011 BIS e-Infrastructure funds announced - £150/160M Janet £26M – Janet6 preparations already well under-way ~2M earmarked for Aurora/dark-fibre facility – Janet funds fibre for one more year from operational budget 2012 – EPSRC prepare to tender for dark-fibre mid-range facility – December - EPSRC launch tender process for mid-range facility 2013 – January – Academic consortium forms to respond to tender UCL Lead, with Bristol, Cambridge and Southampton. – May – Academic consortium supported by Janet wins bid Transition period - funding uncertainties

2013 – Janet runs procurement for new fibres – Old contract could not be extended further – Autumn New supplier wins and contracts signed It will be a completely new infrastructure ! Deployment planning begins 2014 – Deployment continues – Target service date end July Year stable forward look to mid 2019 Fibre-infrastructure and exploitation funding aligned ! Success !

Highlights of more recent research results

First Gridless Network Field Trial over NDFIS and BT dark Fibre Network – ECOC 2011 PostDeadline BT Tower, London 40G, 100G Coherent Tx 10G / 40G, 100G Coherent Tx Flex-grid OXC Flex-grid OXC 40G, 100G Coherent Rx 40G, 555G Tx DC M Optical Back Plane SSS Spli t SD 2 DE M Flexible-Architecture 10G, 40G, 555G Rx BT Labs, Ipswich JANET Aurora Dark Fibre Link PoP, Chelmsford University of Essex, Colchester 205 km 50 km 55 km SD 1 Node-1 Node-3Rx-1 Tx-1 Tx-2 Tx-3 RX-2 Node-2 Link km Link km Link-3 50 km Link-4 50 km SSS: Spectrum Selective Switching DEM: Optical Demultiplexing SD1: Spectrum defragmentation 1 SD2: Spectrum defragmentation 2 10G Tx Tx-4

2. Polarisation-, transmission rate- and format-independent optical wavelength conversion

Black-box wavelength conversion of WDM signals tested on the Aurora testbed  A wavelength converter capable of handling a number of WDM channels at the same time, regardless of their modulation format, speed or state of polarisation was tested on the Aurora transmission link  The total length of the fibre link was 600 km  Channels were transmitted through the link, some of them were wavelength converted and re-transmitted through the same link  The quality of the channel was assessed (i) after first passage, (ii) after the wavelength conversion and (iii) after the second passage Aurora System-under-test

Example constellation diagrams Baud Rate Modulation Constellations 600 Kmafter WC1200 Km 10 G BPSK QPSK 20 G BPSK QPSK

First Experimental SDM Network Demonstration over 680 Km NDFIS dark fiber – ECOC 2013 Postdeadline AoD Node 1 AoD Node 3 SDM DEMUX MCFMCF SDM MUX MCFMCF Add: 18x10G 45x42.7 G 3x555 G Drop Add: 10G Drop AoD Node 4 AoD: Architecture on Demand MCF: Multi-Core Fiber AoD Node 2 Switching 1.Fixed 2.Flex Grid, 3.Space 4.Time Multi- λ Conversion Switching 1.Fixed 2.Space Switching 1.Space Dispersion Comp. Switching 1.Space Add : 45x42.7 G 555 G Node km installed fibre Drop 340 km installed fibre Janet Aurora Dark Fibre Add: 4x10 G 2x 40 G BPSK / QPSK

First Experimental SDM Network Demonstration over 680 Km NDFIS dark fiber – ECOC 2013 Postdeadline V. J. Rancaño, et. al, JTh2A.19/OTh1C.2, OFC 2013 Four AoD nodes and two 7-core fibre links. Elastic multi-granular switching in space, frequency and time. Multi Wavelength Converter

Aurora2 - The New Infrastructure

Uniform engineering across the network Flexible and remotely configurable – SDN inspired Coherent management across the network Must deliver “services” to other users/projects – Research Council requirement – Wants to see facility exploited by other disciplines/applications – Outreach to encourage involvement by – Standard “testbed” as a service concepts with inherent tensions Extend reach through Janet Layer 2 services – “Lightpaths” – VPNs Principles

NDFIS Topology

Installation – SSET at UCL

Software Defined Transmission Network

Intermediate node

Illustrating capabilities of remote fibre switch Slides from Martyn Fice at UCL Flexibility of remote fibre switching

In (east) In (west) Out (east) Out (west) DCM Drop to NetFPGA Drop to L2 Switch Add from NetFPGA Add from L2 Switch 8x8 switch connections NB: ‘east’ and ‘west’ correspond to direction of traffic flow on connected fibre, e.g.: DCM Froxfield In (east) In (west) Out (east) Out (west)

In (east) In (west) Out (east) Out (west) DCM Drop to NetFPGA Drop to L2 Switch Add from NetFPGA Add from L2 Switch Pass through – no amp DCM Froxfield In (east) In (west) Out (east) Out (west)

In (east) In (west) Out (east) Out (west) DCM Drop to NetFPGA Drop to L2 Switch Add from NetFPGA Add from L2 Switch Pass through – with amp DCM Froxfield In (east) In (west) Out (east) Out (west)

In (east) In (west) Out (east) Out (west) DCM Drop to NetFPGA Drop to L2 Switch Add from NetFPGA Add from L2 Switch Pass through – 2-stage DCM DCM Froxfield In (east) In (west) Out (east) Out (west)

In (east) In (west) Out (east) Out (west) DCM Drop to NetFPGA Drop to L2 Switch Add from NetFPGA Add from L2 Switch Loop back – no amp DCM Froxfield In (east) In (west) Out (east) Out (west)

In (east) In (west) Out (east) Out (west) DCM Drop to NetFPGA Drop to L2 Switch Add from NetFPGA Add from L2 Switch Loop back – with amp DCM Froxfield In (east) In (west) Out (east) Out (west)

Bristol High Performance Networks Group Lab

Nipper !

Broader collaboration – NPL & Europe

SDNs / Openflow Strong theme at the conference (other talks) Desire to collaborate Broader UK discussions Maybe 10+ UK Universities interested in some way Janet will help – modest Openflow overlay to interconnect/interact Research Council funded projects – TOUCAN – EPSRC Programme Grant Bristol, Edinburgh, Lancaster £12M (£6M+£6M) – 5 Years Photonic – Layer2 – Access (Wireless +??) – Other smaller individual projects Connecting “Testbeds”

Stable forward-look for 5 years – Build strong collaborations, nationally and Internationally – Community steering committee for governance – Consortium plus Janet for management Build on very good track-record of high-profile results Success at last ! – Persistence – Good results – Good fortune Timely convergence of two funding streams – Infrastructure – Exploitation – Good team UCL Lead, Bristol – Technical direction Cambridge, Southampton - Janet – infrastructure support Summary

Funding – JISC, Janet, BIS, EPSRC University Research Groups – some of the people involved – Bristol (Essex) Professor Dimitra Simeonidou, Dr Reza Nejabati, Dr George Zervas, Mehdi Rashidi – Cambridge Professor Ian White, Professor Richard Penty, Adrian Wonfor – Southampton Professor David Richardson, Professor Periklis Petropoulos – UCL Professor Alwyn Seeds, Professor Polina Bayvel, Dr Martyn Fice, Dr Seb Savory – NPL Dr Giuseppe Marra & colleagues Acknowledgements