Keeping up the momentum: The importance of innovation to keep networks fit for the future Dr Tim Whitley MD Research & Innovation, BT Distinguished Engineer Context Trends Challenges for BT • Increasing consumer and business data usage: • To provide enough capacity to keep up with demand. • To devices that customers want to use, wherever they are and when they want to use it. • At reasonable cost. • To offer seamless connectivity requires integration of fixed and mobile networks. • Manage implications for core network – Device penetration (tablets, smartphones, smart TVs); – Video usage across all device types; – Internet of things, ‘wearables’ etc.. • Increased bundling of services: – Convergence of content/TV and broadband; – Convergence of fixed and mobile; • Voice heading towards being an application on IP. 2 Access – BT’s Broadband Heritage • • • • • • • • • • August 2000: BT Launched 500kbps ADSL broadband September 2004: BT removes distance limitations for ADSL March 2006: BT Launches 8Mbps DSL Max April 2008: BT Launches 24M Wholesale Broadband Connect January 2009: BT Infinity / Fibre Broadband launched April 2012: 80Mbps FTTC launched November 2012: 10G PON trial in Cornwall August 2013: 300Mbps Infinity product launched Spring 2014: Scale Vectoring Trial Summer 2014: G.Fast technology trial G.fast VDSL ADSL Broadband Narrowband Copper speeds: advances over time 3 The access network 5600 locations <2.2MHz, Typ 3.5km length ~89k locations <17MHz, Typ 350m length Fibre to the Cab ~4M locations <106MHz, Typ ~35m length Fibre to the DP ~29M locations Fibre to the Premise 4 Access – broadband Innovations Driving faster speeds Extending reach • • • • • We are innovating to extend benefits of faster access to the final c. 10% of coverage • Copper rearrangement • Broadband regenerators for ADSL • Microwave wireless to the cabinet • VDSL amplifiers • Fibre to the remote node (mini DSLAMs) FTTC – Vectoring FTTRN – VDSL closer to the customer FTTDP – G.Fast FTTP Vectoring Cabinet 5 The impact of crosstalk and vectoring Crosstalk generation within a cable Signal 1 Signal 2 line 1 Signal 3 line 3 line 2 Cable binder Vectoring gain (laboratory measurement) Maximum Downstream (kbit/s) bit-rate Max Attn Bit Rate (kbps) • Crosstalk is due to coupling between pairs in a multi-pair copper cable and limits FTTC speeds. • Increasing cable fill through customer take-up increases crosstalk and reduces line-rates. • By calculating a “pre-compensation” factor for each line, the crosstalk can be cancelled in the DSLAM. • The technique for doing this is called “Vectoring”, and is conceptually similar to noise cancelling headphones. • Openreach has trialled six vectored cabinets with c.1000 live customers in Braintree and Barnet. Phase 2 trials are now underway. 120000 100000 80000 60000 40000 20000 0 200 300 400 600 900 1200 Cable Length (m) Performance without Vectoring Performance with Vectoring 6 Pushing fibre deeper into the network: the Distribution Point • Beyond the green cabinets, BT has overhead and underground DPs with: • • An average of ~8 customers connected; Can have >16 customers in some cases. Typical pole DP serving 12 customers • Space is limited and they are exposed to the elements. • The typical distance from the DP to the customer is ~35m making it the ideal location for delivering very high speeds. • A new technology (G.fast) is being standardised in the ITU: • • Enables speeds of up to 1 Gbit/s; Commercial G.Fast equipment is expected in H2 15/16. Typical distribution footway joint box 7 We have trialled G.Fast in the network • Trial had three customers connected to one DP • The network was overlaid to prevent harm to existing services, but used existing duct infrastructure • Chart shows best and worst case performance • Physics of G.Fast worked as expected • Vectoring proved to be critical in delivering the top speeds 19m 19m 66m 10G backhaul 66m 19m 66m 47m 8 BT is opening a new, purpose-built, ultrafast broadband lab at our Adastral Park R&D centre in Ipswich • G.Fast has demonstrated potential that now needs to be tested and verified. • We will study the full technical capabilities of G.Fast hardware designed by leading system vendors. • We will continue to play a proactive role in developing G.Fast standards in the ITU. • It’s crucial that we stay ahead of the curve for the benefit of our customers and shareholders. Emulation network enables replication of different realworld final-drop topologies DP and CPE installations allow a full range of performance and operations tests 9 FTTP – 10G and beyond • GPON provides 2.5Gbit/s downstream, 1.2Gbit/s upstream • XG-PON1 provides 10Gbit/s downstream, 2.5Gbit/s upstream • Exploiting existing Fibre Network with separated wavelengths • XG-PON and GPON coexist on the same fibre • No change to Optical Distribution Network (ODN) • WDM1r (Filter) component already in place • We are working on NG-PON2 which provides 40Gbit/s downstream and 10Gbit/s upstream on a single PON GPON system 2.5Gbit/s & 10Gbit/s (1490nm & 1577nm) GPON OLT XG-PON1 OLT GPON ONT XG-PON1 ONT WDM1r XG-PON1 system Power splitter 1.24Gbit/s & 2.5Gbit/s (1310nm & 1270nm) GPON ONT 10 Access fibre deployment and new services will drive future core bandwidth demand • Current core traffic growth CAGR = 45%. • Bandwidth demand analysis to 2033 suggests core requirement of c.12Tbit/s by 2020. – – – – – – – – Over the Top content; Video on Demand; 4k/8k content delivery; Cloud-based applications; HD Video-calling; Rich media webpages; Software package size growth; Customer expectations of speed. Future demand mitigation: – – Improved video coding; Time-shift delivery techniques. 70 60 Modelling suggests c 25 to 30 fold core capacity increase over the next ten years Core traffic (Tbit/s) Key future demand drivers: Time of day activity scenarios have been used to model bandwidth demand of different user types Low Scenario Mid Scenario 50 High Scenario 40 30 20 10 2013 2018 2023 2028 2033 Potential long-term core capacity requirements 11 BT’s upgrades core transmission roughly every ten years... 1Tbit/s Bandwidth 100Gbit/s 10Gbit/s 2.5Gbit/s 155Mbit/s 40Gbit/s 1G, 10G, 40G, 100G SDH MESH 622Mbit/s PDH SDH rings Megastream 2-155 Mbit/s PSTN Private Circuits Kilostream PSTN 155M -2.4G Broadband, ATM & IP We are now deploying 100Gbit/s coherent optics 64Kbit/s 1980 1G-10G 1G-10G 34Mbit/s 2Mbit/s 21C WDM UBB NG-WDM 1990 2000 Year 2010 2020 12 The Next Generation: 400G & Terabit Trials 2T & Beyond 2T S/W Configurable modulation formats 1.4T Alien Flexgrid Super channel 1T 800G Super channel 400G & 200G (16-QAM) Cambridge TNOC Alien Wavelengths 100G Coherent 40/100G Norw ich 83km 77km New market Cambridge Ipswich 24km 13.5km 68km Adastral park 50.7km 29km Bishops Stortford 2010 2011 2012 2013 2014 2015 2016 Colchester 58km 52km 64km Wood green 14km 62km Basildon BT Adastral Park BT Tow er BT Tower 13 The Next Generation: Gridless & Optical Superchannels • Current WDM has rigid 50GHz grid structure. • Future gridless infrastructure permits improved fibre utilisation (spectral efficiency). Example: 1.4T “Superchannel” comprising 7 x 200G (16-QAM) subchannels 14 Pushing the boundaries • We have demonstrated real-time 2.2T Transport over managed Flexgrid infrastructure: 400GHz 2.2T Superchannel – 11 x 200G 16-QAM Superchannel; 2.2T (11 x 200G 16-QAM) – 33.5GHz subcarrier spacing giving a spectral efficiency of 5.97bits/s/Hz, more than 49% improvement over conventional 50GHz grid. • We are working with partners to develop our understanding of capacity management, for example: – Demonstrating configurable Superchannel alongside multiple 100G 50GHz grid channels; – Working with Huawei U2000 management system. 100G 100G Superchannel Configurable Superchannel operating alongside 100G grid channels 15 The next challenges are control, flexibility and vendor interoperability... SDN datacentre & network orchestration Datacentre with general purpose H/W suitable for NfV SDN orchestrating the provision of virtualised network functions Software Defined Networking (SDN) Network functions Virtualisation (NfV) • Open standard control and orchestration equipment • Reduced costs through consolidating functions • Setting up optical and Ethernet circuits on the fly • Reduced operational costs • Control of flexgrid bandwidth and optical modulation formats • Rapid provision of new services globally • Multilayer operation allowing time of day router bypass • Software oriented innovation (including Open Source) 16