Photonics in Broadband Access and the Next Generation Network Leo Spiekman
The Next Generation Network Architectural evolution in core and access One network for all information and services Everything transmitted as packets
Internet Traffic Growth Near exponential increase in bandwidth use Organic growth increase in customer base Shift in content, away from static pages, toward multimedia embedded content and streaming video Larger bandwidth demand of content providers and other high end users (P. Magill, AT&T Labs, LEOS Annual Mtg. 2007)
Traffic Type Is Changing Packets (data) have overtaken circuits (voice) 1000 Voice Data Total Traffic (Gb/s) 800 600 400 200 0 1990 1995 2000 2005 2010 (R. Tkach, Alcatel Lucent, OFC 2008)
Economies of Unified Management
Three Ways to Increase Capacity Increase bitrates More wavelengths More fibers New technologies are introduced if and when they become cost effective
25.6 Tb/s Transmission C1..C79 3-dB Coupler AWG ODD Spans ( 3) C PC Delay C/L TX 1 C/L C/L I PC 80km SSMFDCF L N AWG C T I PC L1..L79 N C/L C/L C/L EDFAs C2..C80 I T L Raman N AWG EVEN C PBS PC T PC C/L C/L TX 2 C/L L 50GHz/100GHz Interleavers AWG PC Concatenated passband: 40 GHz L2..L80 MZ Relative Power (db) MZ L 42.7 Gb/s Clock/2 Clock/2 42.7 Gb/s C π/2 Control 160 Channels Two Polarizations 2 x 42.7 Gb/s RZ DQPSK -10-20 -30-40 -50 1520 1540 1560 1580 Wavelength (nm) 1600 (A. Gnauck, Alcatel Lucent, OFC 2007)
DWDM is Hard Linear Impairments Chromatic Dispersion Polarization Mode Dispersion Nonlinear Impairments Self Phase Modulation Four Wave Mixing Cross Phase Modulation
DWDM is Really Hard Dispersion (J. Perino, Electronics Letters 1994) FWM (F. Forghieri, Photonics Technology Letters 1994) Gets worse with higher bitrates and channel densities!
Network Layering Wildly Successful Ethernet: 1973 WWW: 1989 Google: 1998 DWDM transmission: 1987
Network Convergence LAN MAN WAN SONET/ SDH Voice Video Data Ethernet/ IP
Line Speed Convergence TbE? 1000 100GbE Gb / s 100 SONET OC 3072? OC 192 10 1 Ethernet OC 48 OC 768 10GbE OC 12 OC 3 0.1 0.01 10BASE T GbE 100BASE T/ 100BASE FX Generations
Why the Push for Ethernet? Rigorous standardization Vendor interoperability Enables fierce competition Rapid evolution Everyting that is not allowed is prohibited Media, speeds Installed base preservation
Convergence to Packets Core converges to IP or Ethernet Increasingly, services independent from transport layer But: Ethernet was not designed for video LAN MAN WAN SONET/ SDH Voice Video Data Ethernet/ IP
Different Traffic has Different Needs Type of video High speed Long distance (video download) QoS (streaming) Latency (interactive)
DWDM is Hard... Also in the Next Generation Network Linear Impairments Non Linear Impairments Bottom three layers of OSI model cross connect add/drop
100 Gb/s is Next? (G. Raybon, Alcatel Lucent, OFC 2008) 100G Ethernet will come, driven by applications (e.g., video on demand) 100 Gb/s transport will have several flavors Ethernet transport (IEEE) for local area and access networks OTN transport (ITU T) for wide area networks We're upgrading our network to 40 Gbit/s, but we will go to 100 Gbit/s as soon as possible and we hope to deploy early next year," Fred Briggs, executive vice president for network operations and technology at Verizon, Lightreading September 2007.
100 Gb/s Standardization Activity IEEE-Higher Speed Study Group (HSSG) Ethernet transport for local area and access networks Presentations found at http://grouper.ieee.org/groups/802/3/hssg/index.html Optical Internetworking Forum (OIF) Common Electrical Interface (CEI) enable high speed signaling for backplanes and chip to chip communications International Telecommunications Union-Telecommunications Standardization Sector (ITU-T) Define standards for Optical Transport Network (OTN) IEEE OIF OIF ITU-T IEEE IEEE (G. Raybon, Alcatel Lucent, OFC 2008)
100 Gb/s per channel 100GE MAC 100G parallel transport (= inverse multiplexing) 10 x 10 Gb/s MUX 100 Gb/s (10 λ s) 100GE MAC 100G serial transport 10 x 10 Gb/s 10 x 10 Gb/s Modulators (electrical optical) 10 ps / bit Laser 100 Gb/s 100 Gb/s (or 2 x 50) (1 λ ) 100 Gb/s Modulator (electrical optical) MUX Choice depends on lowest cost per bit: Targeted system capacity (spectral efficiency) Targeted system reach Wavelength management and networking aspects (ROADMs, etc.) (P. Winzer, Alcatel Lucent, OFC 2008)
100 Gb/s per wavelength 1 bit/symbol 107 131 Gbaud (OOK, DPSK, DB/PSBT, ) 2 bits/symbol 4 bits/symbol and up 54 66 Gbaud 27 33 Gbaud (DQPSK, pol-muxed OOK, ) (pol-muxed (D)QPSK, 16-QAM, ) RZ-DQPSK (G. Raybon, Alcatel Lucent, OFC 2008) 64-QAM
Towards 1 Tb/s OTDM receiver 640 Gb/s OTDM transmitter 640 Gb/s 4 MLFL 1 4 9.953 GHz PRBS 7 2 1/231 1 Generator 16 500m DF HNF EDFA 640 Gb/s <1ps R C 40 Gb/s SC based pulse compressor iso DM X 50 km SMF+IDF 2 ps 1553 nm ~ 640 40 Gb/s demultiplexer τ 1 τ 2 40 Gb/s preamp detector BPF1 1.5nm BPF2 1nm SOA 40 Gb/s detector Prec CW 1565 nm BER test 1 ps 1538 nm 640 40 Gb/s demultiplexer 40 GHz Electro optical PLL CR 640 Gb/s <1 ps pulses COBRA 2ps/div (E. Tangdiongga, Technical Univ. Eindhoven, ECOC 2007)
1 Tb/s = a lot of optical bandwidth holding beam CW data 640 Gb/s BPF transfer function 0 Relative Power [db] clock 40 GHz 10 5.12nm SOA out 20 1 Tb/s modulated = 16 nm of BW SOA in 30 40 1530 1536 1542 1548 1554 1560 Wavelength [nm] COBRA (E. Tangdiongga, Technical Univ. Eindhoven, ECOC 2007) 1566
NGN in Japan NGN: Next Generation Network. Everything over IP NTT East taking orders as of March 31, 2008 NWGN: New Generation Network. New paradigm in R&D Development of core network control and management technologies to support Peta bit/s class large scale new generation networks, unify path/packet networks, and lead international standardization activities (http://nag.nict.go.jp/index_e.html)
NGN in the US All IP based, but OC 192 over DWDM in core (P. Poll, Qwest, OFC 2008)
NGN in Europe British Telecom: 21st Century Network Convergence to everything over IP MPLS in Core
NGN in Europe The Netherlands: All IP Core network based on Ethernet
IP over MPLS Multi Protocol Label Switching IP Packet L1 IP Packet L2 Label Switching Routers (LSR) LSR Label Switching Path (LSP) Label Edge Routers (LER) (G. Cincotti, Roma Tre University, OFC 2006)
Future: Optical Packet Switching Outputs packets Input packets cw λ 1 cw λ 2 cw λ 2 N Label Extractor Nth label 1st label Labels Processor Fast Label Recognition 1 160 Gb/s payload Wavelength converter A W G : 2N Control signal Fast All Optical Switching Packets at output (N. Calabretta, Technical Univ. Eindhoven, OFC 2008)
Why All Optical? Same reason ROADMs / OXCs offer express lanes in circuit switched networks: Less complexity Lower power consumption
Fast All Optical Switching SOA SOA SOA SOA SOA ISM14 Integrated Switch SOA SOA SOA SOA SOA SOA ISM22 Integrated Switch SOA Switching Elements High Extinction Ratio Nanosecond Switching
Fast All Optical Label Recognition Single-Chip Multiport Encoder / Decoder ACP OC1 CCP LSR node controller label swapping controller switch controller Label Payload E/D E/D label switch optical packet switch E/D (G. Cincotti, Roma Tre University, OFC 2006) Time (100ns/div) label remover buffering
Optical Packet Switching Field Trial 160Gbps WDM Optical Packet with label A EH #1 EN #1 EH #2 EN #2 Field SMF 16km JGN2 testbed network (http://www.jgn.nict.go.jp) Japan Label B OPS system RDF(11.5km) 1x2 Switch 1x2 Switch 2x1 Coupler Kyoto 2x1 Coupler Keihanna RDF(11.5km) EH #3 EN #3 EH #4 EN #4 Keihanna (N. Wada, NiCT, OFC 2008) Osaka Nara Field SMF 16km Nara
Access Networks: PON or P t P? GPON a maturing technology ready for roll out PON Point to multipoint Optical Network Termination Optical Line Termination (DSLAM like Optical Splitter Fiber Fiber Point to Point Optical Network Termination Passive Optical Network (PON) has 3 main advantages Less capex in optics & equipment Less opex in provisioning Wholesale offer available November 14th, 2006 page 9 All rights reserved 2006 France Telecom
PON is High Density in CO 10 Linecards x 4 PON ports x 128 ONTs = 5120 ONTs per shelf
PON Roadmap Next Gen OAN Bitrate (bit/s) (C,D)WDM PON??? Long Reach OAN??? t er en d Un opm ITU T G.984.1..4 vel 2,5/2,5G, 60km De G PON 2,5G 1G B PON 100M 10M 1M TDM PON E PON 1/1 G, 20km IEEE 802.3ah APON 1,2G/622M 20 km 622/155M Standard 20 km FSAN, ITU T G.983.1amd1, G.983.3...10 Standard FSAN, ITU T G.983.1 Standards approved and public 100k APON...ATM PON B PON...Broadband PON E PON...Ethernet PON G PON...Gigabit PON CWDM...Coarse WDM DWDM...Dense WDM 1997 1998 2000 2001 2002 2003 2004 2005 2006 (A. Srivastava, OneTerabit, 2008)... Year
State of the Art: GPON Central Office Outside Plant Voice Single Fiber Customer Premises ONT 1 ONT 2 Video (D. Piehler, Alphion, 2008) GPON Optical Line Terminal (OLT) 1 x N Passive Optical Splitter ONT 3 Data 1490 nm 1310 nm ONT N GPON Optical Network Terminal (ONT)
GPON with Reach Extender ONT 1 1:32 PON OLT ONT 2 TDM down 1480 1500 nm TDMA up 1260 1360 nm ONT 32 20 km Increased reach (will require an OEO or optical amp based reach extender) Near term (tactical): avoid current practice of remoting the OLT to go beyond practical 20 km limit Long term (strategic): may permit consolidation of central offices (local exchanges)* * D. Payne and R. Davey, The future of fibre access systems?, BT Technol. J., 20, 104 114, 2002. (P. Iannone, AT&T, OFC 2008)
GPON with Reach Extender ONT 1 1:128 PON OLT Reach Extender ONT 2 TDM down 1480 1500 nm TDMA up 1260 1360 nm ONT 128 60 km Increased reach (will require an OEO or optical amp based reach extender) Near term (tactical): avoid current practice of remoting the OLT to go beyond practical 20 km limit Long term (strategic): may permit consolidation of central offices (local exchanges)* * D. Payne and R. Davey, The future of fibre access systems?, BT Technol. J., 20, 104 114, 2002. (P. Iannone, AT&T, OFC 2008)
Strategic Use of Reach Extender Long reach PONs enable CO consolidation Elementary PON Idealized geographic distribution of central offices (CO) CO Serving Area (20 km radius) CO Central Office Red lines represent subset of individual PONs within each CO serving area (P. Iannone, AT&T, OFC 2008) (20 km reach)
Strategic Use of Reach Extender Long reach PONs enable CO consolidation (60 100 km reach) Idealized geographic distribution of central offices New CO Serving Area (60 km radius) Eliminate majority of central offices Powered extender box on feeder Possibly increase split per PON Saves on: Powering OpEx Real estate This strategy benefits from WDM or TDM muxing between OLT and PON extender Lots of technology options (, CWDM DWDM, higher rate TDM, λ conversion, etc) New Central Office case for all optical extender box may be more compelling for multi wavelengths Bidirectional reach extender (P. Iannone, AT&T, OFC 2008)
Power and Space Savings
Future: PON Capacity Increase OCDMA WDMA TDMA Code Division Multiple Wavelength Division Time Division Multiple Access Access Multiple Access T (N. Kataoka, NiCT, OFC 2008) t Channel 1 Channel 2 Channel N T Amplitud Wavelength Channel N Channel 2 Channel 1 Intensity Channel N t Channel 1 Channel 2 T t
WDM PON Remote Node ONT CO OLT Tx DFB LD 1530 nm DFB LD 1550 nm DFB LD CO OLT Rx 1290 nm Rx Rx 1330 nm Rx 1350 nm Rx 60 km C W D M 1310 nm C W D M 1510 nm 1.5 µ m Hybrid Amp C W D M DFB LD C W D M 1490 nm 1.5 µm Rx 1.3 µ m Hybrid Amp 2:2 1:16 2:2 2:2 2:2 1.3/1.5 µ m Mux (P. Iannone, AT&T, 2008) 1:16...... 1.3 µm Tx 1490 nm down 1310 nm up 1510 nm down 1290 nm up 1530 nm down 1330 nm up 1550 nm down 1350 nm up
OCDMA PON Downlink (i) DPSK signal (ii) 20 [ps/div] (iii) 20 [ps/div] Central station (Koganei) PPG 20 [ps/div] Multi port VOA PC Encoder 1 OCDM Tx 1 λ :1551nm 10.3125GH ｚ (iv) Network 10GbE OCDM Tx 2 Analyzer λ :1551nm HDV PC3 OCDM Tx 3 Camera λ :1551nm HDV PC4 OCDM Tx 4 Camera λ :1551nm 3 5 7 8 9 11 13 (v) (vi) 20 [ps/div] 20 [ps/div] 20 [ps/div] ONU 1 50km Installed SMF OCDM Rx 1 50km Installed SMF Otemachi DCF ONU 2 OCDM Rx 2 Koganei Koganei 15 1:8 splitter ONU 8 OCDM Rx 8 Tokyo Otemachi Central station (Koganei) Multi port Decoder OCDM Rx 1 1 OCDM Rx 2 3 5 ONU 1 50km Installed SMF Otemachi DCF 50km Installed SMF 7 8 9 13 OCDM Rx 8 (xi) 20 [ps/div] (N. Kataoka, NiCT, OFC 2008) OCDM Tx 2 λ :1551nm 1:8 splitter ONU 8 15 (xii) OCDM Tx 1 λ :1551nm 10.3125GH ｚ ONU 2 Koganei 11 PPG (x) 20 [ps/div] (ix) 20 [ps/div] OCDM Tx 3 λ :1551nm PC3 HDV Camera OCDM Tx 4 λ :1551nm PC4 HDV Camera (viii) 20 [ps/div] 10GbE Network Analyzer (vii) DPSK signal 20 [ps/div] 20 [ps/div] Uplink
Summary Convergence to packet networks Standardization is essential IP in the core already here Future: Optical packet switching??? Access networks are running on GPON Extended reach GPON is coming Future: WDM PON? OCDMA PON?
Thanks: David Piehler, Alphion Corp., USA A. Gnauck, G. Raybon, P. Winzer, Alcatel Lucent, USA E. Tangdiongga, N. Calabretta, TU Eindhoven, Netherlands G. Cincotti, Roma Tre University, Italy N. Wada, NiCT, Japan P. Iannone, K. Reichmann, AT&T Labs Research, USA A. Srivastava, OneTerabit, USA