Tera-Bit Optical Submarine Networks - Meeting the Market's Capacity Demands at Lowest Overall Cost Tera-Bit Optical Submarine Networks - Meeting the Market's Capacity Demands at Lowest Overall Cost Katsutoshi Tamura, General Manager Submarine Networks Business Division International Telecommunications Business Group Fujitsu Limited Tatsuo Matsumoto, Senior Director Submarine Telecommunications Engineering Division Transport Systems Group Fujitsu Limited Colin Anderson, Manager Business Development Submarine Networks Sales & Marketing International Telecommunications Business Group Fujitsu Limited PTC2000 Hawaii: A New Vision for the 21 st Century Session T Tuesday 1 February 2000

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 2 Introduction l Demand for international traffic continues driven by the Internet l Vendors strive to meet capacity and cost demands l Fortunately technology has enabled both capacity increases and cost reductions l Focus of this paper is “cost” rather than “capacity” l What have been the price implications of the technologies recently deployed ? l What will be the likely impacts of the next generation of 'enabling technologies' on price as well as capacity ? l Which technologies will be best for the future submarine networks ?

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 3 Typical WDM Optical Submarine Network Configuration Terminal Station Equipment WDM: N channels of traffic onto N wavelengths on a single fiber Terminal Station Equipment WDM Evolution: 8 x 2.5 Gb/s x 2.5 Gb/s … 16 x 10 Gb/s x 10 Gb/s … 64 x 10 Gb/s x 10 Gb/s... ? 8 x 40 Gb/s x 40 Gb/s... ? Up to 200 Cascaded Optical Amplifiers Span between Terminals: 500 km ~ 10,000 km (span between “optical - electrical” & “optical - electrical” conversion) 40 ~ 80 km between Repeaters

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 4 Key Enabling Technologies l Erbium Doped Fiber Optical Amplifer l Study mid 1960's l Practical reality in laboratories mid-1980's l Practical in commercial networks early 1990's l Slow start perhaps, but a dramatic impact in latter part of 1990's l Dense DWM Optical Devices l Wavelength-Locked Lasers l Tunable lasers l Passive optical devices (filters, multiplexers, etc...) l etc

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 5 History of WDM Optical Submarine Networks l 1995: 1 wave of 2.5 Gb/s or 5.0 Gb/s l 1998: 8 waves x 2.5 Gb/s or 16 waves x 2.5 Gb/s l 1999 / 2000: 32 waves x 10 Gb/s being contracted l Systems with 64 waves x 10 Gb/s will be commercialised in the next two years l Foreseeable future: 128 x 10 Gb/s using C-Band and L-Band l N x 40 Gb/s systems will follow l Currently up to 4 fiber pairs in submerged plant l 6 and 8 fiber pair systems by 2002

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 6 Figure 1: Transmission Capacity per Optical Fiber (8 x 2.5 Gb/s ~ 32 x 40 Gb/s) Nomenclature: "10 x 32 x 4" means "10 Gb/s x 32 waves x 4 fiber pairs" 1,000 Gb/s = 1.0 Tb/s N x 10 Gb/s N x 40 Gb/s

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 7 Figure 2: Transmission Capacity per Cable System (8 x 2.5 Gb/s ~ 32 x 40 Gb/s) Nomenclature: "10 x 32 x 4" means "10 Gb/s x 32 waves x 4 fiber pairs" 10,000 Gb/s = 10 Tb/s

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 8 History of Submarine Cable Capacity l Period from 1989 to 1999 l eg: TPC 3 = 2 x 280 Mb/s Optical Regenerator System l Japan - US Cable = 16 x 10 Gb/s x 4 fiber pairs l Greatest increase in capacity with introduction of WDM technology l Extrapolation to Year 2010 ? l For example using the 'rule-of-thumb' growth rate prediction of "2 times per year" from 1999 base ?

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 9 Figure 3: Submarine Cable Capacity verses Time, 1989 ~ 2010 ? 1,200 m 120 m 12 m 1,200, ,000 Equivalent number of voice circuits (uncompressed)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 10 Price History of Submarine Cable Systems l Breakdown of price has been changing as capacity has increased l In past, large percentage of total price was in submerged plant, and capacity was fixed from initial deployment l Increasing number of waves of WDM has led to increased percentage of the total price is terminal equipment l < 8 x 2.5 Gb/s: submerged 50 ~ 65 %; terminal 8 ~ 25 % l 32 x 10 Gb/s: submerged 20 ~ 40 %; terminal 50 ~ 60 % (fully equipped) l (major variation is with SLTE - SLTE span) l Future terminal equipment approaching 70 % fully equipped? l Also an increase in floor space for terminal equipment

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 11 Price per Unit Capacity Comparison l Price per unit of traffic capacity has dramatically decreased ("price-per-bit" or "price-per-STM-1" etc) l One of the factors stimulating cable deployment l Internet provided traffic demand (pull), and technology has reduced the cost per bit faster than market decreases in selling price per bit l For example l 8 x 2.5 Gb/s to 16 x 2.5 Gb/s l ~ 40 % decrease in cost per STM-1 due to technology l 16 x 2.5 Gb/s to 16 x 10 Gb/s: ~ 65 % decrease in cost per STM-1 l 32 wave systems: perhaps 30 % ~ 35 % lower than 16 x 10 Gb/s ? l Full information in Figure 4 (2,000 km) and Figure 5 (8,000 km)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 12 Figure 4: Overall Price per STM-1 over 2,000 km Submarine Link 1, 2, 3, 4,... 6,... 8 pairs

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 13 Figure 5: Overall Price per STM-1 over 8,000 km Submarine Link 2, 4, 6, 8 pairs

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 14 Technology History, Current & Future Technology Trends l Optical Amplifier Bandwidth & Amplitude Response l Traditionally used optical C-band (centered on 1,550 nm) l L-Band becoming available (new EDFA) l Bandwidth and flatness improvements l Terrestrial systems announced in mid-1999: 80 x 10 Gb/s in C-Band + 90 x 10 Gb/s in L-Band (1.7 Tb/s per fiber) l For submarine systems: C-Band = 26 nm, L-Band = 30 nm useable? l Number of WDM Channels, Bit Rate, Channel Spacing l WDM wave spacing: 1.6nm  0.8 nm  0.4 nm l 0.3 nm possible ? 0.2 nm unlikely ? l 0.4 nm allows > 64 waves in C-Band plus > 64 waves in L-Band

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 15 Optical Fiber Spectrum & Types of Optical Amplifier 1,450nm 1,490nm 1,530nm 1,570nm 1,610nm 1,650nm S+ Band S Band C Band L Band L+ Band RFA TDFAEDTFA GS-EDFAEDFA Erbium Doped Fiber Amplifier Gain-Shifted Erbium Doped Fiber Amplifier Tellurite-Based Erbium Doped Fiber Amplifier Thulium Doped Flouride-Based Fiber Amplifier Raman Fiber Amplifier Total ~ 200 nm: 500 ~ 1,000 waves ? 80 nm: ~ 200 waves ? 40 nm 1,550nm1,580nm Potential of Optical Fiber: perhaps 250 waves x 100 Gb/s = 25,000 Gb/s = 25 Tb/s ?

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 16 Technology History, Current & Future Technology Trends l Number of WDM Channels, Bit Rate, Channel Spacing (cont) l As channel numbers increase, total power must be kept constant and so power per wave decreases l Repeaters need to be closer together (price and noise increase) l Eventually, increasing the number of repeaters to give closer repeater spacing gives worse performance (noise increase overwhelms other gains). Limit of the technology is reached. l Optical Amplifier Pumping Technologies l Traditionally 1,480 nm pumping lasers (cost & reliability) l 980 nm lasers now available for lower noise in pre-amplifier stages l combination of 980 nm and 1,480 nm in 'forward' and 'reverse' pumping directions currently optimum

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 17 Forward & Reverse Pumping Using 980 nm & 1,480 nm Pumping Lasers Erbium Aluminum Doped Optical Fiber L = 10 ~ 80m, Er ~ 500 ppm = 0.05 % 980nm Pump Laser Diode 20:1 Coupler 1,480nm Pump Laser Diode Rear Modulator Reflector / Isolator PIN InputOutput Long Period Fiber Grating SV Monitor & Control Circuiit LPG PIN 20:1 Coupler Input Level Monitor Photo-Diode Output Level Monitor Photo-Diode DC Input Power: 9 V 0.87 A ~ 8 W typ 3dB Coupler WDM MUX 3dB Coupler 980nm Pumping 1,480nm Pumping

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 18 Technology History, Current & Future Technology Trends l Optical Amplifier Pumping & Output Power l Use of two 980 nm Pump Lasers and two 1,480 nm Pump Lasers is now not only cost effective, but further benefits reliability against hardware failures of lasers l Fiber non-linearities (not the amplifiers in the repeaters) now limit the maximum output power l Optical Amplifier Noise Figure l Current schemes have reduced noise figure of the amplifiers from 6.7 dB to around 5.5 dB resulting in increased spans between repeaters and lower overall costs

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 19 Technology History, Current & Future Technology Trends l Non-Linear Effects / Optical Fiber Effective Area l Non-Zero DSF has relatively small "effective area" compared to regular "Single Mode Fiber" (SMF) l Concentration of the light energy causes non-linear effects in the optical fiber l Several "Large Effective Area" optical fibers now available l "Large Effective Area Fiber" is itself more expensive, but used in the first half of the span it allows higher output powers (without non- linear distortions) l Hence increase repeater spacing (overall cost savings)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 20 Technology History, Current & Future Technology Trends l Dispersion Compensation l Non-Zero DSF (or Large Effective Area optical fiber) + positive dispersion fiber, to give overall average zero dispersion l But only at one wavelength! Imperfect correction at other wavelengths l Increasing numbers of waves of WDM mean increased band-widths, and the current dispersion compensation schemes are not perfect over large band-widths.

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 21 Technology History, Current & Future Technology Trends l Amplitude-Slope Compensation l Amplitude-slope is introduced by the fiber itself as well as the amplifiers l Current technologies only partially compensate l Active Gain-Slope Correction l New technology - remotely provisionable over the lifetime of the system. Reduce initial margins, and hence repeater cost savings

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 22 Effect of Gain Slope in the Network Noise Floor Degraded Optical SNR (Signal to Noise Ratio) Degraded SNR Input Signal eg: 32 x 10 Gb/s After Transmission (Case 1) After Transmission (Case 2) Before Transmission Uniform Signal to Noise Ratio (SNR)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 23 Advances in Terminal Equipment l Modulation Techniques l Traditionally Non-Return-to-Zero coding (NRZ) was preferred l Recently significant advances in modulation hardware devices have meant that Return-to-Zero modulation coding is simpler and more cost effective for 10 Gb/s WDM systems l However other schemes (Optical Duo-Binary, etc) hold even further promise for 40 Gb/s systems (improved dispersion tolerance, etc) l Forward Error Correction l Redundant information to allow error correction at the far end l Bit rate is increased, but improvements in SNR far outweigh this penalty l Currently 4 ~ 6 dB of improvement (7 % bit rate increase)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 24 Advances in Terminal Equipment l Forward Error Correction (cont) l Next generation "Super FEC" gives 7 ~ 10 dB of improvement (equivalent to > 4 x number of WDM waves) l Increased repeater spacing and significant cost savings l Increased maximum spans l Dispersion Compensation l Reverse Dispersion Fibers (RDF or +D / -D) l Improved technical performance as well as space savings at terminal stations (less DCF)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 25 Advances in Terminal Equipment l Tunable Lasers l Big savings for customer in spares l Savings for manufacturer in number of different component types l Eventually multiple wavelength arrays - further cost savings l Floor Space Requirements l Dense WDM systems require increasing terminal station space l Cable station space is a real cost to the customer l Re-locate SLTE to Central Station? (pros & cons) l Separate Cable termination & Power Feed (at shore station) from SLTE (at intermediate site) l Use optical-layer protection instead of SDH protection

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 26 Next Generation 40 Gb/s Systems l Next logical choice for transmission rate after 10 Gb/s is 40 Gb/s l Many technical challenges (much more difficult than the migration 2.5 Gb/s  10 Gb/s) l Key issues include l very high speed optical and electronic components l severe effects of Chromatic Dispersion, Self-Phase Modulation (SPM), and Polarisation Mode Dispersion (PMD) in the optical fiber when transmitting 40 Gb/s l To eventually be successful we know that 40 Gb/s systems will need to offer capacity increase at significantly reduced price per bit, as well as floor space savings l Past historical rule: “... 4 times the capacity for 2 ~ 3 times the price...” ? Assumed in this paper.

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 27 Future Submarine Network Price Trends l System prices modelled for spans of 2,000 km ('short-haul') and 8,000 km ('long-haul') as earlier discussed l In fact N x 40 Gb/s may be limited to less than 8,000 km for some time to come... but we assumed that the hurdles will eventually be overcome l Current market prices used where items exist, and 'best estimate' prices used for future technologies l Hypothetical study, but rational and hopefully useful

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 28 Figure 6: Overall Price per STM-1 over 2,000 km Submarine Link N x 10 Gb/s N x 40 Gb/s 1, 2, 3, 4,... 6,... 8 pairs

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 29 Figure 7: Overall Price per STM-1 over 8,000 km Submarine Link N x 10 Gb/s N x 40 Gb/s 2, 4, 6, 8 pairs

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 30 Price-per-Bit Comparison Summary l 64 x 10 Gb/s compared to 32 x 10 Gb/s (640 Gb/s per fiber pair cf 320 Gb/s per fiber pair) (2x) l Long-haul: approx 25% savings (20 ~ 30%) l Short-haul: approx 23% savings (17 ~ 30 %) l 128 x 10 Gb/s compared to 64 x 10 Gb/s (1,280 Gb/s per fiber pair cf 640 Gb/s per fiber pair) (2x) l Long-haul: 10 ~ 15% increase in price per bit (but capacity doubled) l Short-haul: approx same price per bit (but capacity doubled) l 8 x 40 Gb/s compared to 32 x 10 Gb/s (320 Gb/s per fiber pair in both cases) (1x) l 15 ~ 20 % savings approx (short-haul or long-haul)

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 31 Price-per-Bit Comparison Summary l 16 x 40 Gb/s compared to 64 x 10 Gb/s (640 Gb/s per fiber pair in both cases) (1x) l ~ 25 % savings approx (short-haul or long-haul) l 32 x 40 Gb/s compared to 64 x 10 Gb/s (1,280 Gb/s per fiber pair cf 640 Gb/s per fiber pair) (2x) l ~ 50 % savings approx (short-haul or long-haul) Please correct your hard-copy printout

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 32 Comparison of 40 Gb/s to 10 Gb/s l 64 x 10 Gb/s systems are economical compared to 32 x 10 Gb/s, and will continue to provide good solutions for up to 5 Tb/s per cable (64 x 10Gb/s x 8 fp) at low cost-per-bit l Next step of 128 x 10 Gb/s may not be so attractive from point of view of ‘price per bit’ or floor space requirements l When 40 Gb/s systems become available commercially they will compete well at 320 Gb/s per fiber and above, and will offer best solutions for 320 Gb/s to 10 Tb/s per cable (32 x 40G x 8 fp) l 40 Gb/s systems can be expected to much reduce floor space requirements at terminal stations of very high capacity systems

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 33 Future Network Architectures & Protection Schemes l The above analysis does not include SDH Multiplex or Network protection Equipment l Combined SDH (SIE, MUX & NPE) typically represents approx 15 % of the total network price, fully equipped (& much less for initial sub-equipped configurations; perhaps 3 ~ 7 % ?) l Other drivers are acting - SONET / SDH are excellent for voice networks but somewhat inefficient for data-centric and IP-centric networks: ‘IP over WDM’ vs ‘IP over SDH’ l Separate the MUX / SDH SIE requirement from the NPE requirement ?

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 34 Future Network Architectures & Protection Schemes l Full function Network Protection can be provided by new optical layer NPE equipment without the need for any protocol dependence (SDH, etc), and with lower power consumption and floor space requirements than for SDH l Price is already less than for SDH NPE in some configurations l Increased use of optical layer NPE in terrestrial networks will soon see further price reductions in optical switches and optical NPE's

PTC2000: Tera-Bit Optical Submarine Networks - Meeting the Market's Demand at the Lowest Overall Cost File: Tera-Bit Submarine Networks.ppt Issue January 2000Copyright - Fujitsu ProprietarySlide 35 Summary & Conclusions l We have tried to identify the impacts of recent technology developments on both capacity, price, & price-per-bit for submarine cable networks l In future there seem to be several identifiable promising new key technologies, including 40 Gb/s transmission, which will be able to be exploited to give further capacity increases and at the same time give price-per-bit decreases l The era of ‘Terabit’ Submarine Cable Networks is certainly already with us - and the same kind of technology developments which made those networks feasible seem likely to be able to continue to offer the future solutions which the market-place demands, and at affordable prices