Super-PON Scale Fully Passive Optical Access Networks to Longer Reaches and to a Significantly Higher Number of Subscribers Claudio DeSanti Liang Du Cedric Lam Joy Jiang
Agenda Super-PON Idea Why Super-PON? Super-PON PMD Claudio DeSanti 2
IEEE 802.3 EPON Architecture Up to 64 IEEE 802.3 Clause 64 Up to 20Km Claudio DeSanti 3
Super-PON Scalability Passive Optical Network Up to 1024 Up to 50Km Claudio DeSanti 4
From Here To Here 1 1 2 2 n n 1 1 2 n... DWDM m Cyclic AWG... q 2 n 1 1 2 2 Claudio DeSanti n 5 n
The Full Picture 1 Amplification enables longer reach (target: up to 50Km) DWDM enables more subscribers (target: n=64 x m=16 = 1024) The optical network is fully passive 2 n 1 10G-EPON l1 10G-EPON l2 10G-EPON lm MUX DMUX booster preamp BAND MUX... m Cyclic AWG 2 n 1 2 n Claudio DeSanti 6
Super-PON Architecture Residential New PMD: Power, l, 1 2 MUX/Amplifier Power, l n 1 10G-EPON l1 10G-EPON l2 10G-EPON lm MUX DMUX booster preamp BAND MUX... m Cyclic AWG FSRs, l 2 n 1 New PMD: Duplex, Power, l, Power, l n Claudio DeSanti 7 2
Super-PON Architecture Point to Point Support MUX/Amplifier Pt2Pt 10G l1 booster Pt2Pt 10G l2 Pt2Pt 10G lq MUX DMUX preamp BAND MUX... q Cyclic AWG FSRs, l... P2P 1 P2P 2 P2P q New PMD: Duplex, Power, l, Power, l New PMD: Power, l, Claudio DeSanti 8
Super-PON Architecture Complete 1 10G-EPON l1 MUX/Amplifier 2 n 10G-EPON l2 10G-EPON lm Pt2Pt 10G l1 Pt2Pt 10G l2 MUX DMUX booster preamp BAND MUX... m+q Cyclic AWG... P2P 1 P2P 2 P2P q A: up to 40 Km B: up to 20 Km A + B 50 Km 1 2 n 1 Pt2Pt 10G lq n Claudio DeSanti 9 2
Agenda Super-PON Idea Why Super-PON? Super-PON PMD Claudio DeSanti 10
Conventional PON Coverage 20 Km 50 Km Claudio DeSanti 11
Super-PON Coverage Centralized 20 Km 50 Km Larger serving area Fewer number of s Smaller fiber bundles Claudio DeSanti 12
Real Example Conventional PON: 16 s Super-PON: 3 s 10 Km 10 Km Feeder fiber Significantly smaller number of s Better fiber utilization Much less backbone and feeder fiber Lower OSP building cost CAWG feeder fiber Splitter feeder fiber Claudio DeSanti 13
Advantages Fewer fiber strands exiting a Enables smaller/fewer cables From 432-fiber cables to 12/48-fiber cables Lower OSP building cost Smaller cables can be longer and are easier to bend/handle Allows use of micro-trenching and directional boring techniques Easier to repair consolidation The same number of feeder fibers can serve a much greater area Less s à less OPEX Claudio DeSanti 14
About Trenching Traditional Trenching Micro Trenching Directional Boring Claudio DeSanti 15
and Repairs 6 feeder cables (432-fiber) A 432-fiber cable: Contains 36 ribbons of 12 fibers ~10 min to splice a ribbon ~6 hours total to splice a broken cable Additional ~2 hours for cable manipulation Average time to repair a cable damage: ~8 hours A 24-fiber cable: ~40 mins total to splice a broken cable Additional ~1 hour for cable manipulation Average time to repair a cable damage: ~1 hour 40 Claudio DeSanti 16
Super-PON Applicability Well suited for new optical plants (OSPs) developments Significant savings in cabling and building cost Valuable as a retrofit to existing OSP for (5G) cellular deployments Integrated support for both point-to-point and residential customers Can be used to consolidate s leveraging existing fiber plants A RD (Central Office Redesigned as a Data-center) enabler Increased typical utilization of ports Claudio DeSanti 17
Cellular Support Backend Central Office with Space and Power Point to Point Links BBU Optical MUX / Amp Fronthaul Backhaul Cellular Tower Integrated support for both residential and point-to-point deployments Use point-to-point for cellular support Micro-trenching in the tower neighborhood Claudio DeSanti 18
Consolidation Operators find advantageous co-locating data centers with access networks for value added services Current industry trend is toward Data Center consolidation A small set of large data centers rather than a large set of small data centers This also pushes for Central Offices consolidation Upcoming RD (Central Office Re-architected as a Data center) architectures further push toward consolidation Claudio DeSanti 19
RD: Where? a A data-center in each is not cost effective consolidation is needed Claudio DeSanti 20
Consolidation Claudio DeSanti 21
Consolidation Claudio DeSanti 22
Consolidation Let s add a CAWG in peripheral s Claudio DeSanti 23
Consolidation 2:2 Let s add a CAWG in peripheral s Claudio DeSanti 24
Consolidation 2:2 2:2 2:2 2:2 2:2 2:2 2:2 2:2 2:2 Claudio DeSanti 25
Consolidation 2:2 2:2 2:2 2:2 2:2 2:2 Let s use the backbone fiber as CAWG feeder 2:2 2:2 Enable the new service 2:2 Claudio DeSanti 26
Consolidation 2:2 2:2 2:2 2:2 2:2 Remove the old service 2:2 2:2 2:2 Claudio DeSanti 27
Consolidation Remove the old service Claudio DeSanti 28
Consolidated Consolidated RD is now cost effective Claudio DeSanti 29
Super-PON Applicability Summary Green field: Optical fiber plant build simplification (lower CAPEX and TTM) Support for both residential and point-to-point applications Brown field (optical fiber plant already in place): consolidation for RD Re-use existing fiber plant and transform peripheral s from (managed) active sites to (unmanaged) holders of passive components Increased typical utilization of ports Point-to-Point support for (5G) cellular and specific subscribers OSP expansion (i.e., additional OSP build to complement the existing OSP) Claudio DeSanti 30
Super-PON Residential Market Opportunity 2.5Gb/s GPON will continue to be mainstream technology until 2020 10Gb/s PON becomes significant starting from 2020 The transition to 10Gb/s can be a significant market opportunity for Super-PON Multiple 10Gb/s channels support Super-PON may actually help the transition to 10Gb/s Enables infrastructure optimizations Claudio DeSanti 31
Super-PON Point-to-Point Market Opportunity The current growth of cellular networking brings significant opportunity for the point-to-point support offered by Super-PON Source: Ericsson Mobility Report Claudio DeSanti 32
Agenda Super-PON Idea Why Super-PON? Super-PON PMD Claudio DeSanti 33
PMA PMA Super-PON PMD Residential SIGNAL_DETECT SIGNAL_DETECT PMD PMD MDI TP1 TP2 TP3 TP4 Active Optical MUX/Amp TP11 TP12 SMF cable TP5 TP6 Channel: Fiber Optic Cabling Optical MUX/Amp Cyclic AWG Optical Splitter Passive Cyclic AWG Fiber optic cabling, active optical MUX/Amp, passive cyclic AWG, and passive optical splitter (Channel) Claudio DeSanti 34 SMF cable SMF cable Passive Optical Splitter Passive Optical Splitter SMF cable SMF cable TP7 SMF cable SMF cable TP8 MDI PMD PMD PMD PMD Tx_Enable SIGNAL_DETECT Tx_Enable SIGNAL_DETECT TP9 Tx_Enable SIGNAL_DETECT Tx_Enable SIGNAL_DETECT PMA PMA TP10 PMA PMA
Super-PON PMD Point to Point MDI TP1 TP2 TP3 TP4 SMF cable P2P PMD Tx_Enable PMA TP11 TP12 SIGNAL_DETECT PMA Pt2Pt PMD TP7 TP8 TP9 TP10 PMA SIGNAL_DETECT SIGNAL_DETECT Pt2Pt PMD Active Optical MUX/Amp SMF cable TP5 TP6 Channel: Fiber Optic Cabling Optical MUX/Amp Cyclic AWG Passive Cyclic AWG SMF cable MDI P2P PMD Tx_Enable SIGNAL_DETECT PMA Fiber optic cabling, active optical MUX/Amp, and passive cyclic AWG (Channel) Claudio DeSanti 35
Existing 802.3 PMD Definitions Channel: Fiber Optic Cabling Optical Splitter Claudio DeSanti 36
Cyclic AWG Exhibits the same behavior across its FSRs This enables seamless upgrades upstream downstream FSR 4 FSR 2 FSR 4 FSR 2 l FSR 3 FSR 1 l Cyclic AWG l l l Gen Y Gen X FSR 3 FSR 1 Claudio DeSanti 37
Gen X Service 10G-EPON l1 MUX/Amplifier Gen X Service: downstream l FSR 1 upstream l FSR 3 1 2 10G-EPON l2 n FSR 1 & FSR 3 10G-EPON lm Pt2Pt 10G l1 Pt2Pt 10G l2 Pt2Pt 10G lq MUX DMUX booster preamp BAND MUX... m+q Cyclic AWG... P2P 1 P2P 2 P2P q 1 2 n 1 2 n Claudio DeSanti 38
Upgrade to Gen Y Service 10G-EPON l1 MUX/Amplifier Gen X Service: downstream l FSR 1 upstream l FSR 3 Gen Y Service: downstream l FSR 2 upstream l FSR 4 1 2 10G-EPON l2 n FSR 1 & FSR 3 FSR 2 & FSR 4 10G-EPON lm Pt2Pt 10G l1 Pt2Pt 10G l2 Pt2Pt 10G lq 10G-EPON l1 10G-EPON l2 10G-EPON lm Pt2Pt 10G l1 MUX DMUX booster preamp BAND MUX... 2x(m+q) Cyclic AWG... P2P 1 P2P 2 P2P q 1 2 n 1 Pt2Pt 10G l2 2 Pt2Pt 10G lq n Claudio DeSanti 39
Gen Y Service MUX/Amplifier Gen Y Service: downstream l FSR 2 upstream l FSR 4 1 2 n FSR 2 & FSR 4 10G-EPON l1 10G-EPON l2 10G-EPON lm Pt2Pt 10G l1 MUX DMUX booster preamp BAND MUX... m+q Cyclic AWG... P2P 1 P2P 2 P2P q 1 2 n 1 Pt2Pt 10G l2 2 Pt2Pt 10G lq n Claudio DeSanti 40
Defining the Optical Parameters Using EDFAs as amplifiers implies using the C- and L-bands for wavelengths C-band: 1530.. 1565 nm, upstream L-band: 1565.. 1625 nm, downstream Split the bands in two ~equally sized ranges to support speed upgrades Gen X upstream: ~1530.. 1546 nm Gen Y upstream: ~1547.. 1563 nm Gen X downstream: ~1565.. 1581 nm Gen Y downstream: ~1583.. 1599 nm These ranges define the FSRs of the cyclic AWG Within each range, define a set of wavelengths to use for DWDM transmission 20 channels using a nominal channel spacing of 100 GHz Claudio DeSanti 41
Wavelength Plan EPON upstream EPON downstream Gen Y Super-PON l Gen X Gen Y Gen X 1260 1270 1280 1290 1300 1310 1320 1330 1340 1350 1360 1370 1380 1390 1400 1410 1420 1430 1440 1450 1460 1470 1480 1490 1500 1510 1520 1530 1540 1550 1560 1570 1580 1590 1600 l upstream downstream 10G-EPON upstream 10G-EPON downstream 1260 1270 1280 1290 1300 1310 1320 1330 1340 1350 1360 1370 1380 1390 1400 1410 1420 1430 1440 1450 1460 1470 1480 1490 1500 1510 1520 1530 1540 1550 1560 1570 1580 1590 1600 l 25G/50G/100G-EPON Claudio DeSanti 42
Example of Residential and Point-to-Point l Residential: Point-to-Point: Gen Y Gen X Gen Y Gen X 1530 1540 1550 1560 1570 1580 1590 1600 l upstream downstream No need to specify in the standard which wavelengths are for what Can be deployment/implementation specific Claudio DeSanti 43
Preliminary PMD Requirements Continuous-mode wide-band receiver Cooled wavelength-stabilized burst-mode laser transmitter 100GHz nominal channel spacing to enable operation without a wavelength locker Laser transmitter can be: Not tunable (i.e., one l) Partially tunable (e.g., four adjacent l) Fully tunable (e.g., 16 l, easier than full C-band) l C-band {1530.. 1565 nm} PMD speeds can be: Symmetric: 10Gb/s upstream, 10Gb/s downstream Asymmetric: 1Gb/s upstream, 10Gb/s downstream or 2.5Gb/s upstream, 10Gb/s downstream Relaxed power budget because of amplification in the MUX/Amp 1 Gb/s 2.5 Gb/s 10 Gb/s Launch power ~[-1 to 4] dbm ~[-1 to 4] dbm ~[4 to 9] dbm Receiver sensitivity - - ~-28.5 dbm (FEC) - 1Gb/s & 2.5Gb/s from PX10-U - 10Gb/s from PR-U3 Assuming ~14.5 db US MUX/Amp gain and ~7.5 db effective noise figure - From PR-U3 Assuming ~12 db DS MUX/Amp gain Claudio DeSanti 44
Optics Cost Trend Cooled lasers are expected to have today a ~10X cost over uncooled ones Also 1G-EPON optics were ~10X of today s cost when they were introduced Cost is strongly related to volumes Millions 120 100 Source: Infonetics (2004-2007), Ovum (2008-2014) 1G-EPON Units (cumulative) 88.6 106.1 80 60 48.0 70.4 See diagram on page 2 of the NG-EPON Call For Interest 40 30.2 20.6 20 15.4 9.1 4.6 1.8 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Claudio DeSanti 45
Preliminary PMD Requirements Duplex (i.e., 2 fibers) to connect to the MUX/Amp module Burst-mode unfiltered receiver No filter required - MUX/Amplifier performs diplexing and filtering functions Cooled wavelength-stabilized continuous-mode 10Gb/s laser transmitter Single l l L-band {1565.. 1600 nm} PMD can be: Symmetric: Asymmetric: 10Gb/s upstream, 10Gb/s downstream 1Gb/s upstream, 10Gb/s downstream or 2.5Gb/s upstream, 10Gb/s downstream Relaxed power budget because of amplification in the MUX/Amp 1 Gb/s 2.5 Gb/s 10 Gb/s Launch power - - ~[0 to 4] dbm Receiver sensitivity ~-28 dbm (No FEC) ~-28 dbm (No FEC) ~-28 dbm (FEC) - Relaxed from PR-D1 Assuming ~12 db DS MUX/Amp gain - From PR-D3 Assuming ~14.5 db US MUX/Amp gain and ~7.5 db effective noise figure Claudio DeSanti 46
Preliminary Point-to-Point PMD Requirements ( side) Duplex (i.e., 2 fibers) to connect to the MUX/Amp module Continuous-mode unfiltered receiver No filter required - MUX/Amplifier performs diplexing and filtering functions Cooled wavelength-stabilized continuous-mode 10Gb/s laser transmitter Single l l L-band {1565.. 1600 nm} Symmetric speed (i.e., 10Gb/s upstream, 10Gb/s downstream) Relaxed power budget because of amplification in the MUX/Amp ~[-10 to -6] dbm launch power ~-20 dbm (no FEC) receiver sensitivity - Relaxed laser power Assuming ~12 db DS MUX/Amp gain - From typical ZR SFP+ specs Assuming ~14.5 db US MUX/Amp gain Claudio DeSanti 47
Preliminary Point-to-Point PMD Requirements (Customer side) Bidi (i.e., 1 fiber) Continuous-mode wide-band receiver Cooled wavelength-stabilized continuous-mode 10Gb/s laser transmitter Single l Partially tunable Fully tunable l C-band {1530.. 1565 nm} Symmetric speed (i.e., 10Gb/s upstream, 10Gb/s downstream) Relaxed power budget because of amplification in the MUX/Amp ~[-10 to -5] dbm launch power ~-23 dbm (no FEC) receiver sensitivity - Relaxed laser power Assuming ~14.5 db US MUX/Amp gain - From typical ZR SFP+ specs Claudio DeSanti Assuming ~12 db DS MUX/Amp gain 48
Preliminary MUX/Amp Requirements Downstream parameters*: Residential channels power: ~+12 dbm per l Point-to-point channels power: ~+2 dbm per l Port-to-Port small signal gain: >12 db Port-to-Port effective noise figure: <12 db Upstream parameters*: Gain clamped EDFA Port-to-Port small signal gain: 14.5 to 17.5 db Port-to-Port effective noise figure: <7.5 db Dispersion Compensation needed: 1Gb/s 2.5Gb/s 10Gb/s 25Gb/s MUX DMUX DCM DCM booster preamp BAND MUX DML No No Yes N/A EML No No No Yes *: Gain, noise figure, and power values are computed to be consistent with the / PMD parameters Claudio DeSanti 49
Preliminary CAWG Requirements (b) Wavelength shift (CB band) Bidirectional Athermal Operational temperature range: -40 C to 65 C Storage temperature range: -40 C to 85 C Cyclic FSRs: FSR 1: ~1599.. 1583 nm FSR 2: ~1581.. 1565 nm FSR 3: ~1563.. 1547 nm FSR 4: ~1546.. 1530 nm Adjacent channel attenuation: >20 db (c) Wavelength shift (CH5 in CB) Claudio DeSanti 50
Speed Considerations For residential support, an EML laser in the and a DML laser in the allow: 10Gb/s downstream 2.5Gb/s upstream if the MUX/Amp does not contain dispersion compensation 25Gb/s downstream 10Gb/s upstream if the MUX/Amp does contain dispersion compensation For point-to-point support, EML lasers on both ends allow: 10Gb/s symmetric if the MUX/Amp does not contain dispersion compensation 25Gb/s symmetric if the MUX/Amp does contain dispersion compensation The 2.5Gb/s Ethernet speed is already defined for UTP cabling and is being defined for backplane operations IEEE 802.3bz and IEEE P802.3cb Defining it for optical operations seems a very doable effort Down clocking the 10Gb/s specification Enables to leverage for Ethernet the existing 2.5Gb/s GPON optical ecosystem Claudio DeSanti 51
Why Now? The RD architecture is getting real Implies consolidation Super-PON complements in the access network the consolidation made possible in compute and services by RD Super-PON point-to-point support helps 5G cellular deployments Technology advancements made cooled lasers and tunable cooled lasers more affordable than before Enables narrow DWDM channel bands Claudio DeSanti 52
Items for Standardization Larger scale optical architecture Including amplification and cyclic AWG Additional PMD specifications New channels Optical parameters (Wavelength plan, power budgets, etc.) Wavelength-stabilized lasers (with optional tunability) Speeds: Symmetric: 10Gb/s upstream, 10Gb/s downstream Asymmetric: 1Gb/s upstream, 10Gb/s downstream Asymmetric: 2.5Gb/s upstream, 10Gb/s downstream Protocol parameters (if any) Claudio DeSanti 53
1 10G-EPON l1 MUX/Amplifier 2 n Summary 10G-EPON l2 10G-EPON lm Pt2Pt 10G l1 Pt2Pt 10G l2 MUX DMUX booster preamp BAND MUX... m+q Cyclic AWG... P2P 1 P2P 2 P2P q 1 2 n 1 2 Pt2Pt 10G lq n Super-PON introduces new technologies in the EPON standard ecosystem DWDM, amplification, mux/demux (e.g., EDFAs and CAWG) It operates in a different region of the spectrum in respect to existing EPON DWDM C- and L-band, requiring cooled lasers Cooled lasers do not have to be expensive Technology is fine, it is all a matter of volumes Super-PON may help bringing down the cost and enable them for other EPON environments The 2.5Gb/s Ethernet speed could make sense for Super-PON Enables leveraging the existing 2.5Gb/s GPON optical ecosystems for Ethernet Let s put some effort in studying these technologies and their greater implications Is there interest in performing this study and eventually prepare a CFI presentation? Claudio DeSanti 54
Thank you Claudio DeSanti 55