pon pmd timing - ieee 802 · lg.gpon.pmd has been submitted as a white contribution to sg -15...
TRANSCRIPT
![Page 1: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/1.jpg)
PON PMD Timing
Frank Effenberger
![Page 2: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/2.jpg)
List of Supporters / Contributors
l Ajung Kim Samsungl Bob Deri Terawavel Brian Ford BellSouthl David Cleary Optical Solutionsl Eyal Shraga Flexlightl Francois Fredricx Alcatell Frank Effenberger Quantum Bridgel Hiroki Yanagisawa NECl John George OFS l Kent McCammon SBCl Meir Bartur Zonul Michael Wirtz Maximl Oren Marmur Flexlightl Raanan Ivry Broadlightl Richard Michalowski Sprint – Local Telecom Divisionl Walt Soto Agere
![Page 3: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/3.jpg)
Outline
l PON timing overviewl General design considerationsl Current ITU-T PMD situationl Gigabit PON PMD in practicel Strategyl Conclusion
![Page 4: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/4.jpg)
PON Timing Overview
l PON timing can be described by 6 attributes– Tx On time, Ton– Tx Off time, Toff– Rx Dynamic Sensitivity Recovery time, Tdsr– Rx Level Recovery time, Tlr– Rx Clock Recovery time, Tcr– Rx Delimiter time, Tdl
l Note that in 802.3ah, Tcr is in domain of PMA, and Tdl is in domain of PCS and MAC
![Page 5: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/5.jpg)
Our focus here is PMD
l Transmitter timesl Ton limited by driver
circuitry stabilization of laser bias
l Toff limited by speed of ‘switching element’
l Laser is never limiting element
l Receiver timesl Tdsr limited by
– Circuit slewing rates– PIN diffusion tails
l Tlr limited by– Accurate measurement of
threshold level – Speed of track and hold
circuits (if used)
![Page 6: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/6.jpg)
General Design Considerations,Transmitter
Imod Ibias
DataData
Vcc Vcc
TxOffTxOn
Vcc
![Page 7: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/7.jpg)
General Design Considerations,Receiver
![Page 8: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/8.jpg)
Current status of ITU-T
l G.GPON.pmd is the outcome of extensive deliberations of the world’s PON experts – It provides a comprehensive set of specifications for
PON optics at many rates, including symmetric Gigabit per second
l G.GPON.pmd has been submitted as a white contribution to SG-15 plenary– It will be consented on Jan 31, 2003
![Page 9: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/9.jpg)
What G.GPON.pmd says
l The document specifies normatively:– Ton = 16 bits (12.9 ns)– Toff = 16 bits (12.9 ns)– Total burst overhead time = 96 bits (77.2 ns)
l The document specifies informatively:– Guard Time = 32 bits– Preamble Time = 44 bits– Delimiter Time = 20 bits
![Page 10: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/10.jpg)
How to map timing parameters
l The guard time includes– Laser On and Off times– Timing inaccuracies of protocol (small for GPON)– Dynamic sensitivity recovery time (Tdsr)– All of these can overlap to some degree
l The preamble time includes– Level recovery time (Tlr)– Clock recovery time (modest in GPON)
![Page 11: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/11.jpg)
“Equivalent” values for EPON
l Ton and Toff are close to 2 symbol times– Ton = 16 ns– Toff = 16 ns
l If we subtract out the time for delimiter (20 bits) and clock recovery (12 bits) in GPON, we obtain 64 bits for PMD related functions– Tdsr+Tlr = 50 ns
![Page 12: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/12.jpg)
PON Timing in Practice
l The following slides give illustrative examples of Gigabit rate PON optics that achieve ITU-T timing
![Page 13: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/13.jpg)
Example: FlexLight-Network
l Burst control at the ONT Tx– No cost penalty associated with Ton/Toff times of 16 ns
l Data and clock recovery at OLT Rx– Less than 12 ns (10 bits @ 1.244 GHz) lock time
![Page 14: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/14.jpg)
Example: Alcatel
l “Alcatel, together with partners, is currently developing GPON modules for meeting the ITU-T spec at 1.244 Gbit/s, including the timing parameters.”
![Page 15: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/15.jpg)
Example: Maxim Semiconductor
l Burst Mode Laser drivers @ 1.2 & 2.5 G– Ton and Toff < 2 ns with a PON design– Ton and Toff ~10 ns with existing P2P design
and auxiliary S&H circuits
l Burst Mode Receiver @ 1.2 & 2.5 G– Tdsr = 6.4 ns, 8 bits at 1.2G
l Requires arming signal given before burst starts
![Page 16: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/16.jpg)
ZONU
l PON ONU transceivers at 0.155, 0.622, and 1.25 G– Ton and Toff < 3 ns, regardless of speed
![Page 17: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/17.jpg)
Example: BroadLight
l 1.25 Gb/s ONT Transmitter implementation– Ton/Toff times of 4 bits
![Page 18: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/18.jpg)
Vendor “X”
l PON ONU transceivers– Ton =10~14 ns – Toff = 3~5 ns
l Work is still ongoing to improve these values
![Page 19: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/19.jpg)
Example: NEC
l Laser turn-on/turn-off time at ONU Tx– Already developed with Ton=<16bits and Toff=<1bits– No training procedure is required
l Data recovery time at OLT Rx– Targeting Tdsr+Tlr =< 40 ns with reset signal– Circuit design is finished (LSI chip is under fabrication)
Tx Enable
Data In
Optical(2GHz filtered)
Ton=<16bits Toff=<1bits
16bits(12.8ns)
![Page 20: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/20.jpg)
EPON PMD Strategy
l The current PMD timing ‘debate’ goes back to March 2002
l The two sides (‘tight’ and ‘loose’) have approached the problem from completely different directions
![Page 21: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/21.jpg)
The ‘Tight’ approach
l Determine what are the best times achievable while still being cost effective
l Maintain maximal compatibility between all relevant standards
l Drive towards a single market for PON components to boost volume
l This approach leads one to the ITU values
![Page 22: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/22.jpg)
The ‘Loose’ approach
l Determine what are the worst times permitted by efficiency considerations
l This approach leads one to long values for the PMD times
![Page 23: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/23.jpg)
What the numbers really mean
l Really, the timing numbers presented by both camps are limits, not equalities
l Tight limits: Timing can’t get shorter than X, otherwise component cost goes up
l Loose limits: Timing can’t get longer than Y, otherwise efficiency goes down
l The correct choice lies between
![Page 24: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/24.jpg)
Benefits of ITU concordance
l Optics vendors will have a common set of specifications to use– The decision to build PON parts will be easier– More optics companies will participate
l There will be a single pool of PMDs and component parts for all PON systems– Volume effects will be larger and faster– Market uncertainties will be reduced
![Page 25: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/25.jpg)
Think about it
l The Copper track has chosen to defer action pending the line-code decision of T1E1
– The T1E1 decision hasn’t even been made yet– Doing so has delayed 802.3ah’s schedule in a big way– Even then, the benefits of concordance outweigh the costs and
risks of doing so
l The Optics track must now make the same choice– In contrast, the ITU-T parameters are ready now– We can obtain all the benefits at no cost
![Page 26: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/26.jpg)
A Compromise
l A compromise exists (option C), where – ONT timing follows the ITU-T values (16 ns for both)– OLT timing is ‘negotiable’
l This is favorable because– the laser on/off spec is available from several vendors today
(technically feasible) – no ONU parameter negotiation during discovery – it avoids multiple "flavors" of ONTs (important) – OLT does not need to keep an ONT parameter table – it matches ITU-T specs at the ONT
![Page 27: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/27.jpg)
Compromise (cont.)
l Common ONT gives us 95% of the benefit – The ONT is always much more numerous than the
OLT, and is cost sensitive
l A flexible OLT avoids 95% of the trouble– OLT timing is the harder physics problem– Architectural restrictions tend to limit implementation
options in the EPON setting
![Page 28: PON PMD Timing - IEEE 802 · lG.GPON.pmd has been submitted as a white contribution to SG -15 plenary](https://reader031.vdocuments.mx/reader031/viewer/2022022519/5b15817e7f8b9ac7128cd5ad/html5/thumbnails/28.jpg)
Conclusions
l Option C is optimal choice– Broad Market Potential – ü– Compatibility – ü– Distinct Identity – ü– Technical Feasibility – ü– Economic Feasibility – ü