1

25G/50G/100G EPON wavelength plan C

Ed Harstead, member Fixed Networks CTO Dora van Veen, Vincent Houtsma, Bell Labs

Supporters:

John Johnson, Broadcom

Sept. 2016

2

3

Only 25G in O-band: Plan C

kramer_3ca_5_0716.pdf

4

25G EPON TRx λ0

WM

Simple diplexer

ODN

S/C/L band

O band 25G ONU

50G/100G ONU

25G OLT TRx λ2

25G OLT TRx λ3

25G OLT TRx λ4

25G OLT TRx λ1

Plan C: high level optical architecture. Focus on 1+4

Step 1: what wavelengths in the S/C/L band for λ1, λ2, λ3, λ4? Step 2: what wavelengths in the O-band for λ0?

5

λ1 - λ4 in the S/C/L band: dispersion tolerance

0

50

100

150

200

250

300

1460 1510 1560 1610

Dis

pe

rsio

n (

ps/

nm

)

G.652 spectrum (nm)

Fiber dispersion and dispersion tolerance25 Gb/s, 10 km, 1 dB penalty

w/o dispersion compensation

Fiber dispersion, max abs val

duobinary tolerance, EML

NRZ tolerance, EML

Must design for worst case: NRZ detection

No dispersion compensation required for NRZ detection up to 10 km for wavelengths up to ~1560 nm

• 25G dispersion tolerance values based on simulations.

• Experimental data is welcome.

6

Cost-optimized 100G EPON will require an optical pre-amp in the OLT to avoid optical post-amps in the ONU

• EDFAs have lower noise figure than SOAs.

• An EDFA preamp *might* be required to avoid post-amps in 100G ONUs, for PR30

– There are EDFAs that support burst mode

• The need for EDFA might be even stronger for PR40

• EDFA is not an option if 100G wavelengths are in the O-band (Plans A, B, D, E)

100G OLT

λ1 Rx

λ2 Rx

λ3 Rx

λ4 Rx

λ1 Tx

λ2 Tx

λ3 Tx

λ4 Tx

λ d

em

ux

m

ux

Diplexer

-24 dBm

-26.5 dBm

R

OA

7

Choosing wavelengths for λ1 - λ4

No dispersion compensation required for

≤10 km

EDFA

1510 1520 1530 1540 1550 1560 1570 1580 1590 1600 1610 1620 1460 1470 1480 1490 1500

S-band C-band L-band

NG-PON2 US ITU-T

RF

1524-1560 nm Put US λs here

Sweet spot

10G EPON DS

short end of 10G EPON ONU filter guard band

No co-existence requirements

GPON/ GE-EPON DS

8

Wavelength plan for λ1 - λ4. Assume 800 GHz CS

Sweet spot

Pure 100G EPON deployments can take advantage of lower fiber loss at these wavelengths

1510 1520 1530 1540 1550 1560 1570 1580 1590 1600 1610 1620 1460 1470 1480 1490 1500

Can maximize DS/US distance for lowest cost filtering

All 800 GHz CS, conforming to industry norm Should be no wavelength drift issue

upstream downstream

9

Choosing wavelengths for λ0

1310 1320 1330 1340 1350 1360 nm 1260 1270 1280 1290 1300

Plan C2 down up

Plan C4 down up

up Plan C1 down

TDM co-existence with 10G EPON

WDM co-existence with 10G EPON

Plan C5 down up

Plan C allows for increased flexibility to optimize 25G EPON

– Cost-optimization

– TDM or WDM co-existence with 10G EPON

– Co-existence with gen 1 PONs

The only option for Plan B

Low downstream DML TDP

uncooled DML in ONU

up Plan C3 down EPON US (DFB); GPON

25G co-existence with GE EPON (DFB) and GPON*

OLT DML still an option?

*may support CTC objective: “Promote IEEE and ITU-T coordination” including “Shared Unified wavelength plan” zhang_3ca_1b_0716.pdf

10

Dispersion compensation

11

If >10 km PONs must be supported, DCF can be deployed in 1+4 (not in 1+3 if DS/US λ0 is in the O-band)

Should be able to support any differential distance.

Characteristics

– Optical loss: e.g. spec: 1.2 dB (10 km)

– Size: e.g. module 9.2”x9.6”x0.75”

– Cost: low, relative to the cost of the 100G OLT transceiver

Dispersion compensation fiber (DCF) for PONs > 10 km

12

DCF deployment

50G/100G TxR 50G/100G

ONU

50G/100G TxR 50G/100G

ONU

50G/100G TxR 50G/100G

ONU DCF

OLT

0 km 10 km

PONs ≤10 km don’t require DCFs

• A DCF is installed at the OLT only for PONs >10 km

• The cost and space impact of DCFs is proportional to the % of PONs of >10 km reach. For most (all?) operators this will be small.

• Or, just create a 10 km spec and make the issue go away (e.g. miguelez_3ca_1a_0516.pdf)

13

Power budget

14 Public

Optical assumptions

• 100G mux and demux nominally add ~2.5 dB insertion loss each.

• Cost-optimized 100G EPON puts all optical amplification in the OLT

• Downstream TDP = 1.5 dB, upstream TDP = 3 dB (same as 10G EPON, but to be confirmed)

• PR30 loss budget

• There will be 1 dB FEC coding improvement in downstream relative to 10G EPON (to be confirmed)

• 1 dB improvement in PR30 APD receiver performance vs. 10G EPON + 5 dB penalty for 25G (per NeoPhotonics yield analysis, in harstead_3ca_1a_0516.pdf.)

• Launch power values from harstead_3ca_2a_0716.pdf

AVPmin (dBm) ER (dB)

EML 5 8

cooled DML 8 6

15

Downstream optical levels (per harstead_3ca_1_0916).

25G λ1 Tx ODN 25G λ1 Rx

25G ONU

100G OLT

S R

λ1 Rx

λ2 Rx

λ3 Rx

λ4 Rx

λ1 Tx

λ2 Tx

λ3 Tx

λ4 Tx

λ d

em

ux

m

ux

Diplexer

λ1 Rx

λ2 Rx

λ3 Rx

λ4 Rx

λ1 Tx

λ2 Tx

λ3 Tx

λ4 Tx

λ d

em

ux

m

ux

Diplexer ODN

25G OLT

100G ONU

-24.5 dBm +5 dBm

-24.5 dBm

-22 dBm +7.5 dBm

Assume EMLs

S R SOA Common OA post amp shown. Per channel OAs can also be considered.

16

Upstream optical levels (per harstead_3ca_1_0916).

25G λ1 Rx ODN 25G λ1 Tx

25G ONU

100G OLT

R S

λ1 Rx

λ2 Rx

λ3 Rx

λ4 Rx

λ1 Tx

λ2 Tx

λ3 Tx

λ4 Tx

λ d

em

ux

m

ux

Diplexer

λ1 Rx

λ2 Rx

λ3 Rx

λ4 Rx

λ1 Tx

λ2 Tx

λ3 Tx

λ4 Tx

λ d

em

ux

m

ux

Diplexer ODN

25G OLT

100G ONU

-24 dBm

Assume DMLs

+8 dBm

-24 dBm +8 dBm

+5.5 dBm -26.5 dBm

R S

OA

17

18

25 Gb/s dispersion tolerance (simulated)

Noise contributions, laser linewidth

and non-linearities were not taken into

account in the simulation.

Duobinary:

19

10G PON ONU blocking filter

20

If GE-EPON and GPON co-existence were required for 100G EPON

1450(I) 1441(I)

(I) = Informative

1415/1400(I) 1530 1539 1580-1625

22 (I)

7 (I)

ONU tolerance to interferers would constrain NG-EPON downstream wavelengths to ≥ 1539 nm.