Evaluatio ion of Ela lasti tic Modulation Gain ins in in - - PowerPoint PPT Presentation

evaluatio ion of ela lasti tic modulation gain ins in in
SMART_READER_LITE
LIVE PREVIEW

Evaluatio ion of Ela lasti tic Modulation Gain ins in in - - PowerPoint PPT Presentation

Evaluatio ion of Ela lasti tic Modulation Gain ins in in Microsofts Optical Backbone in North Americ ica Monia Ghobadi Jamie Gaudette, Ratul Mahajan, Amar Phanishayee, Buddy Klinkers ( Microsoft ), Daniel Kilper ( University of Arizona)


slide-1
SLIDE 1

Evaluatio ion of Ela lasti tic Modulation Gain ins in in Microsoft’s Optical Backbone in North Americ ica

Monia Ghobadi Jamie Gaudette, Ratul Mahajan, Amar Phanishayee, Buddy Klinkers (Microsoft), Daniel Kilper (University of Arizona)

slide-2
SLIDE 2

Demand is increasing

Conventional wisdom to increase capacity

  • add more wavelengths
  • light more fiber

2

To support the exploding demand in the cloud, we need to efficiently use the deployed fiber.

slide-3
SLIDE 3

Data

3

Is there ability to carry more bits?

  • All fiber paths in Microsoft’s North America

backbone

  • Three-months (Feb-April 2015)
  • Poll signal quality (Q-factor) for 100Gbps PM-

QPSK line cards

  • 1000s of line cards
  • Segments length range: 5km - 2600km
  • Fiber type: LEAF, SSMF
  • 15-min bin samples: min, max, average
slide-4
SLIDE 4

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5 6 7 8 9 10 11 12 13 14 15 16 17

Cumulative Distribution Function

Signal-to-Noise Ratio (dB)

Higher order modulation

4

  • Performance of each channel
  • Margins for higher order

modulations

  • Convert Q-factor to SNR

100Gbps QPSK 150Gbps 8-QAM 200Gbps 16-QAM

slide-5
SLIDE 5

Higher order modulation

5

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5 6 7 8 9 10 11 12 13 14 15 16 17

Cumulative Distribution Function

Signal-to-Noise Ratio (dB) 100Gbps 150Gbps

99% 43%

200Gbps

100G 150G 200G

Gain: 70%

Using the same fiber paths, we get more bits

slide-6
SLIDE 6

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5 6 7 8 9 10 11 12 13 14 15 16 17

Cumulative Distribution Function

Signal-to-Noise Ratio (dB)

Higher order modulation

6

100G 100 G 150G 200G

Gain: 45-70%

Propagation penalty 2.2 dB Propagation penalty 1.5 dB

12%

100Gbps 150Gbps 200Gbps

78%

slide-7
SLIDE 7

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5 6 7 8 9 10 11 12 13 14 15 16 17

Cumulative Distribution Function

Signal-to-Noise Ratio (dB)

Higher order modulation

7

125Gbps 175Gbps 225Gbps 250Gbps

150G 175G 200G 225G

Gain: 86%

100G 100 G 150G 200G

Gain: 45-70%

100Gbps 150Gbps 200Gbps

slide-8
SLIDE 8

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5 6 7 8 9 10 11 12 13 14 15 16 17

Cumulative Distribution Function

Signal-to-Noise Ratio (dB)

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 5 6 7 8 9 10 11 12 13 14 15 16 17

Cumulative Distribution Function

Signal-to-Noise Ratio (dB)

Higher order modulation

8

Gain: 99%

150G 175G 200G 225G

Gain: 86%

100G

Gain: 45-70%

125Gbps 175Gbps 225Gbps 250Gbps 100Gbps 150Gbps 200Gbps

100G 150G 200G

slide-9
SLIDE 9

How to deploy higher order modulations

  • Should we use the same modulation for all segments?
  • Different segments have different SNRs.
  • Should we use the same modulation for all wavelengths in a segment?
  • Different wavelengths have different SNRs.
  • Should the modulation for a wavelength be static?
  • SNR varies over time.

9

slide-10
SLIDE 10

How to deploy higher order modulations

  • Should we use the same modulation for all segments?
  • Different segments have different SNRs.
  • Should we use the same modulation for all wavelengths in a segment?
  • Different wavelengths have different SNRs.
  • Should the modulation for a wavelength be static?
  • SNR varies over time.

10

slide-11
SLIDE 11

SNR variation across wavelengths

11

10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 1545 1550 1555 1560 1565 Signal-to-Noise Ratio (dB) Wavelength (nm) 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1 2 3 4 5 6 CDF over all segments (max – min) SNR (dB)

0.8 dB 3.8 dB

0.9 dB

Different wavelengths need different modulation formats even though the path is shared

slide-12
SLIDE 12

How to deploy higher order modulations

  • Should we use the same modulation for all segments?
  • Different segments have different SNRs.
  • Should we use the same modulation for all wavelengths in a segment?
  • Different wavelengths have different SNRs.
  • Should the modulation for a wavelength be static?
  • SNR varies over time.

12

slide-13
SLIDE 13

SNR variation over time

13

10.5 11 11.5 12 12.5 13 13.5 14 14.5

Signal-to-Noise Ratio (dB)

Time Jumps are due to:

  • Occasional network changes
  • Removal of legacy 10G OOK
  • Removal of old optical gear
  • Maintenance
slide-14
SLIDE 14

SNR variation over time

14

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

1 2 3 4 5 6 7 8 9 10

CDF over channels

SNR (dB)

Per channel max - min

10.5 11 11.5 12 12.5 13 13.5 14 14.5

Signal-to-Noise Ratio (dB)

Time

SNR changes over time, depending on changes in infrastructure

5.8 dB

slide-15
SLIDE 15

How to deploy higher order modulations

  • Should we use the same modulation for all segments?
  • Different segments have different SNRs.
  • Should we use the same modulation for all wavelengths in a segment?
  • Different wavelengths have different SNRs.
  • Should the modulation for a wavelength be static?
  • SNR varies over time.

15

Bandwidth Variable Transponders

slide-16
SLIDE 16

Conclusions

  • Existing fiber can support higher order modulation
  • Deployment should be realized using bandwidth variable transponders

16