professor gary m. carter dr. anthony lenihan and professor ... · dr. anthony lenihan and professor...

1

High Speed Optical NetworksHigh Speed Optical Networks----ExperimentsExperiments

Professor Gary M. Carter Dr. Anthony Lenihan

and Professor Curtis Menyuk

University of Maryland Baltimore County

UMIACS/LTS Review 6/18/04

2

Collaborators

Robert RunserTelcordia, LTS

L. MercerNaval Research Laboratory

T. Gibbons, P. SchulzMIT Lincoln Laboratory

3

Goals—Past year(s)

• Carry out experiments on installed optical communication systems

• To identify and investigate issues of interest in helping establish high speed operation of optical networks

4

This year’s effort

• We have shown in laboratory experiments that dispersion distribution is important for communications performance

• We moved these experiments from the lab to ATDNET/BossNet

• Implications for network management • Investigated novel formats that can increase

performance and relax constraints

5

Nonlinear Interaction

• Dispersion causes adjacent pulses to spread in time and overlap for part of the transmission distance

• The nonlinear interaction due to the optical fiber between adjacent pulses leads to timing jitter

• The dispersion distribution can alter the amount of timing jitter

• Large amounts of timing jitter causes errors

6

Nonlinear Interaction—Frequency Shift

TL uuu +=

tu

zf TL

dd

dd

2

γ−=

FWHM = 100 ps

0

0.01

0.02

600 800 1000

Time (ps)

Pow

er (m

W)

Lu

LuTuIntensity variation of shifts frequency of

Tu

7

Residual compensation all at the end

8

Nearly closed eye and large error rate due to timing jitter

10 Gb/s

9

Split compensation between beginning and end

10

Open eye end low bit-error rate

10 Gb/s

11

Laboratory Results

• Intra-channel cross-phase modulation → timing jitter

• Dependent on dispersion map configuration– Most work to date on undersea transmission systems– Mitigated by symmetric pre- and post-compensation

• Laboratory investigation of terrestrial system– Xu, et al., IEEE Phot. Tech. Lett. 16, 314 (2004)

12

Experiments on ATDNET/BossNet

• Measure timing jitter at 10 Gb/s on an installed terrestrial fiber path

• Obtain excellent qualitative agreement between measured values and numerical simulations

• Arrangement of dispersion compensation important for obtaining best transmission

13

Installed fiber path

Segment 1 (ATDNet):

Segment 3 (BOSSNet):

Washington, D.C.

• TrueWave RS (+4.5 ps/nm-km), 41 km• Uncompensated

• TrueWave Classic (+2.8 ps/nm-km)– To Wilmington: 186 km– To New York City: 389 km

• Compensated at 1550 nm• Uneven EDFA spacing

Segment 2• AllWave (+17 ps/nm-km), 16 km• Uncompensated

One way distances:

=Tx/Rx

Wilmington, DE

New York, NY

14

Experimental setup

InstallednetworkCW Laser

@1558.98nm EAM EOM

DC

NYC

BERT

Oscilloscope

Clock/DataRecovery

OBF

10 GHz Clock

10 Gb/sPRBS

Pre-Compensation

Post-Compensation

OSA

Rx:

Tx:

• Tx/Rx co-located in D.C.

EDFA

15

Dispersion mapsLoopback at Wilmington, DE Loopback at New York, NY

• Not including pre- and post-compensation• Total accumulated dispersion for round trip at 1558.98 nm

= EDFA

Accu

mul

ated

Dis

pers

ion

(ps/

nm)

Distance (km) Distance (km)

0

800

1600

0 6000

900

1800

0 1000

16

Numerical model

• Full nonlinear Schrödinger equation• 200 Monte Carlo simulations• Utilizing best estimate of system parameters:

– Dispersion map– Amplifier noise figures– Power map

Timing jitter results compared with numerical simulation:

17

Eye diagramsW

ilmin

gton

, DE

New

Yor

k, N

Y

50 % Pre-compensation 90 % Pre-compensation

18

Wilmington loopback results

• Total transmission distance ~520 km• Symmetric dispersion compensation minimizes timing jitter• Excellent qualitative agreement with theory

0.5

1.5

2.5

3.5

0 25 50 75 100

Pre-Compensation (%)

Tim

ing

Jitte

r (ps

)

ExperimentTheory

7.0

7.5

8.0

8.5

9.0

0 25 50 75 100

Pre-Compensation (%)Q

Fact

or

Experiment

19

New York loopback results

• Total transmission distance ~930 km• Jitter increased, but similar trend

1.5

2.0

2.5

3.0

3.5

4.0

4.5

0 25 50 75 100

Pre-Compensation (%)

Tim

ing

Jitte

r (ps

)

ExperimentTheory

6.0

6.5

7.0

7.5

8.0

8.5

9.0

0 25 50 75 100

Pre-Compensation (%)Q

Fact

or

Experiment

20

Summary

• Timing jitter measured on installed terrestrial link

• Qualitative agreement with simulations

• Ability to control dispersion compensation layout will be important in typical terrestrial systems may influence network management when paths change

21

Novel Formats

• Explore Novel formats • DPSK• QPSK• QPSK and PolMux• Others

22

DPSK

• Phase modulated signal• 3 dB advantage over ASK• Constant amplitude pulses• Less “patterning”• Security implications

23

DPSK Transmitted Intensity Pattern

10 Gb/s

24

Receiver

25

Demodulated Eye Pattern—10Gb/s

26

DPSK Loop Back from LPS to ISIError Rate < 1 10−13

27

DPSK

• 3 dB performance advantage over ASK• Less Pattern Dependence in Transmission—

maybe smaller timing jitter• More difficult to detect—need specialized

receiver• Provides different information to network

management

28

New Directions for Upcoming Year

• We are working with LTS to adapt measurements and theory to interactions with system and optical control plane

• LTS Personnel: Mark Ciccarello , Dave Hardesty, and Walter Kaechele