Vadim Winebrand

Faculty of Exact Sciences

School of Physics and Astronomy

Tel-Aviv University

Research was performed under a supervision of Prof. Mark Shtaif

Outline

• Design of long haul fiber optic communication systems• Signal propagation in the optical fiber• Introduction to polarization effects in the systems• Emulation with help of optical recirculating loop• Simulations vs. Experiments• Measurements performed to show

Polarizations/Nonliniarities interactions• Fiber optic DPSK systems

DCM DCM DCM DCM

Introduction to WDM long haul fiber optic communication systems

TX

TX

TX

TX

...

MUX

RX

RX

RX

RX

...

MUX

LossDispersionPolarizationNon-liniarities

Noise

Degrees of freedom

• Transmitted waveform (modulation format)

• Optical power

• Dispersion management

Loss management

5

Q f

acto

r dB

Input power dBm

The Q factor grows linearly with input power

ASE domina

ted

But non-linear effects become significant

Non-linear dominated

1

2 2

20log10( )dB

QBER erfc

Q Q

System design – Loss management

6

For given average optical power

OS

NR

dB

Number of amplifiers

Acc

dis

pers

ion

(ps/

nm)

Length (km)

Dispersion management

Acc

dis

pers

ion

(ps/

nm)

Length (km)

Under-compensation

Over-compensation

Acc

dis

pers

ion

(ps/

nm)

Length (km)

Exact-compensation

Propagation in optical fibers

22

2 22 2

A i Ai A A A

z T

A is envelope of the signal

Dispersion of the signal

non-linear interaction

Loss of the signal

Non linear Schrödinger equation NLSE

NLSE Dynamics

Characteristic length-scales

0

20

2

1NL

D

LP

TL

Nonlinear length

Dispersion length

Non-linear effect self phase modulation (SPM)

With negligible dispersion

2( , ) exp( ) ( ,0)effA T L i A T L A T

SPM

• SPM induces chirp on the signal ( )d

tdt

NL DL L

Group velocity dispersion(GVD)

When neglecting non-linearities

22( , ) exp( ) ( ,0)

2

iA L L A

• GVD induces chirp as the pulse propagates

D NLL L

Dispersion

• When both Non-liniarities and Dispersion are present things cannot be described analytically.

• They get complicated….

Combined Effect of SPM and GVD

WDM system considerations – Four wave mixing

Each 3 frequencies generate 4thijk i j k

Pow

er

Spectrum

1

FWM noise

14

WDM system considerations – cross phase modulation(XPM)

Phase of the signal depends on neighboring channels

expM

j j eff j mm j

A P i L P P

SPM XPM

15

WDM system considerations – cross phase modulation (XPM)

XPM causes timing jitter and power fluctuations

16

WDM system considerations – Raman crosstalk

Pow

er

Spectrum

It depletes higher frequenciesAmplifies lower ones

17

WDM system considerations – Raman crosstalk

It depletes higher frequenciesAmplifies lower ones

It causes power fluctuations

Brillouin scattering

• The power is scattered back once the Brillouin threshold is passed

• Negligible in communication systems

18

Pow

er

Spectrum

Brillouin threshold

CW case

Modulated signalcase

Polarization and Nonlinearity• In most of the existing literature – these two

phenomena are separated.

• In the new generation of high-data-rate terrestrial systems this neglect is no longer possible.

• One of the goals of this work was to demonstrate and characterize polarization effects in long nonlinear systems.

Lack of cylindrical symmetry in fibers

Polarization Mode dispersion (PMD)Polarization dependent loss (PDL)

The outcome:

Polarization effects

=

To 1st order in bandwidth

Position dependent birefringence - PMD

NLSE with PMDIn each segment the Coupled Nonlinear

Schrödinger Equations (CNLSE) are solved:

22 2

2

22 2

2

1 2

2 3

1 2

2 3where:

- signals along the two PSPs

- group velocity difference between PSPs

u u ui i u u v uz t t

v v vi i v v u vz t t

u,v

Penalties of PMD/Non linear interactions

• Penalties are shown with cumulative Q distribution

Optical recirculating loop scheme

Eigen

75km SMF

75km SMF

75km SMF

50/50

Eigen

CW

Scope

CK I/PData I/PError

detector

FilterAmpPre-Filter

AmpPost

modulator

Amp3

DCM

Eigen

DCM

Amp2

Amp4

Wide bandfilter

80/20

OSA

80% 20%

CDR

Amp1

Modulator

PC

Amp Bias

Pulse carver

PC

Amp Bias50% RZ pulse

DCM

Eigen

PC

1x2switch

90/10

10%

90%

Measurement methods – Bit error rate

BER = p(1)p(0/1)+p(0)p(1/0)

1

2 2

20log10( )dB

QBER erfc

Q Q

PD

F

Voltage

V0 V1

Measurement methods – eye diagram

Eye-diagram is a bit chain that is folded to a single bit slot

Measurement methods-optical spectrum

Power spectral density provides significant information

Pow

er d

B

Spectrum

Signal power

Noise level

OSNR

Bandwidth

Simulations vs. Experiments

Criterions for comparisons• Bandwidth evolution

• Optical spectrum

• Eye-diagram - difficult.

• Q factor – difficult.

Comparisons results

0 1000 2000 3000 4000 5000

5.5

6

6.5

7

7.5

8

8.5

x 109 bandwidth vs length

Length(km)

Ba

nd

wid

th

SimulationExpirimental

0 1000 2000 3000 4000 5000

6.4

6.6

6.8

7

7.2

7.4

x 109

Length(km)

Ba

nd

wid

th(H

z)

Bandwidth vs length

SimulationExpirimental

1000 2000 3000 4000 5000

6.6

6.8

7

7.2

7.4

7.6

x 109

length(km)

Ba

nd

wid

th (

Hz)

bandwidth vs length

numericalExpirimental

Comparison between theoretical and experimental spectrums

2dBm power and no precompensations 2dBm power and -precompensator of 290ps/nm

3dBm power and -precompensator of 290ps/nm

30

PMD/Non linear measurements – Idea

• Changes in dispersion map will worsen effects of PMD

• But will not affect average Q factor

PMD/Non linear interactions–experimental setup to measure penalties

• The Q statistics was gathered• The Idea is to find that small change in dispersion map increases penalties

Difficulties measuring Q penalty of non-linear PMD

• Periodic PDL & EDFA amplifiers causes BER fluctuations

32

• Periodicity does not allow true PMD measurement

• Requires high accuracy in measuring BER

PMD&PDL states in the recirculating loop are constant

PMD states in the real system are random, but in the recirculating loop they are periodic

33

Real system case

Recirculating loop case

Periodic PDL in the recirculating loop

34

Different states of polarizations lead to different OSNR levels

Orthogonal noise is attenuated – increasing OSNR

PDL element

Orthogonal signal is attenuated – decreasing OSNR

PDL element

Periodic amplifiers in the recirculating loop

35

PDL causes gain fluctuations

PDL element

Amplifiers experience polarization dependent gain

Amplifiers are calibrated for the first cycle only

36

Solution (?) - Polarization scrambler - at the transmitter

• Polarization scrambler makes polarized light to un-polarized

• Effects of PDL are averaged out –but effects of PMD are unchanged

• OSNR variations transformed to amplitude jitter

• Gain and noise levels of the amplifiers are more stable

Eye diagram at 1e-8 Eye diagram at 1e-5

Solution (?) - Loop synchronous polarization controller

• Changes input polarization to a random state

• Break periodicity of the PMD and PDL states

• Problems with LSPC

• Does not break periodicity of the amplifiers and PDG

DPSK - introduction• The data is stored in the phase of adjacent bits.• Reception is performed with delay interferometer

Modulation scheme of the signal Scheme of the reception system

OOKDPSK

MZDI Balanced receiver

Re{E}

Im{E}

Re{E}

Im{E}

DPSK – transmitter Transmitter experimental setup

Eye diagram at the output

Scheme of the DPSK modulator Lasermodulat

orCarver DCA

Bit stream

Sinusoidal signalRe{E}

Im{E}

Requires additional bandwidth

DPSK reception system

MZDI 11 cos 2

2out inI I f T

Frequency response of the interferometer

Problems

• Exact one bit delay• Phase mismatch• Polarization match• Controllable environment

DPSK – combining all the system together

Output

OOK vs. DPSK

Lasermodulat

orCarver MZDI

Bit stream

Sinusoidal signal

DCA

Many thanks to Prof. Mark Shtaif

Many thanks for Prof. Moshe Tur

Many thanks to Chen Rabiner and Efi Shahmon

Many thanks to all members of the laboratory