Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX olarization-Multiplexed System Outage due to Nonlinearity- Induced Depolarization Marcus Winter,

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX

olarization-Multiplexed System Outage due to Nonlinearity-

Induced Depolarization

Marcus Winter, Dimitar Kroushkov, and Klaus Petermann

ECOC MMX / Th.10.E.3

P

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 2

inter-channel nonlinear polarization effects

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 3

typical DWDM system with a nonlinearity probe

► CW probe is unaffected by linear effects / SPM ► other channels are 10 Gbps OOK in 50 GHz grid

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 4

SOP evolutionback-to-back (fully polarized)

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 5

SOP evolution(without amplifier noise)

significant nonlinear depolarizationrapid (symbol-to-symbol) fluctuations of the SOP

can we model this and can this become a problem?

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 6

cross-polarization modulation (XPolM)► basics

► statistical models

► XPolM and polarization multiplex

► open questions

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 7

XPolM basics

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 8

XPolM is closely related to XPM

nonlinear variation of therefractive index refractive index difference

proportional to sum of interfering channels‘powers

(polarization-dependent)Stokes vectors

results in the modulation of signalphase polarization

(phase difference)

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 9

nonlinear polarization effects known since at least 1969 ► e.g. Kerr shutter (Duguay and Hansen, APL, pp. 192+, 1969)

XPolM first described in its „current version“ in 1995 ► Stokes space Manakov equation ► collision of two solitons ► Mollenauer et al., Optics Letters, pp. 2060+, 1995

many-channel formulation in 2006 ► Menyuk and Marks, JLT, pp. 2806+, 2006 ► Karlsson and Sunnerud, JLT, pp. 4127+, 2006

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 10

Poincaré sphereprobe channelDWDM interferersStokes vector sumnonlinear rotation

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 11

statistical models

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 12

(interferer) Stokes vectors are not constant

► length (intensity) varies due to walk-off► (interferer and probe group velocity differs)

► direction (SOP) varies due to PMD ► (interferer and probe birefringence differs)

► both effects are random

various models have been proposed to describe this behavior

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 13

► carousel model (Bononi et al., JLT, pp. 1903+, 2003)

► pump and probe rotate when both carry a mark two-channel system, no PMD►

► diffusion model (Winter et al., JLT, pp. 3739+, 2009)

► SOPs evolve as random walk ensemble mean values only►

► Karlsson‘s statistical model (JLT, pp. 4127+, 2006)

► influence on PMD compensation mostly two-channel system, no PMD dependence►

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 14

SOP distribution resembles diffusion

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 15

DWDM power/channel threshold for mean probe DOP=0.97

► resonant dispersion map, 10 × 10 Gbps OOK interferers► @ 50 GHz spacing

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 16

depolarization of probe vs. number of 3 dBm interferers

► difficult to simulate, expensive to measure► saturates at about 20

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 17

XPolM and polarization multiplex

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 18

► selective upgrade: 10G NRZ » 100G PolDM RZ-QPSK ► fits into 50 GHz grid

a typical PolDM system

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 19

polarization DEMUX must be aligned to PolDM subchannels(visualization in Jones space)

► otherwise crosstalk occurs from x to y and vice versa

► crosstalk increases with misalignment angle and with► length of field vector

detected field at y-Rx:

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 20

modern coherent receivers can handle subchannel SOP changes with PMD time constants

► DCF abuse with a screwdriver: 280 µrad/ns(Krummrich and Kotten, OFC 2004, FI3)

XPolM causes symbol-to-symbol fluctuations around mean SOP

► cannot be (fully) compensated (again like XPM)

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 21

interleaving RZ-shaped symbols minimizes crosstalk generation

time

field

am

plitu

de a

t y-R

x

aligned subchannels

interleaved subchannels

► crosstalk is never zero because pulses at Rx are no longer RZ (accumulated GVD, PMD, noise, filtering)► synchonized sampling necessary at Rx

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 22

10 × 10G NRZ interferers w/ 100G PolDM-RZ-QPSK probe

► 256 ps/nm RDPS, 10 interferers, SSMF, no PMD ► power/channel threshold is reduced by up to 2 dB

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 23

the statistical ensemble (mean DOP = 0.975)

► DOPs and ROSNRs spread over large range ► for DOPs < 0.98 (0.97), ROSNR penalties become significant

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 24

open problems

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 25

outage probability ► progress made when defining outage via DOP

► (see proceedings paper, figure shows 10-5 probability) ► outage via ROSNR much more interesting / relevant

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 26

influence of CMA (polarization demux) window length ► SOP correlation over several symbols possible ► systems with little or no inline GVD compensation ► correlation can be used to reduce crosstalk ► fast algorithms needed

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 27

dispersion map / interferer modulation format ► especially uncompensated spans ► GVD pulse distortion no longer negligible ► PSK formats no longer ideal

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 28

regimes of dominance ► SPM vs. XPM vs. XPolM ► depends on many factors

► (dispersion map, modulation format, PMD, GVD, …) ► first results by Bononi

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 29

summary

Winter et al.: XPolM in PolDM Systems, Th.10.E.3, ECOC MMX 30

► XPolM in DWDM systems causes depolarization

► diffusion model correctly predicts simulated behavior► in many situations

► depolarization creates detrimental PolDM crosstalk

slides available at http://www.marcuswinter.de/research/ECOC2010

► there are still many open questions about XPolM