Title: Investigation of the combined effect of polarizationmode dispersion and polarizationdependent loss o
1Investigation of the combined effect of
polarization-mode dispersion and
polarization-dependent loss on system performance
- Hua Jiao
- Ph.D. dissertation defense
- Department of Computer Science and Electrical
Engineering, University of Maryland Baltimore
County - Apr. 13 , 2007
2Acknowledgements
3Outline
- Motivation
- Introduction to PMD and PDL
- Literature review on receivers
- Our receiver model
- Validation of the receiver model
- Conclusions
41. Motivation
- Polarization effects (PMD and PDL) are some of
the major effects that cause impairments in high
bit rate optical fiber communication systems. - Polarization effects are random and may vary over
time scales of milli-seconds to hours. - The combined effects of PMD and PDL on system
performance can be very complicated. - It is important for system designers to
understand how they interact together to affect
system performance.
5Tools required and available
- Tools required
- The transmission modeling (available)
- The receiver model for a depolarized signal and
- partially polarized noise
- Available Receiver models
- Polarized signal and partially polarized noise
- e.g. systems with low bit-rate and long distance
- Depolarized signal and unpolarized noise
- e.g. systems with high bit-rate and short distance
6Goal of the dissertation
-
- Develop a receiver model for a depolarized
- signal and partially polarized noise
- Validate the receiver model
72. Introduction Effect of PMD on signal
1. Equal power in two PSP
2. Unequal power in two PSP
PSP
PSP
PSP
PSP
8Effect of PDL
Electric field of noise at input
High loss axis
Electric field of noise at output
Low loss axis
out
in
PDL Device
DOP of the noise
93. Literature Review on receivers
- Marcuse 1, Humblet and Azizoglu 2
Rectangular optical - filter, square-law photodetector and
integrate-and-dump electrical filter - Winzer et. al. 3 Format comparisons, optimum
receiver bandwidth and receiver sensitivity - Rebola and Cartaxo 4 Power penalty due to
optical filtering, optical filter detuning,
extinction ratio and eye closure - Assumptions in their work
- Signal is polarized (no PMD effect)
- Noise is unpolarized or polarized and parallel
with the signal
10Literature Review (cont)
- Yu Sun et. al.5 (Theory and experiments)
- Polarized signal and partially polarized noise
- Ivan Lima et. al. 6 (Theory and simulations)
- Polarized signal and partially polarized noise
- Depolarized signal and unpolarized noise
- In this dissertation
- Signal is depolarized
- Noise is partially polarized
- Model validated by simulations and experiments
114. Our receiver model
- System and receiver structure
- System performance measure
- The formula for the variance of the received
signal due - to signal-noise beating
12System and receiver structure
13System performance measure
- Q-factors obtained from the mean and variance of
the received signal
R Responsivity of the photodetector ho, he
Impulse response of the optical and electrical
filter NASE Total power spectral density of the
noise Bo Noise equivalent bandwidth of the
optical filter
14Theory on the receiver model
- Signal pulse distortion due to PMD
Principal states of polarization in Jones space
t Differential group delay (First order
PMD) u(t) Electric field of the signal from the
transmitter
Power splitting ratios (
)
Principal state of polarization in Stokes space
(Slow mode)
Principal state of polarization in Stokes space
(Fast mode)
A complex vector
(i0, 1, 2, 3) are Pauli spin matrices
15Theory (cont)
The new formula for the variance due to
signal-noise beating
Stokes vector of the polarized part of the
noise
16Variables in the new formula
- Constants R, NASE (OSNR 15 dB), ? 36 ps,
and S - Variables
- Power-splitting-ratio
- DOP of the noise
- Angle between the SOP of the polarized part of
the noise and the signal
175. Validation of the receiver model
- How to vary the angle between the signal and the
noise
(b) Parallel case, (c2 0.5)
(a) Orthogonal case, (c2 0.5)
z
z
?
?
a
a
y
y
x
x
Su Average Stokes vector of the signal
18Monte Carlo simulations
- Orthogonal case, (c2 0.5)
Symbols simulations Lines From formula
19Monte Carlo simulations (cont)
(b) Parallel case, (c2 0.5)
? 0
? p
? p/4
? 5p/8
Symbols simulations Lines From formula
20Monte Carlo simulations (cont)
- Orthogonal case, (c2 0.25)
Symbols simulations Lines From formula
21Monte Carlo simulations (cont)
(b) Parallel case, (c2 0.25)
? p/4 ---- ? 5p/8 - - ? p ? ?
p/4 ? ? 5p/8 ? ? p
Symbols simulations Lines From formula
22Summary of Monte Carlo simulations
- The agreement between theory and Monte
- Carlo simulations is excellent
- Two power-splitting-ratios
- Two schemes of varying the SOP of the noise
- A few angles for the average SOP of the signal
-
23Experimental validations
- Issues in the experiments
- Drift of the SOP of the signal
- Measurement uncertainties
- Pulse width due to thermal drift of optical
modulator (12ps) - OSNR due to power fluctuations (up to 0.2 dB)
- Electrical bandwidth (a few GHz)
- Experimental results
- Sensitivity of the performance to the
measurement - uncertainties
24Experimental setup
Transmitter
PM Fiber
OSA
Tunable laser
EOM
EAM
Polarization controller
10 Gb/s
10 GHz
Receiver
Partially polarized noise
BERT
ASE
Polarimeter
Polarizer
25Q-factor in the experiment
Q-factor is estimated from the BER margin
measurement
102
Valid BER
109
Minimum BER
Optimum threshold
26Experimental cases
- Vary the angle between the SOP of the signal
- and the noise in the following cases
- Case 1 c2 0, DOPn 0.5 and 1
- Case 2 c2 0.5, DOPn 0.5 and 1
- Case 3 c2 0.3, DOPn 0.5 and 1
27Case 1 (c2 0)
- Signal is launched along the PSP
- SOP of the polarized part of the noise is varied
z
PSP
a
Su
y
x
28Experimental results (case 1)
(a) DOPn 0.5
(b) DOPn 1
29Case 2 (c2 0.5)
- SOP of the polarized part of the noise is fixed
- SOP of the signal drifts on the blue circle
z
Su
a
PSP
y
x
30Experimental results (case 2)
(a) DOPn 0.5
(b) DOPn 1
31Case 3 (c2 0.3)
- SOP of the polarized part of the noise is fixed
- SOP of the signal drifts on the blue circle
z
Su
a
PSP
y
x
32Experimental results (case 3)
(a) DOPn 0.5
(b) DOPn 1
33Inherent measurement uncertainty of the Q-factor
- Back-to-back measurement (no transmission fiber
between the - transmitter and the receiver)
- Polarized signal and unpolarized noise
?Q 0.3
34Sensitivity to pulse width
?Q 0.4
35Sensitivity to OSNR
?Q 0.3
36Sensitivity to electrical filter
?Q 0.1
?Q 0.1
376. Conclusions
- We developed a new receiver model that can
calculate the - system performance of a depolarized signal and
partially - polarized noise
- We validated the new model through Monte Carlo
simulations - and experiments.
- Performance sensitivities to measurement
uncertainties were - investigated.
38 39References
- D. Marcuse, IEEE J. Lightwave Technol., vol. 8,
pp. 18161823, 1990. - P. A. Humblet and M. Azizoglu, IEEE J. Lightwave
Technol., vol. 9, pp. 15761582, 1991. - P. J. Winzer et al., IEEE J. Lightwave Technol.,
vol. 19, pp. 12631273, 2001. - J. L. Rebola and A. V. T. Cartaxo, IEEE J.
Lightwave Technol., vol. 20, pp. 401408, 2002. - Yu Sun et al., IEEE Photon. Technol. Lett., vol.
15, pp. 16481650, 2003. - I. T. Lima Jr. et al., IEEE J. Lightwave
Technol., vol. 23, pp. 14781490, 2005.
40Jones space
- Jones vector a two-element complex vector, with
elements - represented by the magnitude and phase of the
electric - field of the signal
- Jones matrix a complex 2?2 matrix for a device
- eout J ein.
- Simple in theory, not easy to measure in
experiments
41Stokes space
- Stokes vector a three-element real vector, with
elements - represented by the powers of the signal
- S1 Power through linear horizontal polarizer
(LHP) - S2 Power difference between light transmitted
through a linear 45o polarizer (L 45o) and
through a linear 45o polarizer (L 45o) - S3 Power difference between light transmitted
through a right circular polarizer (RCP) and a
left circular polarizer (LCP)
42Signal-noise beating for a depolarized signal and
partially polarized noise
(0,0,1)
(0,0, 1)
Average SOP of signal (1, 0,
0) Principal states (0, 0, 1) (0,
0, 1)
(1, 0, 0)
(1, 0, 0)
43Sensitivity to optical filter
44Poincaré sphere
Angle between two vectors in Stokes space is two
times of that in Jones space!
45Principal State of Polarization (PSP)
- Two orthogonal polarization states of the fiber
- Pulse un-distorted if launched in PSP
- Well defined in PM fiber
46Assumptions in our receiver model
- We ignore the transmission effects of chromatic
dispersion - and nonlinearity (can be included)
- We assume first order PMD dominate the penalties
induced - PMD.
- We ignore the PMD effect on the noise and the
PDL effect - on the signal (can be included in the model).
47Drift of the SOP of the signal
- Due to short beat length in PM fiber
Su
PSP
48Drift of the SOP of the signal
- Drift was effectively reduced by putting PM
fiber in water
Su
PSP
49Future work
- Limitation of the new model to second-order PMD
- To determine the effect of second-order of PMD
to the model. - Study to be based on Monte Carlo simulations