Professional Documents
Culture Documents
Optical Communication
Course Outline
Introduction [4]
• Optical communication
• Basics of Optics, EM Theory
and Waveguide
• Bandwidth and Bit rate
• dB scale of measurement
Optical Fibers [6]
• Geometry
Step index
Graded index
• Numerical Aperture
• Modes as solutions of
Maxwell’s equations in glass
fiber medium
Wave propagation in fiber [2]
• Step index
• Graded index
Loss Mechanism [6]
• Attenuation through absorption
• Dispersion of light in a
waveguide
• Material dispersion
• Waveguide dispersion
• Loss due to bend in fibers
• Polarization mode dispersion
Fiber fabrication and cabling [3]
• Material
• Various techniques
• Cabling schemes
Optical Amplifiers [6]
• Types of OAs
Erbium Doped Fiber Amplifier
• Basic concepts of EDFAs
• Physics of signal amplification
• Modelling of EDFA
• Qualitative discussion of
Gain Spectrum & bandwidth
Gain Saturation
Amplifier Noise
Optical Receivers [4]
• Characteristics of receivers, viz.,
Responsivity, Rise time, bandwidth
• receivers suitable for optical
communication, photodetectors
• P-N, P-I-N and Avalanche
• Receiver Noise
• SNR calculation and optimization
• Quantum limit of receivers
• Receiver Performance
Optical Transmitter [4]
• Basics of laser theory and rate
equation
• LEDs
• Semiconductor Lasers
• Transmitter design
Design Consideration [4]
• Loss compensating
mechanisms
• Dispersion compensating
fibers
• System design and
performance issues
• Installation issues
Class Timing
Class Room
CEP105
Evaluation
Assignments
Quizzes
Bonus
Recommended Books
1. Fiber Optic Communication Systems
by G.P. Agrawal (Wiley)
2. Fiber Optic Communications by J.C.
Palais (Pearson Education)
3. Introduction to Fiber Optics by A.
Ghatak & K. Thyagarajan (Cambridge)
4. Optical Fiber Communications by Gerd
Keiser (MGH)
Recommended Books
5. Fiber Optic Communications
Technology by Mynbaev &
Scheiner (Pearson Education)
6. Fiber Optic Cable System
Installation by E. R. Pearson
(Delmar)
7. Optical Networks by R. Ramaswami
& K. N. Sivarajan (Elsevier)
Overview of previous year’s class
performance, impact of attendance etc.
9
8.43 Total: 66
8
7
Class Profile:
6.23 6.09 Backlog yr – 2,
6
Final yr – 51
5.1
5
4.8
5
3rd yr – 13
4
Final Grade:
3
3 AA: 4, AB: 5,
BB: 6, BC: 9,
2
CC: 11, CD: 10,
1 DD: 14, DE: 6,
0
F: 1
99-90 80-89 70-79 60-69 50-59 40-49 30-39
avg. grade
OVERALL CLASS PERFORMANCE
• Class Size = 66
• Highest mark obtained was 38.7 (out of 100)
• Class average: 19.9
• Lowest: 4.3
• Below class average: 37 students
• Only 9 students were above 30.
• Some students could get advantage of earning
some bonus marks during the semester.
• A not-so-direct but somewhat related correlation
was observed for performance (grade) vs.
regularity in class (attendance)
OBJECTIVE:
To transmit data at the highest capacity over longest
possible distance.
EARLY DAYS:
# mid-1960s – information encoded in light signals could
be transmitted over a glass fiber waveguide.
Evolution of optical communication era
1950s: it was realized that an overall increase of several
orders of magnitude in BL product would be possible, if
optical waves were used as carrier.
Bottleneck: Source and medium not available.
Ref.
http://www.corning.com/opticalfiber/innovation/fortyyear
soffiber/index.aspx
Law of Reflection
θi = θr
WHAT HAPPENS WHEN LIGHT GOES
FROM ONE MEDIUM INTO ANOTHER?
Snell’s Law
n1 sin θi = n2 sin θr n1
Refractive indices n2
of both media
influence θr
In this case light is coming
from rarer to denser
medium, n > 1.
n
Rarer medium
Denser medium
When a ray moves from a denser to rarer
medium
• it moves away from the normal. In other
words, it moves closer to the interface.
• It (refracted ray) comes closer and closer as
angle of incidence becomes larger and larger.
• at a critical angle (of incidence) it moves just
along the interface, hence θr = 900.
• For all θi > θc the ray will come back to the
original medium, hence manifesting the
phenomenon of total internal reflection.
Representation of the critical angle
and the total internal reflection at a
glass-air interface.
The phenomenon of guidance of light by
multiple total internal reflections was, in
fact, demonstrated by John Tyndall (1854).
But the real development started after the
works of Hopkins and Kapany (1950s; UK)
and of van Heel (Holland).
Ref.:
K.C. Kao and G.A. Hockham, Proc. IEEE, 113, 1151 (1966)
F.P. Kapron, D.B. Keck, and R.D. Mauer, Appl. Phys. Lett., 17,
423 (1970).
From the geometry of this figure, note that if we
have to make φ > φc, it implies that we have to
bring θr smaller. WHY?
In turn, θi will have to be further smaller. WHY?
2θi is called acceptance angle and a solid angle
of that is called acceptance cone.
Numerical Aperture
• Refraction at the air–glass interface: n0 sin θi = n1 sin
θr
[ ( )]
sin θr < 1 -
n2
n1
n1 – n2
D= n1
Typically for a multimode step-index silica fiber,
n1 ≈ 1.48, n2 ≈ 1.46, a ≈ 25 mm, b ≈ 62.5 mm
that gives
D= 0.0135
NA = 0.2431
V < 2.405
• Geometrical Optics
• Classical theory of Electromagnetic
Radiation
• Quantum theory
Geometrical Optics
• Wavelengths involved are too small compared
with the dimensions of equipment available for
their study
• Photon energies are small compared with the
energy sensitivity of the equipment
Classical theory of Electromagnetic Radiation
• Wavelengths are comparable to the dimensions
of equipment available (difficult to arrange with
visible light but easier with radiowaves)
• Photon energies are still negligibly small
Quantum theory
• Realm of very short wavelengths
• Wave nature of radiation does not hold good
• Photons have very large energy compared with the
energy sensitivity of the equipment
HOW DO WE SEE?
• Light goes in straight lines, if there is nothing in the
way
• The rays do not seem to interfere with one-other
• Light is made up of arrows which go through each
other!!!
Wavelength of light used in optical
communication typically ranges within 0.5 – 2.0
mm.
Core radius of a single-mode fiber ~ 2.5 – 3.0
mm.
80% of the optical fibers deployed in various
systems are single-mode fibers.
So, for all our analyses, we need to select
Classical Theory of Electromagnetic
Radiation to study light-matter interaction.
Within vast range of ELECTROMAGNETIC
spectrum, 3 major regions of approximation
which are interesting:
Geometrical Optics
Classical theory of Electromagnetic
Radiation
Quantum theory
Geometrical Optics
Wavelengths involved are too small compared
with the dimensions of equipment available for
their study
Photon energies are small compared with the
energy sensitivity of the equipment
Classical theory of Electromagnetic Radiation
Wavelengths are comparable to the dimensions
of equipment available (difficult to arrange with
visible light but easier with radiowaves)
Photon energies are still negligibly small
Quantum theory
Realm of very short wavelengths
Wave nature of radiation does not hold good
Photons have very large energy compared with the
energy sensitivity of the equipment
HOW DO WE SEE?
Light goes in straight lines, if there is nothing in the
way
The rays do not seem to interfere with one-other
Light is made up of arrows which go through each
other!!!
Wavelength of light used in optical
communication typically ranges within 0.5 – 2.0
mm.
Core radius of a single-mode fiber ~ 2.5 – 3.0
mm.
80% of the optical fibers deployed in various
systems are single-mode fibers.
So, for all our analyses, we need to select
Classical Theory of Electromagnetic
Radiation to study light-matter interaction.
Maxwell’s Equations in an optical fiber are
xE=-∂B/∂t
xH= ∂D/∂t
where
.D =0
. B = 0 ε0 is the vacuum
permittivity, μ0 is the
vacuum permeability,
Constitutive Relations and P and M are the
D = ε0 E + P induced electric and
magnetic polarizations,
B = μ0 H + M
respectively.
For optical fibers
M = 0 because of the non magnetic nature of
silica glass.
Evaluation of P requires Quantum mechanical
treatment.
Essential only when optical frequency is near a
medium resonance.
For optical fibers, we operate in the range l =
0.5 – 2 μm.
Far from resonance. Hence QM treatment is not
required for evaluation of P.
An optical mode refers to a specific
solution of this wave equation that satisfies
the appropriate boundary conditions and
has the property that its spatial distribution
does not change with propagation.
∂ 2Ez ∂ Ez ∂ 2
E ∂ 2
Ez
+ 1ρ 1 z 2 k 2 E = 0,
+ 2 + + n
∂ ρ2 ∂ρ ρ ∂ φ2 ∂z 2 0 z
where
{
n1; ρ ≤ a,
n=
n2; ρ > a.
• In fact, it is a set of six equations (three for E
and three for H) .
d2 F + 1 dF (n 2 k 2 – b 2 – m2 / ρ2 ) F = 0.
dρ 2 ρ dρ + 0
β and m are two constants (m must be an integer).
Z = exp(iβz),
Φ = exp(imφ ).
where p2 = n1k02 − β 2
and q2 = β 2 − n2 k02
where p2 = n1k02 − β 2
and q2 = β 2 − n2 k02
n 2
m2 1+1 1+ 2 1
= a2 p2 q2 p2 n12 q2
and
V = k0a√ (n12-n22)
Modes are quantified in terms of b(w) or b(V)
2 π an √ 2 D
V=
λ 1
V < 2.405
core
When Ez = 0, TE mode
When Hz = 0, TM mode
Two types of rays propagate in fiber
• Meridional rays (those pass through fiber axis)
• Skew rays (does not pass through axis)
Skew rays outnumber the meridional rays
cladding
Skew rays take
helical path. core
The cylindrical waveguide is bounded in two
dimensions within fiber, hence we require two
subscripts to identify each of them.
TEmn or TMmn modes correspond to
meridional rays.
Hybrid modes (where Ez and Hz are non-
zero) occur for skew rays.
These are designated by HEmn and EHmn
depending upon if components of H or E
contribute larger to transverse field.
Optical fibers used for communication have
very small D.
Usually less than 3% (0.03)
Meaning coupling angle, q, is too small.
This causes a dominant transverse field
component in guided modes.
Hence all HE, EH, TE, TM modes may be
represented by only two linearly polarized (LP)
modes.
Note that these LP modes are not the exact
modes of the fiber except the fundamental
mode, i.e., HE11 mode.
Basically these LPmn modes are superposed
degenerate EH, HE mode pairs. These
degenerate modes have almost similar
propagation constants.
LP modes Exact modes
LP01 HE11
LP02 HE12
Therefore
or
V = (2pa/l0) (NA)
Note that HE11 (or LP01) mode has no cut-off.
It ceases to exist only when the core dia is
zero, or V=0.
Therefore,
V ≤ 2.405 represents condition for a fiber to
be a single mode fiber.
The cut-off condition of various modes is
determined by V number.
A mode reaches cut-off when q = 0.
Total number of guided modes in a multimode
step index fiber may be given by
Ms ≈ V2/2
This is a good approximation for a fiber of
very large V number.
Prove the relation:
Ms ≈ V2/2
where all symbols have their usual meaning.
Solution:
sin qimax -> q for qimax -> 0.
Therefore the solid acceptance angle for this
fiber is
= pq2 = p (n12 – n22)
From the laser theory we know that for a
wavelength l, no. of modes per unit solid
angle is 2A/l2, where A is the area the mode
is leaving or entering.
core
Meridional Rays
• Skew rays (do not pass through the core axis)
Skew rays outnumber the meridional rays of the
total number of guided rays.
cladding
Skew Rays
core
Skew rays take
helical path.
Main reason of the loss at that time was
Pulse Broadening or Pulse Dispersion
We had multimode fibers.
Multimode fibers may be dealt with
geometrical optics, i.e., different rays take
different times to propagate through a given
length of the fiber.
Multimode fibers can be treated by
Geometrical/Ray Optics.
We say that a graded index fiber performs
better than a step index fiber in multimode
class of fibers.
WHY?
Intermodal (Pulse) Broadening
Different rays take different times to propagate through
a given length of the fiber.
For a step-index multimode fiber
qr
A C
= (L/sinφc – L)/v
= Ln1/c (1/sinφc – 1)
From the relation n1 cosφc = [n12 – n22] ½
Find out sinφc
That gives
Tf = Ln1/c
Ts = Ln1/(c sinφc) = Ln12/cn2
which yields
= Ts – Tf = Ln1/c (n1/n2 – 1) = (Ln12/cn2) D
n1 – n2
where, D = n1
In one of the modulation schemes
TB = 1/B
Therefore, < 1/B => B < 1
Substituting from the earlier equation gives
BL < [c/n1] [n2/(n1-n2)]
That leads to
BL < 4 x 108 bps-km = 0.4 Mbps-km.
Take another example of a typical fiber:
n1 = 1.5, n2 = 1.485,
[(n1 – n2) / n2] = 0.0101,
L = 1 km (assume), c = 3 x 108 m/sec
With BL < [c/n1] [n2/(n1-n2)]
Attenuation
Pulse Broadening or Pulse Dispersion
Material absorption:
1. Intrinsic absorption loss
2. Extrinsic absorption (absorption by
impurity ions) loss
Other sources of attenuation/loss are
–
Presence of OH ions also contributes to this
loss.
0.6 dB/km
Dry Fiber
marketed
under the
brand name
AllWave by
Lucent
The overall attenuative loss picture of fused
silica fiber
So, to get best result from optical
fiber as a medium for long-haul
communication system we need to
operate
near 1.55 mm, or at the most in the
range of 1.3 mm – 1.6 mm.
Intramodal (Pulse) Dispersion
Graded index fiber performs almost 1000 times
better than a step index fiber in case of multimode
operation.
Note that during 1st and 2nd generation of optical
communication systems multimode fibers were
deployed extensively because of the ease of
coupling of signal in the fiber.
But the best solution for eliminating intermodal
dispersion loss is to use SINGLE MODE FIBERS.
We still focus on the wavelength range
of 1.3 mm – 1.6 mm.
Intramodal Dispersion
n – n2 , 0<b<1
b=n –n
1 2
Material (Fiber) Dispersion arises because
the propagation constant b is not
proportional to the angular freq. w,
db
dw
≠ constant (independent of w)
db
b1
dw
db 1 ñ + w dñ ñg
dw
= c [ dw
] = c
c
Vg = ñ
g
read all ñ as n
where ñg is the group index. Since ñg is
different for different frequency component, so
is Vg, hence leading to pulse broadening as
each spectral component travels with different
group velocity.
Let Dw be the spectral width.
Then the extent of pulse broadening for a fiber
of length L is dT
DT= Dw
dw
Substitute T = L/Vg
d
DT=
dw
( VL ) Dw
g
= L b 2 Dw
d2b
where b 2 =
dw 2
and is known as GVD (Group Velocity Dispersion)
parameter. It determines how much a pulse would
broaden for a fixed L and Dw.
Normally spectral width is given in terms of l.
That is, instead of Dw one must find relation in Dl.
2pc dw _ 2pc
Dw= _ Dl ( )=
2
l dl l2
d 1 _ 2pc b
D=
dl
( )
Vg
=
l2
2
D is called Dispersion Parameter.
units of D is ps/(km – nm).
Criteria for Reliable Detection of information (or
estimation of effect of dispersion) is
B Dt < 1.
B L |D| Dl < 1.
Waveguide Dispersion
Material Dispersion
_ 2p d ng
1 d ng
DM = = c
l2 dw dl
_ 2pD ng2 V d2 (V b) d ng d (V b)
DW = [ n2 w 2 +
dl
]
l2 dV dV
b/k0 – n2 ñ – n2
b= n –n = n –n
1 2 1 2
In which case
bending is more?
1 2
R1 = 5 mm
R2 = 15 mm
Bend Loss
Higher the R, lesser the bend loss.
Bend loss < 0.01 dB/km for R > 5 mm.
This is not
FWHM.
This is where
the maxima
falls by 1/e.
For any profile of the fiber it has been found
that the typical fundamental mode field
distribution resembles almost a Gaussian
shape given by:
y (r) = A e - r 2/w2
Longitudinal
Transverse
Angular
Even if there is no misalignment, loss may
occur due to nonidentical field distributions for
dissimilar SMFs.
Transverse Misalignment
In general the ratio may be given as
T = [(2w1w2)/(w12+w22)]2 exp [-(2u2)/(w12+w22)]
Transverse Misalignment
The maximum power coupling occurs at u=0.
Tmax = [(2w1w2)/(w12+w22)]2
For w1 = w2, i.e., fibers having the identical
Gaussian fundamental modes, this is
maximum.
For perfectly aligned fibers, i.e., u=0
a (dB) = -20 log [(2w1w2)/(w12+w22)]
The loss
will be less
than 0.1 dB
provided:
0.86 < w1/w2
< 1.16
ml
3 2 1 0 -1 -2 -3
MS=1/2+1/2+1/2 so Smax=3/2
ML = 3+2+1 so Lmax= 6.
The term symbol is written as 2S+1L where L is a
label for the orbital angular momentum according
to the historical key:
L= 0 1 2 3 4 5 6 ···
label S P D F G H I ···
E2
E1
We have 2 intensities
What are those?
We have 2 intensities
Ip(r,z) + Is(r,z)
And 3 cross sections
Can you name them?
We have 2 intensities
Ip(r,z) + Is(r,z)
And 3 cross sections
spa, ssa, and spe.
Now let’s assume:
N1 – number density of lower energy level E1
N2 – number density of lower energy level E2
And try to see what processes do take place
at these 2 levels.
N1 – number density of lower energy level E1
N2 – number density of lower energy level E2
E2
E1
Rapid thermalization of Er3+ ions do occur at
4I
13/2 level and hence emission takes place
from bottom of the 4I13/2 level to top of the 4I15/2
level.
For this reason we consider
ssa ≠ sse.
Selection of dopants affects absorption-emission spectrum
At 1550 nm, se(n) = 3.41 units and sa(n) = 2.54, whereas at
1480 nm absorption cross section in much higher.
Selection of dopants affects absorption-emission spectrum.
This is for EDFA whose core is doped by germania (GeO2).
N1 encounters 4 processes –
1. Pump absorption (E1 E3)
2. Signal absorption (E1 E2)
3. Signal stimulated emission (E2 E1)
4. Signal spontaneous emission (E2 E1)
For analytical
analysis, pulse can
be approximated by
Gaussian envelope.
This is useful for
modeling.
Variation of pump power w.r.t. length. Note that
around input, pump power is high and the decrease
is almost linear, whereas for large z it becomes
exponential.
Variation of signal power with length. Input signal
power is 1 mW. Beyond an optimum length (which is ~
7m for Ppin = 5 mW), the signal gets attenuated.
Optimum length depends upon input pump power.
Variation of gain w.r.t. length. For maximum gain,
the length of the fiber must be chosen equal to the
optimum value, Lopt.
Variation of optimum length with the input pump
power.
Variation of gain with input pump power for different
length of the doped fiber. For any given fiber length
there is a threshold pump power for transparency of
the fiber. Beyond this, gain increases with pump
initially and then saturates.
Variation of gain with Input signal power for an
EDFA with different input pump powers.
For small input signal powers, gain is almost
independent of signal power.
Variation of gain with output signal power for an EDFA
with different input pump powers.
For small input signal powers, gain is almost
independent of signal power.
Variation of output amplified signal power with input
signal power.
Variation of gain with signal wavelength of a 980 nm pumped
EDFA for different input pump powers. Fiber exhibits gain for
ls greater than a specific l. As pump power increases, this
specific l shifts towards shorter wavelengths.
Major conclusions:
•Threshold pump power to achieve gain
• Optimum length for maximum gain, after
which gain reduces; it is a function of input
pump power
• Under small signal regime, gain is almost
independent of input signal power
• two noise components are there:
ASE
ESA
• 980 and 1480 nm pump schemes do not
show ESA and hence best suitable pump
schemes for EDFAs
• the flat gain profile within 1520 – 1580 nm
wavelength range makes EDFAs versatile for
WDM application.
Silica fiber is just a host medium.
EDFAs are bit-rate transparent, hence are
scalable.
Theory of EDFA came around 1964. But
Fabrication and characterization techniques
got perfected by 1987.
First EDFA deployed in 1993 by AT&T.
Typical gain ~ 40-50 dB
Insertion loss of EDFAs is very low.
Noise added by a typical EDFA ~ 3-4 dB.
Optical Amplifiers are not available for 1300
nm based systems. The only suitable
element is Praseodymium (Pr, Z=59).
However no commercial thrust was found for
this to develop.
Doping level usually remains ~ 200 ppm =
1025 ions/m3.
Three types based on the location of use:
1. Power Amplifiers: Psin ~ -3 dBm to 0
dBm. Though overall gain remains low,
Psout ~ high. Noise figure is also high.
2. Preamplifiers: Psin ~ -40 to -50 dBm (0.1
to 0.01 mW), small signal or unsturated
regime. ASE becomes a major concern.
Backward directional pumping at 980 nm
is preferred.
3. Line Amplifiers:
Er3+ ions excited to higher energy levels
quickly relax down to 4I13/2 level. At this level
due to rapid thermalization of Er3+ ions, the
emission takes place from the bottom of 4I13/2
level to 4I15/2 level.
Therefore emission cross section shifts
towards higher wavelength no matter what are
the co-dopants.
Al2O3 is found to be most suitable for WDM.
Pump Efficiency
1. Initial lasers – Ar-ion (514 nm-green and
488 nm-blue), Nd:YAG ( 532 nm), very low
efficiencies
2. GaAs (820 nm), pump efficiency 1 dB/mW,
absorption 4I15/2 --- 4I9/2.
3. AlGaInP (670 nm), 33 dB @ 27 mW,
absorption 4I15/2 --- 4F9/2.
4. at 1480 nm, population inversion remains
low, and so is pump efficiency. WHY???