Exercises week 5: Fiber communication
Problem 1 (FoP 9.3‐3): Scattering loss. At λ0=820 nm the absorption loss of a fiber is
0.25 dB/km and the scattering loss is 2.25 dB/km. If the fiber is used instead at
λ0=600 nm, and calorimetric measurements of the heat generated by light absorption
give a loss of 2 dB/km, estimate the total attenuation at λ0=600 nm.
Problem 2 (FoP 23.2‐1): Cascaded MZI MUX/DEMUX. Three Mach‐Zehnder interferometers
(MZIs) are cascaded as shown in Fig. 23.2‐6 to multiplex or demultiplex four
wavelength channels with wavelength separation Δλ=0.2 nm and central wavelength
1550 nm. Determine the necessary path length differences Δd in each interferometer if
the refractive index is n=2.3.
Problem 3 (FoP 23.3‐1): Power loss and crosstalk. A 4x4 switch may be implemented by use
of five 2x2 switches. If each of these switches introduces a power loss of 0.5 dB and a
crosstalk of ‐30 dB, determine the worst case power loss and crosstalk for a 4x4 switch.
Problem 4: Interconnects, routers, and switches.
An integrated Mach‐Zehnder interferometer is used to separate two wavelengths, λ1
and λ2, incident at one input port to either of the output ports (see fig. 1). λ1=1550 nm,
the path length difference is Δd=500 µm, and the refractive index is n=1.5. What must
the wavelength λ2 be in order to go to the other port? Are there several choices for λ2?
What must a third wavelength λ3 be in order to go to the same port like λ1? Is the total
length d0 of the device important for its function?
A nonlinear Mach‐Zehnder interferometer is used for intensity‐based routing (see
fig. 2). Pulses with intensity I1 go to output port 1 and pulses with intensity I2=I1/2 go to
output 2. The nonlinear element, placed in one of the interferometer arms, has a
nonlinear refractive index of n2=10‐11 cm2/W and is 1 mm long. At low intensity both
interferometer arms have identical length. Find the intensities I1 and I2. Assume
λ=1550 nm. (Remember that the refractive index changes as n=n0+n2I.)
Fig 1: 2x2 optical coupler used as multiplexer. Fig. 2: Intensity‐based routing
Problem 5: Fibre optical communication systems.
Power budget and time budget: We look at a fiber communication link operating at
λ=1550 nm with a data rate of B0=5 Gb/s. The link is spliced together from 10 km long
pieces of fiber with losses of 0.2 dB/km and a dispersion coefficient Dλ=17 ps/km‐nm.
The coupling losses between the fiber pieces are 0.2 dB per connection, the in/out‐
coupling losses are 0.5 dB and we consider a power margin of 6 dB. The source is a
distributed feedback laser with spectral width σλ=0.1 nm, a response time of 20 ps, and
a power of 1 mW. The detector requires 1.3 µW of optical power at the output of the
link and it has a response time of 100 ps.
Calculate the power and time budget and find out, if the system is attenuation or
dispersion limited. Use the criterion that the response time of the overall system does
not exceed 70% of the bit time. (Use σsσν=0.375 to find the Fourier limited pulse
duration σs of the source.)
For the above link, if we replace the fiber by a dispersion shifted fiber with
Dλ=1 ps/km‐nm, what is the maximum dispersion limited length of the link? The link
now obviously is attenuation limited. How many 20 dB gain Erbium amplifier segments
would one need to splice into the link in order to remove the attenuation limitation
(Let’s forget about splicing losses and noise from the amplifier here!)?
Short answers:
Problem 1 (FoP 9.3‐3): 9.85 dB/km at 600 nm.
Problem 2 (FoP 23.2‐1): d 2.6 mm for the first interferometer and d 1.3 mm for the
second interferometer.
Problem 3 (FoP 23.3‐1): maximum loss of 1.5 dB. Maximum crosstalk ‐ 25.2 dB.
Problem 4: 1.6 nm (at vacuum). 2 1 m with m 1,3,... , e.g. λ2=1551.6, 1548.4,
1554.8, 1545.2 nm. 3 1 2m with m 1,2,... ., e.g. 1553.2, 1546.8 nm. The total length
d0 of the device is not important for its functioning (as long as it is shorter than the
coherence length of the source).
I1 1.55 10 8 Wcm‐2.
n2 d
Problem 5: Power budget: L 100.3 km, time budget: L 53.6 km. The system is dispersion
limited.
With the dispersion‐shifted fiber: L=911 km, 9 EDFA amplifiers are needed along the way.
Detailed solutions
Problem 1 (FoP 9.3‐3): We know that the scattering loss scales as 1/λ4. Knowing that the
scattering loss at 820 nm is 2.25 dB/km, we obtain 7.85 dB/km at 600 nm, resulting in a total
attenuation of 9.85 dB/km at 600 nm.
Problem 2 (FoP 23.2‐1): The first interferometer should separate wavelength channels with a
spacing of Δλ=0.2 nm from each other (i.e. λ1 from λ2 and λ3 from λ4). Note that the channel
spacing Δλ=Δλ0 normally refers to the vacuum wavelength, whereas the wavelength in the
material is different. The frequency difference is c0 2nd (FoP 23.2‐1). One can use
0 0 c 0 0 20 to obtain d 20 2n 0 2.6 mm. The second
interferometer separates channels with spacing 2Δλ0=0.4 nm. The physical length difference
is d 1.3 mm.
Problem 3 (FoP 23.3‐1): In the worst case the signal has to pass three 2x2 switches, resulting
in a maximum loss of 1.5 dB. The maximum crosstalk is 3 10 3 , corresponding to
10 log 3 10 3 25.2 dB.
Problem 4: Like in problem 2 we use c0 2nd 200 GHz. With
0 0 c 0 0 20 we obtain 1.6 nm (at vacuum). Possible choices for
the wavelength that goes to the other output port are 2 1 m with m 1,3,... , e.g.
λ2=1551.6, 1548.4, 1554.8, 1545.2 nm. Wavelengths that go to the same port like λ1 can be
3 1 2m with m 1,2,... ., e.g. 1553.2, 1546.8 nm. The total length d0 of the device is
not important for its functioning, if the source is sufficiently coherent.
We are looking for the intensities I1 and I2 that lead to an increase of the path length of the
interferometer arm with the nonlinear element by a full wavelength and half a wavelength,
2
respectively. We can write for the nonlinear phase shifts 1 n 2 I 1 d and
0
2
2 n 2 I 2 d . The difference between the two phase shifts must be π:
0
2 I
1 1 n2 d I1 1 I1 1.55 10 8 Wcm‐2.
2 n 2 d
Problem 5: We start by writing the source and detection powers in dB using the definition on
page 1091 (FoP). Ps 0 dB and Pd 28.86 dB. Note that the definition how optical
power is translated to dB is totally arbitrary. We can now write down the power budget:
L
Ps 2 Pio L 1 Pc Psaf Pd
10
Ps 2 Pio Psaf Pd Pc
L 100.3km
Pc 10
The coupling losses between the 10 km long fiber pieces was taken into account in a
rudimentary manner. The particular way of how this is done might change the result slightly.
In order to write down the time budget, we need the duration of the source σs, which can be
obtained from the time‐bandwidth product s 0.375 with we obtain
c0 12.5 GHz and the transform‐limited duration of the source is s 30 ps. We
20
can now write down the time budget:
2
0.7
0.7T
2
s
2
tx
2 2
rx
2
B0
2
0.7
LD s2 tx2 rx2
2
B0
2
0.7
s2 tx2 rx2
B0
L
D
L 53.6 km
The system is dispersion limited.
If we use a dispersion‐shifted fiber with Dλ=1 ps/km‐nm instead, the dispersion limited
length suddenly becomes L=911 km. Now the system obviously is attenuation limited. If we
look at the power budget with the new length, we recognize that an amplification of
178.3 dB is needed for the signal to travel the new length. Thus, 9 EDFA amplifiers with
20 dB amplification each need to be distributed along the way.