SOLUTIONS ELEC0020-A5U 2019

List of physical constants:

Velocity of propagation of an electromagnetic wave in vacuum: c = 3  108 ms−1

Permittivity of vacuum or air: 0 = 8.85 × 10−12 Fm−1
Permeability of vacuum or air: 0 = 4× 10−7 Hm−1
Planck constant: h = 6.63 × 10−34 Js
Elementary charge: e = 1.60 × 10−19 C
Boltzmann constant: k = 1.38 × 10−23 JK-1

Answer TWO questions from Section A and TWO questions from Section B

SECTION A

1. Consider the baseband digital signal given by:

∞

𝑥(𝑡) = ∑ 𝑋𝑘 ∙ 𝑝(𝑡 − 𝑘𝑇𝑠 ),

𝑘=−∞
where p(t) is a unit-energy square pulse with duration Ts and Ts is the signalling interval. It is
assumed that the sequence of symbols {Xk} is independent. It is also assumed that the symbols Xk
are drawn from an alphabet with equal probability.

(a) Assume that one wishes to convey the sequence of bits 11011000.

(i) Sketch the waveform associated with the digital signal assuming that one uses polar binary
transmission. Specify the symbol alphabet associated with this case (assume symbols have
an amplitude, A).

[4 marks]

Page 1 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

Sketch [2 marks]

Alphabet [2 marks]

𝒜 = {−A, A} OR 𝒜 = {A, −A}

(ii) Sketch the waveform associated with the digital signal assuming that one uses 4-ary
(polar) signalling. Write down the symbol alphabet, and show the bit to symbol mapping
which minimises bit error rate (assume innermost symbols have an amplitude, A).

[6 marks]

Page 2 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

Sketch [2 marks]

Any valid Gray labelling [2 marks]

Alphabet 𝒜 = {−3A, −A, A, 3A} OR any permutation of this alphabet. [2 marks]

(iii)Write down the name of the bit-to-symbol mapping used in part (ii).

SOLUTION

Gray Coding OR Gray Mapping OR Gray Labelling OR Gray [2 marks]

(b) Assuming a target data rate of 1 Gbits-1, determine the power density spectrum and the
bandwidth efficiency of the digital signalling scheme associated with the case where:
(i) One conveys information using polar binary signalling;

[5 marks]

Page 3 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

The (average) power density spectrum of a digital signal is given by:

𝟏
𝑺𝒙 (𝒇) = ∙ 𝑺 (𝒇) ∙ |𝑷(𝒇)|𝟐
𝑻𝒔 𝑿

where (for a unit-energy square pulse)

∞ +𝑻𝒔 ⁄𝟐
−𝒋𝟐𝝅𝒇𝒕
𝟏
𝑷(𝒇) = ∫ 𝒑(𝒕) ∙ 𝒆 𝒅𝒕 = ∫ ∙ 𝒆−𝒋𝟐𝝅𝒇𝒕 𝒅𝒕 = √𝑻𝒔 ∙ 𝒔𝒊𝒏𝒄(𝒇𝑻𝒔 )
−∞ −𝑻𝒔 ⁄𝟐 √𝑻𝒔
∞

𝑺𝑿 (𝒇) = ∑ 𝑹𝑿 (𝒌) ∙ 𝒆−𝒋𝟐𝝅𝒌𝒇𝑻𝒔

𝒌=−∞

and

𝑹𝑿 (𝒌) = 𝑬{𝑿𝒊+𝒌 ∙ 𝑿𝒊∗ }

Assume that the binary alphabet consists of symbols – A and A. Then,

𝟐
𝑹𝑿 (𝒌) = {𝑨 𝒌=𝟎
𝟎 𝒌≠𝟎
so

𝑺𝑿 (𝒇) = 𝑨𝟐

and

𝑺𝒙 (𝒇) = 𝑨𝟐 ∙ [𝒔𝒊𝒏𝒄(𝒇𝑻𝒔 )]𝟐

The bandwidth efficiency corresponds to the ratio between the data rate and the
bandwidth occupancy of the digital signal. The data rate is given by:
𝟏
𝑹 = 𝑻 = 𝟏 Gbit⋅s-1
𝒔

The bandwidth occupancy is given by:

𝟏
𝑩 = 𝑻 = 𝟏 GHz
𝒔

(we define bandwidth to be equal to the width of the interval from frequency equal to zero
to the very first frequency where the power density spectrum is equal to zero). Then, the
bandwidth efficiency is equal to:

𝜼 = 𝟏 bit⋅s-1⋅Hz-1 (OR, equivalently, 1 bit)

(ii) One conveys information using 4-ary (polar) signalling.

[5 marks]
Page 4 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

The (average) power density spectrum of a digital signal is given by:

𝟏
𝑺𝒙 (𝒇) = ∙ 𝑺 (𝒇) ∙ |𝑷(𝒇)|𝟐
𝑻𝒔 𝑿

where (for a unit-energy square pulse)

∞ +𝑻𝒔 ⁄𝟐
−𝒋𝟐𝝅𝒇𝒕
𝟏
𝑷(𝒇) = ∫ 𝒑(𝒕) ∙ 𝒆 𝒅𝒕 = ∫ ∙ 𝒆−𝒋𝟐𝝅𝒇𝒕 𝒅𝒕 = √𝑻𝒔 ∙ 𝒔𝒊𝒏𝒄(𝒇𝑻𝒔 )
−∞ −𝑻𝒔 ⁄𝟐 √𝑻𝒔
∞

𝑺𝑿 (𝒇) = ∑ 𝑹𝑿 (𝒌) ∙ 𝒆−𝒋𝟐𝝅𝒌𝒇𝑻𝒔

𝒌=−∞

and

𝑹𝑿 (𝒌) = 𝑬{𝑿𝒊+𝒌 ∙ 𝑿𝒊∗ }

Assume that the binary alphabet consists of symbols [-3A, – A, A and 3A]. Then,
𝟐 𝟐
𝑹𝑿 (𝒌) = {(𝟗𝑨 + 𝑨 )/𝟐 𝒌 = 𝟎
𝟎 𝒌≠𝟎
so

𝑺𝑿 (𝒇) = 𝟓𝑨𝟐

and

𝑺𝒙 (𝒇) = 𝟓𝑨𝟐 ∙ [𝒔𝒊𝒏𝒄(𝒇𝑻𝒔 )]𝟐

The bandwidth efficiency corresponds to the ratio between the data rate and the
bandwidth occupancy of the digital signal. The data rate is given by:
𝟏
𝑹 = 𝑻 = 𝟏 Gbit⋅s-1
𝒔

The bandwidth occupancy is given by:

𝟏
𝑩 = 𝑻 = 𝟓𝟎𝟎 MHz
𝒔

(we define bandwidth to be equal to the width of the interval from frequency equal to zero
to the very first frequency where the power density spectrum is equal to zero). Then, the
bandwidth efficiency is equal to:

𝜼 = 𝟐 bit⋅s-1⋅Hz-1 (OR, equivalently, 1 bit)

Page 5 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
(c) To increase bandwidth efficiency, the size of the signal alphabet can be further increased.
Comment on the practical limitations of this approach.

SOLUTION

The signal-to-noise ratio requirement will increase [2 marks] for a target bit (or symbol) error
probability [1 mark].

OR

To use the channel bandwidth more efficiently, the signal energy must be increased [2 marks]
for a given signal performance [1 mark]

Page 6 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
2.
(a) Sketch a diagram containing the main blocks of a passband digital transmitter. Describe the
function of each particular block.

[6 marks]

SOLUTION

Sketch [3 marks]

Mapper: Maps bits from a binary input sequence to symbols from the symbol alphabet.

Modulator: Modulates symbols onto a signal by applying a pulse shaping filter.

Up-Converter: Provides a frequency shift to the signal converting it from baseband to

passband.

[3 marks]

(b) Consider the passband modulation format, sometimes known as set (1,3), with a signal
alphabet specified by:
1 √3 1 √3
𝒜 = {0, A, (− + 𝑗) 𝐴, (− − 𝑗) 𝐴},
2 2 2 2
where 𝐴 is an arbitrary scalar greater than zero, and 𝑗 = √−1.
(i) For this modulation format, how much information is encoded per transmitted symbol?

[2 marks]

SOLUTION

log 2 (𝑀) = log 2 (4) = 2 bit/symbol [2 marks] {only 1 mark without units}

OR

log 2 (𝑀) = log 2 (4) = 2 bits [2 marks] {only 1 mark without units}

(ii) Sketch the constellation diagram, and the maximum likelihood decision boundaries.
Page 7 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
[6 marks]

SOLUTION

As per figure: Plot of constellation [2 marks], axis labels [1 mark], ML decision boundaries
[3 marks]

(iii) Calculate the minimum Euclidean distance for this modulation format.

[4 marks]

SOLUTION

Compute the two distances in the constellation; 𝐴, and 𝐴√3 [2 marks]

Identify A as the minimum Euclidean distance [2 marks]

(c) Some modulation formats, for example quadrature phase shift keying (QPSK), permit a bit-
to-symbol mapping where only a single bit changes between adjacent symbols.

Page 8 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
(i) By labelling the constellation in part (b), or otherwise, argue that such a bit-to-symbol
mapping is not possible for the format described in part (b).

[5 marks]

Page 9 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

Due to its rotational symmetry, there are only two possible ways to label this constellation.
The zero-valued symbol can encode two equal bits, or two different bits (see the following
two figures). One can argue, by these (or all) bit-to-symbol mappings that a Gray labelling
is not possible. Alternatively, it can be stated that the zero-valued symbol has three nearest
neighbours, but only two bits to encode, therefore a Gray coding is not possible.

 Any valid labelling [2 marks]

 Correct argument [3 marks].

Page 10 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019

(ii) What impact will this have on the relationship between symbol error probability and bit
error probability versus, for example, QPSK?

[2 marks]

SOLUTION

Some nearest-neighbour symbol errors will encode two bit errors, instead of just one [2
marks]

Page 11 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
3. Consider a passband digital communications system where the channel output y(t) is related to
the channel input x(t) as follows:

𝑦(𝑡) = 𝑥(𝑡) + 𝑛(𝑡),

where n(t) is additive white Gaussian noise (AWGN). A digital signal is transmitted through
this channel,
∞

𝑥(𝑡) = ∑ {Re{𝑋𝑘 } ∙ 𝑝(𝑡 − 𝑘𝑇𝑠 ) cos(2𝜋𝑓𝑐 𝑡) − Im{𝑋𝑘 } ∙ 𝑝(𝑡 − 𝑘𝑇𝑠 ) sin(2𝜋𝑓𝑐 𝑡)},
𝑘=−∞
where 𝑝(𝑡) is a unit-energy square pulse of duration 𝑇𝑠 , 𝑇𝑠 is the signalling interval, and 𝑓𝑐 , is
the carrier frequency.
(a) Two processing blocks of a digital passband receiver are the down-converter and the filter.
(i) Sketch the down-conversion process for the received signal, 𝑦(𝑡).

[4 marks]

SOLUTION

As per figure [4 marks].

(ii) For the received signal, 𝑦(𝑡), which filter shape will maximise the signal-to-noise ratio
after sampling? What is the name given to this type of filter?

[2 marks]

SOLUTION

The receiver filter should be equivalent to the transmit filter. I.e., this should be the
matched filter, 𝒑(𝒕 − 𝒌𝑻𝒔 ) [2 marks]

(b)
(i) Show that, for an AWGN channel, the Maximum Likelihood (ML) decision rule is
simply given by:

Page 12 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
𝑋̂𝑘 = argmin|𝑌𝑘 − 𝑋𝑘 |2 ,
𝑋𝑘 ∈𝒜
where 𝑋𝑘 and 𝑌𝑘 are the received samples, respectively, and 𝒜 is the symbol alphabet.

[6 marks]

SOLUTION

For an AWGN channel, noise samples after the matched filter are given by

𝑵𝒌 = 𝒀𝒌 − 𝑿𝒌

[2 marks]

And the maximum likelihood decision rule is:

̂ 𝒌 = 𝐚𝐫𝐠𝐦𝐚𝐱 𝒑(𝒀𝒌 |𝑿𝒌 ) = 𝐚𝐫𝐠𝐦𝐚𝐱 𝒑(𝑵𝒌 )

𝑿
𝑿𝒌 ∈𝓐 𝑿𝒌 ∈𝓐

[2 marks]

Thus,
|𝒀 −𝑿 | 𝟐
𝟏 − 𝒌 𝒌
̂ 𝒌 = 𝐚𝐫𝐠𝐦𝐚𝐱
𝑿 ∙ 𝒆 𝟐𝑵𝟎 = 𝐚𝐫𝐠𝐦𝐢𝐧|𝒀𝒌 − 𝑿𝒌 |𝟐
𝑿𝒌 ∈𝓐 𝟐𝝅𝑵𝟎 𝑿𝒌 ∈𝓐

[2 marks]

(ii) Describe, in words or with the aid of a sketch, a geometric interpretation of this result.

[2 marks]

Page 13 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

This result states that the selected symbol decision is the symbol which is geometrically
closest to the received sample, 𝒀𝒌 [2 marks]

OR

Anything similar to the following sketch (the modulation format does not matter) with an
appropriate description [2 marks]

(c) Consider the modulation formats 4-ary amplitude shift keying (4ASK), with a symbol
alphabet 𝒜 = {−3A, −A, A, 3A}, and binary phase shift keying (BPSK), 𝒜 = {−A, A}.
Assuming that, for each format, the symbols are transmitted with an equal probability (and
the signal is received with a finite signal-to-noise ratio):
(i) Sketch the constellation diagram for these formats, showing the ML and Maximum a
posteriori (MAP) decision boundaries.

[4 marks]

Page 14 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

The MAP and ML decision boundaries are the same in this case. See each figure, below [4
marks]

Page 15 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
(ii) Sketch the ML and MAP decision boundaries for 4ASK, assuming that the symbols
−3A and 3A are transmitted with a lower probability than the other two symbols.

SOLUTION

Boundaries move closer to -3A and +3A as per figure [2 marks].

[2 marks]

(iii)For the BPSK modulation format, derive a general expression for the MAP decision
boundaries, starting with the definition of the MAP receiver, and compare this with the
expression for ML given in part (b).

[5 marks]

Page 16 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

Derivation [3 marks]

MAP receiver definition:

𝑋̂𝑘 = argmax 𝑃(𝑋𝑘 |𝑌𝑘 )

𝑋𝑘 ∈𝐴

Using Bayes’ rule:

𝑃𝑋𝑘 (𝑋𝑘 )𝑃𝑌𝑘|𝑋𝑘 (𝑌𝑘 |𝑋𝑘 )

𝑋̂𝑘 = argmax = argmax 𝑃𝑋𝑘 (𝑋𝑘 )𝑃𝑌𝑘|𝑋𝑘 (𝑌𝑘 |𝑋𝑘 )
𝑋𝑘 ∈𝐴 𝑃𝑌𝑘 (𝑌𝑘 ) 𝑋𝑘 ∈𝐴

Define problem as inequality: when should symbol A be selected:

𝑃(𝑋𝑘 = 𝐴)𝑃(𝑌𝑘 |𝑋𝑘 = 𝐴) > 𝑃(𝑋𝑘 = −𝐴)𝑃(𝑌𝑘 |𝑋𝑘 = −𝐴)

Define 𝑃(𝑋𝑘 = 𝐴) = 𝑃𝐴 and 𝑃(𝑋𝑘 = −𝐴) = 𝑃−𝐴

𝟐
(𝑌 −𝑋 )
1 − 𝑘 𝑘
Note that 𝑃(𝑌𝑘 |𝑋𝑘 ) = 𝑒 𝑁0
√𝜋𝑁0

Thus
(𝑌𝑘 −𝐴)𝟐 (𝑌𝑘 +𝐴)𝟐
− −
𝑃𝐴 𝑒 𝑁0 > 𝑃−𝐴 𝑒 𝑁0

−(𝑌𝑘 −𝐴)𝟐 +(𝑌𝑘 +𝐴)𝟐 𝑃−𝐴

𝑒 𝑁0 >
𝑃𝐴

𝑃−𝐴
−(𝑌𝑘2 + 𝐴2 − 2𝑌𝑘 𝐴) + (𝑌𝑘2 + 𝐴2 + 2𝑌𝑘 𝐴) > N0 log ( )
𝑃𝐴

N0 𝑃−𝐴
𝑌𝑘 > log ( )
4A 𝑃𝐴

This solution shows that when both symbol probabilities are equally likely, or, when the
signal-to-noise ratio is high, the ML and MAP decision boundaries become the same [2
marks]

Page 17 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
SECTION B

4.
(a) In general, the refractive index (𝑛) and relative permittivity (𝜀𝑟 ) of a material can be
represented as complex numbers (where 𝑗 = √−1):

𝑛 = 𝑛′ − 𝑗𝐾 and 𝜀𝑟 = 𝜀𝑟′ − 𝑗𝜀𝑟′′ .

In the case of titanium dioxide (TiO2), at a free-space wavelength of 340 nm (in the
ultraviolet part of the electromagnetic spectrum), 𝑛′ = 5.0 and 𝐾 = 1.1.
(i) State the mathematical relationship between 𝑛 and 𝜀𝑟 . Assuming this relationship holds
when 𝑛 and 𝜀𝑟 are complex, find the values of 𝜀𝑟′ and 𝜀𝑟′′ for TiO2 at 340 nm.
[5 marks]

SOLUTION

Relationship between relative permittivity and refractive index is: 𝜀𝑟 = 𝑛2 . [1 mark]

𝑛2 = (𝑛′ − 𝑗𝐾)2 = (𝑛′ )2 − 𝐾 2 − 𝑗. 2𝑛𝐾, so

𝜀𝑟′ = (𝑛′ )2 − 𝐾 2 and 𝜀𝑟′′ = 2𝑛𝐾. [2 marks]

Substituting the values gives 𝜀𝑟′ = 23.79 and 𝜀𝑟′′ = 11. [2 marks]

(ii) Find the attenuation of a 50-nm-thick layer of TiO2 at a wavelength of 340 nm. Express
your answer in dB.
[5 marks]

SOLUTION

The extinction ratio, 𝐾, is related to the field attenuation coefficient, 𝛾, as 𝐾 = 𝛾/𝛽0 .

So the power attenuation coefficient is given by:

4𝜋
𝛼 = 2𝛾 = 2𝐾𝛽0 = 𝐾 [2 marks]
𝜆0

Substituting values gives 𝛼 = 4.07x107 m-1. [1 mark]

Transmission through 𝐿 = 50 nm is exp(−𝛼𝐿) = exp(−2.03) = 0.131 = -8.8 dB. [2 marks]

(b) A semiconductor symmetric slab waveguide with dimensions and core and cladding
refractive indexes shown in Figure 4.1 has been designed to guide light with a wavelength of
800 nm. The electromagnetic wave equation can be solved to find the TE and TM modes of
the waveguide.
(i) Explain what is meant by a mode in this context.
Page 18 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
[4 marks]

SOLUTION

A mode can be considered in one sense to be ‘one of the ways in which light can propagate through
the waveguide’. It describes a specific spatial variation of the EM field across the waveguide (i.e.
perpendicular to the direction of propagation), which is one solution of the wave equation. It could
also be described in terms of light being totally internally reflected down the waveguide at one of a
set of discrete angles of incidence. [4 marks]

(ii) Define the orientation of the electric and magnetic fields of the TE modes with respect
to the directions 𝑥 and 𝑧 shown in Figure 4.1.
[3 marks]

SOLUTION

Both the TE and TM modes have E and H fields that are perpendicular to each other and to the
direction of propagation 𝑧 [1 mark]. The TE (transverse electric) modes have the E field parallel to
the interface between the waveguide core and cladding layers, i.e. in the direction perpendicular to
both 𝑥 and 𝑧 directions [1 mark]. The H field is thus in the 𝑥 direction [1 mark].

(iii) Sketch the amplitude distribution in the 𝑥 direction for the TE2 mode.
[3 marks]

SOLUTION

That is, the TE2 mode has cosinusoidal variation in the core, with 2 zeros, and decaying exponential
distribution in the cladding layers [3 marks for diagram that shows these main points].

Page 19 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
(iv) Calculate the V-number for the waveguide and hence find the number of modes this
waveguide can support. Explain how you obtain your answer by reference to a diagram
illustrating how the mode parameters can be solved for graphically.
[5 marks]
SOLUTION

2𝜋𝑎 2
𝑉= (𝑛1 − 𝑛22 )1/2
𝜆
where 𝑎 is half the width of the waveguide core.

Substituting the values gives 𝑉 = 4.5. [2 marks]

The number of modes is 𝑀 = Int(𝑉/(𝜋/2)) + 1 = 3. [1 mark for stating equation and giving 𝑀]

This is obtained from the following graphical solution, with the number of modes being the number
of intersections between a circle of radius 𝑉 with the other curves. [2 marks for diagram]

pd  hd tanhd 
6

5 pd  hd cot hd 

4

 pd 2  hd 2   0 2 n12  n2 2 d 2
pd 3

1
2 
0 0  
0  2

4 3 6  c
2 2
hd

x
𝑛2 = 3.2
𝑛1 = 3.4
1 µm z
𝑛2 = 3.2

Figure 4.1: Cross-section through symmetric slab waveguide (light propagation in z direction).

Page 20 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019

5.
(a) Explain, using a diagram, how a system with three energy levels can be used as the gain
medium for a laser. Briefly explain why a system with at least three levels is required to
obtain gain and what its limitations are.
[6 marks]

SOLUTION

Non-radiative
transitions (phonons)

𝐸3

𝐸2
Metastable state

Stimulated
Pumping Emission
𝑓21
𝑓13

𝑓21
𝐸1 [3 marks for diagram]

Pumping with high frequency photons at 𝑓13 from energy level 1 to energy level 3, followed by
non-radiative transition to metastable (long-lifetime) energy level 2 allows population inversion of
level 2 over level 1. This allows net stimulated emission (optical gain) between level 2 and level 1
for photons with frequency 𝑓21 . [1 mark]

Using a 3-level system and pumping through level 3 allows population inversion to be achieved in a
way that is not possible with a 2-level system, for which only equal populations of states at 𝐸2 and
𝐸1 can be obtained. [1 mark]

However, once population inversion has been achieved (so that the number of states at 𝐸1 is small),
pumping cannot take place effectively until stimulated emission has occurred, re-populating level 1,
thus making continuous lasing difficult in a 3-level system. [1 mark]

(b)
(i) Show that the threshold gain (𝑔𝑡ℎ ) of a Fabry-Perot (FP) laser is given by:

1 1
𝑔𝑡ℎ = 𝛼 + ln ( ),
2𝐿 𝑅1 𝑅2

where 𝛼 is the loss coefficient, 𝐿 is the length of the laser cavity, and 𝑅1 , 𝑅2 are the
power reflectances of the two mirrors forming the cavity.
[6 marks]

Page 21 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

Length, 𝐿

𝑃4 𝑃2
𝑃 𝑡, 𝑃 𝑡,𝑟
Forward wave Reverse wave
Power Power
reflectance 𝑃1 𝑃3 reflectance
= 𝑅1 = 𝑅2

 For the forward wave: 𝑃2 = 𝑃1 exp 𝑔 − 𝛼 𝐿

 At the right-hand reflector: 𝑃3 = 𝑅2 𝑃2 [and 𝑃 𝑡,𝑟 = 1 − 𝑅2 𝑃2 ]
 For the reverse wave: 𝑃4 = 𝑃3 exp 𝑔 − 𝛼 𝐿
 At the left-hand reflector: 𝑃1 = 𝑅2 𝑃4 [and 𝑃 𝑡, = 1 − 𝑅1 𝑃4]
 So for unity round-trip gain: 𝑃1 = 𝑃1 × 𝑅1 𝑅2 exp[2 𝑔𝑡ℎ − 𝛼 𝐿]
1 1
 𝑔𝑡ℎ = 𝛼 + ln
2𝐿 𝑅1 𝑅2

[1 marks for each of the above steps, total 6 marks]

(ii) Calculate the threshold gain of a helium-neon (He-Ne) FP laser of length 20 cm with
mirror reflectances of 90% and 100% if the loss coefficient is 0.06 m-1.
[3 marks]

SOLUTION
1 1
𝑔𝑡ℎ = 0.06 + 2×0.2 ln (0.9×1) = 0.323 m-1

[3 marks]

(c) The external quantum efficiency (𝜂𝑒𝑥𝑡 ) of a photodiode (PD) is given by:

𝜂𝑒𝑥𝑡 = 𝜂𝑖𝑛𝑡 (1 − 𝑅)[exp(−𝛼𝑥1 ) − exp(−𝛼𝑥2 )],

where 𝜂𝑖𝑛𝑡 is the internal quantum efficiency, 𝑅 is the reflectance at the surface where the
light is incident, 𝛼 is the absorption coefficient, and 𝑥1 and 𝑥2 are the distances from the
surface to the top and bottom of the depletion region.

(i) Define the internal and external quantum efficiencies for a photodiode.
[4 marks]

Page 22 of 26
CONTINUED
 SOLUTIONS ELEC0020-A5U 2019
SOLUTION

The internal quantum efficiency is the number of EHPs generated that contribute to the
photocurrent for each photon absorbed in the PD. [2 marks]

The external quantum efficiency is the ratio of the number of electrons generated in the external
circuit to the number of incident photons. [2 marks]

(ii) A silicon photodiode is illuminated through its top surface, which is anti-reflection
coated to give a reflectance of 3%. Under reverse bias, the depletion region starts
0.3 µm below the top surface and is 5 µm thick. The absorption coefficient of silicon at
a wavelength of 633 nm is 4000 cm-1.

If the responsivity of the photodiode is measured to be 0.35 AW-1 at a wavelength of

633 nm, calculate the internal quantum efficiency of the photodiode.
[6 marks]

SOLUTION
𝐼
Responsivity, 𝑅𝑠 = 𝑃𝑜
𝑜

𝐼 /𝑒 ℎ𝑐
𝜂𝑒𝑥𝑡 = 𝑃 𝑜/ℎ𝑓 = 𝑅𝑠 𝑒𝜆 = 0.35 × 1.96 = 0.686. [3 marks]
𝑜

To calculate 𝜂𝑖𝑛𝑡 , use 𝑅 = 3%, 𝑥1 = 0.3 µm, 𝑥2 = 5.3 µm in equation given.

𝛼𝑥1 = 0.12; 𝛼𝑥2 = 2.12, so [exp(−𝛼𝑥1 ) − exp(𝛼𝑥2 )] = 0.8869 − 0.1200 = 0.7669.

𝜂𝑒𝑥𝑡 = 𝜂𝑖𝑛𝑡 × 0.97 × 0.7669 = 𝜂𝑖𝑛𝑡 × 0.744 = 0.686.

0.686
Therefore, 𝜂𝑖𝑛𝑡 = 0.744 = 92.2%. [3 marks]

Page 23 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
6.
(a) Group velocity dispersion (GVD) contributes to chromatic dispersion (CD) in optical fibres,
which causes pulse spreading as signals propagate through the fibre.

(i) What other form of dispersion, when combined with GVD, determines the CD?
[1 mark]

SOLUTION

Waveguide dispersion [1 mark]

(ii) The temporal pulse spreading (Δ𝑡) is usually represented by the equation

Δ𝑡 = 𝐿. |𝐷𝑚 |. Δ𝜆

where 𝐿 is the length of the optical fibre, 𝐷𝑚 is the material dispersion coefficient, and
Δ𝜆 is the spectral width of the signal expressed in terms of wavelength.
𝑑𝑛
Given that the group refractive index, 𝑛𝑔 , is given by 𝑛𝑔 = 𝑛 − 𝜆0 𝑑𝜆 , derive an
0
expression relating 𝐷𝑚 to the refractive index, 𝑛, and the free-space wavelength, 𝜆0 .
[5 marks]

SOLUTION

[5 marks]

(b)
(i) Binary data is transmitted over an optical fibre using non-return-to-zero pulses that are
shaped so that the spectral bandwidth of the signal is equal to the bit rate, 𝐵. If the
maximum pulse-spreading due to chromatic dispersion that can be tolerated is one bit
period, show that the bit-rate distance product is given by

Page 24 of 26
END OF PAPER
 SOLUTIONS ELEC0020-A5U 2019
1
𝐵∙𝐿 =
Δ𝜆 ⋅ 𝐷
where 𝐷 is the chromatic dispersion coefficient of the optical fibre, and 𝐿 and Δ𝜆 are as
defined in (a)(ii) above.
[5 marks]

SOLUTION

1 1
Δ𝑡(max) = 𝐷𝐿Δ𝜆 = 𝑇𝑏 = →𝐵∙𝐿 =
𝐵 Δ𝜆 ⋅ 𝐷
[5 marks]

(ii) If such a system operates at a bit rate of 12.5 Gbits-1 and a wavelength of 1550 nm, and
the chromatic dispersion coefficient of the optical fibre used is 17 psnm-1km-1 at
1550 nm, what is the maximum dispersion-limited distance over which the system can
operate? (You may use the fact that a wavelength difference of 1 nm corresponds to a
frequency difference of 125 GHz at 1550 nm.)
[5 marks]

SOLUTION
1
𝐿𝑚𝑎𝑥 = Δ𝜆⋅𝐷∙𝐵 with Δ𝜆 = 12.5/125 = 0.1 nm, 𝐵 = 12.5 × 109 Hz, 𝐷 =17 ps/nm/km.

Hence 𝐿𝑚𝑎𝑥 = 47 km.

[5 marks]

(c) A semiconductor laser is directly modulated with non-return-to-zero (NRZ) data. The mean
bias current applied to the laser is 50 mA and the data modulation current is 40 mA peak-to-
peak. The laser has a threshold current of 25 mA and a slope efficiency of 0.15 WA-1.

Calculate the Q parameter at the receiver if the signal is transmitted through 120 km of
optical fibre with loss 0.25 dBkm-1. Assume the receiver is thermal-noise limited, and that
its front end consists of a photodiode with responsivity 0.8 AW-1 connected to a 50 Ω load.
The receiver temperature is 300 K and its bandwidth is 1 GHz.

[9 marks]

SOLUTION

Fibre loss = 120 × 0.25 = 30 dB, i.e. a factor of 1/1000. [1 mark]

Power transmitted in zeros: 𝑃0 = (30 − 25) × 0.15 = 0.75 mW [1 mark]

Page 25 of 26
TURN OVER
 SOLUTIONS ELEC0020-A5U 2019
Power transmitted in ones: 𝑃1 = (70 − 25) × 0.15 = 6.75 mW [1 mark]
𝜇1 − 𝜇0 𝜇1 − 𝜇0
𝑄= =
𝜎1 + 𝜎0 2𝜎𝑡ℎ

where 𝜎𝑡ℎ is the thermal noise standard deviation. [1 mark]

In terms of currents:

𝜇0 = 𝑅𝑠 × 𝑃0 (𝑟𝑒𝑐𝑒𝑖𝑣𝑒𝑑) = 0.8 × 0.75 × 10−3 /1000 = 0.6 × 10−6 A. [1 mark]

𝜇1 = 𝑅𝑠 × 𝑃1 (𝑟𝑒𝑐𝑒𝑖𝑣𝑒𝑑) = 0.8 × 6.75 × 10−3 /1000 = 5.4 × 10−6 A. [1 mark]

2 4𝑘𝑇𝐵 4×1.38×10−23 ×300×1×109

𝜎𝑡ℎ = = = 3.312 × 10^ − 13 A2.
𝑅 50

𝜎𝑡ℎ = 5.75 × 10−7 A. [2 marks]

4.8×10−6
Therefore 𝑄 = 2×5.75×10−7 = 4.17. [1 mark]

Page 26 of 26
END OF PAPER