Signals, Systems, and

Spectral Analysis

Signals & Systems UOIT: Communication Systems (ELEE3130U) 1

 Signals

- Any time-varying physical phenomenon that conveys information is referred to as a signal

 Examples of signals with useful information, such as human voices, MP3 music, JPEG images,
seismic signals, CT scans, ECG signals, and MPEG videos

 Examples of signals that are considered useless and unwanted, such as noise and interference

- The basic operations on signals are divided into the following two distinct categories, however, a mixed set of
operations from both categories can be simultaneously applied to signals:

 i) operations performed on the dependent variable 𝑔 𝑡

 ii) operations performed on the independent variable 𝑡

Signals & Systems UOIT: Communication Systems (ELEE3130U) 2

 Amplitude-Scaling Operation: 𝒚 𝒕 𝒌𝒈 𝒕

𝑔 𝑡 𝑦 𝑡
𝑘 1.5 → 𝑦 𝑡 1.5𝑔 𝑡

1.5
1
Amplification

𝑡 𝑡
0 1 0 1

𝑦 𝑡 𝑦 𝑡
𝑘 0.5 → 𝑦 𝑡 0.5𝑔 𝑡 𝑘 0.5 → 𝑦 𝑡 0.5𝑔 𝑡

Attenuation
0 1
0.5 𝑡
𝑡 – 0.5
0 1

Signals & Systems UOIT: Communication Systems (ELEE3130U) 3

 Addition and Multiplication Operations

𝑔 𝑡 𝑔 𝑡

2
1
𝑡 𝑡
0 1 2 3 4 0 2

𝑦 𝑡 𝑧 𝑡

𝑦 𝑡 𝑔 𝑡 𝑔 𝑡 𝑧 𝑡 𝑔 𝑡 𝑔 𝑡

e.g., audio mixer e.g., AM signal

3
2 2
1
𝑡 𝑡
0 1 2 3 4 0 1 2

Signals & Systems UOIT: Communication Systems (ELEE3130U) 4

 Differentiation and Integration Operations

𝑔 𝑡 𝑔 𝑡

1 1
1
1 1
𝑡 𝑡
0 0

𝑧 𝑡
𝑦 𝑡
𝑑
𝑦 𝑡 𝑔 𝑡 𝑔 𝑡 𝑧 𝑡 𝑔 𝑡 𝑑𝑡
𝑑𝑡
1

𝑡 𝑡
0 0
–1
e.g., an inductor e.g., a capacitor

Signals & Systems UOIT: Communication Systems (ELEE3130U) 5

 Time-Reversal 𝒚 𝒕 𝒈 𝒕 and Time-Shifting Operations 𝒚 𝒕 𝒈 𝒕 𝝉

𝑔 𝑡 𝑦 𝑡

2 2 𝑦 𝑡 𝑔 𝑡
1 1

𝑡 𝑡
–1 0 1 –1 0 1

𝑧 𝑡 𝑤 𝑡

𝑧 𝑡 𝑔 𝑡 1 2 2 𝑤 𝑡 𝑔 𝑡 1

Time-delayed Time-advanced

𝑡 𝑡
0 2 –2 0

Signals & Systems UOIT: Communication Systems (ELEE3130U) 6

 Time-Scaling Operation 𝒚 𝒕 𝒈 𝜷𝒕

𝑔 𝑡 𝑦 𝑡

2 2 𝑦 𝑡 𝑔 2𝑡
1
1
Compression
𝑡 𝑡
–1 0 1
–0.5 0 0.5

𝑧 𝑡 𝑤 𝑡

𝑡
𝑧 𝑡 𝑔 𝑤 𝑡 𝑔 2𝑡
2 2 2
1 1
Expansion
𝑡 𝑡
–2 0 2 –0.5 0 0.5

Signals & Systems UOIT: Communication Systems (ELEE3130U) 7

 Precedence Rule for Time-Shifting and Time-Scaling Operations: 𝒚 𝒕 𝒈 𝜷𝒕 𝝉

𝑝 𝑡 𝑝 𝑡 𝑝 𝑡 1

1 1

𝑡 𝑡
–1 0 1 0 1 2

𝑝 𝑡 𝑝 2 𝑡 1 𝑝 2𝑡 2
𝑝 𝑡 𝑝 2𝑡 𝑝 2𝑡 1

1 Incorrect:
Correct: 1 First time scaling,
First time shifting, then time shifting
then time scaling
𝑡
𝑡 0 0.5 1 1.5
0 0.5 1

Signals & Systems UOIT: Communication Systems (ELEE3130U) 8

 Continuous/Discrete Value/Time Signals
𝑔 𝑡 𝑔 𝑛
Continuous-value, continuous-time signal
Continuous-value, discrete-time signal

Sampling
𝑡 𝑛

𝑔 𝑛 𝑔 𝑡
Discrete-value, discrete-time signal Discrete-value, continuous-time signal

111 101 100 101 111 110

Encoding
𝑛 𝑡

𝒚-axis (vertical axis) 𝒙-axis (horizontal axis)

• Continuous value: any finite or infinite values • Continuous time: any time, a real number
• Discrete value: a finite number of values • Discrete time: discrete instants of time, an integer
Signals & Systems UOIT: Communication Systems (ELEE3130U) 9
 Analog and Digital Signals

- The terms analog and digital describe the nature of the signal amplitude (value) on the vertical axis, as analog and
digital signals are both continuous-time signals defined for all time 𝑡

- For an analog signal, the dependent variable on the vertical axis can be any real number ∞, ∞ , examples of
analog signals include signals representing light and sound intensity and multi-dimensional position

- For a digital signal, on the other hand, over any finite interval of time, also known as a bit or symbol duration, the
continuous-time waveform belongs to a finite set of possible waveforms, a good example of digital signals is touch-
tone signalling, in which a touch-tone telephone simultaneously transmits down the subscriber line a certain pair of
audible sinusoidal frequencies for each of the twelve possible keys

- The sampling process can transform an analog signal (a continuous-value, continuous-time signal) into a sampled
signal (a continuous-value, discrete-time signal). The quantization (rounding) process can then transform a sampled
signal into a quantized signal (a discrete-value, discrete-time signal), and the encoding process can finally transform
a quantized signal into a digital signal (a discrete-value, continuous-time signal)

Signals & Systems UOIT: Communication Systems (ELEE3130U) 10

 Periodic and Nonperiodic Signals

- A periodic signal repeats itself in time, which satisfies the periodicity condition 𝑔 𝑡 𝑔 𝑡 𝑇 for all time 𝑡,
where 𝑡 starts from minus infinity and continues forever, and 𝑇 is a positive number; the smallest value of 𝑇 that
satisfies this condition is called the period

- A time-shift to the right or to the left by 𝑇 results in exactly the same periodic signal 𝑔 𝑡 and with 𝑝 𝑡 as a time-
limited signal that defines only one period of the signal 𝑔 𝑡 , 𝑔 𝑡 can be analytically expressed as follows:

𝑔 𝑡 𝑝 𝑡 𝑘𝑇

- The reciprocal of the period is called the fundamental frequency of the periodic signal, and its multiples are called
harmonics

- Any signal for which no value of 𝑇 satisfies the above condition is then a non-periodic signal

Signals & Systems UOIT: Communication Systems (ELEE3130U) 11

 Even and Odd Signals

- Even signals are symmetric about the vertical axis, a real signal is said to be an even signal if we have:

𝑔 𝑡 𝑔 𝑡

- Odd signals are symmetric about the origin, a signal is said to be an odd signal if we have:

𝑔 𝑡 𝑔 𝑡

- A signal, which is by definition a single-valued function, cannot be both odd and even and a real physical signal is
rarely an even or an odd signal in the time domain

- An arbitrary real signal can always be decomposed into an even part 𝑔 𝑡 and an odd part 𝑔 𝑡 , as reflected
below:
𝑔 𝑡 𝑔 𝑡 𝑔 𝑡

𝑔 𝑡 𝑔 𝑡
𝑔 𝑡
2
𝑔 𝑡 𝑔 𝑡
𝑔 𝑡
2

Signals & Systems UOIT: Communication Systems (ELEE3130U) 12

 Energy and Power Signals

Total energy of the non-periodic signal 𝑔 𝑡 𝐸 |𝑔 𝑡 | 𝑑𝑡

1
Average power of the non-periodic signal 𝑔 𝑡 𝑃 lim |𝑔 𝑡 | 𝑑𝑡
→ 2𝑆

/
Average power of the periodic signal 𝑔 𝑡 1
𝑃 |𝑔 𝑡 | 𝑑𝑡
𝑇
/

Energy Signal: 𝟎 𝑬 ∞ (finite total energy) and 𝑷 𝟎 (zero average power)

Power Signal: 𝟎 𝑷 ∞ (finite average power) and 𝑬 ∞ (infinite total energy)

Signals & Systems UOIT: Communication Systems (ELEE3130U) 13

 Energy and Power Signals

- A signal cannot be both an energy signal and a power signal, if it is one, it cannot be the other

- A signal with infinite power, such as a unit ramp signal can be neither an energy signal nor a power signal

- No physical signal can have infinite energy or infinite average power

- In signal analysis, according to strict mathematical definitions, signals, such as sinusoidal, have infinite energy

- Every signal observed in real life is an energy signal

- Periodic signals and random signals are usually viewed as power signals

- Signals that are deterministic and non-periodic are usually viewed as energy signals

- Signal energy and power are generally measured in Joules (𝐽) and Watts (𝑊), respectively

- Due to a very wide range of values for energy, the logarithmic scale is used to represent signal energy and power

Signals & Systems UOIT: Communication Systems (ELEE3130U) 14

 Time-Limited and Band-limited Signals

- An (absolutely) time-limited signal is exactly zero outside a finite interval of time

- An (absolutely) band-limited signal has no spectrum (frequency contents) outside a finite band of frequencies

- An absolutely band-limited signal cannot be absolutely time-limited, and vice versa

- A signal may be neither time-limited nor frequency-limited, such as a Gaussian pulse

- All real (physical) signals are time-limited, as such they are not absolutely band-limited

- A real signal may be considered to be band-limited for all practical purposes in the sense that its amplitude
(magnitude) spectrum may have a negligible level above a certain frequency

Signals & Systems UOIT: Communication Systems (ELEE3130U) 15

 System Analysis, Synthesis, and Design

- A system is defined mathematically as a unique transformation that maps an input into an output, and a system—
made up of various inter-related, inter-dependent, and interacting components—may transform a signal into another
signal with properties and characteristics different from those of the input signal

- System analysis is the determination of the output signal when the input signal and the system characteristics (what
fully defines the system) are known and the solution is always unique

- System synthesis is the identification (characterization) of the system when the input and output signals are both
known and there may exist infinitely many solutions or sometimes none at all

- System design includes analytical methods, along with practical heuristic procedures—mainly based on all or a
combination of trial-and-error techniques, parametric simulations, limited experimentations, numerical analysis,
optimization/sub-optimization of subsystems, and above all the experience of the designer

- In design, the goal is to carry out, in an iterative fashion, an array of design trade-offs through a host of system
parameters and variables, so as to accommodate a multitude of conflicting design constraints to the extent possible

Signals & Systems UOIT: Communication Systems (ELEE3130U) 16

 Causal and Non-causal Systems

- A system is said to be causal if it does not respond before the input is applied, in a causal system, the output at any
time depends only on the values of the input signal up to and including that time and does not depend on the future
values of the input

- The output signal of a non-causal system depends on one or more future values of the input signal

- All physically realizable systems are causal, if delay is allowed to be incorporated in a system, then a non-causal
system may become physically realizable

Signals & Systems UOIT: Communication Systems (ELEE3130U) 17

 Linear and Nonlinear Systems

𝑥1 𝑡 𝑦1 𝑡

𝑥1 𝑡 𝑦1 𝑡
Linear
𝑡 System 𝑡

Linearity: 𝑥2 𝑡 𝑦2 𝑡
If 𝑥 𝑡 → 𝑦 𝑡 and 𝑥 𝑡 → 𝑦 𝑡 ,
then 𝛼𝑥 𝑡 𝛽𝑥 𝑡 → 𝛼𝑦 𝑡 𝛽𝑦 𝑡 𝑥2 𝑡 Linear 𝑦2 𝑡
where 𝛼 0 and 𝛽 0 𝑡 System 𝑡

𝛼𝑥 𝑡 𝛼𝑦 𝑡
𝛽𝑥2 𝑡 𝛽𝑦2 𝑡
𝛼𝑥 𝑡 𝛼𝑦 𝑡
𝛽𝑥2 𝑡 Linear 𝛽𝑦2 𝑡
𝑡 System 𝑡

Signals & Systems UOIT: Communication Systems (ELEE3130U) 18

 Time-Invariant and Time-Varying Systems

𝑥 𝑡 𝑦 𝑡

Time- 𝑦 𝑡
𝑥 𝑡
Invariant
System
Time-Invariance:
𝑡 𝑡
If 𝑥 𝑡 → 𝑦 𝑡 ,
then 𝑥 𝑡 𝜏 → 𝑦 𝑡 𝜏
where 𝜏 0 𝑥 𝑡 𝑦 𝑡

Time- 𝑦 𝑡 𝜏
𝑥 𝑡 𝜏
Invariant
System
𝜏 𝑡 𝜏 𝑡

Signals & Systems UOIT: Communication Systems (ELEE3130U) 19

 Linear Time-Invariant (LTI) Systems

- In a linear time-invariant (LTI) system, both linearity and time-invariance conditions must be satisfied and the
input-output relationship is straightforward

- In an LTI system, the system impulse response, i.e., the system output when the system input is an impulse (a
signal with extremely short duration and extremely large amplitude), can completely characterize the system and
provide all relevant information to describe the system behavior for any input

Signals & Systems UOIT: Communication Systems (ELEE3130U) 20

 Sinusoidal Signals: Characteristics

𝐴 cos 2𝜋𝑓 𝑡 𝜑 𝑨𝟐 cos 2𝜋𝑓 𝑡 𝜑

All sinusoidal signals are related

to one another through basic 𝑡 𝑡

operations on signals, that is,

when there are two different
sinusoidal signals whose
amplitudes, frequencies 𝐴 cos 2𝜋𝑓 𝑡 𝝋𝟐
𝐴 cos 2𝜋𝒇𝟐 𝑡 𝜑
(periods), and initial phases are
all different, one is basically an
amplitude-scaled, time-shifted,
time-scaled version of the other.
𝑡 𝑡

Signals & Systems UOIT: Communication Systems (ELEE3130U) 21

 Sinusoidal Signals: Benefits & Applications
- The sinusoidal signal is regarded as the most dominant signal in communications

- Nature itself is characteristically sinusoidal: sound and light waves, seismic waves, ripples of the ocean surface
- Man-made devices can have sinusoidal motions, such as sinusoidal variation in the motion of a pendulum, the
oscillation of a spring around the equilibrium, and the vibration of a string
- A simple physical example in electric circuits includes an ideal charged capacitor connected in parallel to an ideal
inductor, where the voltage across the capacitor in the steady state is a sinusoidal signal

- A sinusoidal signal is easy to generate, transmit and distribute and it is the form of electricity generated throughout
the world and supplied to consumers

- The derivative, integral, and power functions of a sinusoid as well as the time-scaled and time-shifted versions of a
sinusoid are themselves sinusoids
- A sinusoidal signal is the only periodic signal where it retains its wave shape when added to another sinusoidal
signal of the same frequency with arbitrary initial phase and amplitude
- Through Fourier analysis, any practical non-sinusoidal periodic signal can be represented by the sum of sinusoids,
each with its own amplitude, frequency, and initial phase

Signals & Systems UOIT: Communication Systems (ELEE3130U) 22

 Sinusoidal Signals in LTI Systems

𝑥 𝑡 𝑨 cos 2𝜋𝒇𝟎 𝑡 𝝋 𝑦 𝑡 𝑩 cos 2𝜋𝒇𝟎 𝑡 𝜽

𝑥 𝑡 Linear 𝑦 𝑡
𝑡 Time-Invariant 𝑡
(LTI) System

In an LTI system, for a sinusoidal input signal, the output is also a sinusoidal
signal; however, the amplitude and initial phase of the output signal may be
different from those of the input signal, but their frequencies remain the same:
LTI systems hold the property of sinusoidal signal in, sinusoidal signal out

Signals & Systems UOIT: Communication Systems (ELEE3130U) 23

 Relationship between Time-Domain and Frequency-Domain Properties

Time-domain property Fourier representation Frequency-domain property

Continuous & periodic Fourier series Discrete & nonperiodic

Continuous & nonperiodic Fourier transform Continuous & nonperiodic

Discrete & periodic Discrete-time Fourier series Discrete & periodic

Discrete & nonperiodic Discrete-time Fourier transform Continuous & periodic

Signals & Systems UOIT: Communication Systems (ELEE3130U) 24

 Fourier Series

- Fourier series is an expansion of a periodic signal in terms of summing of an infinite number of sinusoids or
complex exponentials, as any periodic signal of practical nature can be approximated by adding up sinusoids with
the properly chosen frequencies, amplitudes and initial phases

- To apply the Fourier series representation to an arbitrary periodic signal 𝑔 𝑡 with the period 𝑇 , it is sufficient, but
not strictly necessary, that the following conditions, known as Dirichlet’s conditions, are satisfied:
i) The periodic signal is single-valued, with a finite number of maxima and minima and at most a finite number
of discontinuities within the interval of one period
ii) The signal is absolutely integrable over the interval of one period, that is,
⁄
| 𝑔 𝑡 |𝑑𝑡 ∞
⁄

- Note that a periodic pulse train does not satisfy these conditions; nevertheless it has a Fourier series representation.
The signals usually encountered in communication systems meet the Dirichlet’s conditions, as the physical
possibility of a periodic signal is a sufficient condition for the existence of a converging Fourier series

Signals & Systems UOIT: Communication Systems (ELEE3130U) 25

 Fourier Series

Trigonometric Fourier Series Complex Exponential Fourier Series

2𝜋𝑛𝑡 2𝜋𝑛𝑡 𝑇 𝑇
𝑔 𝑡 𝑎 𝑎 cos 𝑏 sin , 𝑡
𝑇 𝑇 2 2

1 ⁄ 𝑗2𝜋𝑛𝑡 𝑇 𝑇
𝑎 𝑔 𝑡 𝑑𝑡 𝑔 𝑡 𝑐 exp , 𝑡
𝑇 ⁄
𝑇 2 2

⁄ /
2 2𝜋𝑛𝑡 1 𝑗2𝜋𝑛𝑡
𝑎 𝑔 𝑡 cos 𝑑𝑡, 𝑛 1, 2, … 𝑐 𝑔 𝑡 exp 𝑑𝑡, 𝑛 0, 1, 2, …
𝑇 ⁄ 𝑇 𝑇 / 𝑇

⁄
2 2𝜋𝑛𝑡
𝑏 𝑔 𝑡 sin 𝑑𝑡, 𝑛 1, 2, …
𝑇 ⁄ 𝑇

Signals & Systems UOIT: Communication Systems (ELEE3130U) 26

 Fourier Series: An Example

𝑔 𝑡

|𝑐 |

𝑡
𝑇 𝑇 0 𝑇 𝑇
∡𝑐
2 2

𝜋 1
𝑇
1 1 2 3 4 5
𝑇 𝑇 𝑇 𝑇 𝑇 𝑇
∎∎∎∎∎ ∎∎∎∎∎ ∎∎∎∎∎∎∎∎∎ ∎∎∎∎∎ ∎∎∎∎∎ 𝑓
5 4 3 2 1 0
𝑇 𝑇 𝑇 𝑇 𝑇
𝑓
5 4 3 3 1 0 1 2 3 4 5 𝜋
𝑇 𝑇 𝑇 𝑇 𝑇 𝑇 𝑇 𝑇 𝑇 𝑇

Signals & Systems UOIT: Communication Systems (ELEE3130U) 27

 Fourier Transform

- To represent the frequency components of a non-periodic deterministic signal, the Fourier transform is employed
- The spectrum consists of a continuous range of frequencies, and may include a continuum of frequencies ranging
from ∞ to ∞

- A signal is Fourier transformable if it satisfies the Dirichlet’s conditions. These conditions, which are sufficient but
not strictly necessary, are as follows:
i) The signal is single-valued, with a finite number of maxima and minima and a finite number of discontinuities
in any finite time interval.
ii) The signal is absolutely integrable over the entire time, that is,

| 𝑔 𝑡 |𝑑𝑡 ∞

- Note that a sinc function does not satisfy Dirichlet’s condition, as it is not absolutely integrable, yet it possesses the
Fourier transform, the physical existence of a signal, i.e., a signal with finite energy, is a sufficient condition for the
existence of its Fourier transform

Signals & Systems UOIT: Communication Systems (ELEE3130U) 28

 Fourier Transform

𝐺 𝑓 𝐹𝑔 𝑡 𝑔 𝑡 𝑒 𝑑𝑡

𝑔 𝑡 𝐹 𝐺 𝑓 𝐺 𝑓 𝑒 𝑑𝑓

𝑔 𝑡 ↔𝐺 𝑓 𝐺 𝑓 𝑒

When 𝑔 𝑡 is a real signal, its Fourier transform has the following property:

𝐺 𝑓 𝐺 𝑓 , i. e. , magnitude spectrum is an even function

∗
𝐺 𝑓 𝐺 𝑓 →
𝜃 𝑓 𝜃 𝑓 , i. e. , phase spectrum is an odd function

Signals & Systems UOIT: Communication Systems (ELEE3130U) 29

 Fourier Transform Properties
1 Linearity 𝑔 𝑡 ↔ 𝐺 𝑓 & 𝑔 𝑡 ↔ 𝐺 𝑓 ⟹ 𝛼𝑔 𝑡 𝛽𝑔 𝑡 ↔ 𝛼𝐺 𝑓 𝛽𝐺 𝑓

2 Time-frequency duality 𝑔 𝑡 ↔𝐺 𝑓 ⟹𝐺 𝑡 ↔𝑔 𝑓
1 𝑓
3 Time scaling 𝑔 𝑡 ↔ 𝐺 𝑓 ⟹ 𝑔 𝛼𝑡 ↔ 𝐺
𝛼 𝛼
4 Time reversal 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝑡 ↔𝐺 𝑓

5 Time shifting 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝑡 𝑡 ↔𝐺 𝑓 𝑒

6 Frequency translation 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑒 𝑔 𝑡 ↔𝐺 𝑓 𝑓

7 Linear modulation 𝑔 𝑡 ↔ 𝐺 𝑓 ⟹ 𝑔 𝑡 cos 2𝜋𝑓 𝑡 ↔ 0.5 𝐺 𝑓 𝑓 𝐺 𝑓 𝑓

8 Time convolution 𝑔 𝑡 ↔𝐺 𝑓 &𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝑡 ∗𝑔 𝑡 ↔𝐺 𝑓 𝐺 𝑓

9 Time multiplication 𝑔 𝑡 ↔𝐺 𝑓 &𝑔 𝑡 ↔𝐺 𝑓 ⟹𝑔 𝑡 𝑔 𝑡 ↔𝐺 𝑓 ∗𝐺 𝑓

Conjugate Functions 𝑔 𝑡 ↔ 𝐺 𝑓 ⟹ 𝑔∗ 𝑡 ↔ 𝐺 ∗ 𝑓

Time autocorrelation 𝑔 𝑡 ↔𝐺 𝑓 &𝑅 𝜏 𝑔∗ 𝑡 𝑔 𝑡 𝜏 𝑑𝑡 ⟹ 𝑅 𝜏 ↔ |𝐺 𝑓 |

𝑑
Time differentiation 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝑡 ↔ 𝑗2𝜋𝑓𝐺 𝑓
𝑑𝑡
𝐺 𝑓 𝐺 0
Time integration 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝜏 𝑑𝜏 ↔ 𝛿 𝑓
𝑗2𝜋𝑓 2
𝑗 𝑑
𝑛th moments 𝑔 𝑡 ↔ 𝐺 𝑓 ⟹ 𝑡 𝑔 𝑡 𝑑𝑡 ↔ 𝐺 𝑓
2𝜋 𝑑𝑓
Time average 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝑡 𝑑𝑡 𝐺 0

Frequency average 𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝐺 𝑓 𝑑𝑓 𝑔 0

Parseval’s relation 𝑔 𝑡 ↔𝐺 𝑓 &𝑔 𝑡 ↔𝐺 𝑓 ⟹ 𝑔 𝑡 𝑔∗ 𝑡 𝑑𝑡 𝐺 𝑓 𝐺 ∗ 𝑓 𝑑𝑓

Signals & Systems UOIT: Communication Systems (ELEE3130U) 30
 Fourier Transform: An Example

𝑔 𝑡 𝐺 𝑓
𝐴𝑇

1
𝑡 𝑇
𝑇 𝑇 𝑓
0 0 1
2 2 𝑇

|𝐺 𝑓 |
∡𝐺 𝑓
𝐴𝑇
𝜋
1 2 4
𝑇 𝑇 𝑇
𝑓
4 2 10
𝑇 𝑇 𝑇
𝑓
5 3 1 0 1 3 5 𝜋
𝑇 𝑇 𝑇 𝑇 𝑇 𝑇

Signals & Systems UOIT: Communication Systems (ELEE3130U) 31

 Fourier Transform: Examples

𝑔 𝑡 𝐺 𝑓

𝑡 𝑓
0 0

𝑔 𝑡 𝐺 𝑓

𝑡 𝑓
0 0

Signals & Systems UOIT: Communication Systems (ELEE3130U) 32

 Fourier Transform: Examples

𝑔 𝑡 𝐺 𝑓

𝑡 𝑓
0 𝑓 0 𝑓

1
𝑓
𝑔 𝑡 𝑗𝐺 𝑓

𝑡 𝑓
0 𝑓
0 𝑓
1
𝑓

Signals & Systems UOIT: Communication Systems (ELEE3130U) 33

 Time & Frequency Relations

- A signal can be fully defined either in the time domain or in the frequency domain, but not in both

- The time duration and frequency spectrum of a signal are inversely related, as a contraction in one domain results in
an expansion in the other domain

- Bandwidth is a very important measure of performance in digital communication systems

- The bandwidth of a signal reflects a range of positive frequencies with significant spectral content. All practical
signals are time-limited, their spectra thus extend to infinity

- Shifting the spectral content of a lowpass signal by a sufficiently large frequency—through a process known as
modulation—to produce its corresponding bandpass signal has the effect of increasing the bandwidth of the signal

- The term bandwidth, denoted by 𝐵, may be defined, as the difference (in Hz) between two nominal frequencies, 𝑓
and 𝑓 , i.e., 𝐵 𝑓 𝑓 , where 𝑓 𝑓 0, and 𝑓 and 𝑓 are determined by one of the following definitions of
bandwidth

Signals & Systems UOIT: Communication Systems (ELEE3130U) 34

 Bandwidth Definitions

Absolute Bandwidth – Absolute bandwidth provides a theoretical definition. Assuming the spectrum is zero beyond
𝑓 𝑓 0, we have 𝐵 𝑓 𝑓 .

3-dB (or Half-Power) Bandwidth – The 3-dB (or half-power) bandwidth is one of the widely-used definitions.
Assuming the maximum (peak) value of the magnitude spectrum occurs at a frequency inside the band 𝑓 , 𝑓 , we have
𝐵 𝑓 𝑓 , where the magnitude spectrum at any frequency inside the band falls no lower than 1/ 2 times the peak
value. The signal power at 𝑓 or 𝑓 is thus 3 (≅ 20 log 1/ 2)) dB lower than the peak signal power.

Fractional-Power Bandwidth – The occupied bandwidth, as adopted by FCC, defines a band of frequencies with 99%
of the signal power, where 0.5% of the signal power is above the upper frequency limit and 0.5% of the signal power is
below the lower frequency limit. This definition is primarily focused on passband signals and filters.

Null-to-Null Bandwidth – The null-to-null (zero-crossing) bandwidth is a commonly-used definition. For bandpass
signals, when the magnitude spectrum has a main lobe (the lobe with the peak value) bounded by nulls (the frequencies
𝑓 and 𝑓 at which the magnitude spectrum is zero), we have 𝐵 𝑓 𝑓 , as the main lobe is centered on the frequency
𝑓 𝑓 𝑓 /2.

Signals & Systems UOIT: Communication Systems (ELEE3130U) 35

 Distortionless Channel

𝐻 𝑓 𝛽 𝑓

𝑘 2𝜋𝑡

f f
0 0

In a distortionless transmission, the input and output signals in the time domain have identical shapes, except for a
possible change of amplitude and a constant delay, for the input signal 𝑥 𝑡 , the output signal 𝑦 𝑡 is defined by:

𝐻 𝑓 𝑘
𝑌 𝑓
𝑦 𝑡 𝑘𝑥 𝑡 𝑡 →𝑌 𝑓 𝑘𝑋 𝑓 𝑒 →𝐻 𝑓 𝑘𝑒 →
𝑋 𝑓
∡𝐻 𝑓 𝛽 𝑓 2𝜋𝑓𝑡

where the constant 𝑘 1 reflects the transmission attenuation, and the constant 𝑡 0 accounts for the
transmission delay

Signals & Systems UOIT: Communication Systems (ELEE3130U) 36

 Nonlinear Distortion

- A nonlinear system cannot be described by a transfer function, and to evaluate the nonlinear distortion, the
common procedure is to approximate the input-output relation by a power series of the input 𝑥 𝑡 :

𝑦 𝑡 𝑎 𝑥 𝑡 𝑎 𝑥 𝑡 𝑎 𝑥 𝑡 …→𝑌 𝑓 𝑎 𝑋 𝑓 𝑎 𝑋 𝑓 ∗𝑋 𝑓 𝑎 𝑋 𝑓 ∗𝑋 𝑓 ∗𝑋 𝑓 …

- When 𝑥 𝑡 is band-limited to 𝑊 Hz, 𝑥 𝑡 is band-limited to 2𝑊 Hz, 𝑥 𝑡 is band-limited to 3𝑊 Hz, and so on

and so forth, the nonlinearities thus create new output frequency components that are not present in the input, with
appropriate filtering, these out-of-band frequency components (|𝑓| 𝑊) can be suppressed; however, the 2nd, 3rd
and the higher order nonlinearities all produce undesirable in-band frequency components (|𝑓| 𝑊), which lie in
the frequency band of interest, cannot be removed, we have nonlinear distortion

Signals & Systems UOIT: Communication Systems (ELEE3130U) 37

 Magnitude and Phases Responses of Ideal Low-Pass Filter and Band-Pass Filter

|𝐻 𝑓 |

∡𝐻 𝑓

𝑊
𝑓 𝑓
𝑊 0 𝑊 𝑊 0

|𝐻 𝑓 |

∡𝐻 𝑓

𝑓
𝑓 𝑓
𝑓 0 𝑓 𝑊 𝑓 𝑓 𝑊 𝑓 0

Signals & Systems UOIT: Communication Systems (ELEE3130U) 38

 Tolerance Diagram for a Physically-Realizable Analog (Continuous-Time) LPF

- The magnitude response in the passband must lie between 1 and 1– 𝜀 for 0 |𝑓| 𝑓 , where 𝑓 is the passband
cut-off frequency and 0 ≪ 𝜀 1 is a tolerance parameter
- The magnitude response in the stopband must not exceed 0 𝜀 ≪ 1 for |𝑓| 𝑓 , where 𝑓 is the stopband-edge
frequency
- The magnitude response in the transition bandwidth must lie between 1– 𝜀 and 𝜀 for 𝑓 𝑓 𝑓
- The fluctuation in the passband is 20 log 1 𝜀 dB and the fluctuation in the stopband is 20 log 𝜀 dB

After specifying the filter parameters 𝜀 , 𝜀 , 𝑓 , 𝑓 , the design |𝐻 𝑓 |

of filter consists of the following two steps, where they can be Passband
implemented in different ways, thus highlighting the non- 1
Transition band
uniqueness of the filter design problem: 1 𝜖
i) The approximation of a frequency response (i.e.,
magnitude response or phase response or both) by a
realizable transfer function that represents a stable and
causal system, and Stopband
ii) The realization of the approximating transfer function by 𝜖
f
a physical system 0 𝑓 𝑓

Signals & Systems UOIT: Communication Systems (ELEE3130U) 39