Signal Processing EE2S31

Digital Signal Processing - Lecture 1: Introduction

Borbala Hunyadi

Delft University of Technology, The Netherlands

25 April 2023
Outline

● Course Organization
● Introduction
● Applications
● Recap: Ideal sampling and reconstruction

2 / 32
Course organization - DSP track

● Information
● Website: general overview
● Brightspace: more detailed information, quiz, forum
● Organization
● DSP 1x a week(±) on Tuesday or Thursdays
● Exam comprises both tracks

3 / 32
Study materials - DSP track

● Theory
● Lectures
● Book (Proakis, Manolakis: Digital Signal Processing)
● Collegerama videos
● Important notes:
1. Studying the slides is not sufficient, you need to read the book!
2. Attending lectures is important, we will solve more exercises
during lectures compared to last year
● Practice
● Brightspace Quiz (easy)
● Exercises from book (more advanced)
● Past exams on website

4 / 32
Contents Digital Signal Processing Track

● Lecture 1: Introduction
● Lecture 2: Non-ideal sampling and reconstruction
● Lecture 3: Sampling in frequency domain: DFT
● Lecture 4: Spectral analysis and filtering using DFT
● Lecture 5: Efficient implementation of DFT: FFT + Exercises
● Mid-term exam (Lectures 1-4)
● Lecture 6: A/D conversion: quantization, coding, sigma-delta
● Lecture 7: Round-off effects and filter structures
● Lecture 8: Multirate signal processing
● Exercise session
● Final exam (Lectures 5-8)

5 / 32
Introduction

What is a signal?

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier
2 mechanical: acceleration of a car

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier
2 mechanical: acceleration of a car
3 acoustic: air pressure measured by microphone

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier
2 mechanical: acceleration of a car
3 acoustic: air pressure measured by microphone
4 biological: body temperature

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier
2 mechanical: acceleration of a car
3 acoustic: air pressure measured by microphone
4 biological: body temperature
Classification of signals

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier
2 mechanical: acceleration of a car
3 acoustic: air pressure measured by microphone
4 biological: body temperature
Classification of signals
1 analog (continuous-time) vs digital (discrete time)

6 / 32
Introduction

What is a signal?
Any measurable quantity that conveys information

Examples
1 electrical: voltage output of amplifier
2 mechanical: acceleration of a car
3 acoustic: air pressure measured by microphone
4 biological: body temperature
Classification of signals
1 analog (continuous-time) vs digital (discrete time)
2 unquantized (continuous-amplitude) vs quantized (discrete
amplitude)

6 / 32
Introduction

Digital Signal Processing

Processing of analog signals by means of discrete-time operations
implemented on digital hardware

7 / 32
Analog vs Digital Signal Processing

⇓ instead

Pro: Contra:
● accuracy ● extra complexity
● flexibility ● limited bandwidth
● ease of data storage ● quantization effects

8 / 32
Digital signal processing

sampling filtering reconstruction

quantization spectral analysis

9 / 32
DSP applications

● Digital communication
● Audio signal processing
● Speech signal processing
● Image Processing
● Medical applications

10 / 32
DSP applications (1)

Mobile communication:

11 / 32
DSP applications (2)

Analogy: audio information processing in the brain 12 / 32

DSP applications (2)

Connection with DSP

Analogy: audio information processing in the brain 12 / 32

DSP applications (2)

Connection with DSP

● Speech coding for cochlear implant

Analogy: audio information processing in the brain 12 / 32

DSP applications (2)

Connection with DSP

● Speech coding for cochlear implant
● EEG signal processing

Analogy: audio information processing in the brain 12 / 32

DSP applications (2)

Connection with DSP

● Speech coding for cochlear implant
● EEG signal processing
● Brain-computer interface

Analogy: audio information processing in the brain 12 / 32

DSP applications (2)

Connection with DSP

● Speech coding for cochlear implant
● EEG signal processing
● Brain-computer interface
● Early detection of
developmental disorders

Analogy: audio information processing in the brain 12 / 32

DSP Applications (3)
EEG processing for epileptic seizure detection:

13 / 32
DSP Applications (3):
Seizure detection pipeline:

14 / 32
Quiz

Go to:

www.kahoot.it

15 / 32
Quiz - question 1

The spectrum of a continuous-time signal is depicted above.

Which one of the figures below represent the spectrum of the sampled
version of the signal?

16 / 32
Quiz - question 2

Aliasing occurs when we....

a. oversample a signal, i.e. with a sampling rate Fs >> 2FH
b. sample an aperiodic signal
c. sample below the Nyquist rate

17 / 32
Quiz - question 3

The reconstruction of an analog signal from its samples can happen

using
a. a highpass filter in frequency domain
b. a sinc function in the time domain
c. the inverse fourier transform

18 / 32
Ideal sampling and reconstruction

Under which conditions can

we reconstruct xa (t)?

19 / 32
Ideal sampling and reconstruction

Under which conditions can

we reconstruct xa (t)?

To answer this question, we

will investigate the form of
the digital signal in the
frequency domain.

19 / 32
Ideal sampling and reconstruction

20 / 32
Ideal sampling and reconstruction

Can we express the DTFT of the sampled signal using the FT of the analog signal?

20 / 32
Recap: Fourier Transform in continuous and discrete time
FT DTFT
∞
∞ X (ω) = ∑ x[n]e −jωn
X (F ) = ∫ x(t)e −j2πFt
dt n=−∞
−∞

Inverse FT Inverse DTFT

∞
1 π
x(t) = ∫ X (F )e j2πFt dF x[n] = ∫ X (ω)e dω
jωn
−∞ 2π −π

F [Hz]: frequency f [cycles/sample]: normalized frequency

Ω [radians/s]: angular frequency ω [radians/sample]: normalized angular frequency
Ω = 2πF ω = 2πf

Ω = ω/T
F = f ⋅ Fs

21 / 32
DTFT of digital vs FT of analog signal

Let’s express the digital signal in terms of its spectrum using the inverse
DTFT!

1 sF
2 1 2
x[n] = ∫ X (f )e j2πfn df = ∫ X (F )e j2πF /Fs n dF
− 12
Fs
Fs − 2

22 / 32
DTFT of digital vs FT of analog signal

Let’s express the digital signal in terms of its spectrum using the inverse
DTFT!

1 sF
2 1 2
x[n] = ∫ X (f )e j2πfn df = ∫ X (F )e j2πF /Fs n dF
− 12
Fs
Fs − 2

Inverse transform
f = ω/2π

22 / 32
DTFT of digital vs FT of analog signal

Let’s express the digital signal in terms of its spectrum using the inverse
DTFT!

1 sF
2 1 2
x[n] = ∫ X (f )e j2πfn df = ∫ X (F )e j2πF /Fs n dF
− 12
Fs
Fs − 2

Inverse transform
Change of variables:
f = ω/2π
f = FFs and
df /dF = 1/Fs

22 / 32
DTFT of digital vs FT of analog signal

Let’s express samples of the analog signal in terms of its spectrum using
the inverse FT!
Inverse transform

F
∞ ∞ kFs + 2s
∞
j2πFnTs j2πF /Fs n
xa (nTs ) = ∫ Xa (F )e dF = ∫ Xa (F )e dF = ∑ ∫ Xa (F )e j2πF /Fs n dF
k=−∞ F
−∞ −∞ kFs − 2s

Fs Fs
∞ 2 2 ∞
j2π(F −kFs )/Fs n j2πF /Fs n
= ∑ ∫ Xa (F − kFs )e dF = ∫ ∑ Xa (F − kFs )e dF
k=−∞ F F k=−∞
− 2s − 2s

23 / 32
DTFT of digital vs FT of analog signal

Let’s express samples of the analog signal in terms of its spectrum using
the inverse FT!
1
Inverse transform Ts = Fs

F
∞ ∞ kFs + 2s
∞
j2πFnTs j2πF /Fs n
xa (nTs ) = ∫ Xa (F )e dF = ∫ Xa (F )e dF = ∑ ∫ Xa (F )e j2πF /Fs n dF
k=−∞ F
−∞ −∞ kFs − 2s

Fs Fs
∞ 2 2 ∞
j2π(F −kFs )/Fs n j2πF /Fs n
= ∑ ∫ Xa (F − kFs )e dF = ∫ ∑ Xa (F − kFs )e dF
k=−∞ F F k=−∞
− 2s − 2s

23 / 32
DTFT of digital vs FT of analog signal

Let’s express samples of the analog signal in terms of its spectrum using
the inverse FT!
1 divide the infinite interval to a sum
Inverse transform Ts = Fs of Fs long intervals

F
∞ ∞ kFs + 2s
∞
j2πFnTs j2πF /Fs n
xa (nTs ) = ∫ Xa (F )e dF = ∫ Xa (F )e dF = ∑ ∫ Xa (F )e j2πF /Fs n dF
k=−∞ F
−∞ −∞ kFs − 2s

Fs Fs
∞ 2 2 ∞
j2π(F −kFs )/Fs n j2πF /Fs n
= ∑ ∫ Xa (F − kFs )e dF = ∫ ∑ Xa (F − kFs )e dF
k=−∞ F F k=−∞
− 2s − 2s

23 / 32
DTFT of digital vs FT of analog signal

Let’s express samples of the analog signal in terms of its spectrum using
the inverse FT!
1 divide the infinite interval to a sum
Inverse transform Ts = Fs of Fs long intervals

F
∞ ∞ kFs + 2s
∞
j2πFnTs j2πF /Fs n
xa (nTs ) = ∫ Xa (F )e dF = ∫ Xa (F )e dF = ∑ ∫ Xa (F )e j2πF /Fs n dF
k=−∞ F
−∞ −∞ kFs − 2s

Fs Fs
∞ 2 2 ∞
j2π(F −kFs )/Fs n j2πF /Fs n
= ∑ ∫ Xa (F − kFs )e dF = ∫ ∑ Xa (F − kFs )e dF
k=−∞ F F k=−∞
− 2s − 2s

change of variables
by shifting the inter-
val by kFs

23 / 32
DTFT of digital vs FT of analog signal

Let’s express samples of the analog signal in terms of its spectrum using
the inverse FT!
1 divide the infinite interval to a sum
Inverse transform Ts = Fs of Fs long intervals

F
∞ ∞ kFs + 2s
∞
j2πFnTs j2πF /Fs n
xa (nTs ) = ∫ Xa (F )e dF = ∫ Xa (F )e dF = ∑ ∫ Xa (F )e j2πF /Fs n dF
k=−∞ F
−∞ −∞ kFs − 2s

Fs Fs
∞ 2 2 ∞
j2π(F −kFs )/Fs n j2πF /Fs n
= ∑ ∫ Xa (F − kFs )e dF = ∫ ∑ Xa (F − kFs )e dF
k=−∞ F F k=−∞
− 2s − 2s

change of variables
periodicity of the complex exponen-
by shifting the inter-
tial: e j2π(F −kFs )/Fs = e j2πnF /Fs
val by kFs

23 / 32
DTFT of digital vs FT of analog signal

Let’s express samples of the analog signal in terms of its spectrum using
the inverse FT!
1 divide the infinite interval to a sum
Inverse transform Ts = Fs of Fs long intervals

F
∞ ∞ kFs + 2s
∞
j2πFnTs j2πF /Fs n
xa (nTs ) = ∫ Xa (F )e dF = ∫ Xa (F )e dF = ∑ ∫ Xa (F )e j2πF /Fs n dF
k=−∞ F
−∞ −∞ kFs − 2s

Fs Fs
∞ 2 2 ∞
j2π(F −kFs )/Fs n j2πF /Fs n
= ∑ ∫ Xa (F − kFs )e dF = ∫ ∑ Xa (F − kFs )e dF
k=−∞ F F k=−∞
− 2s − 2s

change of variables
periodicity of the complex exponen- interchange integral
by shifting the inter-
tial: e j2π(F −kFs )/Fs = e j2πnF /Fs and summation
val by kFs

23 / 32
DTFT of digital vs FT of analog signal

Fs Fs
2 2
1 ∞
x[n] = X (F )e j2πF /Fs n dF = ∫ ∑ Xa (F − kFs )e
j2πF /Fs n
dF = xa (nTs )
Fs ∫ k=−∞
F F
− 2s − 2s
∞
X (F ) = Fs ∑ Xa (F − kFs )
k=−∞

24 / 32
Ideal sampling and reconstruction

25 / 32
Ideal sampling and reconstruction

25 / 32
Spectrum of the sampled signal

26 / 32
Spectrum of the sampled signal

26 / 32
Spectrum of the sampled signal

Sampling theorem
If the signal is bandlimited, it is possible to reconstruct the original
signal from the samples, provided that the sampling rate is at least twice
the highest frequency contained in the signal (i.e. the Nyquist rate).

26 / 32
Ideal reconstruction in frequency domain

Define an ideal low-pass filter G (f ):

⎧
⎪
⎪F −1 , if ∣F ∣ ≤ F2s
G (F ) = ⎨ s
⎪
⎪
⎩0, otherwise

Apply it G (F ) to X (F ):

Xa (F ) = G (F )X (F )

27 / 32
Ideal reconstruction in time domain

Define an ideal interpolator g (t):

sin(πt/T )
g (t) =
πt/T

Apply it to x[n]:
∞
xa (t) = ∑ x[n]g (t − nT )
n=−∞

28 / 32
The ideal interpolator

x[n]

29 / 32
The ideal interpolator

sin(π/Ts )(t − 1Ts )

x[1]
(π/Ts )(t − 1Ts )

29 / 32
The ideal interpolator

sin(π/Ts )(t − 2Ts )

x[2]
(π/Ts )(t − 2Ts )

29 / 32
The ideal interpolator

sin(π/Ts )(t − 3Ts )

x[3]
(π/Ts )(t − 3Ts )

29 / 32
The ideal interpolator

sin(π/Ts )t ⎧
⎪
⎪F −1 , if ∣F ∣ ≤= F2s
g (t) = G (f ) = ⎨ s
(π/Ts )t ⎪
⎪
⎩0, otherwise

29 / 32
The ideal interpolator

sin(π/Ts )t ⎧
⎪
⎪F −1 , if ∣F ∣ ≤= F2s
g (t) = G (f ) = ⎨ s
(π/Ts )t ⎪
⎪
⎩0, otherwise

Ideal interpolator
In time domain, the ideal interpolator takes the linear combination of
the time-shifted versions of the interpolation function g(t) weighted
with the sample values x[n]. In frequency domain, the ideal interpolator
scales the spectum with Fs−1 and removes frequencies ∣F ∣ > Fs /2, i.e.
removes the discrete-time spectral periodicity.

29 / 32
Ideal sampling and reconstruction

30 / 32
Ideal sampling and reconstruction

30 / 32
Non-ideal sampling and reconstruction

1
3

31 / 32
Sampling and reconstruction in practice

1 Delta pulse train for sampling: non-zero duration in practice

2 Signals often are non-low pass, non-bandlimited
● How to sample non-bandlimited signals?
● How to sample bandpass signals?
3 Sinc interpolation in practice is not possible: nondeterministic,
infinite length

32 / 32