Вы находитесь на странице: 1из 48

ECE 3331b

Introduction to Signal Processing


Lecture 1

Instructor: Dr. Ilia G. Polushin

Department of Electrical & Computer Engineering

Faculty of Engineering

The University of Western Ontario

Winter 2017
Objectives of the Course:

To present an introduction to the basic analysis tools and techniques for


Digital Signal Processing (DSP):

Converting analog signals to digital form


Reconstruction of analog signals from quantized digital signals
Mathematical tools for analysis and processing of discrete-
time signals

To prepare students for more advanced topics in digital signal processing


covered in ECE4429a/b.
Content:
Introduction: Signals, Systems and Signal Processing, Classification of Signals, The Concept
of Frequency in Continuous-Time and Discrete-Time Signals, Analog-to-Digital and Digital-to-
Analog Conversion.

Discrete-Time Signals and Systems: Discrete-Time Signals, Discrete-Time Systems, Anal-


ysis of Discrete-Time Linear Time-Invariant (LTI) Systems, Discrete-Time Systems Described
by Dierence Equations, Implementation of Discrete-Time Systems

The z-Transform and its Application to the Analysis of LTI Systems: The z-Transform,
Properties of the z-Transform, Rational z-Transforms, Inversion of the z-Transform, Analysis of
LTI Systems in the z-Domain, The One-sided z-Transform

Frequency Analysis of Signals: Frequency Analysis of Continuous-Time Signals Frequency


Analysis of Discrete-Time Signals Properties of the Fourier Transform for Discrete-Time Signals

Frequency-Domain Analysis of LTI Systems: Frequency-Domain Characteristics of LTI


Systems, Frequency Response of LTI Systems

The Discrete Fourier Transform: Frequency-Domain Sampling: The Discrete Fourier Trans-
form (DFT), Properties of the DFT

The Fast Fourier Transform


Feature:

Is this yet another course that is heavy on applied maths?


YES!

Todays lecture is the only one without maths


Textbook:
Required Textbook: John G. Proakis & Dimitris K. Manolakis,
Digital Signal Processing, Prentice Hall, 4th Edition, 2007.

Recommended Reference: Dimitris G. Manolakis & Vinay K. Ingle,


Applied Digital Signal Processing, Cambridge University Press, 2011.

Lecture slides will be posted on WebCT


Instructor:

Name: Ilia G. Polushin

Associate Professor,
Department of Electrical and Computer Engineering,
University of Western Ontario

Office: TEB 357

E-mail: ipolushi@uwo.ca

Phone: 519-661-2111 ext. 88575

Area of Research: Robotics & Control


Schedule: Lectures & Office Hours
Lectures:

Wednesday 8:30 a.m. - 9:30 a.m., SEB 1200


Friday 8:30 a.m. - 10:30 a.m., SEB 1200

Office Hours:

Day/Time: suggestions? please submit by e-mail before January 13th


Random drop-by is welcome, availability is not guaranteed
Office: TEB 357
Schedule: Laboratories
Matlab tutorials: January 23rd January 27th

4 Lab sessions:

Lab 1: January 30th February 3rd


Lab 2: February 13th February 17th
Lab 3: March 13th March 17th
Lab 4: March 27th March 31st

Reports are due one week after the lab session. Should be submitted to locker
# 206, 2nd floor of TEB
Homework

At the end of each topic, a homework will be given (6 homeworks in total).

Homeworks will not be marked.

The solutions will be posted approximately one week after the homework is
given.
Midterm and Final Examinations
Midterm Exam:

Friday, March 3rd, during the regularly scheduled lecture hours


(8:30 am 10:30 am, SEB 1200).
Duration: 1 hour 50 minutes.
Closed book exam. Necessary equations are provided.
Use of nonprogrammable calculators only is permitted. No other
electronic devices are allowed.

Final Exam:

Will be scheduled during the regular Winter term examination pe-


riod in April.
Duration: 3 hours.
Closed book exam. Necessary equations are provided.
Use of nonprogrammable calculators only is permitted. No other
electronic devices are allowed.
Grading Scheme
Midterm examination 20 %

Final examination: 60 %

Laboratory: 20 %

To obtain a passing grade in the course a mark of 50 % or more must be


achieved on the final examination.

In order to pass the course, passing grades (50 % or more) must


be achieved in both the lecture component (Midterm + Final
Exam) and the laboratory component.

?
What is a Signal?

Definition (http://www.merriam-webster.com/dictionary/signal):
A signal is a detectable (measurable) physical quantity
or impulse (as a voltage, current, or magnetic field strength)
by which messages or information can be transmitted.

Key characteristics:

Signals are physical quantities


Signals are measurable/detectable
Signals contain information
What is Signal Processing?

A System (in the context of our course, as more general


definitions can be given!) is a physical device that performs
an operation on a signal.

Signal Processing: passing a signal through a system

The primary (but not the only) reason for signal processing
is to extract information!

Signal Information
System
Reasons for Processing Real World Signals

The primary reason: to extract the most relevant information


contained in the signal

Information can exists in the form of

Signal amplitude
Signal frequency or spectral content
Signal phase
Relationship with respect to other signals
Processing of Real World Signals: Examples
Example: Electrocardiography (ECG)

Electrocardiography (ECG) is the recording of the electrical activity of the heart.


ECG signal contains various types of information that the physician can analyze to
make conclusions about the heart conditions.

ECG signal is typically distorted by interference with other signals (for example, 60-Hz
power lines). These interferences can be removed using signal processing techniques.
Processing of Real World Signals: Examples
Example: Electroencephalography (EEG)

Electroencephalography (EEG) is the recording of electrical ac-


tivity along the scalp produced by the firing of neurons within the
brain.

EEG contains information that can be used for diagnosis of of


epilepsy, encephalopathies, tumors, stroke and other brain disor-
ders.
Processing of Real World Signals: Examples
Example: Speech Processing

Speech analysis: automatic speech recognition, speaker verification, speaker iden-


tification.
Speech synthesis: reading machines for the automatic conversion of written text
into speech, computer-generated speech.
Speech analysis and synthesis: voice scrambling for secure transmission.
Speech enhancement: enhancing the intelligibility and/or perceptual quality of a
speech signal (noise reduction, etc.)

Source: http://oxygen.lcs.mit.edu/Speech.html
Processing of Real World Signals: Examples
Example: seismic signal processing:

Seismogram is a record of the ground mo-


tion resulting from an earthquake, a volcanic
eruption, or an underground explosion.
Contains information that allows to determine
the magnitude of an earthquake or explosion
and the location of the source.
Allows to monitor, analyze and predict earth-
quakes.
Used in the geophysical exploration, in par-
ticular, for oil and gas.
Processing of Real World Signals: Examples
Time series analysis:

Analyzing historical data and developing a model for forecasting


future behavior
Example: stock market forecast - forecasting the direction of
prices through the study of past market data
Processing of Real World Signals: Examples
Example: Sound processing applications:

Compressors and limiters


Expanders and noise gates
Equalizers
Noise reduction
Delay and reverberation
Special eects
Processing of Real World Signals: Examples

Example: Image Processing - Weather Forecasting


Information vs. Noise

Information: the part of the signal that is of interest

Noise: the part of the signal that is not of interest

Noise does not have to be noisy!


Example: in a signal that contains speech and music,
if only the speech is of interest then the music can be
regarded as noise to be filtered out
Often, noise is noisy! e.g., in measurements of signals
such as forces measured by a force sensor
Information vs. Noise
Signal Processing: Analog vs. Digital
Analog signals: Signals that vary continuously (with time
or other independent variable(s))

(Loosely speaking) Real-life signals are analog

Analog signals can be processed directly

Analog
Analog Input Signal Analog Output Signal
Signal Processor
Signal Processing: Analog vs. Digital

Alternatively, an analog signal can be

converted into digital form (A!D);


processed digitally;
the result is converted back into analog form (D!A).
Signal Processing: Analog vs. Digital

Source: http://courses.ece.illinois.edu/ECE101/SP09/
Signal Processing: Analog vs. Digital

Conversion into digital form (A!D):

Sampling
Quantization
Coding
Signal Processing: Analog vs. Digital
Sampling:

Conversion from continuous-time to discrete-time


Obtained by taking samples of the continuous-time
signal at discrete-time instants
Signal Processing: Analog vs. Digital
Quantization:

Conversion of a discrete-time continuous-valued signal


into a discrete-time discrete-valued signal
The value of the signal is represented by a value selected
from a finite set of possible values
Introduces a (quantization) error
Signal Processing: Analog vs. Digital
Coding:

Each discrete value (quantization level) is assigned a


unique (binary) number
Signal Processing: Analog vs. Digital

Digital-to-Analog Conversion:

Connects the dots of a digital signal by performing


some sort of interpolation
Signal Processing: Analog vs. Digital

Why Digital?:

Noise robustness: the signal is either 0 or 1. Makes exact copies


possible.
Digital circuits is fairly independent of external parameters, such as
temperature.
Digital signals are easy to store and transmit.
Software implementation: digital signal processing algorithms
can be implemented in software, which makes it cheaper and much
more flexible (easy to change and/or reconfigure)
Digital signal processing allows for sharing of a given processor
among a number of signals by timesharing: reducing the cost of
processing per signal.
Signal Processing: Analog vs. Digital
Signal Processing: Analog vs. Digital

Disadvantages of digital signal processing:

Increased system complexity: the need of additional pre- and post-


processing devices and complex digital circuitry.
Limited range of frequencies: sampling rate must be at least twice
higher than the highest frequency in the signal (we will learn this soon).
Signals with wide bandwidth require very fast sampling rate and very fast
processors.
Power consumption: Digital signal processing devices are active devices
as they consume power. On the contrary, analog signal processing devices
can frequently be implemented using passive circuits (resistors, capacitors,
and inductors).

Advantages typically outweigh the disadvantages!


Topic 1: Introduction and Background

Signals and their classification

The Concept of Frequency in Continuous-Time and Discrete-


Time Signals

Analog-to-Digital and Digital-to-Analog Conversion


Signals and Their Classification

Definition (http://www.merriam-webster.com/dictionary/signal):
A signal is a detectable (measurable) physical quantity or impulse
(as a voltage, current, or magnetic field strength) by which messages or
information can be transmitted.

Mathematically, a signal is represented as a function of one or more


independent variables (such as time, spatial variables, frequency, etc.).
Examples:
Classification of Signals

Continuous-Time vs. Discrete-Time

One-Channel vs. Multi-Channel

One-Dimensional vs. Multi-Dimensional

Continuous-Valued vs. Discrete-Valued

Deterministic vs. Random


Continuous-Time vs. Discrete-Time Signals

Continuous-time Discrete-time
Continuous-Time vs. Discrete-Time Signals
Continuous-time Discrete-time

Intuitively:

A continuous-time signal is defined for all values of time in a time


interval.

A discrete-time signal is defined only at certain specific (isolated)


values of time in a time interval.
Continuous-Time vs. Discrete-Time Signals

Mathematically, continuous-time and discrete-time signals dier in


that they are defined on dierent sets of time

A continuous-time signal:

is defined on a continuous time interval, such as (a, b), [0, +1),


( 1, +1), etc.
can be represented as a function of a continuous variable, such
as x(t), t 2 [0, +1)

A discrete-time signal:

is defined on a set of integer numbers, such as {0, 1, , 2, . . .},


{. . . , 2, 1, 0, 1, , 2, . . .}, etc.
can be represented as a sequence, such as x[n] := {x(0), x(1), x(2), . . .}.
Continuous-Time vs. Discrete-Time Signals

Discrete-time signals usually arise from sampling of continuous-time


signals at discrete-time instants.

Example:
Single Channel vs. Multi-Channel Signals
Single channel signals:

Usually generated by one source


Represented by scalar functions (of time or other independent variables), such as

s(t) = a sin(2t).

Real-life example?

Multi-channel signals:

Usually generated by multiple sources (sensors)


Represented by vector-valued functions (of time or other independent variables),
such as 2 3
s1 (t)
s(t) := 4 s2 (t) 5 2 R3
s3 (t)
Real-life example?
One-Dimensional vs. Multi-Dimensional Signals

One-Dimensional Signal is a function of one independent variable


(for example, time)

N-Dimensional Signal is a function of N independent variables

Examples?
One-Dimensional vs. Multi-Dimensional Signals

Example: Color TV picture

Q. How many channels does this signal have?


A. Three. Each point characterized by brightness of three principal
colors: red, green and blue
Q. How many dimensions does this signal have?
A. Three. Color of each point on a screen is a function of two
spatial coordinates (x, y) and time t.

2 3
Ir (x, y, t)
I(x, y, t) = 4 Ig (x, y, t) 5
Ib (x, y, t)
Continuous-Valued vs. Discrete-Valued Signals
Continuous-Valued Signal takes on all possible values from its con-
tinuous range

Discrete-Valued Signal takes on values from a finite set

The process of converting a continuous-valued signal into a discrete-


valued signal is called quantization
Continuous-Valued vs. Discrete-Valued Signals

Digital Signal = Discrete-Valued + Discrete-Time

From: Proakis & Manolakis, 2007


Deterministic Signals vs. Random Signals

A signal is called deterministic if it can be uniquely described by an


explicit mathematical expression or well-defined rule.

Past and future values of a deterministic signal are assumed to be


known precisely without any uncertainty.

On the contrary, the values of a random signal are assumed to be not


known exactly but rather described in terms of their probability dis-
tributions.

In real life, there is no strict boundary between deterministic and


random signals. A signal can be considered deterministic or random
depending on the mathematical model available.
Classification of Signals
Problem (Problem 1.1 from Proakis & Manolakis, 2007): Classify the following signals
according to whether they are (1) one- or multi-dimensional; (2) single or multichannel; (3)
continuous-time or discrete-time; and (4) analog or digital (in amplitude). Give a brief expla-
nation.

Q1. Closing prices of utility stocks on the New York Stock Exchange
A1. One dimensional, multichannel, discrete time, and digital.

Q3. Position of the steering wheel of a car in motion relative to cars reference frame
A3. One dimensional, single channel, continuous-time, analog.

Q4. Position of the steering wheel of a car in motion relative to ground reference frame
A4. One dimensional, multichannel, continuous-time, analog

Q5. Weight and height measurements of a child taken every month


A5. One dimensional, multichannel, discrete-time, digital