You are on page 1of 102

Dr.

Zulfiqar H Khan
zulfiqar@alumni.chalmers.se
Lecture 1

Course Objective
To establish the idea of using computing techniques to
alter the properties of a signal for desired effects
Understanding of fundamentals of discrete-time, linear,
shift-invariant signals and systems in

Representation: sampling and quantization;


Processing: filtering and transform techniques;
Processing System Design: filter & processing algorithm design.
Efficient computational algorithms and their implementation.

Understanding to real world scenario

What is DSP?

A-to-D conversion: bandwidth control, sampling and quantization


Computational processing: implemented on computers or ASICs
with finite-precision arithmetic
Basic numerical processing: add, subtract, multiply (scaling,
amplification, attenuation), mute,
Algorithmic numerical processing: convolution or linear
filtering, non-linear filtering (e.g., median filtering), difference
equations, DFT, inverse filtering, MAX/MIN,
D-to-A conversion: re-quantification* and filtering (or
interpolation) for reconstruction

Course Introduction: What is Signal Processing

Digital Signal Processing: Historical Context

Basic Interests & Issues in DSP


Digital Filters: Filter design, noise analysis, structures
Fourier Analysis: Spectrum estimation, FFT, cosine
transform, short-time FT
Signal Modeling and Analysis: Linear prediction, wavelets

Hardware and Software: Minicomputers, array


processors, DSP chips, workstations, PCs, MATLAB, RTOS
Applications: Speech, radar, image, video, data, ...

Book Authors in this DSP course

DSP Algorithms
Common algorithms
Linear filtering: FIR, IIR
FFT, cosine transform, filterbanks
Correlation
Matrix calculations
Common features
Lots of multiply/accumulate operations
Block processing is often appropriate
Fixed-point arithmetic for economical solutions

Digital Signal Processing: Advantages

Digital Signal Processing: Some Disadvantages

Some Notations

Examples of Signals: Speech Waveform

Example of 1D Signal

Digital Speech Signal

Voice frequency range: 20Hz ~ 3.4 KHz


Sampling rate: 8 KHz (8000 samples/sec)
Quantization: 8 bits/sample
Bit-rate: 8K samples/sec * 8 bits/sample = 64 Kbps (for
uncompressed digital phone)
In current Voice over IP (VOIP) technology, digital speech
signals are usually compressed (compression ratio: 8~10)

Examples of 1-D Signals: Seismic Wave

Examples of 1-D Signals: Electrocardiogram

Examples of 2-D Signals: Image

Digital Image Signal


One mega-pixel image (1024x1024)
Quantization: 24 bits/pixel for the RGB full-color space,
and 12 bits/pixel for a reduced color space (YCbCr)
Bit-rate: 1024x1024 samples/sec * 12 bits/pixel = 12
Mbits = 1.5 Mbytes (for uncompressed digital phone)
How many uncompressed images can be stored in a 2G
SD flash-memory card?

Examples of 3-D Signals: Video

Textbook
Discrete-Time Signal processing, A. V. Oppenheim, R. W.
Schafer, and J. R. Buck, Pearson Prentice Hall, US, 1999.:
Signal Processing First, J. H. McClellan, R. W. Schafer,and
M. A. Yodar, Pearson Prentice Hall, US, 2003.
Statistical Digital Signal Processing and Modeling by
Monson H. Hayes

Signals and Systems


Signals
Something that carries information
Speech, audio, image, video, biomedical signals, radar
signals, seismic signals, etc.

Systems
Something that can manipulate, change, record, or
transmit signals
CD, VCD/DVD

Signals: Review of Basic Concepts

Signal

Signals Types
Most of the signals we will talk about are functions of time.
There are many ways to classify signals. This class is organized
according to whether the signals are continuous in time, or
discrete.
A continuous-time signal has values for all points in time in
some (possibly infinite) interval.
A discrete time signal has values for only discrete points in
time.

Signals can also be a function of space (images) or of space


and time (video), and may be continuous or discrete in each
dimension

Discrete-Time Signal vs Digital Signal


Discrete-Time signal
A sampled version of a continuous signal
What should be the sampling frequency which is
enough for perfectly reconstructing the original
continuous signal?
Nyquist rate (Shannon sampling theorem)
Digital Signal
Sampling + Quantization
Quantization: use a number of finite bits (e.g., 8bits) to
represent a sampled value

Discrete Time Signals


Fundamentally, a discrete-time signal is sequence of samples,
written,
[]
where is an integer over some (possibly infinite) interval.
Often, at least conceptually, samples of a continuous time
signal
[] = ()
where is an integer, and is the sampling period.

Example

Signal Characteristics and Models


Operations on the time dependence of a signal
Amplitude Scaling
Unit Step Sequence
Time scaling
Unit Rectangle
Time reversal
Unit Ramp
Time shift
Unit Triangle Signal
Combinations
Exponential Signals
Even and Odd Symmetry
Sinusoidal Signals
Discrete Amplitude Signals
Complex Signals
Impulse Sequence

Amplitude Scaling (Continuous-Time)

The scaled signal ax(t) is x(t) multiplied by the constant a

Amplitude Scaling (Discrete Time)


The scaled signal ax[n] is x[n] multiplied by the constant a

Time Scaling (Continuous Time)


A signal x(t) is scaled in time by multiplying the time variable
by a positive constant b, to produce x(bt).
A positive factor of b either expands (0 < b < 1) or compresses
(b > 1) the signal in time.

Time Scaling (Discrete Time)


The discrete-time sequence [] is compressed in time by
multiplying the index n by an integer , to produce the timescaled sequence [].
This extracts every sample of [].
Intermediate samples are lost.
The sequence is shorter.

Called downsampling, or decimation

Time Scaling (Discrete Time)


The discrete-time sequence [] is expanded in time by
dividing the index n by an integer m, to produce the timescaled sequence [ = ].
This specifies every sample.
The intermediate samples must be synthesized (set to zero, or
interpolated).
The sequence is longer.

Called upsampling, or interpolation

Time Reversal
Continuous time: replace with , time reversed signal is
()

Discrete time: replace n with , time reversed signal is [].

Same as time scaling, but with b = -1.

Time Shift: Continuous Time


For a continuous-time signal (), and a time 1 > 0,
Replacing t with 1 gives a delayed signal 1
Replacing t with + 1 gives an advanced signal x( + 1 )

May seem counterintuitive. Think about where 1 is zero

Time Shift: Discrete Time


For a discrete time signal [], and an integer 1 > 0
[ 1 ] is a delayed signal.
[ 1 ] is an advanced signal.
The delay or advance is an integer number of sample times.

Again, where is 1 zero?

Combination of Operations

Time scaling, shifting, and reversal can all be combined.


Operation can be performed in any order, but care is required.
This will cause confusion.
Example: (2( 1))
Scale first, then shift
Compress by 2, shift by 1

Combination of Operations
Example (2( 1)), continued
Shift first, then scale
Shift by 1, compress by 2

Shift first, then scale


Rewrite (2( 1)) = (2 2)
Shift by 2, scale by 2

Combination of Operations: HW

Event and Odd Symmetry


An even signal is symmetric about the origin

An odd signal is antisymmetric about the origin

Even and Odd Symmetry


Any signal can be decomposed into even and odd components

and that

Even and Odd Symmetry


Example

Same type of decomposition applies for discrete-time signals.

Even and Odd Symmetry: HW


The decomposition into even and odd components depends on
the location of the origin. Shifting the signal changes the
decomposition.
Plot the even and odd components of the previous example,
after shifting () by 1/2 to the right.

Discrete Amplitude Signals


Discrete amplitude signals take on only a countable set of
values.
Example: Quantized signal (binary, fixed point, floating point).
A digital signal is a quantized discrete-time signal.
Requires treatment as random process, not part of this
course.

Impulse Sequence (Discrete-time)

Expressing Sequence as Sum of Unit ImulseSequences

Impulse Signal (Continuous - Time)

Unit Step Sequence (Continuous Time)


The unit step function u(t) is defined as

Also known as the Heaviside step function.


Alternate definitions of value exactly at zero, such as 1/2..

Uses for the Unit Step (Continuous Time)


Extracting part of another signal. For example, the piecewisedefined signal

Can be written as

Uses for the Unit Step (Continuous Time)


Extracting part of another signal. For example, the piecewisedefined signal

Can be written as

Unit Step Sequence (Discrete-Time)

Unit Rectangle

Unit Ramp

Unit Triangle Signal

Exponential Signals
An exponential signal is given by

If < 0, this is exponential decay.


If > 0 this is exponential growth.

Damped Exponential Signals


Exponential growth ( > 0)or decay( < 0), modulated by
a sinusoid.

Discrete Exponential Sequence

Sinusoidal Signals
A sinusoidal signal is of the form

where the radian frequency is , which has the units of


radians/s.
Also very commonly written as

where f is the frequency in Hertz.


We will often refer to as the frequency, but it must be kept in
mind that it is really the , and the
frequency is actually .

Sinusoidal Signals
The period of the sinusoid is

With the units of seconds.


The phase or phase angle of the signal is , given in radians.

Complex Signals

Complex Exponential Sequence

Non-Uniqueness of Discrete Time Signals


Suppose

Discrete Time Periodicity

More Complex Signals


Many more interesting signals can be made up by
combining these elements.
Example: Pulsed Doppler RF Waveform

More Complex Signals


Start with a simple rect(t) pulse

Scale to the correct duration and amplitude for one


subpulse

Combine shifted replicas

This is the envelope of the signal.

More Complex Signals

System: Review of Basic Concepts

Systems: Intuitive Description

Systems: Definition

Systems: Block Diagram

System: Examples

System Examples

System Examples

System Examples: Multiple Inputs

System Interconnection

System Interconnection

System Interconnection: Integrator with feedback

System: Linearity

System: Linearity

System: Linearity

System Memory

System: Time Invariance

System: Time Invariance

System: Stability

System Stability

System Inevitibility

System Invertibility

System Invertibility

System: Described by Differential Equation

System: Described by Differential Equation

System: Examples

System Examples: 2nd order Circut

System Examples: Mechanical Systems

System Examples: Mechanical Systems

System Types
Systems are classified according to the types of input and output signals
Continuous-time system has continuous-time inputs and outputs.
AM or FM radio
Conventional (all mechanical) car

Discrete-time system has discrete-time inputs and outputs.


PC computer game
Matlab
Hybrid systems are also very important (A/D, D/A converters).
You playing a game on a PC
Modern cars with ECU (electronic control units)
Most commercial and military aircraft

Discrete Type Systems

Discrete Time Systems


We typically denote the input of a discrete-time system
as x[n] and the output as y[n].

The system T computes the outputs as a function of the


inputs. In general, the output at time n may depend on
several sa0mples of the input sequence x.
Example: Moving average

Causal Discrete-Time Systems

Stable Discrete-Time Systems

Input-Output Description: Capabilities & Limitations

Impulse Response Description: Capabilities & Limitations

LTI Discrete-Time Systems

Discrete-Time Convolution

You might also like