Review - DTFT

Review
 Continuous-time signals
 Discrete-time signals and systems
 Discrete-time Fourier transform (DTFT)
 The z-transform
 Difference equations

2
 Multimedia Signal Processing At A Glance…
• Some of the specifications of this MP3 player are:
 1” x 2” package  8 Gbytes memory
 music, video, photo  audio recording
 FM radio  USB interface
• Advances in integrated circuits and digital signal processing
technology enabled the creation of this device.
• Low power and large inexpensive memories were crucial to
the commercial viability of the device.
• The signals stored in this device are digital signals, because
they consist of discrete-time signals whose amplitudes are
represented as a finite set of numbers.
• The signals are compressed to save space using software-
based compression (MP3).
• An FM radio demodulates an electromagnetic wave and
converts it to a digital audio signal.
• The audio recorder uses a microphone to convert a sound pressure wave
into an electrical signal, which is then converted to an MP3 digital signal.
• A lot of signal processing technology for $30!
 Examples of Signals
• Electrical signals (e.g., voltages and currents in a circuit)
• Acoustic signals (e.g., audio or speech signals)
• Video signals (e.g., intensity variations in an image)
• Biological signals (e.g., sequence of bases in a gene)
• The Internet is introducing many new forms of “signals” as information streams (e.g., multimedia
news broadcasts).
• Noise: unwanted components in your signals that are often random.
• Video signals are interesting because they
are discrete in several dimensions:
 The image has a particular horizontal and
vertical resolution (e.g., 320 x 240 pixels).
 The image also is sampled in the temporal
dimension (e.g., 30 frames per second).
 Each pixel is represented by 8 to 24 bits of
color (e.g., 8 bits for Red, Green and Blue).
 The audio and video are compressed.
 Independent Variables

• Time is often the independent variable

for a signal. x(t) will be used to represent
a signal that is a function of time, t.
• A temporal signal is defined by the
relationship of its amplitude
(the dependent variable) to time
(the independent variable).

• An independent variable can be 1D (time), 2D

(space), 3D (space) or even something more
complicated.
• The signal is described as a function of this
variable.
• There are many types of functions that can be used
to describe signals (continuous, discrete, random
are just a few of the concepts we will encounter this
semester).
6
Classification of signals
 According to the nature of the independent variables
 Values of the function defining the signals
 Examples:
 discrete / continuous function
 discrete / continuous independent variable
 Real / complex valued function
 Scalar / vector
 Single / multi trial
 Dimensionality based on the number of independent variables
(1D/2D/3D)
 Even / odd
7
‫اﻧﻮاع ﺳﯿﮕﻨﺎل‬

‫‪8‬‬
 Continuous Time (CT) Signals
• Most of the signals in the physical world are CT signals, since the time scale
is infinitesimally fine (e.g., voltage, pressure, temperature, velocity).
• Often, the only way we can view these signals is through a transducer, a
device that converts a CT signal to an electrical signal.
• Common transducers are the ears, the eyes, the nose… but these are a little
complicated.
• Simpler transducers are voltmeters, microphones, and pressure sensors.
 Discrete-Time (DT) Signals

• We can write a collection of numbers (1, -3, 7, 9) representing a signal as a

function of a discrete variable, n. x[n] represents the amplitude, or value of
the signal as a function of n, which takes on integer values.

• Many human-generated signals are discrete (e.g., MIDI codes, stock market
prices, digital images).
• In this course, we will show that most of the properties that apply to CT
signals apply in a similar manner to DT signals.
 Signals With Symmetry: Periodic

• CT Periodic signals:
x(t ) = x(t + T )

• DT Periodic signals: x ( n) = x ( n + N )

• Note: The sum of two CT signals is periodic only if the ratio of their periods can be
written as the ratio of two integers. We will exploit this fact when building signals
out of sums of periodic signals (e.g., Fourier series).
Signals With Symmetry: Even/Odd

• Even: CT: x(t ) = x(−t )

DT: x(n) = x(−n)

• Odd: CT: x(t ) = − x(−t )

DT: x(n) = − x(−n)
Properties of Even/Odd Signals

• Any signals can be expressed as a

sum of even and odd signals. That is:

x(t ) = xeven (t ) + xodd (t )

where :
xeven (t ) = [ x(t ) + x(−t )] / 2
xodd (t ) = [ x(t ) − x(−t )] / 2

• This is demonstrated to the right for

a signal referred to as a unit step.
Other Types Of Symmetry

• A right-sided signal is zero for t < T and a left-sided signal is zero for t > T,
where T can be positive or negative.

• A signal can be bounded or unbounded depending on the stability of the

system.
 Real and Complex Signals
• Complex signals are an important abstraction in many disciplines such as
communications and multidimensional signal processing.
• In general, x is a complex quantity and has:
 a real and imaginary part, or equivalently
 a magnitude and a phase angle.
• A very important class of signals is complex exponentials:
 CT signals of the form x(t) = est
 DT signals of the form x[n] = zn
where z and s are complex numbers.
• For example, suppose s = jπt/8 and
z = e jπ/8, the exponentials are purely
imaginary, then the real parts are:
Real and Complex Signals (Cont.)

For example, suppose s = σ + jω

σ>0:
for a CT signal:

{ } { }
ℜ x(t ) = e st = ℜ e (σ + jω )t = e σt cos( jωt ) σ<0:

σ>0:

For example, suppose z = e(σ + jω) for

σ<0:
a DT signal:

{ } { }
ℜ x(t ) = z n = ℜ z (σ + jω ) n = e σn cos(ωn)
 Energy and Power

∞
• Signal Energy:
x(t ) dt
2
Ef = ∫
−∞

• Signal Power: 1
T /2

T →∞ T ∫
Pf = lim x (t ) 2
dt
−T / 2

• Comments:
 Power is the time average of energy. Why?
 What is an example of a signal for which the energy integral is bounded? We refer to such
signals as energy signals.
 What is an example of a signal for which the energy integral is unbounded but the power
integral is bounded? We refer to these as power signals.
 What is an example of a signal for which the power integral is unbounded? Such signals are
often described as being unstable. Can such a signal exist in the real world?
 Later will we relate these to the same concepts of energy and power that you have used in
electrical circuits. We will also relate these concepts to familiar statistical measures of
randomness such as mean, standard deviation and root mean square (RMS) value.
Discrete-time signals
 Discrete-time signals (or simply sequences) may be inherently discrete or they may
be obtained by sampling a continuous-time signal.

18
Basic Sequences

19
Basic Sequences

20
Basic Sequences

 Differently from continuous-time sinusoids, discrete-time sinusoids...

 are periodic only if is rational.
 have maximum frequency

 Complex exponentials:

21
Classification of discrete-time signals
 Energy and power signals
 Energy signals have finite energy

 Power signals have infinite energy (e.g., periodic signals), but they have finite
average power

22
Classification of discrete-time signals
 Symmetry

 Any signal x[n] can be decomposed as a sum of an even (xe[n]) and an odd (xo[n])
component:

23
Discrete- time systems

 Some important properties

 Linearity or superposition

 Time invariance or shift invariance

 A time shift of the input causes an equal time shift of the output 24
Linear time-invariant (LTI) systems
 LTI systems are completely characterized by their impulse response.

25
Linear time-invariant (LTI) systems
 LTI systems are completely characterized by their impulse response.

26
Linear time-invariant (LTI) systems
 LTI systems are completely characterized by their impulse response.

27
The convolution sum

28
29
30
31
Other properties of discrete-time systems
 Memorylessness
 A system is memoryless if its output at time n, y[n], depends only on the present input x[n].
 Causality
 A system is causal if its output at time n, y[n], depends only on the present and past samples of
the input {x[n]; x[n 1]; x[n 2]; … }. All physical systems are causal.
 For LTI systems, causality implies h[n] = 0; n < 0

 (Convolution sum for causal systems)

32
 Other properties of discrete-time systems
 Stability
 A system is BIBO stable if every bounded input (|x[n]| < 𝐵𝐵𝑥𝑥 < ∞) produces a bounded output
(|y[n]| < 𝐵𝐵𝑦𝑦 < ∞)
 For LTI systems,
 Convolution sum

 Triangle inequality

 Bounded input

 only if

 Therefore, an LTI system is BIBO stable only if its impulse response h[k] is absolute summable.
33
34
35
Example

36
37
Let’s practice!

38
Frequency-domain representation of LTI systems
 Let's use the convolution sum to compute the output of an LTI system to a complex exponential:

 𝐻𝐻(𝑒𝑒 𝑗𝑗𝑗𝑗 ) is a complex number that depends only on the impulse response of the LTI system
 The output of an LTI system to a complex exponential is also a complex exponential of same
frequency, but with possibly different amplitude and phase
 Conclusions
 Complex exponentials are eigenfunctions of LTI systems, and 𝐻𝐻(𝑒𝑒 𝑗𝑗𝑗𝑗 ) are the corresponding eigenvalues.
 By representing signals as a sum of complex exponentials, we can readily calculate their output to an LTI
system:
39
Discrete-time Fourier transform (DTFT)
 Definition
 Direct transform

 Inverse transform

 Important
 The DTFT is periodic with period

 If x[n] is absolute summable i.e., , then the DTFT exists. This is a

sufficient condition, but it is not a necessary condition.

 Absolute summability also implies that the DTFT converges uniformly to a continuous function of ω
40
41
42
Frequency response of the moving average

43
44
Example: the ideal low pass filter
 Time domain Frequency domain

45
Example: the ideal low pass filter
 Time domain Frequency domain

46
Example: the ideal low pass filter
 Time domain Frequency domain

47
Convergence of the DTFT
 The sinc function is not absolute summable, but it is square summable

 In this case, the DTFT does not converge uniformly; it converges in the mean-square sense
 The oscillations are known as Gibbs phenomenon, and they occur whenever there is a discontinuity in the
frequency domain
 Interestingly, as 𝑀𝑀 → ∞ the oscillations become more rapid, but the size of the ripples does not decrease
 The DTFT may exist even when sequences are neither absolute summable nor square summable.
Examples: a constant, unit step, complex exponentials

48
 DTFT properties

 Linearity

 Time shift, delay for nd > 0, advance for nd < 0

 Frequency shift (modulation)

 Convolution

 Multiplication of sequences (windowing)

49
 DTFT properties
 Linear weighting

 Time reversal

 Parseval's theorem

 if x[n] = y[n] (signal energy)

 Deterministic autocorrelation function

50
51
Example

52
53
54
55