Advance Signal

Processing
ESE 752
Dr. Idnin Pasya bin Ibrahim
 Lecturers Profile
Name: Idnin Pasya bin Ibrahim
Education: PhD in Radio Communication Eng. (Tokyo Denki University, Japan)
B. Eng, M. Eng in Information and Communication Eng. (TDU, Japan)
Major: Wireless Communication, Antenna and Propagation, Radar Systems

Work Experience:
2006 – 2009 Engineer (Snr.), Toshiba PC & Network, Tokyo, Japan
2009 – 2010 Snr. Engineer, Pradonet Technology, KL.
2010 – Present UiTM Shah Alam

Email: idnin@salam.uitm.edu.my
Tel: 018-2716560
• A2-12-3C
• Microwave Technology Center
Chapter 1
Introduction to
Advance Signal
Processing
Discussion of background theory of signals
Course outline
 Course outcomes
 (CO1)Describe the concept of advance digital signal
processing
-> PO1
 (CO2)Apply DSP Techniques in Communication and Related
Areas
-> PO3
 (CO3)Design and evaluate selected problems in
communications system
-> PO5
Roadmap of ESE752
Digital Signal
Signal and Systems
Processing

Signals definition Analog to digital conversion

Continuous and discrete systems Digitization, Sampling Theorem
Linear, non-linear systems Transformation Techniques:
Impulse response, convolution Fourier (DFT, FFT), Z Transforms
Digital Filter Design: FIR and IIR filters

Software based signal processing

Applications of signal processing in
Communication Engineering
Continuous and discrete system
Signals conversion

Discrete signals or
digitized signals

Digital
Input Output
ADC signal DAC
signal signal
processor
Signals
• What is a signal?
Is this a signal?

Readings of human’s heart beat

Is this a signal?

Recordings of human’s speech

Can a noise be considered as a
signal?

Measurements of noise in a circuit

Signals
• What is signal?
• Expression of a certain variable of information

• In this course, we define a signal as “a varying physical

quantity” or “an expression of a varying physical quantity in
the form of symbols”

• Hence, a signal consists of both the physical and symbol s

characteristics
 Signals (cont)

Signal
Physical
Information
dimension
dimension

Fig. 1. Signal consists of both the physical quantitative information characteristics

Signal (cont)
• In the technology/engineering field, a signal has the following
properties:
I. Variation of a physical quantity
II. Symbols that represents variation of physical quantity
III. Equation that represents both (i) and (ii)
IV. Includes transmission of information
Functions of signal
• Varying physical quantity => relationship of a quantity against
another physical quantity
(Description of how one parameter is related to another
parameter)

• Signal can be expressed in a form of ‘function’

• Example: the most common type of signal in analogue
electronics is a voltage (v) that varies with time (t)

- Taking the signal as a function of t, f(t)

One dimension (1D)
- Taking the signal as a function of space, f(x, y) or f(x, y, z)
Two dimension(2D) Three dimension
Continuous and Discrete Signals
• Most signals are continuous in nature

• Computer and digital systems cannot deal with continuous

signals
-> continuous signals require infinite numbers/samples in
order to be expressed without loss of information

• Computer/digital systems have to deal with discrete signals

based on ‘sampling ’ and ‘digitization’
Systems
• System -> Combination of multiple elements (devices) to
achieve certain objectives
• Block diagram represents the flow of a system

x y
f( ) + f( )
x y
+
g( ) g( )

(a) (b)

Try to derive the equation expressing the relationship

between the input x and output y of these systems
Roadmap
Digital Signal
Signal and Systems
Processing

Signals definition Analog to digital conversion

Continuous and discrete systems Digitization, Sampling Theorem
Linear, non-linear systems Transformation Techniques:
Impulse response, convolution Fourier (DFT, FFT), Laplace Z Transform
Digital Filter Design: FIR and IIR filters

Orthogonal signals
Applications of signal processing in
Software based signal processing
Communication Engineering
Linear Systems
Concept of linear systems
• First rule: Principle of superposition applies on all the input
and outputs

I. ‘Adding the inputs’ produces ‘added outputs’

L{x1 (t )  x2 (t )}  L{x1 (t )}  L{x2 (t )} (1.1)

II. ‘Multiplying the inputs’ produces ‘multiplied outputs’

(1.2)
L{ax(t )}  aL{x(t )}

Combining (1.1) and (1.2)

 
L ak xk (t )   a k Lxk (t )
k  k
Linear systems (cont)

Fig. 1.2. Principle of superposition in Fig. 1.2 Signal expressed in

linear systems. addition of ‘weighted’
fundamental signal.
Linear systems (cont)
• For the sake of clarification, we always treat a linear system by
‘decomposing’ them into multiple fundamental signals

• By doing this, it is easier to predict the behaviour of the system

• Main principles of signal analysis:

I. Decompose signal into multiple fundamental signals
II. System characteristics is expressed by ‘response to the
fundamental signal’
III. Define the relationship between input and output based on
superposition of responses from each fundamental signal.
Delta function 
• Fundamental signal of any signal is defined as an impulse
signal

• Impulse signal -> significantly narrow width

• We define a linear systems input output relationship based on

the delta function .
Signal decomposition base on
delta function 
Fig. 1.4. Example of a signal
decomposed into square
wave with a width of t.

Consider a square wave at

t  nt n = (…, 1, 2, 3, …)

The signal can be approximated by


x(t )   x(nt ) D
n  
t (t  nt )t
Delta function  (cont)
• Expression of the square wave

1  t 
 t  t  
Dt    2
0  t 
t  
  2
• Approximating the the square wave to zero

x(t )  lim  x(nt ) D
t 0 n  
t (t  nt )t

Delta function 


x(t )   x(t ' ) (t  t ' )dt '

Delta function  (cont)
• Expression of delta function  (broad definition)

 (t )  lim Dt (t )
t  0

or

 t  0
 (t )   

0 t  0 

 (t )dt 1

Integration equals to 1
Impulse response
• We have understand that every linear system can be modeled
by linear superposition of delta function .

• AND

• We can expressed a system characteristics by the output of

the system (based on multiple fundamental signal)
• The output of a system which based on an impulse signal as
the input is called the ‘impulse response’

A linear system can be fully characterized / expressed by impulse response

Impulse response
• What can we do with the knowledge of impulse response?
- Prediction of a systems output, and vice versa

• Time invariant systems and variant systems

• Mathematical expression :

h(t )  L{ (t )}
Impulse response and
convolution
• Consider a time invariant system with impulse response h(t),
the output y(t) of the system is

y(t )   x(t ' )h(t  t ' )dt '


• This is called convolution of the input signal and the impulse

response.
Characteristics of convolution
involving delta function
• Three main characteristics:

I. Integration of a multiplication of a delta function  and a

function


f (t ) (t  t0 )dt0  f (t0 )

II. Convolution of a delta function

f (t )   (t  t0 )  f (t  t0 )
III. Inverse convolution

x(t )  y (t )  y (t )  x(t )
Analysis based on sinusoidal
waves
• Lets consider the fundamental signal of as a sinusoidal wave.
• Radial frequency:
Frequency response
• When a sinusoidal wave is considered as an input into a linear
and time invariant system -> the amplitude and phase
undergone changes
• Amplitude (Gain), phase (phase shift)

Gain and phase shifts

Frequency response
• Expressing the frequency response in the form of complex
numbers -> transfer function

G()  H ()
H ( )  G( )e j ( )
 ( )  H ( )

Illustration of Transfer function

• An output signal can be considered as a combination of
multiple sinusoidal wave with different frequency
Summary of delta function and
impulse response
• A signal can be decompose by utilizing a delta function
• An input signal can be considered as a weighted addition of
delta function .
• When considering a delta function as in input signal, the output
signal is an impulse response h(t)
=> the relationship between the input and output by
convolution of
Convolution in discrete systems
• Consider a linear discrete system with input x(n) and output
y(n)
x(n) y(n)
L{  }

• Expression of the system

y (n)  L{x(n)}
• Delta function in a discrete system

1 (n  0)
 ( n)  
0 (n  0)
 Convolution in discrete systems

Expression in the form of impulse signal

• A discrete signal x(n) can be expressed by


x ( n)   x(k ) (n  k )
k  
Convolution in discrete
systems
• Considering linearity of the system

y(n)  L{x(n)}  L{ x(k ) (n  k )}


  x(k )( L{ (n  k )})
k  
• Remember, impulse response of the system is
h(n)  L{ (n)}


y ( n)   x ( n) h( n  k )
k  
Steps of convolution
• Identify input signal x(k) and impulse response h(k)
• Invert impulse response h(k) at n = 0 -> we obtain h(-k)
• Move h(-k) for m samples, and let it be h(m-k)
• Multiply all x(k) with h(m-k) for all k, and let them be the
output signal at time m,
• Do all the above for all m
Example
• Example 4.1 Calculate the output y(n) of a discrete system,
given the impulse response h(n) and the input signal x(n) as
below:
• Impulse response
3 (n  0)
2 (n  1)

h( n)  
1 (n  2)
0 others
• Input signal

1 (n  0,1,2,3)
x ( n)  
0 others
Example (Answers)

Result of convolution
Review questions
Note:
Solve the following review questions (1 to 4)

(1) Plot the following delta functions

(a)  (t) + 2 (t-1)+ 3 (t-2)
(b) 2 (t+2) + 2 (t-2)

(2) Determine the result of convolution of these functions

g(n) = h(n)*x(n)

2 (n  0)  2 (n  0)
 
h(n)  1 (n  1) x(n)   1 (n  1)
0 others  0 others
 
Review questions (cont)
(3) Determine whether the below systems are a linear system or
otherwise, and also whether they are a time invariant systems
or otherwise:
(a) y(n) = x(n) + a (b) y(n) = ay(n-1) + bx(n-2)
(c) y(n) = x2(n)

(4) Consider a case where you have to calculate the output y(n)
of a linear system, which receives a discrete input x(n)
consisting of 40×103 samples, by using either the impulse
response or frequency response (transfer function).
Explain which method is more cost-efficient in terms of
computational resource, and why. Show the comparison of
efficiency in a quantitative evaluation, expressed in actual
numbers or figures.
Hint: Search for the computational complexity of a 1D
convolution vs. multiplication
(1) Refer to notes

(2) {g (2), g (1), g (0), g (1), g (2)}  {1,0,2,0,1}

(3) a- although the plot of y(n) is a straight line the function is

non-linear, time invariant ,
b- linear, time invariant
c- non-linear, time invariant, d- linear, time variant
(4) Using the impulse response requires convolution process which
consume N2 computational complexity, however, using the transfer
function in the frequency domain only requires N + N log2(N). (N for
multiplication, N log2(N) for FFT since we need to transform the
original time domain signal into the frequency domain)
Comparison
10
of computational complexity when using impulse and frequnecy response to predict the output of a system
10

9
10

8
10

7
Computational complexity, [order]

10

6
10

5
10

4
10

3
10

2
10

1
10 Using frequency response
Using impulse response
0
10
0 0.5 1 1.5 2 2.5 3 3.5 4
Number of samples, [n] 4
x 10