5/1/2017

UNIT–IV Part-A
CONVOLUTION AND
CORRELATION OF
SIGNALS
Prof K.Venkat Reddy

Reference Books

1-May-17 Prof K.Venkat Reddy 2

1
 5/1/2017

UNIT-IV Syllabus
 Convolution and Correlation of Signals:
Concept of Convolution in time domain and
frequency domain, Graphical representation of
Convolution, Convolution properties, Cross
correlation and auto correlation of functions,
Properties of correlation function, Energy
density spectrum, Parseval’s theorem, Power
density spectrum, Relation between auto
correlation function and energy/power
spectral density function, relation between
Convolution and Correlation

1-May-17 Prof K.Venkat Reddy 3

Contents
Convolution and Correlation of Signals :
 Concept of Convolution in
 Time domain and
 Frequency domain
 Graphical representation of Convolution
 Convolution properties
 Cross correlation and auto correlation of
functions
 Energy density spectrum
 Parseval’s theorem
 Power density spectrum
 Relation between auto correlation function
and energy/power spectral density function
 relation between Convolution and Correlation
1-May-17 Prof K.Venkat Reddy 4

2
 5/1/2017

CONCEPT OF CONVOLUTION

 The Convolution is one of the

most powerful tools in frequency
analysis.
 It is of two types

1. Time Convolution and

2. Frequency Convolution

1-May-17 Prof K.Venkat Reddy 5

TIME CONVOLUTION

 Given two functions f1(t) and f2(t).

 We form the integral

f (t )   f1   f 2 t    d ......5.1


It is known as time convolution

operation between functions f1(t) and
f2(t). In simple notation it is written as
f1 t * f 2 t 
Where

f1 t * f 2 t    f1   f 2 t    d (5.2)

1-May-17 Prof K.Venkat Reddy 6

3
 5/1/2017

FREQUENCY CONVOLUTION

 Given two functions(Spectrums) F1(ω)

and F2(ω).
 We form the integral

F ( )   F1 u  F2   u  du ......5.3


It is known as frequency convolution

operation betn functions(Spectrums)
F1(ω) and F2(ω). In simple notation it is
written as F1 () * F2 ().
Where

F1 ( ) * F2 ( )   F1 u  F2   u  du (5.4)


1-May-17 Prof K.Venkat Reddy 7

CONVOLUTION RELATIONSHIPS

1. Convolution operation is Commutative

.i.e.
f1 t * f 2 t   f 2 t * f1 t  (5.5)
Proof:

f1 t  * f 2 t    f1   f 2 t    d


Changing the variable τ to t-x, we get


f1 t  * f 2 t    f 2 x  f1 t  x  dx  f 2 t  * f1 t 


1-May-17 Prof K.Venkat Reddy 8

4
 5/1/2017

CONVOLUTION RELATIONSHIPS

2. Convolution operation is Distributive

.i.e.
f1 t *[ f 2 t   f 3 (t )]  f1 t * f 2 t   f1 t * f 3 t  (5.6)

3. Convolution operation is Associate

.i.e.

f1 t *[ f 2 t * f3 (t )]  [ f1 t * f 2 t ] * f 3 t  (5.7)

1-May-17 Prof K.Venkat Reddy 9

GRAPHICAL INTERPRETATION OF
CONVOLUTION
 In linear
systems
graphical
convolution is
very helpful
in analysis, if
f(t) and h(t)
are known
only
graphically.

1-May-17 Prof K.Venkat Reddy 10

5
 5/1/2017

GRAPHICAL INTERPRETATION
OF CONVOLUTION
 Consider f1(t) and f2(t) as rectangular
and triangular pulses respectively. Find
the convolution f1(t)*f2(t) graphically
by definition

f1 t  * f 2 t    f1   f 2 t    d .............( I )


 Solution: The function f1(τ) and f2(-

τ) are shown in Figure 5.1b.
 The function f2(-τ) is obtained by tilting
f2(τ) about vertical axis passing
through the origin.

1-May-17 Prof K.Venkat Reddy 11

GRAPHICAL INTERPRETATION
OF CONVOLUTION
 The term f2(t-τ) represents the function
f2(-τ) shifted by t seconds along the
positive τ-axis.
 Figure 5.1 c and d shows f2(t-τ).
 To value of the convolution integral at
t=t1 is given by the integral in Equ(I)
evaluated at t=t1. This is clearly the
area under the product curve of f1(τ)
and f2(t1-τ).
 This area is shown shaded in
figure.5.1d.
1-May-17 Prof K.Venkat Reddy 12

6
 5/1/2017

GRAPHICAL INTERPRETATION
OF CONVOLUTION
 The value of f1(t)*f2(t) at t=t1 is equal
to this shaded area and is plotted in
Figure.5.1f. We choose different values
of t1 shift the function f2(-τ) accordingly
and find the area under the new
product curve. These areas represent
the value of the convolution function
f1(t)*f2(t) at the respective value of t.
 The plot of the area under the product
curve as a function of t represents the
desired convolution function f1(t)*f2(t).
1-May-17 Prof K.Venkat Reddy 13

SUMMARY OF STEPS IN GRAPHICAL

INTERPRETATION OF CONVOLUTION
1. Take given functions f1(t), f2(t) and
change t-axis to τ-axis so, they
become f1(τ) and f2(τ).
2. Tilt the function f2(τ) about the
vertical axis passing through the
origin of τ-axis and obtain f2(-τ).
3. Consider the tilted function f2(-τ) as a
rigid frame and progress it along the
τ-axis by an amount say t0. The rigid
frame now represents the function
f2(t0-τ).
1-May-17 Prof K.Venkat Reddy 14

7
 5/1/2017

SUMMARY OF STEPS IN GRAPHICAL

INTERPRETATION OF CONVOLUTION
4. The product of the fn represented by
this displaced rigid frame with f1(τ)
represents the fn f1(τ) f2(t0-τ) and the
area under this curve is given by

f (t )   f1   f 2 t0    d  [ f1 (t ) * f 2 (t )]t t0 5.8


5. Repeat this procedure for different

values of t by successively progressing
the frame by different amounts and
find the values of the convolution
function f1(t)*f2(t) at those values.
1-May-17 Prof K.Venkat Reddy 15

SUMMARY OF STEPS IN GRAPHICAL

INTERPRETATION OF CONVOLUTION
Note:
 To find the convolution function
f1(t)*f2(t)
◦ for the positive values of t, we
progress the frame along the positive
τ-axis,
◦ where as for the negative values of t,
the frame is progressed along the
negative τ-axis.

1-May-17 Prof K.Venkat Reddy 16

8
 5/1/2017

CONVOLUTION OF A FUNCTION WITH

A UNIT IMPULSE FUNCTION

 The convolution of a function f(t) with a

unit impulse function δ(t) yields the
function f(t) itself .i.e.

f t  * t    f   t    d  f (t )


 (Using sifting property of an impulse function)

 By using convolution theorem if
f t   F   and  t   1
 Then f t * (t )  F  
 Then consequently f t * (t )  f (t )
1-May-17 Prof K.Venkat Reddy 17

CONVOLUTION OF A FUNCTION WITH

A UNIT IMPULSE FUNCTION

 Since the impulse is concentrated at one

point and has an area of unity, the
convolution integral yields the function
f(t).
 Similarly

f t  *  (t  T )  f (t  T )
f t  t1  *  (t  t 2 )  f (t  t1  t 2 )
 t  t1  *  (t  t 2 )   (t  t1  t 2 )

1-May-17 Prof K.Venkat Reddy 18

9
 5/1/2017

TIME – CONVOLUTION THEOREM

Statement: If f1 t   F1   and f 2 t   F2  
Then 
f1 t  * f 2 t    f   f t    d  F   F  
1 2 1 2


 Solution:



F  T  f1 t  * f 2 t      f1   f 2 t    d  e  jt dt
  
 By changing order of integration

  
F  T  f1 t  * f 2 t    1  f 2 t   e dt  d
f    jt

1-May-17 Prof K.Venkat Reddy 19

TIME – CONVOLUTION THEOREM

 
F  T  f1 t  * f 2 t     f1   e  j  d  F2    F1   F2  
  

 The convolution of two functions in

the time-domain is equivalent to
multiplication of their spectra in
the frequency domain.

F  T  f1 t  * f 2 t   F1   F2  

1-May-17 Prof K.Venkat Reddy 20

10
 5/1/2017

FREQUENCY – CONVOLUTION
THEOREM
Statement: If f1 t   F1   and f 2 t   F2  
Then 
1
f1 t  f 2 t  
1
F1   * F2     F u  F   u  du
2 2
1 2


Proof:
 1

 1 

I  F T F1   * F2    1    F u  F   u  du  e
jt
d
 2  2  2
1 2
 

 By interchanging order of
integration
 1

 1 

I  F T F1   * F2    1 F1 u   2  F   u e
jt
d  du
 2  2
2
 
1-May-17 Prof K.Venkat Reddy 21

FREQUENCY – CONVOLUTION
THEOREM
 Using frequency-shifting property

 1
I  F T F1   * F2    f 2 t  1  F u e
jut
du
 2 2
1
 

 f1 (t ) f 2 t 

 The multiplication of two functions

in the time-domain is equivalent to
convolution of their spectra in the
frequency domain.
I .F  T 21 [ F1   * F2  ]  f1 t  f 2 t 
1-May-17 Prof K.Venkat Reddy 22

11
 5/1/2017

CONVOLUTION THEOREM

Example: Given the unit-impulse

response h(t )  2e 2t ut  and the driving
function f (t )  2e ut  . Evaluate the
t

response of the system.

Solution : The response of a
system is given by
t t
y(t )  f (t ) * h(t )   f ( ) h(t   ) d   2e  2e 2(t  ) d
0 0
t
y(t )  4 e e d  4 e  t
 2t  t
 2t
e  4e  2t
e  4 e 2t (et  1)
0 0 0

y(t )  4 et (1  et ) u(t )

1-May-17 Prof K.Venkat Reddy 23

CORRELATION OF SIGNALS

 Correlation mathematically resembles

convolution.
 Correlation between two signals is to
measure the degree to which the two
signals are similar and thus to extract
some information that depends to a
large extent on the application.
 Correlation of signals is often
encountered in digital communications,
RADAR, Sonar, geology etc.

1-May-17 Prof K.Venkat Reddy 24

12
 5/1/2017

CORRELATION OF SIGNALS

 In case of RADAR application:

 The transmitted signal and the received
signal will be compared.
 The received signal will be the delayed
form of transmitted signal.
 The received signal may be distorted by
the addition of some noise to it.
 By comparison action may be taken
upon the target.

1-May-17 Prof K.Venkat Reddy 25

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 The signal may be compared on the
basis of similarity of waveforms. The
comparison may be based upon the
amount of component of one waveform
contained in the other waveform.
 If f1(t) and f2(t) are two functional
waveforms, f1(t) contain an amount of
C12 f2(t) of that particular waveform
f2(t) in the interval (t1,t2) where
f 1 t  f 2 t  dt
t2

C12 

t1

f 22 t  dt
t2

t1

1-May-17 Prof K.Venkat Reddy 26

13
 5/1/2017

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 The magnitude of the interval in the
numerator might be taken as an
indication of the similarity of the two
signals.
If C12  0  t f1 t  f 2 t  dt  0
t 2

 1

 Then two signals f1(t) & f2(t) are said be

Orthogonal. In general, we compare two
signals for the entire interval (-∞,∞),
then above equation can be written as

 f1 t  f 2 t  dt  0


1-May-17 Prof K.Venkat Reddy 27

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 In order to search for a similarity
between the waveforms, we must shift
one waveform with respect to the other
by various amounts and see whether
similarity exists for some amount of
shift of one waveform with respect to
the other.

Or
12 ( )   f1 t  f 2 t    dt


 Where τ is known as searching

parameter.
1-May-17 Prof K.Venkat Reddy 28

14
 5/1/2017

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 This is a function of time. This integral
of a cross-correlation between f1(t) and
f2(t) is denoted by Ф12(τ). The shift of τ
may be in the -ve direction as before or
may be in the +ve direction and is given
as 
12 ( )   f1 t    f 2 t  dt


The cross-correlation function can be

obtained by shifting the function f2(t)
by τ seconds and multiplying it with
f1(t).
1-May-17 Prof K.Venkat Reddy 29

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 The area under the product curve gives
the value of ɸ12(τ).
 The function f2(t) may be shifted by
different amounts.
 The area under the product curve for
various values of τ (shift) generates
ɸ12(τ).
 There is a similarity between the
operation of correlation and convolution.

1-May-17 Prof K.Venkat Reddy 30

15
 5/1/2017

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 The graphical operation of correlation is
similar to convolution in every respect
except that the function f2(t) is not
tilted back about the vertical axis.
 The correlation of f1(t) and f2(t) is
equivalent to the convolution of f1(t)
and f2(-t).
 The cross correlation fn provides the
measure of similarity between signals
f1(t) and f2(t) as a fn of searching
parameter τ .i.e., the two waveforms
are similar but shifted with respect to τ. 31
1-May-17 Prof K.Venkat Reddy

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 If the cross correlation vanishes
everywhere, the two signals are said to
be uncorrelated.



1-May-17 Prof K.Venkat Reddy 32

16
 5/1/2017

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 Similarly 
 21 ( )   f 2 t  f1 t    dt


 Replace t-τ with x then

 
 21 ( )   f 2 x    f1 x  dx   f1 t  f 2 t  ( )  dt 12 ( )
 

 So  21 ( )  12 ( )
 Also from
 
12 ( )   f1 t  f 2 t    dt   f1 x    f 2 x)  dx
 


12 ( )   f 2 t  f1 t  ( ))  dt  21( )


12 ( )   21 ( )
1-May-17 Prof K.Venkat Reddy 33

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 If f1 t   F1   and f 2 t   F2  
 So that f1  t   F1    and f 2  t   F2   
 [Using scaling property i.e. f at   1 F   
a a
with a=1]
 And 12 ( )  f1 (t ) * f 2 (t )  F1  F2   
 21 ( )  f1 (t ) * f 2 (t )  F1   F2  

11 ( )  f1 (t ) * f1 (t )  F1  F1     F1 ( )   S1 ( )

2

 Where S1 ( ) is the energy density

spectrum of f1 (t )
1-May-17 Prof K.Venkat Reddy 34

17
 5/1/2017

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 The above equation may also be
expressed as

F1 ( ) F2 ( )   12 ( ) e  j  d



F1 ( ) F2 ( )    21( ) e  j  d



F1 ( )  F1 ( ) F1 ( )   11( ) e  j  d
2


1-May-17 Prof K.Venkat Reddy 35

CORRELATION AS A TOOL FOR

SIGNAL COMPARISON
 The Fourier transform of 11( ) yields π
times the energy density function of
f1(t)
 The energy density function

F  
1
S1 ( ) 
2


1  1 
E  2.  F ( ) d   F ( ) d
2 2

2   


E   S ( ) d
0
1-May-17 Prof K.Venkat Reddy 36

18
 5/1/2017

AUTOCORRELATION OF
FUNCTIONS
 The concept of comparison of two
signals may be applied to a signal with
itself, when shifted by an amount of τ
known as Autocorrelation. It can be
defined as
 ( )   f t  f t    dt

11 1 1


 f1 t    f1 t  dt by replacing t   with t


 f1 t  f1 t   )  dt 11 ( )


11 ( )  11 ( )

 Auto-correlation function is an even
function of τ.
1-May-17 Prof K.Venkat Reddy 37

PROPERTIES OF AUTO-
CORRELATION OF FUNCTIONS
 The Fourier transform of auto-
correlation function is
  
 11( )e  j  d    f1 t  f1 t   e  j  dt d
  

f1 t e  jt   f1 t   e j ( t  ) d dt
  


11( )e  j  d  
   

 Substitute t-τ=x then dt=-dx. As τ varies from

-∞ to +∞ x varies from -∞ to +∞
 so
f1 t e  jt   f1 x e jx dx dt
  


11( )e  j  d  
   


 11( )e  j  d  F1 ( ) F1 ( )  F1 ( )
2


1-May-17 Prof K.Venkat Reddy 38

19
 5/1/2017

PROPERTIES OF AUTO-
CORRELATION OF FUNCTIONS
 Hence
1. The Fourier transform of the
autocorrelation function is i.e. energy
density spectrum.
11 ( ) 
 F1 ( )
F .T 2

2. The autocorrelation function at the

origin is equal to the energy of the
signal. 
i.e.  11 ( )   f1 t  f1 t    dt
and 
11 (0)  
2
f1 (t ) dt

1-May-17 Prof K.Venkat Reddy 39

PROPERTIES OF AUTO-
CORRELATION OF FUNCTIONS
3. The autocorrelation function will be
maximum at τ=0 (no shift).i.e.
Consider the

integral
 [ f1 (t )  f1 t   ]2 dt


This integral is a positive, non-zero

quantity for τ≠0.

 [ f1 (t )  f1 t   ]2 dt  0

or
  
 f12 (t ) dt   f12 t    dt  2 f1 (t ) f1 t    dt > 0
  

1-May-17 Prof K.Venkat Reddy 40

20
 5/1/2017

PROPERTIES OF AUTO-
CORRELATION OF FUNCTIONS
 The first two integrals have the same
value and it is given by 11 (0)
211 (0)  211 ( )  0

or 11 (0)  11 ( ) for τ≠0

Therefore 11(0)  11( ) for   0

4. The autocorrelation function becomes
smaller at larger values of τ.
Limit
 ( )  0
   11

1-May-17 Prof K.Venkat Reddy 41

1-May-17 Prof K.Venkat Reddy 42

21