0% found this document useful (0 votes)
40 views48 pages

Frequency Analysis

Uploaded by

Somnath
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
40 views48 pages

Frequency Analysis

Uploaded by

Somnath
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 48

1

What is Frequency Analysis?

• Process of obtaining the Spectrum of a given


signal using basic Mathematical Tools

2
Why Frequency Analysis?

input LTI
output
System

Input may consist of Output depends on


various frequencies frequency of input

3
Tools for Frequency Analysis

Continuous Time Discrete Time


Domain Domain

Laplace Transform Z-Transform

Fourier Transform Discrete Fourier


Transform

4
5
6
A tool for frequency analysis

Simplifies the analysis of signals

Fourier transform
x(n) X()
Inverse
Fourier transform
In time domain In frequency domain

Mathematical representation:

X() =  x(n) . e-jn
n = -

7
Inverse Fourier Transform

Fourier transform
x(n) X()
Inverse
Fourier transform

Mathematical representation:


1
x(n) = 2
X() . e jn
-

8
Magnitude and Phase Plots

Useful in graphically analyzing frequency


content of any signal in frequency domain

Helps to indicate very quickly the


presence or absence of any frequency
components

9
Consider H() = HRR() + j HI()
Real part Imaginary part

Magnitude = ( )2 + ( )2 = | H() |

Phase = tan-1 =  H()

In Polar Form,
H() =
10
Consider H() = 0.5 . (1+cos ) – j . 0.5 . Sin 
Real part Imaginary part
| H() |
1


Magnitude = cos
2

- 0 

 H()
/2

 
Phase = - - 0 
2
- / 2

11
Limitations of Fourier Transform

• X() is a continuous function of 


i.e. X() exists for all values of 
• Digital Computers can only work with
discrete signals
• Therefore, X() cannot be computed using
digital computers.

12
13
The Discrete Fourier Transform
• Digital Computers require discrete signals
• Thus, we sample the Fourier Transform
X() to obtain the Discrete Fourier
Transform (DFT)
• DFT is obtained by sampling the Fourier
Transform
X(k) = X()|k=2k/N k = 0, 1, …, N-1

DFT Fourier Sampling


14
Transform of FT
Mathematical Expression
We have the equation for Fourier Transform:

X() =  x(n) . e-jn
n = -

Here,  ranges from - to +, continuously. Therefore,


it is impossible to calculate X() on a digital computer.

Hence we compute X() only at discrete values of .

X(k) = X()| =2k/N


N-1
=  x(n) . e-j 2kn/N k = 0, 1, …, N-1
n=0
15
Magnitude Plot for |X()|
Fourier Transform
 is continuous

 Frequency

Magnitude Plot |X(k)|


for Discrete k
k k is discrete
k
Fourier Transform
k k
k k

Frequency
16
Samples
The Twiddle Factor

N-1
-j(2 )
X(k) =  x(n) e . kn /N
n=0

Let WN = e -j 2 /N Twiddle Factor

N-1

X(k) =  x(n) . WNkn


n=0

17
Periodicity property WN(k+N) = WNk

Symmetry property WN(k+ N/2) = -WNk

WNm = WN/m

18
19
Statement:

WN(k+N) = WNk
Proof:
We have,
e– j2 = cos 2 - j sin 2
WN = e-j2/N
=1–j0
-j2/N (k+N)
W N
(k+N)
= (e ) = 1 always
= e-j2 (k+N) /N
= (e-j2k /N) . (e-j2 N/N )
= (e -j2 k )
/N
20
21
Statement:
(k+ N/2)
WN = -WNk
Proof:
We have,
e– j = cos  - j sin 
WN = e-j2/N
= -1 – j 0
(k+N/2)
W N
(k+N/2)
= (e -j2/N
) = -1 always
= e-j2 (k+N/2) /N
= (e-j2k /N) . (e-j 2N/2N )
= - (e -j k )
/N
22
To prove that,

WNm = W(N/m)

We have,
WN = e-j2/N

WNm = (e-j2/N)m
= e-j2 m /N
- j 2 / N/m
=e ( )

= W(N/m)
23
Linearity
Periodicity
Circular Symmetry
Time Reversal
Circular Time Shift
Circular Frequency Shift
24
25
DFT
x1(n) N
X1(k)
…Eq.1
DFT
x2(n) N
X2(k)

Statement:

DFT
a1 x1(n) + a2 x2(n)
. .
a1 . X1(k) + a2 . X2(k)
N

26
DFT
x1(n) a1 N-point

a1.X1(k) + a2.X2(k)

a2 DFT
x2(n) N-point

Equal

x1(n) a1

DFT
N-point DFT { a1.x1(n) + a2.x2(n) }

x2(n) a2

27
Example of Linearity
Consider two sequences, x1(n) and x2(n)
x1(n) = {1, 1, 1, 1}

x2(n) = {1, 2, 3, 4}

Let x3(n) be the sum of x1(n) and x2(n)

x1(n) = {1, 1, 1, 1}
x3(n) = +
x2(n) = {1, 2, 3, 4}
28
x1(n)
1 1 1 1

0 1 2 3
x3(n)
Time
5

+ = 3
4

x2(n)
2
4

2 0 1 2 3 Time
1

0 1 2 3 Time

29
x3(n) X3(k)
14

2
2.82 2.82
2

0 1 2 3 Time 0 1 2 3 Frequency
samples

DFT
x3(n) = {2, 3, 4, 5} 4
X3(k) = {14, 2.82, 2, 2.82}

30
The two sequences are
x1(n) = {1, 1, 1, 1}

x2(n) = {1, 2, 3, 4}

Their corresponding DFTs are,

DFT
x1(n) = {1, 1, 1, 1} 4
X1(k) = {4, 0, 0, 0}

DFT
x2(n) = {1, 2, 3, 4} X2(k) = {10, 2.82, 2, 2.82}
4

31
X1(k) 4

0 0 0
X3’(k) 14

0 1 2 3 Frequency
samples

+
10
=
X2(k) 2.82 2.82
2

0 1 2 3 Frequency
2.82 2.82
2 samples

0 1 2 3 Frequency
samples

32
X3(k) 14
X3’(k) 14

2.82 2.82 2.82 2.82


2 2

0 1 2 3 Frequency 0 1 2 3 Frequency
samples samples

Equal

Thus, the DFT is a Linear Operation, i.e.


DFT
a1 x1(n) + a2 x2(n)
. .
a1 . X1(k) + a2 . X2(k)
N
33
34
DFT
x(n) N
X(k) …Eq.1

Statement:
If x(n+N) = x(n) for all n ,

Then X(k+N) = X(k) for all k

35
x(n) = { 1, 2, 3, 4 }

4
3
2
1

0 1 2 3 Time

xp(n) = {…, 1, 2, 3, 4, 1, 2, 3, 4, …}

4 4 4 4
4
3 3 3 3
3
2 2 2 2
2
1 1 1 1
1

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11
Time

36
xp(n) = {…, 1, 2, 3, 4, 1, 2, 3, 4, …}

4 4 4
4
3 3 3
3
2 2 2
2
1 1 1
1

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11
Time
samples

xp(n) shifted by 4 samples

4 4 4
4
3 3 3
3
2 2 2
2
1 1 1
1

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 Time
samples

37
xp(n) = {…, 1, 2, 3, 4, 1, 2, 3, 4, …}

4 4 4 4
4
3 3 3 3
3
2 2 2 2
2
1 1 1 1
1

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 Time
samples

DFT of xp(n)
10 10 10 10 10

2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82

2 2 2 2 2

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 Frequency
samples

38
DFT of xp(n)
10 10 10 10 10

2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82

2 2 2 2 2

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11
Frequency
Xp(k) shifted by 4 samples samples
10 10 10 10

2.82 2.82 2.82 2.82 2.82 2.82 2.82 2.82


2 2 2 2

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 Frequency
samples
39
Representation of periodic signal

40
Consider the sequence x(n) as shown

x(n) = { 1, 2, 3, 4 }

4
3
2
1

0 1 2 3 Time

xp(n) = {…, 1, 2, 3, 4, 1, 2, 3, 4, …}

4 4 4 4
4
3 3 3 3
3
2 2 2 2
2
1 1 1 1
1

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11
Time
41
xp(n) shifted by 4 samples

4 4 4
4
3 3 3
3
2 2 2
2
1 1 1
1

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 11 Time
samples
This sequence is now circularly shifted
x (n) plotted circularly x(0)
by p4 samples since N=4
anti-clockwise
1
4
3
2

The periodic sequence xp(n)


canWe
thusconclude thatas:
be written x(1) 3
2
1
4 4 x(3)
3
2
1
a periodic signal can be
xp(n) = x(N-n)
conveniently plotted
This is represented
along a circle as follows:
x((n))N = x(N-n) 4
3
2
1
x(2)
42
43
Circularly Even Sequence
Original xp(n) 1

For a Circularly Even Sequence,


xp(n) = x(N – n) 2 2

Example:
xp(n) = {…,1, 2, 3, 2, 1, 2, 3, 2,…} 3

Thus,
x (4-0) = xp(0) = 1
x(4-1) = xp(1) = 2
x(4-2) = xp(2) = 3
x(4-3) = xp(3) = 2
x(N-n) = xp(n)
44
Circularly Odd Sequence
Original xp(n) 1

For a Circularly Odd Sequence,


xp(n) = -xp(N – n) 2 -2

Example:
xp(n) = {…,1, 2, 3, -2, 1, 2, 3, -2,…} 3

Thus,
xp(4-0) = xp(0) = 1
xp(4-1) = xp(1) = - (2)
xp(4-2) = xp(2) = 3
xp(4-3) = xp(3) = - (-2)
xp(N-n) = -xp(n)
45
46
DFT
x (n) N
X (k)

Statement:
DFT
x((-n))N = x(N-n) X((-k))N = X(N-k)
N

Consider the following sequence and its DFT

DFT
{ 1, 2, 3, 4 } { 10, 2.82, 2, 2.82 }
N

47
Original x(n) = x1(n) Original X(k) = X1(k)
x1(0) = 1 X1(0) =10

DFT
x1(1) = 2 4 = x1(3) X1(1) =2.82 2.82
4 = X1(3)

x1(2) = 3 X1(2) =2

x2(0) = X2(0) =

DFT
x2(1) = = x2(3) X2(1) =
4 = X2(3)

x2(2) = X2(2) =
Time Reversed x(n) = x((-n))N = x2(n) X2(k) = DFT of x2(n) = X((-k))N
48

You might also like