Vous êtes sur la page 1sur 22

# Fourier Analysis of Signals &

Systems
Assistant Professor, UET Taxila

Fourier Analysis
Decompose the signal in terms of sinusoidal
(or complex exponential) components.
With such decomposition, the signal is said to
be represented in the frequency domain.
Fourier analysis applies to both periodic and
aperiodic signals.
12/7/2010

## Fourier Analysis (cntd.)

When analyzing periodic waveforms, the FOURIER
SERIES applies.
When analyzing aperiodic signals, we apply the
FOURIER TRANSFORM.
Furthermore, signal may be continuous or discrete, so
there are FOUR varieties of the analysis technique, i.e.:
The continuous Fourier series
The continuous Fourier transform
The discrete Fourier series, also called the discrete Fourier
Transform
The discrete time Fourier transform.
12/7/2010

## Fourier Analysis (cntd.)

Fourier analysis allows a signal to be expressed as
a series of sinusoidal in ascending frequency.
These sinusoids are represented as a set of cosine
and sine functions.
But this is not strictly necessary: it is quite
permissible to express the series as a set of sine
function only (or cosine function only), since each
is simply a phase shifted version of the other.

12/7/2010

## Fourier Analysis (cntd.)

Fourier transformation is used in

12/7/2010

Signal restoration
Signal compression and decompression
Signal modulation and demodulation
Phase shifting
Digital wave synthesis
Interpolation and so on
Every time you use your mobile phone, DVD, CD or MP3
player, Fourier is at the heart of the matter.
Dr. Sarmad Sohaib (Signals & Systems)

## The continuous trigonometric Fourier

series for periodic signals
A periodic waveform may be represented as a series
involving cosine and sine coefficients: in general,
although not always, the series in infinite. Thus the
synthesis equation may be written as

k =1

k =1

## where 0 represents the fundamental frequency of

the system, given by
0 =

12/7/2010

2
T

## The continuous trigonometric Fourier

series for periodic signals (cntd.)

where
T represents the time duration of a single period
of the waveform.
A0 coefficient represents the mean signal level
(also known a the DC level or zeroth harmonic),
Bk is the coefficient representing the magnitude
of the cosine wave content of the kth harmonic
Ck is the coefficient representing the magnitude
of the sine wave content of the kth harmonic.

12/7/2010

## The continuous trigonometric Fourier

series for periodic signals (cntd.)
In order to calculate A0, Bk, and Ck, we employ the analysis
equations as follows:
T

1
A0 = x(t )dt
T 0
T

2
Bk = x(t ) cos(k0t )dt
T 0

the limits 0 to T
means single time
period, the limit
does not have to
start specifically
from 0.

2
Ck = x(t ) sin(k0t )dt
T 0

## The Equation for calculating Bk and Ck allows us to calculate how much of a

given sinusoid, at a particular frequency, is presented in the waveform x(t).

12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

Example-1
Calculate the Fourier components of the square wave pulse
train shown below over the range t=- to , which is defined
as

0,
x(t ) =
1,

1
<t <0
2
1
0<t <
2
x(t )

1
0

t
12/7/2010

1/ 2

1/ 2

## Dr. Sarmad Sohaib (Signals & Systems)

t
9

Example-1 (cntd.)
In this example, t ranges from - to , T=1. The mean signal level is
given by
T

1
A0 = x(t )dt
T 0
1/2

1
= x(t )dt
1 1/2
0

1/2

1
= (0)dt + (1)dt =
2
1/2
0
Clearly, we could have obtained this by inspection (DC level =1/2).
12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

10

Example-1 (cntd.)
The cosine terms are given by
1/2

2
2
Bk = x(t ) cos(k0t )dt = x(t ) cos(k0t )dt
T 0
1 1/2
1/2
0

## = 2 (0) cos(k0t )dt + (1) cos(k0t )dt

0
1/2

1/2

1/2

sin(k0t )
sin(k 2 t )
= 2
= 2
k 2 0
k 0 0

From the final line of the above equation we deduce that whatever the
value of k, Bk is always zero:
Bk=0,
k=1,2,3,
12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

11

Example-1 (cntd.)
The sine terms are given by
1/2

2
2
Ck = x(t ) sin(k0t )dt = x(t ) sin(k0t )dt
T 0
1 1/2
1/2
0

0
1/ 2

1/2

cos(k0t )
1
cos(k 2 t )
= 2
=
2

=
[cos(0) cos(k )]

k 0 0
k 2 0
k

1/ 2

have:
2
Ck =
For k = odd
k
12/7/2010

12

## Example-1 (cntd.) + Observations

So the Fourier series for this square wave
pulse train is
2
1
1
1

3
5
7

## Therefore we need infinitely large number of harmonics to

recreate the ideal square wave pulse train.
This is because the waveform exhibits discontinuities where it
changes state from zero to one and vice versa.
12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

13

Observations
The signal in example-1 is odd, therefore has
only sine terms (as Bk=0).
Therefore,
Odd signal all Fourier coefficients would be sine
terms
Even signal all Fourier coefficients would be
cosine terms
Neither even nor odd Fourier series would
contain both cosine and sine harmonics.
12/7/2010

14

## The continuous trigonometric Fourier

series for aperiodic signals
If we have a signal that does not repeat, but
we want to find the Fourier coefficients within
some range, say -T to T, we pretend that it
is cyclical outside of this range.
We then use exactly the same analysis
equations as for continuous periodic signals.

12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

15

Example-2
Find the Fourier coefficients for the function
x(t)=2t over the range - to .
x(t )

1
0
1/ 2

1/ 2

12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

16

Example-2 (contd.)
Construct a series of repeating ramps outside
of the range of interest, shown by dotted
lines.
Function has odd symmetry, which means:
a) the DC term, A0, is zero
b) there are no cosine terms, Bk, in the Fourier
expansion

12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

17

Example-2 (contd.)
The sine terms are thus:
T

1/2

2
Ck = x(t ) sin( k0t )dt = 2 2t sin(k0t )dt
T 0
1/2

## Now the integral here is product of two functions, so we use the

integration by parts formula, i.e.

u dv = uv v du
where, u = 2t , hence

du
= 2, and du = 2dt
dt

12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

1
k0

cos(k0t ).
18

Example-2 (contd.)
Inserting these into formula, we get
1/2

4t

4
Ck =
cos ( k0t ) +
k 0
1/2 k0

1/2

cos(k0t )dt

1/2

1/2

4t

4
1/2
=
cos ( k0t ) + 2 2 [sin(k0t ) ]1/2
k 0
1/2 k 0
Fortunately, the final sine term in the above equation cancels out, leaving

2
2
2
Ck =
cos(k )
cos(k ) =
cos(k )
2 k
2 k
k

12/7/2010

## Dr. Sarmad Sohaib (Signals & Systems)

19

Example-2 (contd.)
Hence when k is odd, we have

2
Ck =
k

Ck =

2
k

x(t ) = 0 +

12/7/2010

2
1
1
1
1

sin(

t
)

sin(2

t
)
+
sin(3

t
)

sin(4

t
)
+
sin(5

t
)
+

0
0
0
0
0

2
3
4
5

20

signals at home.

12/7/2010

21

## Example-3 (home work)

Determine the Fourier series of the
rectangular pulse train signal illustrated below

x(t )

12/7/2010

Tp

0
2
2

Tp

22