Principles of Communication-Part 1

Professor Aditya K. Jagannatham

Department of Electrical Engineering
Indian Institute of Technology Kanpur
Module 1
Lecture No 4
Fourier Transform, Inverse Fourier Transform of Continuous Signals, Example of FT
of Pulse and Sine Function

Keywords: Fourier Transform, Inverse Fourier Transform of Continuous Signals

Hello, welcome to another module in this massive open online course. So in this module let
us start looking at the Fourier transform, remember we are looking at the spectra of signals
and we said that spectra can be of two types, one is the discrete Fourier series, which is for a
periodic signal and the other one is the Fourier transform which is defined for an aperiodic
continuous time signal x(t).

So let us start looking at the Fourier transform, which we can be abbreviated as F.T and
remember this is for the spectrum of an aperiodic continuous time signal x(t), that is the
signal x(t). Previously we had looked at the spectrum that is the discrete Fourier series of a
periodic signal x(t). Now we are looking at the Fourier transform, which gives the spectrum
of a continuous aperiodic times signal x(t), alright.

(Refer Slide Time: 2:55)

So let us consider a signal x(t), then its Fourier transform X(f), which is the Fourier transform


 x(t )e
 j 2 Ft
of x(t), is given as X ( f )  dt where F is the frequency, alright.

(Refer Slide Time: 4:28)

The signal x(t) and its Fourier transform form a Fourier transform pair and it is denoted as,
x (t )  X ( f ) , so this forms a Fourier transform pair. That is the signal x(t) and X(f) forms a
Fourier transform pair. Remember, the signal x(t) is in the time domain and X(f) is in the
frequency domain and these are two totally different domains, alright.

So the signal x(t) is in the time domain and X(f) is in the frequency domain, and note that
even though x(t) might be real but this X(f) is a complex quantity. In fact, e  j 2 Ft is a complex
sinusoid, alright. So the Fourier transform X(f) is in general a complex quantity.
(Refer Slide Time: 6:18)

And therefore we have two components of X(f), one is the magnitude, so I have the
magnitude, this denotes the magnitude spectrum, then we have the angle that is the phase of
X(f) which denotes the the phase spectrum. That is, the magnitude X(f) is in general a
complex quantity and it denotes the magnitude spectrum and the phase of X(f) denotes the
phase spectrum, okay alright. And of course, now what we have found so far is given x(t),
how to find the Fourier transform X(f) in the frequency domain. Of course, we are also
interested in the inverse problem that is given X(f), how to find the Fourier transform X(f),
that is, how to find the corresponding time domain signal and this is given through the
corresponding inverse Fourier transform.
(Refer Slide Time: 7:50)

So what we have shown is given x(t), how to find X(f). And now what we are interested in is
given X(f) what is the corresponding time domain signal x(t) and we are saying this is given
by the inverse Fourier transform the IFT. The inverse Fourier transform can be defined as


 X ( f )e
j 2 Ft
follows that is the signal x(t) corresponding to X(f) is given as x(t )  df .


(Refer Slide Time: 9:22)

So notice that for the Fourier transform we have an e  j 2 Ft , for the inverse Fourier transform
we have an e j 2 Ft . Thus basically we have defined Fourier transform, which gives the
spectrum of a continuous time aperiodic signal x(t) and we have also shown how to derive the
expression for the continuous time signal in the time domain given the Fourier transform in
the frequency domain.

So the Fourier transform of the continuous time signal is given by the expression for the
Fourier transform, the time domain signal from the Fourier transform that is capital X(f) is
given by the inverse Fourier transform, alright. So let us now do a simple example to
compute the Fourier transform, okay.

(Refer Slide Time: 10:14)

So let us now do a simple example to compute the Fourier transform, okay. Consider the
pulse x(t), which is defined as, PT(t). Let me first draw the signal in the time domain. It is the
pulse of width T which can be denoted as follows that is the pulse of height unity from
T T
 to . So the width of the pulse is T and outside of this window of size T, this is 0. So
2 2

x(t )  PT (t )  1,0,
T
t
2
otherwise
, alright.

So this is a pulse or rectangular window of width T, height equal to 1 centered at 0 which

T T
means it expands from  to . This is a very simple signal and of course, it is of lot of
2 2
relevance in the communication system.
(Refer Slide Time: 13:09)

We are going to look at the relevance of this in the coming modules. Let us now find the
Fourier transform of this signal PT(t), okay. We know that the Fourier transform X(f) is given

T T
 x(t )e
 j 2 Ft
as X ( f )  dt and x(t) is nonzero only from  to in which it is 1. Let us first

2 2
T
 2

 P t  e e
 j 2 Ft  j 2 Ft
substitute x(t) as PT(t) so, X ( f )  T dt  dt .
 
T
2
(Refer Slide Time: 14:24)

T
e  j 2 Ft 2
Now evaluating this integral, we have integral of e  j 2 Ft , which is PT ( f )  .
( j 2 F )  T
2

(Refer Slide Time: 15:07)

T T
 j 2 F j 2 F
e e2 2
And now substituting the limits we have PT ( f )  and this is equal to
( j 2 F )

 T
2 j sin  2 F 
 2
.
  j 2 F 
 sin  FT 
The 2 j , 2 j cancels and the 2 here, 2 here goes. So what we have is PT ( f )  ,
F
okay. So this is the expression for the Fourier transform of the rectangular window PT(t) and
remember we can also say that PT(t) and its spectrum PT(f), these form a Fourier transform
pair.

(Refer Slide Time: 18:16)

And in fact, now I can also write this PT(f) as follows that is, I can multiply the numerator
T sin  FT 
and denominator by T. I can now write this as PT ( f )   Tsinc  FT  where
 FT
basically this function sinc(x) is a very interesting function and arises frequently in
sin  x 
communication and signal processing. This is defined as sinc( x)  .
x

(Refer Slide Time: 19:43)

sin  x 
Now look at this, the sinc function at x  0 that is, sinc  0   lim sinc  x  is  lim 1
x 0 x 0 x
. So people simply say that sinc  0  is simply set as equal to 1 although it is not defined at 0

sin  x
because at x  0 is undefined. So we consider the limit as x  0 and therefore
x
sinc  0   lim sinc  x   1 .
x 0
(Refer Slide Time: 21:26)

sin( x)
And further notice that sinc( x)   0 if sin( x)  0   x equals any multiple of k
x
that is,  x  k implies x is equal to any integer k other than 0. So at x  0 we have
sin( k )
sin c( x)  1 except that at all other integers x  k sinc(k )  but sin( k ) for any integer
k
k is 0, so this is 0. So for all integers k  0 sin c( x) that is sinc (k ) is 0, okay.

(Refer Slide Time: 22:46)

So therefore, now look at this we have our function P T(f), now let us go back using these
properties your PT  f   Tsinc  FT  , if at f  0 , this becomes Tsinc  0  this is equal to T.

k  k
Now, at f  , where k is any integer, we have PT  f   Tsinc  F   Tsinc(k )  0 .
T  T

(Refer Slide Time: 24:21)

T sin( FT )
And if you can look at this, this is . So this looks like a sine function, except for
 FT
1
the amplitude which is constantly decreasing due to this factor .
F
(Refer Slide Time: 25:27)

Let us try to plot this sinc function in the frequency domain, this function looks as this is the
frequency axis, this is amplitude and at f  0 , this is T and then it will have an amplitude
1
that is continuously decreasing and these points where it is 0 are integer multiples of , so
T
1 2 1 2
this is , , the first null on the negative axis is  ,  and therefore if you look at this
T T T T
sinc function, it has a decaying envelope and if you look at the null to null bandwidth, that
1 1
will be  to .
T T

So the spacing in the frequency axis between the first null on the negative frequency side and
2
the first null on the positive frequency is known as the null to null bandwidth that is ,
T
where T is the duration of the pulse, alright. So there are several nulls, in fact, at every
1 k
multiple integer multiple k of that is , this is except when k is equal to 0 at F is equal to
T T
0 this is T, alright.
(Refer Slide Time: 28:01)

So we can say that our pulse PT(t) has now its Fourier transform pair, which is Tsinc  FT  ,

alright and this sinc function has very significant applications in communication and it is of
fundamental importance in communication, alright.

So, in this module what we have done is basically we have defined Fourier transform of a
continuous time aperiodic signal, the inverse Fourier transform, then given the Fourier
transform, how to derive the time domain signal, and also illustrated the computation of the
Fourier transform through an example.

(Refer Slide Time: 29:06)

And just to conclude remember one can also now say that using the inverse Fourier transform

we have PT  t   PT ( f ) and therefore you can say that  Tsinc  FT e
j 2 Ft
df will be PT(t),


that is your time domain pulse, alright. So if you take the Fourier transform P T(f) in the
frequency domain compute the inverse Fourier transform you must get back the time domain
pulse, alright. So we will stop this module here and look other aspects subsequent modules,
thank you.