Representation of Signals
mywbut.com
Principles of Communication
CHAPTER 1
Representation of Signals
1.1 Introduction
The process of (electronic) communication involves the generation,
transmission and reception of various types of signals. The communication
process becomes fairly difficult, because:
a) the transmitted signals may have to travel long distances (there by
undergoing severe attenuation) before they can reach the destination i.e.,
the receiver.
b) of imperfections of the channel over which the signals have to travel
c) of interference due to other signals sharing the same channel and
d) of noise at the receiver input1.
In quite a few situations, the desired signal strength at the receiver input
may not be significantly stronger than the disturbance component present at that
point in the communication chain. (But for the above causes, the process of
communication would have been quite easy, if not trivial). In order to come up
with appropriate signal processing techniques, which enable us to extract the
desired signal from a distorted and noisy version of the transmitted signal, we
must clearly understand the nature and properties of the desired and undesired
signals present at various stages of a communication system. In this lesson, we
begin our study of this aspect of communication theory.
1
Complete statistical characterization of the noise will be given in chapter 3, namely, Random
Signals and Noise.
1.1
mywbut.com
Principles of Communication
Signals physically exist in the time domain and are usually expressed as a
function of the time parameter1. Because of this feature, it is not too difficult, at
least in the majority of the situations of interest to us, to visualize the signal
behavior in the Time Domain. In fact, it may even be possible to view the signals
on an oscilloscope. But equally important is the characterization of the signals in
the Frequency Domain or Spectral Domain. That is, we characterize the signal
in terms of its various frequency components (or its spectrum). Fourier analysis
(Fourier Series and Fourier Transform) helps us in arriving at the spectral
description of the pertinent signals.
xp (t ) = xp (t + T ) , (1.1)
1
We will not discuss the multi-dimensional signals such as picture signals, video signals, etc.
1.2
mywbut.com
Principles of Communication
the period of x p ( t ) .
1.3
mywbut.com
Principles of Communication
namely, A e (
j 2 πf t + ϕ)
or A exp ⎡⎣ j ( 2 π f t + ϕ ) ⎤⎦ , where
Both A and f are real and non-negative. As the radian frequency, ω (in
units of radians/ sec), is equal to 2 π f , the complex exponential can also written
j ( ωt + ϕ)
as A e . We use subscripts on A , f (or ω ) and ϕ to denote the specific
values of these parameters.
⎡ ⎛ π ⎞⎤
Fourier analysis uses cos ω t ⎢or sin ω t = cos⎜ ω t − ⎟ ⎥ in the represen-
⎣ ⎝ 2 ⎠⎦
( e j ωt )
∗
e j ωt +
cos ω t = ( ∗ denotes the complex conjugate)
2
e j ωt + e − j ωt
=
2
The term e − j ωt
or e − j 2 πf t
is referred to as the complex exponential at
the negative frequency − ω (or − f ).
1.4
mywbut.com
Principles of Communication
1
Let x p ( t ) be a periodic signal with period T0 . Then f0 = is called the
T0
namely,
∞ j 2 π n f0 t
xp (t ) = ∑ xn e (1.2a)
n = −∞
x p ( t ) as
1 − j 2 π n f0 t
xn = ∫T xp (t ) e dt (1.2b)
T0
0
where ∫ denotes the integral over any one period of x p ( t ) . Most often, we
T0
⎛ T T ⎞
use the interval ⎜ − 0 , 0 ⎟ or ( 0 , T0 ) . Eq. 1.2(a) is referred to as the
⎝ 2 2⎠
Exponential form of the Fourier series.
xn = xn e j ϕn (1.3)
where xn denotes the magnitude of the complex number and ϕn , the argument
1.5
mywbut.com
Principles of Communication
vs. n (or n f0 ) is called the magnitude spectrum, and ϕn vs. n (or n f0 ) is called
the phase spectrum. It is important to note that the spectrum of a periodic signal
exists only at discrete frequencies, namely, at n f0 , n = 0, ± 1, ± 2, ⋅ ⋅ ⋅ , etc.
1 j 2 π n f0 t
x− n = ∫T xp (t ) e dt
T0
0
= xn∗
That is, for a real periodic signal, we have the two symmetry properties, namely,
x− n = x n (1.4a)
ϕ- n = − ϕn (1.4b)
Properties of Eq. 1.4 are part of an if and only if (iff) relationship. That is, if
x p ( t ) is real, then Eq. 1.4 holds and if Eq. 1.4 holds, then x p ( t ) has to be real.
into a cosine term. As an example, let the only nonzero coefficients of a periodic
*
signal be x ± 2 , x± 1 , x 0 . x0 =X 0 implies, x0 is real and let
π π
j
= x2∗ and x = 3 e
j
x− 2 = 2 e 4 3 = x1∗ and x0 = 1. Then,
−1
π π π π
j j − j − j
xp (t ) = 2 e 4 e − j 4 π f0 t
+ 3e 3 e − j 2 π f0 t
+ 1 + 3e 3 e j 2 π f0 t
+ 2e 4 e j 4 π f0 t
1.6
mywbut.com
Principles of Communication
Example 1.1
For the unit amplitude rectangular pulse train shown in Fig. 1.2, let us
compute the Fourier series coefficients.
OFF during the remaining half. The fundamental frequency f0 = 250 Hz.
T0
4
1
= ∫T e− j 2 π n f0 t
dt
T0
− 0
4
⎛ nπ⎞
1 sin ⎜⎝ 2 ⎟⎠
=
T0 n π f0
1.7
mywbut.com
Principles of Communication
⎛nπ⎞
sin ⎜ ⎟
= ⎝ 2 ⎠
nπ
As can be seen from the equation for xn , all the Fourier coefficients are
real but could be bipolar (+ve or –ve). Hence ϕn is either zero or ± π for all n .
Fig. 1.3: Magnitude and phase spectra for the x p (t) of example 1.1
1.8
mywbut.com
Principles of Communication
1
i) x0 , the average or the DC value of the pulse train is . For any periodic
2
T0
2
1
signal, the average value is ∫ x p ( t ) dt .
T0 T
− 0
2
spectrum).
iii) the curve drawn with broken line in Fig. 1.3(a) is the envelope of the
magnitude spectrum. The envelope consists of several lobes and the
maximum value of each lobe keeps decreasing with increase in frequency.
iv) the plot of xn vs. frequency is symmetric and the plot of ϕn vs. frequency is
One of the functions that is useful in the study of Fourier analysis is the
sinc ( ) function defined by
sin ( π λ )
sinc λ = sinc ( λ ) = (1.5)
πλ
A plot of the sinc λ vs. λ along with a table of values are given in
appendix A1.1, at the end of the chapter.
1.9
mywbut.com
Principles of Communication
1 ⎛n⎞
as, xn = sinc ⎜ ⎟ .
2 ⎝2⎠
Exercise 1.1
⎛ τ ⎞
For the x p ( t ) of Fig .1.4, show that xn = ⎜ ⎟ sinc ( n f0 τ )
⎝ T0 ⎠
⎡x ⎤
The spectral plots 1 to 4 give the values of 20 log10 ⎢ n ⎥ for the
⎣⎢ x1 ⎥⎦
waveforms 1 to 4 respectively. The units for the above quantities are in decibels
(dB).
1.10
mywbut.com
Principles of Communication
1.
Waveform 1
Spectral Plot 1
1.11
mywbut.com
Principles of Communication
x1
two lines, namely at ± 10 kHz, and their values are 20 log10 = 0 dB.
x1
2.
Waveform 2
mywbut.com
Principles of Communication
τ 1
Waveform 2: Periodic square wave with = ; T0 = 0.1 msec.
T0 5
Comments on Spectral Plot 2: Values of various spectral components are:
i) fundamental: 0 dB
⎡ sin c ( 0.4 ) ⎤
ii) second harmonic: 20 log10 ⎢ ⎥
⎣⎢ sin c ( 0.2 ) ⎦⎥
= 20 log10 0.809 = − 0.924 dB
⎡ sin c ( 0.6 ) ⎤
iii) third harmonic: 20 log10 ⎢ ⎥
⎣⎢ sin c ( 0.2 ) ⎦⎥
⎡ 0.504 ⎤
= 20 log10 ⎢ ⎥
⎣ 0.935 ⎦
= 20 log10 (0.539) dB
= − 5.362
⎡ sin c ( 0.8 ) ⎤
iv) fourth Harmonic: 20 log10 ⎢ ⎥
⎣⎢ sin c ( 0.2 ) ⎦⎥
= − 12.04 dB
⎡ sin c (1) ⎤
v) fifth harmonic: 20 log10 ⎢ ⎥
⎢⎣ sin c ( 0.2 ) ⎥⎦
= 20 log10 ( 0 ) = − ∞
1.13
mywbut.com
Principles of Communication
3.
Waveform 3
1.14
mywbut.com
Principles of Communication
4.
Waveform 4
1.15
mywbut.com
Principles of Communication
The Example 1.1 and the periodic waveforms 1 to 4 all have fundamental
as part of their spectra. Based on this, we should not surmise that every periodic
signal must necessarily have a nonzero value for its fundamental. As a counter to
the conjuncture, let x p ( t ) = cos ( 20 π t ) cos ( 2000 π t ) .
This is periodic with period 100 msec. However, the only spectral
components that have nonzero magnitudes are at 990 Hz and 1010 Hz. That is,
the first 99 spectral components (inclusive of DC) are zero!
1
xp (t ) d t
2
Px p =
T0 ∫
T0
∞
∑
2
Pxp = xn
n = −∞
1.16
mywbut.com
Principles of Communication
x p ( t ) = xn e (
j 2 π n f0 t + ϕ )
then, Px p = xn2
In other words, Parseval’s power theorem implies that the total average power in
x p ( t ) is superposition of the average powers of the complex exponentials
present in it.
⎛ T ⎞
xp ⎜ t ± 0 ⎟ = − xp (t ) (1.6c)
⎝ 2⎠
With respect to the symmetries defined by Eq. 1.6, we have the following
special forms for the coefficients xn :
1.17
mywbut.com
Principles of Communication
T0 2
1
xp (t ) e
− j 2 π n f0 t
xn =
T0 ∫ dt
− T0 2
1 ⎡ ⎤
0 T0 2
= ⎢ ∫ x p ( t ) e− j 2 π n f0 t
dt + ∫ x p (t)e
− j 2 π n f0 t
d t⎥
T0 ⎢ − T 2 ⎥
⎣ 0 0 ⎦
Changing t to - t in the first integral, and noting x p ( − t ) = x p ( t ) ,
1⎡0 ⎤
T 2 T0 2
xp (t ) e xp (t ) e
j 2 π n f0 t − j 2 π n f0 t
T0 ⎢ 0∫ ∫
= ⎢ dt + d t⎥
⎣ 0 ⎥⎦
2 ⎡0 ⎤
T /2
= ⎢ ∫ x p ( t ) ( cos 2 π n f0 t ) d t ⎥
T0 ⎢ 0 ⎥⎦
⎣
x −n = x n .
ii) x p ( t ) odd:
1⎡ ⎤
0 T0 2
xn = ⎢ ∫ xp ( t ) e− j 2 π n f0 t
dt + ∫ xp ( t ) e−
j 2 π n f0 t
d t⎥
T0 ⎢ − T 2 ⎥
⎣ 0 0 ⎦
Changing t to - t , in the first integral and noting that x p ( − t ) = − x p ( t ) ,
we have
1⎡0 ⎤
T 2 T0 2
= ⎢ ∫ − xp (t ) e j 2 π n f0 t
d t + ∫ xp (t ) e
− j 2 π n f0 t
d t⎥
T0 ⎢ 0 ⎥⎦
⎣ 0
T0 2
1
x p ( t ) ⎡e
− j 2 π n f0 t j 2 π n f0 t ⎤
=
T0 ∫ ⎣
−e
⎦
dt
0
T0 2
2j
= − ∫ x p ( t ) sin ( 2 π n f0 t ) d t
T0 0
1.18
mywbut.com
Principles of Communication
1⎡ ⎤
0 T0 / 2
xn = ⎢ ∫ xp (t ) e− j 2 π n f0 t
dt + ∫ xp (t ) e− j 2 πn f0 t
d t⎥
T0 ⎢ −T / 2 ⎥⎦
⎣ 0 0
1 ⎡0 ⎤
T /2 T0 / 2
− j ( n ω0 t +π n )
xn = ⎢ ∫ x p (t + T0 /2) e d t + ∫ x p ( t )e − j n ω0 t
d t⎥
T0 ⎢ 0 ⎥⎦
⎣ 0
i) ∫ xp (t ) < ∞
T0
That is, the function is absolutely integrable over any period. It is easy to
verify that the above condition results in xn < ∞ for any n .
ii) x p ( t ) has only a finite number of maxima and minima over any period T0 .
iii) There are only finite number of finite discontinuities over any period1.
M
Let xM ( t ) = ∑ xn e j 2 π n f0 t (1.7)
n = −M
1
For examples of periodic signals that do not satisfy one or more of the conditions i) to iii), the
reader is referred to [1, 2] listed at the end of this chapter.
1.19
mywbut.com
Principles of Communication
and eM ( t ) = x p ( t ) − xM ( t )
Dirichlet conditions are sufficient but not necessary. Later on, we shall
have examples of Fourier series for functions that voilate some of the Dirichet
conditions.
∫ xp (t )
2
dt < ∞ , then the series converges in the mean. That is,
T0
∫ eM ( t )
2
lim dt = 0 (1.8)
M → ∞
T0
Note that Eq. 1.8 does not imply that lim eM ( t ) is zero. There could be
M →∞
nonzero values in lim eM ( t ) ; but they occur at isolated points, resulting in the
M →∞
integrable. This is illustrated in Fig. 1.5(a). From the figure, we see that
1.20
mywbut.com
Principles of Communication
1.21
mywbut.com
Principles of Communication
1.22
mywbut.com
Principles of Communication
⎛t ⎞
ii) Triangular pulse, tri ⎜ ⎟
⎝T ⎠
⎧ t
⎛t ⎞ ⎪1− , t ≤ T
tri ⎜ ⎟=⎨ T (1.9b)
⎝T ⎠ ⎪
⎩ 0 , outside
⎛t ⎞
iii) One-sided (decaying) exponential pulse, ex1 ⎜ ⎟ :
⎝T ⎠
⎧ −t
⎪e T , t > 0
⎪
⎛ t ⎞ ⎪ 1
ex1 ⎜ ⎟ = ⎨ , t = 0 (1.9c)
⎝T⎠ ⎪ 2
⎪ 0 , t < 0
⎪⎩
⎛t ⎞
iv) Two-sided (symmetrical) exponential pulse, ex 2 ⎜ ⎟ :
⎝T ⎠
⎧ −t
⎪e T , t > 0
⎛ ⎞ ⎪
t
ex 2 ⎜ ⎟ = ⎨ 1 , t = 0 (1.9d)
⎝T ⎠ ⎪ t
⎪ eT , t < 0
⎩
Fig. 1.6 illustrates specific examples of these pulses.
1.23
mywbut.com
Principles of Communication
∫ x ( t ) dt
2
Ex = (1.10)
−∞
(When no specific signal is being referred to, we use the symbol E without any
subscript to denote the energy quantity.)
ranges), and quite often this range could be ( −∞, ∞ ) . Eq. 1.2(a) expresses x p ( t )
as a sum over a discrete set of frequencies. Its counterpart for the aperiodic case
is an integral relationship given by
1.24
mywbut.com
Principles of Communication
∞
x (t ) = ∫ X (f ) e
j 2 πf t
df (1.11a)
−∞
Eq. 1.11(a) is given the following interpretation. Let the integral be treated
as a sum over incremental frequency ranges of width ∆f . Let X ( fi ) ∆f be the
Eq. 1.11(b) is referred to as the Fourier Transform (FT) relation or, the
analysis equation, or forward transform relation. We use the notation
X ( f ) = F ⎡⎣ x ( t ) ⎤⎦ (1.12a)
x (t ) = F −1
⎡⎣ X ( f ) ⎤⎦ (1.12b)
Eq. 1.12(a) and Eq. 1.12(b) are combined into the abbreviated notation, namely,
x ( t ) ←⎯→ X ( f ) . (1.12c)
X (f ) = XR (f ) + j XI (f )
1.25
mywbut.com
Principles of Communication
= X (f ) e
j θ( f )
X R ( f ) = Real part of X ( f ) ,
X I ( f ) = Imaginary part of X ( f )
X ( f ) = magnitude of X ( f )
= X R2 ( f ) + X I2 ( f )
⎡ X (f ) ⎤
θ ( f ) = arg ⎣⎡ X ( f ) ⎦⎤ = arc tan ⎢ I ⎥
⎢⎣ X R ( f ) ⎥⎦
phase spectrum.
Example 1.2
⎛t ⎞
Let x ( t ) = A ga ⎜ ⎟ . Let us compute and sketch X ( f ) .
⎝T ⎠
T
2
X (f ) = ∫ Ae− j 2 πft
dt = AT sin c (f T ) ,
−
T
2
where
sin ( π λ )
sin c ( λ ) =
πλ
Appendix A1.1 contains the tabulated values of sin c ( λ ) . Its behavior is shown in
⎧1 , λ = 0
Fig. A1.1. Note that sin c ( λ ) = ⎨
⎩0 , λ = ± 1, ± 2 etc.
Fig. 1.7(a) sketches the magnitude spectrum of the rectangular pulse for AT = 1 .
1.26
mywbut.com
Principles of Communication
m m +1
During the interval, < f < , with m odd, sin c ( f T ) is negative. As the
T T
magnitude spectrum is always positive, negative values of sin c ( f T ) are taken
1.7(b).
Remarkable balancing act: A serious look at the magnitude and phase plots
reveals a very charming result. From the magnitude spectrum, we find that a
rectangular pulse is composed of frequency components in the range
−∞ < f < ∞ , each with its own amplitude and phase. Each of these complex
exponentials exist for all t . But when we synthesize a signal using the complex
exponentials with the magnitudes and phases as given in Fig. 1.7, they add up to
1.27
mywbut.com
Principles of Communication
a constant for t < T and then go to zero forever. A very fascinating result
2
indeed!
Fig. 1.7(a) illustrates another interesting result. From the figure, we see
that most of the energy, (that is, the range of strong spectral components) of the
1
signal lies in the interval f < , where T is the duration of the rectangular pulse.
T
Hence, if T is reduced, then its spectral width increases and vice versa (As we
shall see later, this is true of other pulse types, other than the rectangular). That
is, more compact is the signal in the time-domain, the more wide-spread it would
be in the frequency domain and vice versa. This is called the phenomenon of
reciprocal spreading.
Example 1.3
Let x ( t ) = ex1( t ) . Let us find X ( f ) and sketch it.
∞
1
X ( f ) = ∫ e − t e − j 2πf t dt =
0
1 + j 2πf
1
Hence, X ( f ) =
1 + ( 2πf )
2
A plot of the magnitude and the phase spectrum are given in Fig. 1.8.
1.28
mywbut.com
Principles of Communication
1.29
mywbut.com
Principles of Communication
∫ x ( t ) dt < ∞
−∞
∞
X (f ) = ∫ x (t ) e
− j 2 πf t
dt
−∞
∞ ∞
∫ x (t ) e ∫ x ( t ) dt
− j 2 πf t
≤ dt = <∞
−∞ −∞
(ii) x ( t ) is single valued and has only finite number of maxima and minima
W
(s )
( t ) = Wlim X ( f ) e j 2 πf t df , then x( ) (t ) x (t )
→∞ ∫
s
If x converges to
−W
∫ x (t )
2
dt < ∞ (Finite energy signal), then we have the convergence in the
−∞
mean, namely
∞ 2
∫ x ( t ) − x ( ) ( t ) dt = 0
s
−∞
1.30
mywbut.com
Principles of Communication
P1) Linearity
Let x1 ( t ) ←⎯→ X1 ( f ) and x2 ( t ) ←⎯→ X 2 ( f ) .
a1 x1 ( t ) + a2 x2 ( t ) ←⎯→ a1 X1 ( f ) + a2 X 2 ( f )
It is very easy to see the validity of the above transform relationship. This
property will be used quite often in the development of this course material.
If x ( t ) ←⎯→ X ( f ) , then
∞
∫ x ( t ) dt = X ( 0 )
−∞
1.31
mywbut.com
Principles of Communication
∞
If x ( t ) ←⎯→ X ( f ) , then x ( 0 ) = ∫ X ( f ) df
−∞
∞
2πf
Noting that ∫ 1 + ( 2πf )2 df = 0 , we have
−∞
∞ ∞
1 1
∫ X ( f ) df = ∫ 1 + ( 2πf )2 df = , which is the value of ex1( t ) | t = 0
−∞ −∞
2
1.32
mywbut.com
Principles of Communication
Fig. 1.9: A triangular pulse and its time compressed and reversed version
x ( − t ) ←⎯→ X ( − f ) . That is, both the time function and its Fourier transform
ex1( t ) + ex1( − t ) = ex 2 ( t )
ex 2 ( t ) = exp ( − t ) ←⎯→
1 1
+
1 + j 2πf 1 − j 2πf
2
=
1 + ( 2πf )
2
1 ⎛f ⎞
x ( 2t ) ←⎯→
2 ⎜⎝ 2 ⎟⎠
X
1.33
mywbut.com
Principles of Communication
1 ⎛f ⎞
Let us compare X ( f ) with X ⎜ ⎟ by taking x ( t ) = ex1( t ) , and
2 ⎝ 2⎠
y ( t ) = ex1( 2t ) . That is,
⎧ e − 2t , t > 0
⎪
⎪ 1
y (t ) = ⎨ , t =0
⎪ 2
⎪ 0 , t <0
⎩
1 ⎡ 1 ⎤
Y (f ) = ⎢ ⎥
2 ⎣⎢ 1 + j 2π ( f 2 ) ⎥⎦
1
=
2 + j 2πf
Let X ( f ) = X ( f ) e x ( ) and
θ f
θy ( f ) 1
Y (f ) = Y (f ) e , where Y ( f ) =
2 1 + π2 f 2
θy ( f ) = − arc tan ( π f )
Fig. 1.10 gives the plots of X ( f ) and Y ( f ) . In Fig. 1.10(a), we have the
Fig. 1.10(c) gives the plots of θ x ( f ) and θy ( f ) . From Fig. 1.10(b), we see that
⎛f ⎞
of the original signal. (In fact, if X ( f ) is band limited to W Hz, then X ⎜ ⎟
⎝ 2⎠
will be band limited to 2W Hz.)
1.34
mywbut.com
Principles of Communication
1.35
mywbut.com
Principles of Communication
x ( t − t0 ) ←⎯→ e − j 2 πf t0 X ( f )
1.36
mywbut.com
Principles of Communication
∞
= e − j 2 π f t0 ∫ x (λ)e
− j 2 πf λ
dλ
−∞
= e − j 2 π f t0 X ( f )
e ± j 2 πfc t x ( t ) ←⎯→ X ( f ∓ fc )
Proof: Exercise
As an application of the above result, let us consider the spectrum of
y ( t ) = 2 cos ( 2πfc t ) x ( t ) ; that is, we want the Fourier transform of
1.37
mywbut.com
Principles of Communication
P5) Duality
If we look fairly closely at the two equations constituting the Fourier
transform pair, we find that there is a great deal of similarity between them. In
Eq.1.11a, ' f ' is the variable of integration where as in Eq. 1.11b, it is the variable
' t ' . The sign of the exponent is positive in Eq. 1.11a where as it is negative in
Eq. 1.11b. Both t and f are variables of the continuous type. This results in an
interesting property, namely, the duality property, which is stated below.
If x ( t ) ←⎯→ X ( f ) , then
Note: This is one instance, where the variable t is associated with a function
denoted using the upper case letter and the variable f is associated with a
function denoted using a lower case letter.
∞
Proof: x ( t ) = ∫ X ( f ) e df
j 2 πf t
−∞
∞
x (−t ) = ∫ X (f ) e
− j 2 πf t
df
−∞
The result follows by interchanging the variables t and f . The proof of the
second part of the property is similar.
Example 1.4
If z ( t ) = A sin c 2W t , let us use duality to find Z ( f ) .
1.38
mywbut.com
Principles of Communication
A ⎛ t ⎞
We know that if x ( t ) = ga ⎜ ⎟ , then,
2W ⎝ 2W ⎠
X ( f ) = A sin c ( 2W f ) and
X ( t ) = A sin c ( 2W t ) = z ( t ) ⇒
A ⎛ f ⎞
Z (f ) = x ( − f ) = ga ⎜ − ⎟
2W ⎝ 2W ⎠
As ga( − f ) = ga( f ) , we have
⎛ A ⎞ ⎛ f ⎞
Z (f ) = ⎜ ⎟ ga ⎜ ⎟.
⎝ 2W ⎠ ⎝ 2W ⎠
Example 1.5
2
Find the Fourier transform of z ( t ) = .
1+ t 2
2
We know that if y ( t ) = e , then Y ( f ) =
− t
1+( 2πf )
2
Let x ( t ) = y ( αt ) , with α = 2π .
1 ⎛ f ⎞
Then X ( f ) = Y⎜ ⎟
2π ⎝ 2π ⎠
1 2
=
2π 1 + f 2
2
or 2π X ( f ) =
1+ f 2
2
As z ( t ) = with f being replaced by t , we have
1+ f 2
Z ( f ) = 2π x ( − f )
− 2π f
= 2π e
1.39
mywbut.com
Principles of Communication
2 − 2π f
Hence ←⎯→ 2π e
1+ t 2
∞
X ∗ (f ) = ∫ x ∗ ( t ) e j 2 πf t dt
−∞
∞
X ∗ (−f ) = ∫ x ∗ ( t ) e − j 2 πf t dt
−∞
From the time reversal property, we get the additional relation, namely
x ∗ ( − t ) ←⎯→ X ∗ ( f )
that
1
xR ( t ) ←⎯→ ⎡ X ( f ) + X ∗ ( − f )⎤
2 ⎣ ⎦
1
xI ( t ) ←⎯→ ⎡ X ( f ) − X ∗ ( − f )⎤
2j ⎣ ⎦
1.40
mywbut.com
Principles of Communication
x ( − t ) = − x∗ (t ) .
If x ( t ) is real, then x ( t ) = x ∗ ( t ) .
As a result, X ( f ) = X ∗ ( − f ) or X ∗ ( f ) = X ( − f ) .
Hence, the spectrum for the negative frequencies is the complex conjugate of the
positive part of the spectrum. This implies, that for real signals,
X ( − f ) = X (f )
θ ( − f ) = − θ (f )
Going one step ahead, we can show that if x ( t ) is real and even, then
2
X ( f ) is also real and even. (Example: e ). Similarly, if x ( t )
− t
←⎯→
1 + (2 π f )
2
is real and odd, its transform is purely imaginary and odd (See Example 1.7).
x2 ( t ) ←⎯→ X 2 ( f )
∞ ∞
then, x1 ( t ) x2 ( t ) ←⎯→ ∫ X1 ( λ ) X 2 ( f − λ ) d λ = ∫ X 2 ( λ ) X1 ( f − λ ) d λ
−∞ −∞
(Note that ∗ in between two functions represents the convolution of the two
quantities where as a superscript, it denotes the complex conjugate)
1.41
mywbut.com
Principles of Communication
Proof: Exercise
x2 ( t ) ←⎯→ X 2 ( f )
∞
⎣⎡ X1 ( f ) X 2 ( f ) ⎦⎤ = ∫ x1 ( λ ) x2 ( t − λ ) d λ
−1
then, F
−∞
∞
= ∫ x2 ( λ ) x1 ( t − λ ) d λ
−∞
we have
x1 ( t ) ∗ x2 ( t ) ←⎯→ X1 ( f ) X 2 ( f )
Proof: Let x3 ( t ) = x1 ( t ) ∗ x2 ( t )
∞
That is, x3 ( t ) = ∫ x1 ( λ ) x2 ( t − λ ) d λ
−∞
∞ ∞⎡∞ ⎤
X 3 ( f ) = F ⎡⎣ x3 ( t ) ⎤⎦ = ∫ x3 ( t ) e − j 2 πf t
dt = ∫ ⎢ ∫ x1 ( λ ) x2 ( t − λ ) d λ ⎥ e − j 2π f t
dt
−∞ −∞ ⎢
⎣− ∞ ⎥⎦
Hence,
1.42
mywbut.com
Principles of Communication
∞
X3 (f ) = ∫ x1 ( λ ) X 2 ( f ) e
− j 2 πf λ
dλ
−∞
∞
= X2 (f ) ∫ x1 ( λ ) e
− j 2 πf λ
dλ
−∞
= X1 ( f ) X 2 ( f )
Property P7(b), known as the Convolution theorem, is one of the very useful
properties of the Fourier transform.
Example 1.6
⎛t ⎞
In this example, we will find the Fourier transform of T tri ⎜ ⎟ .
⎝T ⎠
⎛t ⎞ ⎛t ⎞
T tri ⎜ ⎟ can be obtained as the convolution of ga ⎜ ⎟ with itself. That is,
⎝T ⎠ ⎝T ⎠
⎛t ⎞ ⎛t ⎞ ⎛t ⎞
ga ⎜ ⎟ ∗ ga ⎜ ⎟ = T tri ⎜ T ⎟
⎝T ⎠ ⎝T ⎠ ⎝ ⎠
⎛t ⎞
As ga ⎜ ⎟ ←⎯→ T sin c ( f T ) , we have
⎝T ⎠
⎛t ⎞
T tri ⎜ ⎟ ←⎯→ ⎡⎣T sin c ( f T ) ⎤⎦
2
⎝T ⎠
1.43
mywbut.com
Principles of Communication
P8) Differentiation
P8a) Differentiation in the time domain
Let x ( t ) ←⎯→ X ( f ) ,
d
then, ⎡ x ( t ) ⎦⎤ ←⎯→ j 2πf ⎡⎣ X ( f ) ⎤⎦
dt ⎣
d n x (t )
←⎯→ ( j 2πf ) X ( f )
n
Generalizing, n
dt
Proof: We shall prove the first part; generalization follows as a consequence
this. We have,
∞
x (t ) = ∫ X ( f ) e df
j 2 πf t
−∞
d ⎡ ⎤
∞
d
⎡⎣ x ( t ) ⎦⎤ = ⎢ ∫ X ( f ) e j 2 πf t df ⎥
dt dt ⎢ − ∞ ⎥⎦
⎣
Interchanging the order of differentiation and integration on the RHS,
∞
d ⎡ d j 2 πf t ⎤
⎡ x ( t ) ⎤⎦ = ∫ X ( f ) ⎢⎣ dt e ⎥⎦ df
dt ⎣ −∞
∫ ⎡⎣ j 2πf X ( f )⎤⎦ e
j 2 πf t
= df
−∞
d
From the above, we see that F − 1 ⎡⎣ j 2πf X ( f ) ⎤⎦ is x ( t ) . Hence the property.
dt
Example 1.7
Let us find the FT of the doublet pulse x ( t ) shown in Fig. 1.12 below.
1.44
mywbut.com
Principles of Communication
d ⎡ ⎛t ⎞⎤
x (t ) = ⎢T tri ⎜ T ⎟ ⎥ . Hence,
dt ⎣ ⎝ ⎠⎦
⎡ ⎛ t ⎞⎤
X ( f ) = j 2πf F ⎢T tri ⎜ ⎟ ⎥
⎣ ⎝ T ⎠⎦
Let x3 ( t ) = x1 ( t ) ∗ x2 ( t )
Then X 3 ( f ) = X1 ( f ) X 2 ( f )
= X1 ( f ) ⎡⎣ j 2πf X 2 ( f ) ⎤⎦
That is,
1.45
mywbut.com
Principles of Communication
d d
⎡⎣ x1 ( t ) ∗ x2 ( t ) ⎤⎦ = ⎡ x1 ( t ) ⎤⎦ ∗ x2 ( t )
dt dt ⎣
d
= x1 ( t ) ∗ ⎡ x2 ( t ) ⎤⎦
dt ⎣
d X (f )
Then, ⎡⎣( − j 2πt ) x ( t ) ⎤⎦ ←⎯→
df
⎡ ⎤ d n X (f )
Generalizing, ( − j 2πt ) x ( t ) ←⎯→
n
⎣ ⎦ df n
Proof: Exercise
The generalized property can also be written as
⎛ j ⎞ d X (f )
n n
t x ( t ) ←⎯→ ⎜
n
⎟
⎝ 2π ⎠ df n
Example 1.8
⎛t ⎞
Find the Fourier transform of x ( t ) = t ex1⎜ ⎟.
⎝T ⎠
1
We have, ex1( t ) ←⎯→
1 + j 2πf
⎛t ⎞ 1
Hence, ex1⎜ ⎟ ←⎯→ T
⎝T ⎠ 1 + j 2πfT
⎛t ⎞ j d ⎡ T ⎤
t ex1⎜ ⎟ ←⎯→
⎝T ⎠ 2π df ⎣ 1 + j 2πfT ⎦⎥
⎢
j − ( j 2πT )
←⎯→ T
2π (1 + j 2πfT )2
1.46
mywbut.com
Principles of Communication
⎡ T2 ⎤
←⎯→ ⎢ ⎥
⎢⎣ (1 + j 2πf T ) ⎥⎦
2
∫ X ( f ) df
2
Ex =
−∞
∫ x1 ( t ) x2∗ ( t ) dt = ∫ X1 ( f ) X 2∗ ( f ) df
−∞ −∞
Proof: We have
∗
∞ ∞ ⎡∞ ⎤
∫ x1 ( t ) x2∗ ( t ) dt = ∫ x1 ( t ) ⎢ ∫ X 2 ( f ) e j 2πf t df ⎥ dt
⎢⎣ − ∞ ⎥⎦
−∞ −∞
∞ ⎡∞ ⎤
= ∫ X 2∗ ( f ) ⎢ ∫ x1 ( t ) e − j 2πf t dt ⎥ df
−∞ ⎢⎣ − ∞ ⎥⎦
∞
= ∫ X 2∗ ( f ) X1 ( f ) df
−∞
If x1 ( t ) = x2 ( t ) = x ( t ) , then
∞ ∞
∫ x (t ) ∫ X ( f ) df
2 2
dt = E x =
−∞ −∞
1.47
mywbut.com
Principles of Communication
∞ ∞ ∞
∫ x1 ( t ) x2 ( t ) d t = ∫ X1 ( f ) X 2 ( − f ) d f = ∫ X 2 ( f ) X1 ( − f ) d f
−∞ −∞ −∞
Example 1.9
Let us find the energy of the signal x ( t ) = 2 AW sin c ( 2W t ) .
⎛ f ⎞
1.4, X ( f ) = A ga ⎜ ⎟ . Hence,
⎝ 2W ⎠
W
Ex = ∫ A2 df = 2W A2
−W
More important than the calculation of the energy of the signal, Rayleigh’s
⎛t ⎞
signal is confined to the interval f ≤ W . Consider the rectangular pulse ga ⎜ ⎟.
⎝T ⎠
1
The first nulls of the magnitude spectrum occur at f = ± . The evaluation of
T
1.48
mywbut.com
Principles of Communication
1 1
T T
X ( f ) df = T sin c 2 ( f T ) df
2
∫1 ∫1 will yield the value 0.92 T , which is 92
− −
T T
⎛t ⎞ ⎛ 1 1⎞
percent of the total energy of ga ⎜ ⎟ . Hence, the frequency range ⎜ − , ⎟ can
⎝T ⎠ ⎝ T T⎠
⎛ 2 2⎞
be taken to be the spectral width of the rectangular pulse. [The interval ⎜ − , ⎟
⎝ T T⎠
may result in about 95 percent of the total energy].
to denote the unit impulse. We define the unit impulse as any (generalized)
function that satisfies the following conditions:
(i) δ ( t ) = 0, t ≠ 0 (1.13a)
(ii) δ ( t ) = ∞, t = 0 (1.13b)
1.49
mywbut.com
Principles of Communication
∞ ε
∫ δ (t ) d t = ∫ δ (t ) d t = 1 (1.13d)
−ε −∞
Let
1 ⎛t ⎞ 1 ⎛t ⎞ 1 ⎛t⎞
(a) p1 ( t ) = ga ⎜ ⎟ (b) p2 ( t ) = tri ⎜ ⎟ (c) p3 ( t ) = sin c ⎜ ⎟
ε ⎝ε⎠ ε ⎝ε⎠ ε ⎝ε⎠
Then, lim pi ( t ) = δ ( t ) , i = 1, 2, 3 . p3 ( t ) is shown below in Fig. 1.13.
ε→0
1.50
mywbut.com
Principles of Communication
⎛ 1 ⎛t⎞ ⎞
⎜ Note that ε sin c ⎜ ε ⎟ ←⎯→ ga ( ε f ) . Hence the area under the time function = 1. ⎟
⎝ ⎝ ⎠ ⎠
From the above examples, we see that the shape of the function is not
very critical; its area should remain at 1 in order to approach δ ( t ) in the limit.
normally shown as a spear (Fig. 1.14) with the weight or area of the impulse
shown in parentheses very close to it.
1.51
mywbut.com
Principles of Communication
∫ p ( t ) δ ( t − t0 ) dt = p ( t0 )
a
consequence.
P2) Replication property
Let p ( t ) be any ordinary function. Then,
p ( t ) ∗ δ ( t − t0 ) = p ( t − t0 )
The proof of this property follows from the fact, that in the process of
convolution, every value of p ( t ) will be sampled and shifted by t0
resulting in p ( t − t0 ) .
1.52
mywbut.com
Principles of Communication
∞
1
δ ( t ) dt
α −∫∞
=
∞
1
=
α ∫ δ ( t ) dt
−∞
∞
1
=
α ∫ δ ( t ) dt
−∞
1
It is easy to show that δ ⎣⎡ α ( t − t0 ) ⎦⎤ = δ ( t − t0 ) .
α
as the even sided delta function. It is also possible to come up with delta
functions as a limiting case of functions that are not even; that is, as a limiting
case of one-sided functions. In such a situation we have a left-sided delta
function or right-sided delta function etc. Left-sided delta function will prove to be
useful in the context of probability density functions of certain random variables,
subject matter of chapter 2.
1.53
mywbut.com
Principles of Communication
Example 1.10
Find the value of
4
∫ t δ ( t − 5 ) dt
3
(a)
−4
5.1
∫ t δ ( t − 5 ) dt
3
(b)
4.9
t 3 δ ( t − 5 ) = 53 δ ( t − 5 ) .
Hence,
5.1 5.1
∫ t δ ( t − 5 ) dt = 5 ∫ δ ( t − 5 ) dt = 125
3 3
4.9 4.9
Example 1.11
⎛t⎞
Let p ( t ) = tri ⎜ ⎟ . Find ⎣⎡ p ( t ) ∗ δ ( 2t − 1) ⎦⎤ .
⎝4⎠
⎡ ⎛ 1 ⎞⎤ 1⎡ ⎛ 1 ⎞⎤
δ ( 2t − 1) = δ ⎢2 ⎜ t − ⎟ ⎥ = ⎢ δ ⎜ t − ⎟⎥
⎣ ⎝ 2 ⎠⎦ 2⎣ ⎝ 2 ⎠⎦
1 ⎡ ⎛ 1 ⎞⎤ 1 ⎛ 1⎞ 1 ⎛t − 1 2⎞
p (t ) ∗ ⎢δ ⎜ t − 2 ⎟ ⎥ = 2 p ⎜ t − 2 ⎟ = 2 tri ⎜ 4 ⎟
2 ⎣ ⎝ ⎠⎦ ⎝ ⎠ ⎝ ⎠
have,
∞
F ⎡⎣δ ( t ) ⎤⎦ = ∫ δ (t ) e
− j 2 πf t
dt = 1 (1.14a)
−∞
1.54
mywbut.com
Principles of Communication
How do we interpret this result? The spectrum of the unit impulse consists
of frequency components in the range ( − ∞, ∞ ) , all with unity magnitude and
zero phase shift, a fascinating result indeed! Hence exciting any electric network
or system with a unit impulse is equivalent to exciting the network simultaneously
with complex exponentials of all possible frequencies, all with the same
magnitude (unity in this case) and zero phase shift. That is, the unit impulse
response of a linear network is the synthesis of responses to the individual
complex exponentials and we intuitively feel that the impulse response of a
network should be able to characterize the system in the time domain. (We shall
see a little later that if the network is linear and time invariant, a simple relation
exists between the input to the network, its impulse response and the output).
Based on Eq. 1.14(a) and Eq. 1.14(b), we make the following observation:
a constant in one domain will transform into an impulse in the other domain.
quantity for the transform because it is zero for f ≠ 0 and its inverse transform
yields the required constant in time (note that only an impulse can yield a
nonzero value when integrated over zero width).
Because of the transform pair,
1 ←⎯→ δ ( f ) ,
e− j 2 π f0 t
←⎯→ δ ( f + f0 ) (1.15b)
1.55
mywbut.com
Principles of Communication
e j 2 π f0 t + e − j 2 π f0 t
As cos ω0 t = , we have,
2
1
cos ω0 t ←⎯→ ⎡δ ( f − f0 ) + δ ( f + f0 ) ⎤⎦ (1.16)
2 ⎣
Similarly,
1
sin ω0 t ←⎯→ ⎡δ ( f − f0 ) − δ ( f + f0 ) ⎤⎦ (1.17)
2j ⎣
Note that the impulses in Fig. 1.15(b) have weights that are complex. It is
fairly conventional to show the spectrum of sin ω0 t as depicted in Fig. 1.15(b); or
else we can make two separate plots, one for magnitude and the other for phase,
where the magnitude plot is identical to that shown in Fig. 1.15(a) and the phase
π π
plot has values of − at f = f0 and + at f = − f0 .
2 2
with a discontinuity that is not finite), and using impulses in the frequency
domain, we have developed the Fourier transforms of the periodic signals such
1.56
mywbut.com
Principles of Communication
as e ± j 2 π f0 t
, cos ω0 t and sin ω0 t , which are neither absolutely integrable nor
square integrable.
∞
= ∑ xn δ ( f − nf0 ) (1.18)
n = −∞
impulses in the spectrum. As such, the differences between the line spectrum of
sec. 1.1 and spectral representation given by Eq. 1.18 are only minor in nature.
They both provide the same information, differing essentially only in notation.
1.57
mywbut.com
Principles of Communication
⎧ T0 T
⎪x (t ) , − < t < 0
x (t ) = ⎨ p 2 2
⎪⎩ 0 , outside
T0
2
1
xn = ∫ xp (t ) e− j 2 π n f0 t
dt
T0 T
− 0
2
⎡ T0 ⎤
1 ⎢ 2
= ⎢ ∫ x (t ) e− j 2 π n f0 t
dt ⎥⎥
T0 T0
⎢⎣ − 2 ⎥⎦
⎛n⎞
1 ⎡ ⎤
∞ ∞ − j 2 π⎜ ⎟t
1
= ⎢ ∫ x (t ) e− j 2 π n f0 t
dt ⎥ = ∫ x (t ) e ⎝ T0 ⎠ dt
T0 ⎢ − ∞ ⎥⎦ T0
⎣ −∞
1 ⎛n⎞
Hence, xn = X⎜ ⎟ (1.19)
T0 ⎝ T0 ⎠
Example 1.12
Find the Fourier transform of the uniform impulse train
∞
xp (t ) = ∑ δ ( t − nT0 ) shown in Fig 1.16 below.
n = −∞
1.58
mywbut.com
Principles of Communication
T0 T
Let x ( t ) be one period of x p ( t ) in the interval, − < t < 0 . Then,
2 2
x ( t ) = δ ( t ) for this example. But as δ ( t ) ←⎯→ 1, from Eq. (1.19) we have,
1
xn = for all n . Hence,
T0
∞
1
X p (f ) =
T0
∑ δ (f − n f0 ) (1.20)
n = −∞
∞
As δ ( − t ) = δ ( t ) , we have, ∫e
− j 2 πf t
df = δ ( t )
−∞
∞
That is, ∫e
± j 2πf t
df = δ ( t ) (1.21)
−∞
−∞
∞ ⎡∞ ⎤
∫ ⎢⎢ ∫ x ( λ ) e
− j 2πf λ
= d λ ⎥ e j 2 π f t df
− ∞ ⎣− ∞ ⎥⎦
∞ ∞
j 2 π f (t − λ )
= ∫ ∫ x (λ) e df dλ
−∞ −∞
∞ ⎡ ∞ j 2 πf t − λ ⎤
( )
= ∫ x ( λ ) ⎢ ∫ e df ⎥ dλ
−∞ ⎢
⎣− ∞ ⎥
⎦
1.59
mywbut.com
Principles of Communication
∞
= ∫ x (λ ) δ (t − λ ) dλ = x (t )
−∞
Let R ⎡⎣ δ ( t ) ⎤⎦ denote the output (response) of the LTI system, when the
input is exited by the unit impulse δ ( t ) . This is generally denoted by the symbol
h ( t ) and is called the impulse response of the system. That is, when δ ( t ) is
invariant, R ⎣⎡ δ ( t − τ ) ⎦⎤ = h ( t − τ ) .
⎡ ∞ ⎤ ∞
and R ⎡⎣ x ( t ) ⎤⎦ = y ( t ) = R ⎢ ∫ x ( τ) δ (t − τ ) d τ⎥ = ∫ x ( τ) h (t − τ) d τ
⎢⎣− ∞ ⎥⎦ −∞
1.60
mywbut.com
Principles of Communication
Note that the properties P1) to P3) are valid even if the independent variable is
other than t .
the frequency response of the system and describes the frequency domain
Y (f )
behavior of the system. (As H ( f ) = , it is also referred to as the transfer
X (f )
as two different plots, namely, the magnitude response: H ( f ) vs. f and the
1.61
mywbut.com
Principles of Communication
Example 1.13
RC-lowpass filter (RC-LPF) is one among the quite often used LTI
systems in the study of communication theory. This network is shown in Fig.
1.17. Let us find its frequency response as well as the impulse response.
One of the important properties of any LTI system is: if the input
x ( t ) = e j 2 π f0 t , then the output y ( t ) is also a complex exponential given by
y ( t ) = H ( f0 ) e j 2 π f0 t .
1
j 2 π f0 C
But, y ( t ) = e j 2 π f0 t
1
R+
j 2 π f0 C
1
or y (t ) = e j 2 π f0 t , when x ( t ) = e j 2 π f0 t
1 + j 2 π f0 R C
Generalizing this result, we obtain,
1
H (f ) = (1.24)
1 + j 2πf RC
That is,
1
H (f ) = (1.25a)
1 + (2 π f R C )
2
1.62
mywbut.com
Principles of Communication
1
Let = F . Then,
2πRC
1
H (f ) = 2
(1.26)
⎛f ⎞
1+ ⎜ ⎟
⎝F ⎠
1.63
mywbut.com
Principles of Communication
1
ex1( t ) ←⎯→
1 + j 2πf
1 ⎛ t ⎞ 1
Hence, ex1⎜ ⎟ ←⎯→
RC ⎝ RC ⎠ 1 + j 2πf RC
That is,
1 ⎛ t ⎞
⎡⎣ h ( t ) ⎤⎦ = ex1 ⎜ ⎟ (1.27)
RC − LPF RC ⎝ RC ⎠
This is shown in Fig. 1.19.
Example 1.14
The input x ( t ) and the impulse response h ( t ) of an LTI system are as
1.64
mywbut.com
Principles of Communication
Fig. 1.20: The input x ( t ) and the impulse response h ( t ) of an LTI system
∞
Let y ( t ) = ∫ h ( τ) x (t − τ) d τ
−∞
we first reverse x ( τ ) to get x ( − τ ) and then shift by t , the time instant for which
1.65
mywbut.com
Principles of Communication
∫ h ( τ) x (t − τ) d τ = 0.
−∞
1.66
mywbut.com
Principles of Communication
1.67
mywbut.com
Principles of Communication
Using the similar arguments, y ( t ) can be computed for t > 2 . The result
even be sticky1.)
Exercise 1.3
⎧⎪ e − αt , t ≥ 0
Let x ( t ) = ⎨
⎪⎩ 0 , otherwise
⎪⎧ e − βt , t ≥ 0
h (t ) = ⎨
⎪⎩ 0 , otherwise
1
Some people claim that convolution has driven many electrical engineering students to
contemplate theology either for salvation or as an alternative career (IEEE Spectrum, March
1991, page 60). For an interesting cartoon expressing the student reaction to the convolution
operation, see [3].
1.68
mywbut.com
Principles of Communication
Exercise 1.4
Find y ( t ) = x ( t ) ∗ h ( t ) where
⎧ 2 , t < 2
x (t ) = ⎨
⎩ 0, outside
⎧⎪ 2 e− t , t ≥ 0
h (t ) = ⎨
⎪⎩ 0 , outside
Exercise 1.5
⎧1
⎪5 , 950 < f < 1050 Hz
⎪
⎪1
Let X ( f ) = ⎨ , − 1050 < f < − 950 Hz
⎪10
⎪0 , elsewhere
⎪
⎩
(a) Compute and sketch Y ( f ) = X ( f ) ∗ X ( f )
⎧ 1, t > 0
⎪
sgn ( t ) = ⎨ 0 , t = 0 (1.28)
⎪ − 1, t < 0
⎩
1.69
mywbut.com
Principles of Communication
⎧1 , t > 0
⎪1
⎪
u (t ) = ⎨ , t = 0 (1.29)
⎪2
⎪⎩ 0 , t < 0
F ⎡⎣ sgn ( t ) ⎤⎦ :
Let x ( t ) = e − α t u ( t ) − e α t u ( − t ) (1.30)
Fig. 1.23.
1 1 − j 4 πf
X (f ) = − =
α + j 2πf α − j 2πf α2 + ( 2 π f )
2
1
F ⎡⎣sgn ( t ) ⎤⎦ = lim X ( f ) = (1.31)
α→0 j πf
F ⎡⎣u ( t ) ⎤⎦
1
As u ( t ) = ⎡1 + sgn ( t ) ⎤⎦ , we have
2 ⎣
1.70
mywbut.com
Principles of Communication
1 1
U (f ) = δ (f ) + (1.32)
2 j 2πf
Properties of FT continued...
P9) Integration in the time domain
Let x ( t ) ←⎯→ X ( f )
t
1 X (0)
Then, ∫ x ( τ) d τ ←⎯→
j 2πf
X (f ) +
2
δ (f ) (1.33)
−∞
Proof:
∞
Consider x ( t ) ∗ u ( t ) = ∫ x ( τ) u (t − τ) d τ . As u ( t − τ ) = 0 for τ > t ,
−∞
t
x (t ) ∗ u (t ) = ∫ x ( τ ) d τ. But F ⎣⎡ x ( t ) ∗ u ( t ) ⎦⎤ = X ( f ) U ( f ) .
−∞
t
⎡ 1 δ (f ) ⎤
Hence, ∫ x ( τ ) d τ ←⎯→ X ( f ) ⎢ + ⎥
−∞ ⎣ j 2πf 2 ⎦
t
X (f ) X (0) δ (f )
(ii) X ( 0 ) ≠ 0 ; then ∫ x ( τ) d τ ←⎯→ +
−∞
j 2πf 2
1 ⎛t ⎞
Let p 1 ( t ) = ga ⎜ ⎟ . Then, we know that lim p1 ( t ) = δ ( t ) .
ε ⎝ε⎠ ε→0
0 0
1
But lim ∫ε p1 ( t ) dt = ∫ p1 ( t ) dt =
ε→0 2
− −∞
2
1.71
mywbut.com
Principles of Communication
ε
2 ∞
and lim ∫ε p1 ( t ) dt = ∫ p1 ( t ) dt = 1.
ε→0
− −∞
2
That is,
t
∫ δ ( τ) d τ = u (t ) (1.34a)
−∞
d u (t )
or = δ (t ) (1.34b)
dt
d u (t )
As = δ (t ) ,
dt
j 2 π f U (f ) = 1
1
or U (f ) =
j 2πf
But this is valid only for f ≠ 0 because of the following argument.
u ( t ) + u ( − t ) = 1. Therefore,
U (f ) + U ( − f ) = δ (f )
U ( 0 ) + U ( − 0 ) = 2 U ( 0 ) = δ ( f ) or
1
U (0) = δ ( f ) . Hence,
2
⎧1
⎪⎪ 2 δ ( f ) , f = 0
U (f ) = ⎨
⎪ 1 , f ≠ 0
⎪⎩ j 2 π f
1.72
mywbut.com
Principles of Communication
Example 1.15
(a) For the scheme shown in Fig. 1.24, find the impulse response (This system is
referred to as Zero-Order-Hold, ZOH).
(b) If two such systems are cascaded, what is the overall impulse response
(cascade of two ZOHs is called a First-Order-Hold, FOH).
a) h ( t ) of ZOH:
When x ( t ) = δ ( t ) , we have
v (t ) = δ (t ) − δ (t − T )
⎛ T⎞
t t
⎜t − 2 ⎟
∫ δ ( τ) d τ − ∫ δ ( τ − T ) d τ = u (t ) − u (t − T ) = ga ⎜ ⎟
−∞ −∞ ⎜ T ⎟
⎝ ⎠
b) Impulse response of two LTI systems in cascade is the convolution of the
impulse responses of the constituents. Hence,
⎡ T⎤ ⎡ T⎤
⎢t − 2 ⎥ ⎢t − 2 ⎥
⎡⎣ h ( t ) ⎤⎦ = ga ⎢ ⎥ ∗ ga ⎢ ⎥
FOH T
⎢ ⎥ ⎢ T ⎥
⎣ ⎦ ⎣ ⎦
⎛t −T ⎞
= T tri ⎜ ⎟
⎝ T ⎠
1.73
mywbut.com
Principles of Communication
∫ δ ( τ) d τ = u (t )
−∞
p ( t ) can be written as
p ( t ) = u ( t + 2 ) − 2 u ( t + 1) + u ( t − 3 )
1.74
mywbut.com
Principles of Communication
Hence,
d p (t )
= δ ( t + 2 ) − 2 δ ( t + 1) + δ ( t − 3 )
dt
which is shown in Fig. 1.25(b). From this result, we note that if there is a step
discontinuity of size A at t = t1 in the signal, its derivative will have an impulse
of weight A at t = t1 .
Example 1.16
Let x ( t ) be the doublet pulse of Example 1.7 (Fig.1.12). We shall find
X (f )
d x (t )
from .
dt
d x (t )
= δ (t + T ) − 2 δ (t ) + δ (t − T )
dt
Taking Fourier transform on both the sides,
j 2 π f X ( f ) = e j 2 π f T − 2 + e− j 2 πf T
( )
2
= e j π f T − e− j πf T
X (f ) = 2 j T
(e j πf T
− e− j πf T
) (e j πf T
− e− j πf T
)
2 j πf T 2j
= 2 j T sin c ( f T ) sin ( π f T )
Example 1.17
Let x ( t ) , h ( t ) and y ( t ) denote the input, impulse response and the
Find a) Y ( f )
1.75
mywbut.com
Principles of Communication
b) y ( t ) = F − 1 ⎡⎣Y ( f ) ⎤⎦
c) y ( t ) = x ( t ) ∗ h ( t )
1
If z ( t ) = e − 2 t u ( t ) , then Z ( f ) = .
2 + j 2πf
⎛ j ⎞ d Z (f ) 1
As x ( t ) = t z ( t ) , we have X ( f ) = ⎜ ⎟ =
⎝ 2 π ⎠ df ( 2 + j 2 π f )2
1
H (f ) =
4 + j 2πf
1 1
Hence, Y ( f ) =
(2 + j 2πf )
2
4 + j 2πf
Y (f ) =
( − 14 ) +
1
2 +
1
4
2 + j 2πf (2 + j 2 π f ) 2
4 + j 2πf
⎛ 1⎞ 1 1 − 4t
Hence, y ( t ) = ⎜ − ⎟ e − 2 t u ( t ) + t e − 2 t u ( t ) + e u (t )
⎝ 4⎠ 2 4
∞
(c) y ( t ) = ∫ x ( τ) h (t − τ) d τ
−∞
∞
− 4( t − τ )
= ∫ τ e − 2τ u ( τ ) e u (t − τ) d τ
−∞
1.76
mywbut.com
Principles of Communication
t
= e− 4 t ∫τ e
2τ
dτ
0
t
⎡ e2 τ e2 τ ⎤
∫ 2 ⎥⎦
− 4t
= e ⎢τ − d τ
⎣ 2 0
⎧ e2 t ⎡ e2 τ ⎤ ⎫⎪
t
− 4t ⎪
= e ⎨t − ⎢ ⎥ ⎬
⎪⎩ 2 ⎣ 4 ⎦ 0 ⎪⎭
e− 2t ⎡ e2 t − 1⎤
= t − e− 4 t ⎢ ⎥, t ≥ 0
2 ⎣ 4 ⎦
= 0 for t < 0
Exercise 1.6
⎧ T0
⎪⎪1 + cos ( 2 π f0 t ) , t <
Let x ( t ) = ⎨ 2
⎪ T
0 , t > 0
⎪⎩ 2
1
where T0 is the period of the cosine signal and f0 = .
T0
(a) Show that
d 3 x (t ) 2 ⎧
⎪ ⎛ T ⎞ ⎛ T ⎞ d x ( t ) ⎫⎪
= ( 2 π f0 ) ⎨δ ⎜ t + 0 ⎟ − δ ⎜ t − 0 ⎟ − ⎬
d t3 ⎪⎩ ⎝ 2⎠ ⎝ 2⎠ dt ⎪⎭
1.77
mywbut.com
Principles of Communication
When we say that there is some correlation between two objects, we imply
that there is some similarity between them. We would like to quantify this intuitive
notion and come up with a formal definition for correlation so that we have a
mathematically consistent and physically meaningful measure for the correlation
of the objects of interest to us.
1.78
mywbut.com
Principles of Communication
∞
Ry x ( τ ) = ∫ y (t ) x (t − τ) d t
∗
(1.35b)
−∞
accounting for the shift in y ∗ ( t ) . Note that the variable of integration in Eq. 1.35
The power signals that we have to deal with most often are of the periodic
variety. For periodic signals, we have the following definitions:
T0
2
1
Def. 1.8(a): Rx p y p ( τ ) = ∫ x p ( t ) y p∗ ( t − τ ) dt (1.37a)
T0 T
− 0
2
T0
2
1
Def. 1.8(b): Ry p x p ( τ ) = ∫ y p ( t ) x p∗ ( t − τ ) dt (1.37b)
T0 T
− 0
2
−∞
= Ry∗ x ( − τ ) (1.38)
1.79
mywbut.com
Principles of Communication
∞
If we compute ∫ x ( t ) y ( t ) dt , we find it to be zero as x ( t ) and y ( t ) do not
−∞
⎛ 1⎞
overlap. However, if we delay x ( t ) by half a unit of time, we find that x ⎜ t − ⎟
⎝ 2⎠
1
and y ( t ) start overlapping and for τ > , we have nonzero value for the
2
integral. For the value of τ 2.75 , we will have a positive value for
1.80
mywbut.com
Principles of Communication
For values of τ > 2.75 , Ry x ( τ ) keeps decreasing, becoming zero for τ > 5 .
system, then we are willing to accept x ( t ) as the likely transmitted signal (we can
treat the received signal as a delayed and distorted version of the transmitted
signal) whereas if z ( t ) is received, it would be difficult for us to accept that x ( t )
could have been the transmitted signal. Thus the parameter τ helps us to find
time-shifted similarities present between the two signals.
−∞
∞
= ∫ x ( t ) y ( t − τ ) dt =
∗
Rx y ( τ ) (1.39a)
−∞
Let E x y ( f ) = F ⎡⎣Rx y ( τ ) ⎤⎦ .
Then, E x y ( f ) = F ⎡⎣ x ( τ ) ∗ y ∗ ( − τ ) ⎤⎦ = X ( f ) Y ∗ ( f ) (1.39b)
1.81
mywbut.com
Principles of Communication
Example 1.18
⎛t⎞
Let x ( t ) = exp1 ( t ) and y ( t ) = ga ⎜ ⎟ .
⎝ 2⎠
Let us find (a) Rx y ( τ ) and (b) Ry x ( τ ) .
From the results of (a) and (b) above, let us verify Eq. 1.38.
(a) R x y ( τ ) :
(i) τ < − 1:
Rx y ( τ ) = 0 for τ < − 1.
ii) − 1 ≤ τ < 1:
1+ τ
Rx y ( τ ) = e − t dt = 1 − e ( )
− 1+ τ
∫
0
1.82
mywbut.com
Principles of Communication
(iii) τ ≥ 1:
τ+1
Rx y ( τ ) = e − t dt = e ( ) − e ( )
− τ−1 − τ+1
∫
τ−1
⎛ 1⎞
= e− τ ⎜ e − ⎟
⎝ e⎠
(b) Ry x ( τ ) :
= e τ ⎡⎣e − τ − e − 1 ⎤⎦
= 1− e ( )
− 1− τ
(iii) For τ ≤ − 1,
1
− (t − τ)
Ry x ( τ ) = ∫e dt
−1
⎛ 1⎞
= eτ ⎜ e − ⎟
⎝ e⎠
1.83
mywbut.com
Principles of Communication
Ry x ( τ ) = R x y ( − τ )
∫ x ( t ) y ( t ) dt
∗
= 0 (1.40)
−∞
Rx y ( τ ) τ=0 = 0 (1.41a)
1.84
mywbut.com
Principles of Communication
Rx p y p ( τ + nT0 ) = Rxp y p ( τ )
A1.2.
signals. Hence Rx ( τ ) can provide some information about the time variations of
the signal.
−∞
∫ x ( t + τ ) x ( t ) dt
∗
= (1.42b)
−∞
1.85
mywbut.com
Principles of Communication
1.42(a), we keep x ( t ) fixed and move x ∗ ( t ) to the right by τ and take the
product; in Eq. 1.42(b), we keep x ∗ ( t ) fixed and move x ( t ) to the left by τ . This
does not change the integral of Eq. 1.42(a), because if we let t = t1 and τ = τ1 ,
quantities are obtained and hence the integral for a given τ1 remains the same.
If x ( t ) is a power signal, then the ACF is special case of Eq. 1.36(a). That
T0
2
1
= ∫ x p∗ ( t ) x p ( t + τ ) dt (1.44b)
T0 T
− 0
2
1.86
mywbut.com
Principles of Communication
Proof: Exercise
Proof: The proof of the above property follows from Schwarz’s inequality;
which is stated below.
Let g1 ( t ) and g 2 ( t ) be two energy signals.
2
∞ ∞ ∞
∫ g1 ( t ) g2 ( t ) dt g1 ( t ) dt g 2 ( t ) dt .
2 2
Then, ≤ ∫ ∫
−∞ −∞ −∞
Let g1 ( t ) = x ( t ) and g 2 ( t ) = x ∗ ( t − τ ) .
Rx ( τ ) ≤ ⎡⎣Rx ( 0 ) ⎤⎦ or
2 2
Rx ( τ ) ≤ Rx ( 0 ) (1.45c)
P4) Let E x ( f ) denote the Energy Spectral Density (ESD) of the signal x ( t ) .
∞
That is, ∫ Ex ( f ) df = Ex
−∞
Then, Rx ( τ ) ←⎯→ E x ( f )
Proof
From Eq. 1.39(b),
Ex y (f ) = X (f ) Y ∗ (f )
1.87
mywbut.com
Principles of Communication
Rx ( τ ) ←⎯→ X ( f )
2
Ex (f ) = Ex x (f ) = X (f )
2
Hence,
Rx ( τ ) ←⎯→ E x ( f ) (1.45d)
resemblance to y ( t ) ; but their ACFs will be the same. In other words, ACF does
not provide a unique description of the signal. Given x ( t ) , its ACF is unique;
but given some ACF, we can find many signals that have the given ACF.
P5) Let x ( t ) = y ( t ) + v ( t ) . Then,
Rx ( τ ) = Ry ( τ ) + Rv ( τ ) + Ry v ( τ ) + Rv y ( τ ) (1.45e)
Proof: Exercise
If Ry v ( τ ) = Rv y ( τ ) ≡ 0 (that is, y ( t ) and v ∗ ( t − τ ) are orthogonal for all
τ ), then, Rx ( τ ) = Ry ( τ ) + Rv ( τ ) (1.45f)
E x ( f ) = E y ( f ) + Ev ( f ) (1.45g)
1.88
mywbut.com
Principles of Communication
T0
2
1
P2) Rx p ( τ ) x p ( t ) d t = Px p
2
τ = 0
=
T0 ∫
T0
(1.46b)
−
2
P3) Rx p ( τ ) ≤ Rx ( 0 ) (1.46c)
where T0 is the period of x p ( t ) . That is, the ACF of a periodic signal is also
∫ Px ( f ) df
p
= Px p . Then,
−∞
Rx p ( τ ) ←⎯→ Px p ( f ) (1.46e)
Exercise 1.7
⎧ T T
⎪xp (t ) , − 0 < t < 0
Let x ( t ) = ⎨ 2 2 and x ( t ) ←⎯→ X ( f ) .
⎪⎩ 0 , outside
2
1 ∞ ⎛n⎞ ⎛ n⎞
Show that F ⎡Rx p ( τ ) ⎤ = Px p ( f ) = 2
⎣ ⎦ ∑ X ⎜ ⎟ δ⎜f − ⎟
T0 n = −∞ ⎝ T0 ⎠ ⎝ T0 ⎠
Example 1.19
a) Let x ( t ) be the signal shown in Fig. 1.29. Compute and sketch Rx ( τ ) .
1.89
mywbut.com
Principles of Communication
b) x ( t ) of part (a) is given as the input to an LTI system with the impulse
a) Computation of Rx ( τ ) :
We know that the maximum value of the ACF occurs at the origin; that is, at
τ = 0 and Rx ( 0 ) = E x .
1
Rx ( 0 ) = 1. 1 + . 2 = 1.5
4
Consider the product x ( t ) x ( t − τ ) for 0 < τ ≤ 1. As τ increases in this range,
the overlap between the positive parts of the pulses x ( t ) and x ( t − τ ) (and also
between the negative parts and these pulses) decreases, which implies a
decrease in the positive value for the integral of the product. In addition, a part of
x ( t − τ ) that is positive overlaps with the negative part of x ( t ) , there by further
⎛ 1⎞
slope) until τ = 1 . The value of Rx (1) is ⎜ − ⎟ . For 1 < τ < 2 , the positive part
⎝ 4⎠
of x ( t − τ ) fully overlaps with the negative part of x ( t ) ; this makes Rx ( τ ) further
negative and the ACF reaches its minimum value at τ = 2 . As can easily be
1.90
mywbut.com
Principles of Communication
⎛ 1⎞
checked, Rx ( 2 ) = ⎜ − ⎟ . As τ increases beyond 2, the Rx ( τ ) becomes less
⎝ 2⎠
and less negative and becomes zero at τ = 3 . As Rx ( − τ ) = Rx ( τ ) , we have all
b) Calculating h ( t ) :
We have y ( t ) = x ( t ) ∗ h ( t )
= Rx ( t − 3 )
= Rx ( t ) ∗ δ ( t − 3 )
Hence, y ( t ) = x ( t ) ∗ ⎡⎣ x ( − t ) ∗ δ ( t − 3 ) ⎤⎦
= x ( t ) ∗ x ⎡⎣ − ( t − 3 ) ⎤⎦
That is, h ( t ) = x ⎡⎣ − ( t − 3 ) ⎤⎦
1.91
mywbut.com
Principles of Communication
Example 1.20
Let x ( t ) = A cos ( ω0 t + θ ) . We will find Rx ( τ ) .
Method 1:
T0
2
1
Rx ( τ ) = ∫ A2 cos ( ω0 t + θ ) cos ⎡⎣ω0 ( t − τ ) + θ ⎤⎦ dt
T0 T
− 0
2
T0
A2 2
=
T0 ∫
1
2
{cos ( 2 ω0 t − ω0 τ + 2 θ ) + cos ω0τ} dt
T
− 0
2
A2
= cos ω0 τ
2
We find that:
(i) Rx ( τ ) is periodic with the same period as x ( t ) .
A2
(iii) The maximum value is which is the average power of the signal.
2
(iv) Rx ( τ ) is independent of θ .
1.92
mywbut.com
Principles of Communication
Method 2:
⎡ A j ω t + θ ) A − j ( ω0 t + θ ) ⎤
A cos ( ω0 t + θ ) = ⎢ e ( 0 + e ⎥
⎣2 2 ⎦
= v (t ) + w (t )
As x ( t ) = v ( t ) + w ( t ) , we have
Rx ( τ ) = Rv ( τ ) + Rw ( τ ) + Rv w ( τ ) + Rw v ( τ ) .
Rx ( τ ) = Rv ( τ ) + Rw ( τ )
Rx ( τ ) = 2 Re ⎡⎣Rv ( τ ) ⎤⎦
⎡
2
T0 ⎤ T0
A ⎢1 2
A2 2
dt ⎥⎥ =
j ⎡⎣ω0 ( t − τ ) + θ ⎤⎦
e ( 0
j ω t + θ) −
Rv ( τ ) = ∫ e ∫ e j ω0 τ dt
4 ⎢T0 4T0
⎢⎣ −
T 0
2
⎥⎦ −
T0
2
A2 j ω0 τ
= e
4
A2
Hence Rx ( τ ) = cos ( ω0 τ )
2
Example 1.21
Let x ( t ) = sin ( 2 π t ) , 0 ≤ t ≤ 2 . Let us find its ACF and sketch it.
In Fig. 1.32, we show x ( t ) and x ( t − τ ) for 0 < τ < 2 . Note that if τ > 2 ,
1.93
mywbut.com
Principles of Communication
Fig. 1.32: (a) Sinusoidal pulse of Example 1.21 and (b) its shifted version
2
Rx ( τ ) = ∫ sin ( 2 π t ) sin ⎣⎡2 π ( t − τ )⎦⎤ dt
τ
2
cos ( 2 π τ ) − cos ( 4 π t − 2 π τ )
= ∫ 2
dt
τ
⎡ sin ( 4 π t − 2 π τ ) ⎤
2 2
⎡ t cos 2 π τ ⎤
= ⎢ ⎥ − ⎢ ⎥
⎣ 2 ⎦τ ⎣ 4π ⎦τ
τ cos ( 2 π τ ) sin ( 8 π − 2 π τ ) − sin ( 2 π τ )
= cos ( 2 π τ ) − −
2 4π
τ cos 2 π τ sin 2 π τ
= cos ( 2 π τ ) − + , 0 ≤ τ ≤2
2 2π
As Rx ( − τ ) = Rx ( τ ) , we have
⎧ τ cos ( 2 π τ ) sin ( 2 π τ )
⎪cos ( 2 π τ ) − + , τ ≤ 2
Rx ( τ ) = ⎨ 2 2π
⎪
⎩ 0 , otherwise
1.94
mywbut.com
Principles of Communication
y (t ) = x (t ) ∗ h (t )
or Y (f ) = X (f ) ∗ H (f )
θ y ( f ) = θ x ( f ) + θh ( f ) (1.47b)
From Eq. 1.47 we see that an LTI system alters, in general, both the magnitude
spectrum and the phase spectrum of the input signal. However, there are certain
networks, called all pass networks, which would alter only the (input) phase
spectrum. That is, if H ( f ) is the frequency response of an all pass network, then
1.95
mywbut.com
Principles of Communication
Y (f ) = X (f )
θ y ( f ) = θ x ( f ) + θh ( f )
An interesting case of all-pass network is the ideal delay, with the impulse
response h ( t ) = δ ( t − td ) . Though θ y ( f ) ≠ θ x ( f ) , phase shift imparted to each
⎧ π
⎪⎪− 2 + θ x ( f ) , f > 0
(ii) θy ( f ) = ⎨
⎪ π + θ (f ) , f < 0
⎪⎩ 2 x
⎛ π⎞
That is, a Hilbert transform is essentially a ⎜ ± ⎟ phase shifter.
⎝ 2⎠
Hence, we define the Hilbert transformer in the frequency domain, with the
frequency response function
H ( f ) = − j sgn ( f ) (1.48a)
⎧ 1 , f > 0
⎪
where sgn ( f ) = ⎨ 0 , f = 0
⎪ −1 , f < 0
⎩
1
As ⎡⎣ − j sgn ( f ) ⎤⎦ ←⎯→ ,
πt
1
h (t ) = (1.48b)
πt
where
1
x̂ ( t ) = x ( t ) ∗
πt
1.96
mywbut.com
Principles of Communication
1
∞
x ( τ)
= ∫ (t − τ) d τ
1
(1.49a)
π −∞
Note: Unlike other transforms, both x ( t ) and x̂ ( t ) are functions of the same
Hilbert Transform (HT) will prove quite useful later on in the study of
bandpass signals and single sideband signals. For the present, let us look at
some examples of HT.
Example 1.22
Hilbert transform δ ( t ) .
As δ ( t ) ←⎯→ 1, we have
F ⎡⎣δˆ ( t ) ⎤⎦ = − j sgn ( f ) ⇒
1
δˆ ( t ) =
πt
⎡ 1⎤ 1
This also establishes the relation, ⎢ δ ( t ) ∗ ⎥ = !!
⎣ πt ⎦ πt
Example 1.23
HT of a cosine signal.
Let x ( t ) = cos ( 2 π f0 t ) . Let us find x̂ ( t ) .
1
This integral is actually Cauchy’s principal value.
1.97
mywbut.com
Principles of Communication
1
X (f ) = ⎡ δ ( f − f0 ) + δ ( f + f0 ) ⎦⎤
2⎣
Xˆ ( f ) = − j sgn ( f ) X ( f ) . That is,
1
Xˆ ( f ) = ⎡ δ ( f − f0 ) − δ ( f + f0 ) ⎤⎦
2j ⎣
Alternatively,
if x1 ( t ) = e j 2 π f0 t , then xˆ1 ( t ) = e
(
j 2 π f0 t − π
2 )
and if x2 ( t ) = e − j 2 π f0 t
, then xˆ 2 ( t ) = e
(
− j 2 π f0 t − π
2 )
Hence,
1 ⎛ π⎞
x ( t ) = cos ( 2 π f0 t ) = ⎡ x1 ( t ) + x2 ( t ) ⎦⎤ has xˆ ( t ) = cos ⎜ ω0 t − ⎟
⎣
2 ⎝ 2⎠
= sin ( ω0 t )
Example 1.24
1
Let x ( t ) = . Let us find x̂ ( t ) .
1 + t2
1
x̂ ( t ) = x ( t ) ∗
πt
∞
1 1
= ∫ dτ
− ∞ (1 + τ ) ( t − τ )
π 2
1 1 ⎡∞ t +τ ∞
dτ ⎤
= ⎢∫ dτ + ∫ t − τ ) ⎥⎥
−∞(
π 1 + t2 ⎢⎣ − ∞ 1 + τ
2
⎦
∞ ∞
τ 1
As ∫ dτ = ∫ d τ = 0 , we have
− ∞1+ τ
2
−∞
t −τ
1.98
mywbut.com
Principles of Communication
1 t ⎡∞ 1 ⎤
xˆ ( t ) = ⎢∫ d τ ⎥
π 1 + t2 ⎢⎣ − ∞ 1 + τ
2
⎥⎦
Exercise 1.8
sin t 1 − cos t
(a) Let x1 ( t ) = . Show that xˆ1 ( t ) =
t t
1
t −
1
(b) Let x2 ( t ) = ga ( t ) . Show that xˆ 2 ( t ) = − ln 2
π t +
1
2
Example 1.25
Let x ( t ) = m ( t ) cos 2 π fc t where m ( t ) is a lowpass signal with
xˆ ( t ) = m ( t ) sin ( 2 π fc t ) = m ( t ) cos ( 2 π fc t ) .
1
X (f ) = ⎡M ( f − fc ) + M ( f + fc ) ⎤⎦
2⎣
1
Xˆ ( f ) = ⎡M ( f − fc ) + M ( f + fc ) ⎤⎦ ⎡⎣ − j sgn ( f ) ⎤⎦
2⎣
But ⎡⎣M ( f − fc ) ⎤⎦ is nonzero only for f > 0
Hence,
⎧1 − j π
⎪⎪ 2 M ( f − fc ) e
2 , f > 0
Xˆ ( f ) = ⎨
π
⎪ 1 M (f + f ) e j 2 , f < 0
⎪⎩ 2 c
1.99
mywbut.com
Principles of Communication
1
Consider m ( t ) sin ( 2 π fc t ) ←⎯→ M ( f ) ∗ ⎡ δ ( f − fc ) − δ ( f + fc ) ⎤⎦
2j⎣
1⎡ ⎤
π π
− j − j
←⎯→ M ( f ) ∗ ⎢δ ( f − fc ) e 2 + ejπ e 2 δ ( f + fc ) ⎥
2 ⎣⎢ ⎦⎥
1⎡ ⎤
π π
− j j
←⎯→ M ( f ) ∗ ⎢ δ ( f − fc ) e 2 +e 2 δ ( f + fc ) ⎥
2 ⎣⎢ ⎦⎥
That is,
⎧1 − j
π
⎪ M ( f − fc ) e 2 , f > 0
⎪2
F ⎣⎡m ( t ) sin ( 2 π fc t ) ⎦⎤ = ⎨
π
⎪1 j
⎪⎩ 2 ( + c)
M f f e 2 , f < 0
xˆ ( t ) = m ( t ) sin ( 2 π fc t ) = m ( t ) cos ( 2 π fc t )
V (f ) .
1.100
mywbut.com
Principles of Communication
As Xˆ ( f ) = − j sgn ( f ) X ( f )
X̂ ( f ) = X ( f ) . Hence E x = E xˆ .
Note: Though Xˆ ( f ) is zero, it will not change the value of the integral; and
f = 0
That is, applying the HT twice on a given signal x ( t ) changes the sign of the
signal. Intuitively, this is satisfying. Each time we perform HT, we change the
phase of a spectral component in X ( f ) by 900 . Hence, performing the
= ( − j sgn ( f ) ) X ( f ) = − X ( f )
2
Example 1.26
1
Let x ( t ) = . We shall find x̂ ( t ) .
t
1
From Example 1.22, we have δˆ ( t ) = .
πt
Hence,
⎡1⎤
HT ⎡⎣δˆ ( t ) ⎤⎦ = − δ ( t ) = HT ⎢ ⎥ . That is,
⎣ πt ⎦
⎡ 1⎤
HT ⎢ ⎥ = − π δ ( t )
⎣t ⎦
1.101
mywbut.com
Principles of Communication
∫ x ( t ) xˆ ( t ) d t = ∫ X ( f ) Xˆ ( − f ) d f
−∞ −∞
∫ x ( t ) xˆ ( t ) d t = ∫ X ( f ) ⎣⎡ − j sgn ( − f ) X ( − f ) ⎦⎤ d f
−∞ −∞
As x ( t ) is real, X ( − f ) = X ∗ ( f ) . Therefore,
∞ ∞
x ( t ) xˆ ( t ) d t = − j sgn ( − f ) X ( f ) d f
2
∫ ∫
−∞ −∞
1
z ( t ) = x ( α t ) ∗ h ( α t ) , then z ( t ) = y (α t ) .
α
If z ( t ) = x ( α t ) ∗ h ( α t ) , then
1 ⎛f ⎞ ⎛f ⎞
Z (f ) = X⎜ ⎟ H⎜ ⎟
⎝a⎠ ⎝a⎠
2
α
1 ⎛f ⎞
Also, y ( α t ) ←⎯→
α ⎜⎝ α ⎟⎠
Y , where
⎛f ⎞ ⎛f ⎞ ⎛f ⎞
Y ⎜ ⎟ = X⎜ ⎟ H⎜ ⎟
⎝α⎠ ⎝α⎠ ⎝α⎠
= α Z (f )
2
1
Hence, y ( α t ) ←⎯→ α Z (f ) = α Z (f )
2
y (α t )
That is, z ( t ) =
α
1.102
mywbut.com
Principles of Communication
1
HT ⎡⎣ x ( α t ) ⎤⎦ = x ( α t ) ∗
πt
⎛ 1 ⎞
= x (α t ) ∗ α ⎜ ⎟
⎝ α πt ⎠
⎡ 1 ⎤
= α ⎢x (α t ) ∗
⎣ α π t ⎥⎦
⎡ 1⎤ ⎡ 1 ⎤ xˆ ( α t )
As ⎢ x ( t ) ∗ ⎥ = xˆ ( t ) , we have ⎢x (α t ) ∗ π α t ⎥ = .
⎣ πt ⎦ ⎣ ⎦ α
α
Hence HT ⎣⎡ x ( α t ) ⎦⎤ = xˆ ( α t ) = sgn ( α ) xˆ ( α t ) .
α
1
As a simple illustration of the property, let x ( t ) = and α = 2 . Then
1 + t2
1 ⎡ 1 ⎤
x (α t ) = . Let us obtain HT ⎢ 2⎥
using P4.
1 + 4t2 ⎣ 1 + 4 t ⎦
t ⎡ 1 ⎤ ⎡ 2t ⎤ 2t
As xˆ ( t ) = , HT ⎢ 2⎥
= sgn ( 2 ) ⎢ ⎥ =
1 + t2 ⎣1 + 4 t ⎦ ⎢⎣1 + ( 2 t ) ⎥⎦ 1 + 4t2
2
1
If α = − 2 , x ( α t ) = which is the same as with α = 2 .
1 + 4t2
⎛ − 2t ⎞
With α = − 2 , sgn ( α ) xˆ ( α t ) = − ⎜ 2⎟
⎝1 + 4t ⎠
2t
= , as required
1 + 4t2
Exercise 1.9
⎛ sin t ⎞ 1 − cos t
Using the result HT ⎜ ⎟ = ,
⎝ t ⎠ t
1.103
mywbut.com
Principles of Communication
Rx xˆ ( τ ) = − Rˆ x ( τ )
Proof: F ⎡⎣Rx xˆ ( τ ) ⎤⎦ = F ⎡⎣ x ( τ ) ∗ xˆ ( − τ ) ⎤⎦
= X ( f ) ⎡⎣ − j sgn ( − f ) X ( − f ) ⎤⎦
= X ( f ) ⎡⎣ − j sgn ( − f ) ⎤⎦ = X ( f ) ⎡⎣ j sgn ( f ) ⎤⎦
2 2
That is,
Rx xˆ ( τ ) = − Rˆ x ( τ )
Exercise 1.10
Show that Rxˆ x ( τ ) = Rˆ x ( τ ) .
1
transmitted signal is S ( f ) = ⎡M ( f − fc ) + M ( f + fc ) ⎤⎦ . If M ( f ) is as shown in
2⎣
Fig 1.34(a), then for a carrier frequency fc = 100 k Hz , S ( f ) will be as shown in
Fig 1.34(b).
1.104
mywbut.com
Principles of Communication
(NBBP) signal. Fig 1.35 shows some more spectra that represent NBBP signals.
1.105
mywbut.com
Principles of Communication
meant only to “carry” the information m ( t ) and is not part of the information) If
1.106
mywbut.com
Principles of Communication
1.8.1 Pre-envelope
Def. 1.10: Let x ( t ) be any real signal with the FT, X ( f ) . We define its pre-
envelope as,
x p e ( t ) = x ( t ) + j xˆ ( t ) (1.50a)
= X ( f ) + sgn ( f ) X ( f )
That is,
⎧2 X ( f ) , f > 0
⎪
X p e (f ) = ⎨ X (0) , f = 0 (1.50b)
⎪
⎩ 0 , f < 0
components only for f ≥ 0 , some authors use the symbol x+ ( t ) to denote the
Consider the signals x1 ( t ) and x2 ( t ) whose spectra are shown in Fig. 1.36.
1.107
mywbut.com
Principles of Communication
The corresponding X1, p e ( f ) and X 2, p e ( f ) are as shown in Fig. 1.37(a) and (b)
respectively.
Fig. 1.37: Fourier transform of (a) x1, p e ( t ) and (b) x2, p e ( t ) of the signals in
1.108
mywbut.com
Principles of Communication
Example 1.27
1
Let x ( t ) = . Let us find X pe ( f ) and x pe ( t ) .
1 + t2
1 − 2 πf
←⎯→ π e .
1+ t 2
⎧1 , f > 0
⎪1
⎪
Hence X pe ( f ) = 2 π e − 2 π f u ( f ) , where u ( f ) = ⎨ , f = 0.
⎪2
⎪⎩ 0 , f < 0
We require F − 1 ⎡⎣ X pe ( f ) ⎤⎦ .
1
As ex1( t ) ←⎯→
1 + j 2πf
1 1
ex1( 2 π t ) ←⎯→ .
2π 1 + j f
1
From duality, 2 π ex1( 2 π f ) = 2 π e − 2 π f u ( f ) ←⎯→ .
1− jt
1 1+ jt
That is, x pe ( t ) = =
1− jt 1 + t2
1 t
= + j .
1+ t 2
1 + t2
t
Then, xˆ ( t ) = , which is a known result.
1 + t2
1.109
mywbut.com
Principles of Communication
in Fig. 1.38(a).
(We can treat fc to be the center frequency in Fig. 1.38(a), by taking the
bandpass spectrum from 94 to 110 k Hz , though the spectrum is zero for the
1.110
mywbut.com
Principles of Communication
xc e ( t ) = x p e ( t ) e − j 2 π fc t
(1.52a)
(We assume that we know the center frequency fc and it is such that
X p e ( f + fc ) is lowpass in nature).
We will show a little later that, both xc ( t ) and xs ( t ) are lowpass in nature. Using
phase component and xs ( t ) , the coefficient of the sine term, as the quadrature
also common in the literature to use the symbol xI ( t ) for the in-phase
1.111
mywbut.com
Principles of Communication
⎡ x (t ) ⎤
ϕ ( t ) = tan−1 ⎢ s ⎥ (1.56b)
⎢⎣ xc ( t ) ⎥⎦
Then,
xce ( t ) = A ( t ) e ( )
jϕ t
(1.56c)
x ( t ) = Re ⎡⎣ x p e ( t ) ⎤⎦ = Re ⎡⎣ xc e ( t ) e j 2 π fc t ⎤⎦ (1.57)
= Re ⎡ A ( t ) e ( ) e j 2 π fc t ⎤
jϕ t
(1.58)
⎣ ⎦
Eq. 1.58 resembles phasor representation of a sinusoid. We know that,
A cos ( 2 π fc t + ϕ ) = Re ⎡⎣ A e j ϕ e j 2 π fc t ⎤⎦ (1.59)
about the amplitude and phase (at t = 0 ) of the sinusoid.) The quantity
1.112
mywbut.com
Principles of Communication
1.113
mywbut.com
Principles of Communication
⎡ xc e ( t ) + xc∗ e ( t ) ⎤
xc ( t ) = ⎣ ⎦
2
X c e ( f ) + X c∗ e ( − f )
or F ⎡⎣ xc ( t ) ⎤⎦ = X c ( f ) =
2
As X c e ( f ) and X c∗ e ( − f ) are zero for f > W , X c ( f ) is also zero for f > W .
from x ( t ) .
In Fig 1.40, x ( t ) a NBBP signal, with the spectrum confined to the interval
f ± f c ≤ W , where W << fc .
v1 ( t ) = 2 x ( t ) cos ωc t
= 2 xc ( t ) cos2 ωc t − xs ( t ) sin 2 ωc t
1.114
mywbut.com
Principles of Communication
⎡1 + cos 2 ωc t ⎤
= 2 xc ( t ) ⎢ ⎥ − xs ( t ) sin 2 ωc t
⎣ 2 ⎦
= xc ( t ) + xc ( t ) cos ( 2 ωc t ) − xs ( t ) sin 2 ωc t (1.60)
response.
⎧⎪1 , f ≤ W
Hl p ( f ) = ⎨
⎪⎩0 , outside
will be filtered out by the ILPF and at the output of the top channel, we obtain
xc ( t ) . Similar analysis will show that the output of the bottom channel is xs ( t ) .
is called the natural envelope (or simply the envelope) of x ( t ) and ϕ ( t ) , its
1.115
mywbut.com
Principles of Communication
Example 1.28
Let x ( t ) = cos [ 2 π fc t ] . Let us find x pe ( t ) , xce ( t ) , A ( t ) and ϕ ( t ) .
1
X (f ) = ⎡ δ ( f − fc ) + δ ( f + fc ) ⎦⎤
2⎣
From Eq. 1.50(b), we have
Hence X pe ( f ) = δ ( f − fc ) ⇒ x pe ( t ) = e j 2 π fc t
As X ce ( f ) = δ ( f ) , we obtain xce ( t ) = 1
Example 1.29
⎛t ⎞
Let x ( t ) = ga ⎜ ⎟ cos ωc t . Assume that fc T >> 1 so that x ( t ) can be
⎝T ⎠
taken as a NBBP signal. We shall find x pe ( t ) , xce ( t ) and A ( t ) .
⎧T
⎪⎪ 2 sin c ⎡⎣( f − fc )T ⎤⎦ , f > 0
X (f ) ⎨
⎪T sin c ⎡( f + f )T ⎤ , f < 0
⎪⎩ 2 ⎣ c ⎦
Hence,
⎛t ⎞ j 2 π fc t
x pe ( t ) = ga ⎜ ⎟e
⎝T ⎠
1.116
mywbut.com
Principles of Communication
xce ( t ) = x pe ( t ) e − j 2 π fc t
⎛t ⎞
= ga ⎜ ⎟
⎝T ⎠
As xce ( t ) is real, we have xs ( t ) = 0 ⇒ ϕ ( t ) = 0 .
⎛t ⎞
A ( t ) = xce ( t ) = g a ⎜ ⎟
⎝T ⎠
Note that for Examples 1.28 and 1.29, the complex envelope is real valued and is
equal to the envelope, A ( t ) .
⎛t ⎞
canonic form with xc ( t ) = ga ⎜ ⎟ , xs ( t ) = 0
⎝T ⎠
⎛t ⎞
Hence, xce ( t ) = ga ⎜ ⎟
⎝T ⎠
And x pe ( t ) = xce ( t ) e j 2 π fc t
⎛t ⎞
= ga ⎜ ⎟ e j 2 π fc t
⎝T ⎠
⎛t ⎞
A ( t ) = ga ⎜ ⎟ and ϕ ( t ) = 0
⎝T ⎠
Example 1.30
x ( t ) is a NBBP signal with X ( f ) as shown in Fig 1.41. Let us find xc ( t )
and xs ( t ) .
1.117
mywbut.com
Principles of Communication
Fig. 1.42.
Fig. 1.42: Spectrum of the complex envelope of the signal of Example 1.30
1.118
mywbut.com
Principles of Communication
Exercise 1.11
Find the pre-envelope of x ( t ) = sin c ( t ) .
Exercise 1.12
Let x ( t ) = ex1( t ) sin ⎣⎡( ωc + ∆ ω) t ⎦⎤ where ωc >> ∆ ω .
Find xce ( t ) .
Exercise 1.13
Find the expression for the envelope of
x ( t ) = ⎣⎡1 + k cos ( ωm t ) ⎦⎤ cos ( ωc t ) where k is a real constant and
Ey ( f ) = Ex ( f ) H ( f )
2
(1.63a)
Ry ( τ ) = Rx ( τ ) ∗ Rh ( τ )
= Rx ( τ ) ∗ ⎡⎣ h ( τ ) ∗ h∗ ( − τ ) ⎤⎦ (1.63b)
1.119
mywbut.com
Principles of Communication
bandpass signal is usually processed a bandpass system; that is, a system with
passband in the interval f ± fc < B where B ≤ W . We would like to study the
Example 1.31
Let x ( t ) be a sinusoidal pulse given by
x (t ) = ⎨ ⎣ ( )
⎧⎪2 cos ⎡2 π 106 t ⎤ , 0 ≤ t ≤ 1 m sec
⎦
⎪⎩ 0 , outside
xce ( t ) = 2 , 0 ≤ t ≤ 1 m sec
h ( t ) = 2 cos ⎡2 π 106
⎣ ( ) (T − t )⎦⎤
{ ( ) ( ) ( ) ( )}
= 2 cos 2 π 106 T cos 2 π 106 t + sin 2 π 106 T sin 2 π 106 t
{ ( ) ( ) ( )
= 2 cos 2 π × 103 cos ⎡2 π 106 t ⎤ + sin 2 π × 103 sin 2 π × 106 t
⎣ ⎦ }
⎣ ⎦( )
= 2 cos ⎡2 π 106 t ⎤ , 0 ≤ t ≤ 1 m sec
1.120
mywbut.com
Principles of Communication
⎧2 , 0 ≤ t ≤ 1 m sec
hce ( t ) = ⎨
⎩0 , outside
⎛t −T ⎞
That is, xce ( t ) = hce ( t ) = 2 ga ⎜ 2⎟
⎜ T ⎟
⎝ ⎠
From Eq.1.64,
⎛t −T ⎞
y ce ( t ) =
1
2
( )
⎡⎣ xce ( t ) ∗ hce ( t ) ⎤⎦ = 2 × 10− 3 tri ⎜
⎝ T ⎠
⎟
y ( t ) = y ce ( t ) cos ( 2 π fc t )
Exercise 1.14
⎣ ( )
Let x ( t ) = sin c ( 200 t ) cos ⎡ 2 π 106 t ⎤ be the input to an NBBP
⎦
system with H ( f ) =
1 ⎡
2j ⎣
( ) (
H1 f − 106 − H1 f + 106 ⎤
⎦ ) (1.65)
1.121
mywbut.com
Principles of Communication
Appendix A1.1
Tabulation of sinc ( λ )
1.122
mywbut.com
Principles of Communication
1.123
mywbut.com
Principles of Communication
Appendix A1.2
Fourier transform of Rxp y p ( τ )
T0
2
1
Rx p y p ( τ ) = ∫ x p ( t ) y p∗ ( t − τ ) dt
T0 T0
−
2
⎧ T0 T
⎪x (t ) , − < t < 0
Let x (t ) = ⎨ p 2 2
⎪⎩ 0 , outside
∞
then, x p ( t ) = ∑ x ( t − nT0 )
n = −∞
∞
Similarly, y ∗p ( t − τ ) = ∑ y ∗ ( t − τ − nT0 )
n = −∞
⎧ ∗ T T
⎪y p (t ) , − 0 < t < 0
where y ( t ) = ⎨
∗
2 2
⎪⎩ 0 , outside
T0
1 2 ⎡ ∞ ⎤
Rx p y p ( τ ) = ∫ x (t ) ⎢ ∑ ( y *
t − τ − nT0 ) ⎥ dt
T0 T0 ⎢⎣ n = −∞ ⎥⎦
−
2
∞ T0 / 2
1
= ∑ ∫ x ( t ) y * ( t − τ − nT0 ) dt
n = −∞ T0 −T0 / 2
T0
As x ( t ) = 0 for t > ,
2
1.124
mywbut.com
Principles of Communication
∞ ∞
1
Rx p y p ( τ ) =
T0
∑ ∫ x ( t ) y ( t − τ − nT0 ) dt
∗
n = − ∞ −∞
∞
1
=
T0
∑ Rx y ( τ + nT0 )
n = −∞
1 ⎡ ∞ ⎤
F ⎡R x p y p ( τ ) ⎤ =
⎣ ⎦
F⎢
T0 ⎢⎣ n
∑ xy R ( τ + nT0 ) ⎥
⎥⎦
= −∞
As Rx y ( τ ) ←⎯→ X ( f ) Y ∗ ( f ) ,
1 ⎡ ∞ ⎤
F ⎡R x p y p ( τ ) ⎤ =
⎣ ⎦
⎢
T0 ⎢⎣ n
∑ X ( f ) Y ∗
( f ) e j 2 π n f T0
⎥
⎥⎦
= −∞
1 ⎡ ∞ ⎤
= ⎢ X (f ) Y ∗ (f )
T0 ⎢⎣
∑ e j 2 π n f T0
⎥
⎥⎦
(A1.2.1)
n = −∞
1 ⎡ ∞ ⎤
= ⎢
T0 ⎢⎣ m
∑ e j 2 π m f0 t ⎥
⎥⎦
= −∞
∞
1 ⎡ ∞ ⎤
∑ δ ( f − m f0 ) = ⎢
f0 ⎢⎣ m
∑ e j 2 π m f T0 ⎥
⎥⎦
m = −∞ = −∞
1
As f0 = , we have
T0
∞ ⎛ ∞
1 m⎞
∑ δ⎜f − T ⎟ = ∑ e j 2 π m f T0 (A1.2.3)
T0 m = −∞ ⎝ 0⎠ m = −∞
1.125
mywbut.com
Principles of Communication
1 ⎡ ∞ ⎛m⎞ ⎛m⎞ ⎛ m ⎞⎤
F ⎡R x p y p ( τ ) ⎤ = 2 ⎢
⎣ ⎦ T0 ⎢⎣ m
∑ X ⎜T ⎟ Y∗ ⎜T ⎟ δ⎜f − ⎟⎥
T0 ⎠ ⎥
(A1.2.4)
= −∞ ⎝ 0⎠ ⎝ 0⎠ ⎝ ⎦
⎛m⎞
where X ⎜ ⎟ = X ( f ) m
f =
⎝ T0 ⎠ T0
⎛m⎞
and Y ∗ ⎜ ⎟ = Y ∗ ( f ) m .
⎝ T0 ⎠ f =
T0
1.126
mywbut.com
Principles of Communication
Appendix A1.3
Complex envelope of the output of a BP system
Let x ( t ) , a BP signal, be applied as input to a BP system with impulse response
We know that,
x ( t ) = xc ( t ) cos ωc t − xs ( t ) sin ωc t ,
xce ( t ) = xc ( t ) + j xs ( t ) ,
x ( t ) = Re ⎡⎣ xce ( t ) e j 2 π fc t ⎤⎦ ,
X (f ) =
1
2
{
X ce ( f − fc ) + X ce
∗
⎡⎣ − ( f + fc ) ⎤⎦ . } (A1.3.1)
Similarly, let
hce ( t ) = hc ( t ) + j hs ( t ) , (A1.3.2)
h ( t ) = Re ⎡⎣hce ( t ) e j 2 π fc t ⎤⎦ ,
2 h ( t ) = hce e j 2 π fc t + hce
∗
e− j 2 π fc t
. (A1.3.4)
But y ( t ) = Re ⎡⎣ y ce ( t ) e j 2 π fc t ⎤⎦ .
Therefore,
Yce ( f − fc ) + Yce
∗
⎡⎣ − ( f + fc ) ⎤⎦
Y (f ) = X (f ) H (f ) = (A1.3.6)
2
Because of Eq. A1.3.1 and A1.3.5, Eq. A1.3.6 becomes
X (f ) H (f ) =
1
4 ⎣
{
⎡Hce ( f − fc ) + Hce
∗
⎡⎣ − ( f + fc ) ⎤⎦ ⎤⎦ ⎡⎣ X ce ( f − fc ) + X ce
∗
}
⎡⎣ − ( f + fc ) ⎤⎦ ⎦⎤
1.127
mywbut.com
Principles of Communication
Hce ( f − fc ) X ce
∗
⎡⎣ − ( f + fc ) ⎤⎦
⎣⎡ − ( f + fc ) ⎦⎤ X ce ( f − fc ) = 0 . Hence,
∗
product is zero. Similarly, Hce
Yce ( f − fc ) + Yce
∗
⎡⎣ − ( f + fc ) ⎤⎦ 1
= Hce ( f − fc ) X ce ( f − fc )
2 4
1 ∗
+ Hce ⎡⎣ − ( f + fc ) ⎤⎦ X ce
∗
⎡⎣ − ( f + fc ) ⎤⎦
4
Yce ( f − fc ) has nonzero spectral components only in the range ( fc − B , fc + B ) .
That is,
1 1
Yce ( f − fc ) = ⎡Hce ( f − fc ) X ce ( f − fc ) ⎤⎦ ,
2 4 ⎣
and
1 ∗
2
Yce ⎡⎣ − ( f + fc ) ⎤⎦ =
1
4
⎡Hce
⎣
∗
( − ( f + fc ) ) X ce
∗
( − ( f + fc ) )⎤⎦ .
In other words,
1
Yce ( f ) = X ce ( f ) Hce ( f ) (A1.3.7)
2
Therefore,
1
y ce ( t ) = ⎡ xce ( t ) ∗ hce ( t ) ⎤⎦ (A1.3.8)
2⎣
From Eq. A1.3.8, we obtain the equations for yc (t ) and ys (t ) .
y ce ( t ) =
1
{ }
⎡ xc ( t ) + j xs ( t ) ⎤⎦ ∗ ⎡⎣hc ( t ) + j hs ( t ) ⎤⎦
2 ⎣
Therefore,
yc (t ) =
1
2
{xc ( t ) ∗ hc ( t ) − xs ( t ) ∗ hs ( t )} (A1.3.9)
ys (t ) =
1
2
{xc ( t ) ∗ hs ( t ) + xs ( t ) ∗ hc ( t )} (A1.3.10)
and, y ce ( t ) = y c ( t ) + j y s ( t ) .
1.128
mywbut.com
Principles of Communication
Exercise A1.3.1
Given the pairs ( xc ( t ) , xs ( t ) ) and ⎡⎣ hc ( t ) , hs ( t ) ⎤⎦ suggest a scheme to
recover y c ( t ) and y s ( t ) .
1.129
mywbut.com
Principles of Communication
References
1. Oppenheim, A. V, Willisky, A. S, with Hamid Nawab, S., Signals and Systems
(2nd Edition), PHI, 1997
2. Ashok Ambardar, Analog and Digital Signal Processing (2nd Edition),
Brooks/Cole Publishing Company, Thomson Asia Pvt. Ltd., Singapore, 1999
3. Lathi, B. P., Signal Processing and Linear systems, Berkeley-Cambridge
Press, 1998
Note: The above three books have a large collection of problems. The student is
advised to try to solve them.
1.130
mywbut.com