UNIT I

Spectral Analysis and Sampling

- DHIRAJ PATEL
UNIT I
Spectral Analysis and Sampling
• Fourier series and fundamentals,
• The Fourier transform,
• signal spectra,
• Energy density spectrum,
• Power density spectrum,
• Auto and cross correlation functions,
• properties of Fourier transform,
• Parseval’s theorem,
UNIT I
Spectral Analysis and Sampling
• Rayleigh Energy theorem,
• LTI system response and distortion less
transmission,
• Band limited and time limited signals,
• sampling theorem in frequency domain and
time domain,
• Nyquist criteria,
• Reconstruction using interpolation filters,
• Ideal, natural, flat top sampling,
• Aperture effect,
Today's lecturer agenda
• Band limited and time limited signals,
• sampling theorem in frequency domain and
time domain,
• Nyquist criteria,
• Reconstruction using interpolation filters,
• Ideal, natural, flat top sampling,
• Aperture effect,
BAND LIMITED AND TIME
LIMITED SIGNALS
Band limited and time limited signals
Time limited signal
• Time limited signal is the signal which exits
only over a certain time slot or duration
• Outside that slot the signal does not exit
Mathematically representation of RP
Rectangular pulse

X(t) = A -T/2≤ t ≤ T/2

=0 else where
Band limited signal
• A band limited signal is the signal having a
frequency spectrum which exits only over a
certain frequency range
• The frequency spectrum of a band limited
signal will be zero outside this frequency
range
• The spectrum mathematically expressed
Band limited signal
A signal can't be band limited & Time
limited simultaneously
SAMPLING
PROCESS(SAMPLIN
G)
Introduction(Sampling
• we know that a message signal can originate
from a digital or analog source
• The PM and DMS,the signal to be
transmitted must be in the discrete time form
• Data is not always in digital communication
• But this is not always the case, the message
signal can be in analog in nature(i.e. speech
and video signal)
Sampling Process
• Sampling is the process of converting a
continuous analog signal to a discrete analog
signal and the sampled signal is the discrete
time representation of the original analog
signal
• Higher the number of samples, the closer is
the representation
• The number of samples depends on the
sampling rate
Sampling Process
 SAMPLING THEOREM
IN FREQUENCY DOMAIN AND TIME
DOMAIN

Claude Elwood Shannon (April 30, 1916 – February 24,

2001) was an American mathematician,
electronic engineer, andcryptographer known as "the
father of information theory”]
 Known for Information Theory

• Shannon–Fano coding
• Shannon–Hartley law
• Nyquist–Shannon sampling theorem
• Noisy channel coding theorem
• Shannon switching game
• Shannon number
• Shannon index BornApril 30, 1916
• Shannon's source coding theorem Petoskey,
• Shannon's expansion Michigan, United States
DiedFebruary 24,
• Shannon-Weaver model of communication 2001 (aged 84)
• Whittaker–Shannon interpolation formula Medford, Massachusetts,
United States
 SAMPLING THEOREM
• sampling theorem was introduce to the
communication theory in 1949 by shannon,there
for this theorem is also called as Shannon's
sampling theorem

sampling theorem - sampling theorem state that if, The

sampling rate in any pulse modulation system exceeds twice
the maximum signal frequency, the original signal can be
constructed in the receiver with minimum distortion

Fs ≥ 2Fm or Fs ≥ 2W
Fs - The sampling frequency
Fm-Maximum frequency present in the signal
 Sampling Theorem
The sampling theorem for finite-energy band-limited signals
can be stated in two equivalent parts related to the transmitter
and receiver of a pulse modulation system:
1. A finite-energy band-limited signal with bandwidth W Hertz,
is completely described in terms of its samples taken at a rate
fs=1/Ts2W samples per second.
2. A finite-energy band-limited signal with bandwidth W Hertz,
may be completely recovered (reconstructed) from its samples
taken at a rate fs=1/Ts2W samples per second.

Note: The minimum sampling rate fs=2W is called Nyquist rate

and its reciprocal (inverse) Ts=1/2W is called Nyquist interval.

03/15/24 18
 sampling theorem in time domain
• To prove the sampling theorem
• we shall show that a signal whose spectrum is
band limited to Fm Hz, can we reconstructed
exactly without any error from its samples
taken uniformly at a rate Fs ≥ 2Fm Hz
• Let us consider a CTS x(t) whose spectrum is
bandlimited to Fm Hz.
 sampling theorem in time domain
• This means that the signal x(t) has no frequency
components beyond Fm Hz
• x(ω) is zero for ⃓ ω⃓ > ωm
• i.e. X(ω) = 0 for ⃓ ω⃓ > ωm
ωm= 2π fm
 Fig shows CTS x(t) .
 Let X(ω) represents its Fourier transform or
frequency spectrum as shown in fig 3.1(b)
sampling theorem in time domain
• Sampling of x(t) at a rate of Fs Hz(Fs sample per
second) may be achieved by multiplying x(t) by
an impulse train δTs (t)
• The impulse train δTs (t) consists of unit
impulses repeating periodically every Ts seconds
fig b shows this impulse train
Where Ts=1/Fs --- sampling period
Fs=1/Ts --- sampling rate
• The multiplication results in the sampled signal
g(t) shown in fig e
• This sampled signal consist of impulses spaced
every Ts second(the sampling interval)
102
 ST in TD
• The resulting or sampled signal may be written as
g(t) = x(t) δTs (t) ……….. 1

• Again, since the impulse train δTs (t) is a periodic signal

of period Ts, it may be express as a Fourier series
• The trigonometric Fourier series expansion of impulse
train δTs (t) is expressed as
δTs(t)= 1/Ts [ 1+ 2 cos ωst + 2 cos 2ωst +2 cos 3ωst + ………]…. 2
Here ωs = 2π /Ts = 2π fs
ST in TD
• Putting the value of δTs (t) from equation… 2
in Eq. …1
g(t) = x(t) 1/Ts [ 1+ 2 cos ωst + 2 cos 2ωst +2 cos 3ωst + ………]
g(t) = 1/Ts [ x(t)+ 2 x(t)cosωst + 2x(t) cos2ωst +2 x(t) cos 3ωst +
………] -----------3
Now obtain G(ω),the Fourier transformation of g(t),we will have
take the FT of right side
FT of x(t) is X(ω)
FT of 2x(t) cosωst is [X (ω- ωs) + X (ω+ ωs)]
Ft of 2x(t) cos2ωst is [X (ω- 2ωs) + X (ω+ 2ωs)] and so on
Therefore by taking FT of Eq. 3 becomes
 ST in TD
G(ω) = 1/Ts[X(ω)+X (ω- ωs) + X (ω+ ωs)+ X (ω- 2ωs)
+ X (ω+ 2ωs) + X (ω- 3ωs) + X (ω+ 3ωs) + …..] ….4

Or

From Eq. 4 & 5 it is clear that the spectrum G(ω)

consist of X(ω) repeating periodically with period
ωs = 2π /Ts rad/sec or Fs = 1/Ts Hz as shown in fig f
• Now if we reconstruct x(t) from g(t),we must be able to
recover X(ω) from G(ω)
• This is possible if there is no overlap between successive
cycles of G(ω) fig f shows that this requires
Fs >2Fm
but the sampling interval Ts = 1/Fs
Hence Ts< 1/2fm
SAMPLING
THEOREM
IN FREQUENCY
DOMAIN
SAMPLING THEOREM IN FREQUENCY
DOMAIN
• In previous section we prove sampling
theorem in time domain.
• That means sampling takes place in the time
domain & signal x(t) which is to be sampled
is bandlimited.
•A time limited signal which has a zero value
for ⃓ t ⃓ > T is uniquely determined by
the samples of its frequency spectrum if
the samples are spaced at intervals less
than (1/2T)Hz
SAMPLING THEOREM IN FREQUENCY
DOMAIN
• The signal x(t) is time limited instead of
being bandlimited as shown in fig 2.2.15(a)

• As the signal x(t) is a rectangular pulse

extended from –T/2 to +T/2 its spectrum is a
sinc pulse as shown in fig 2.15 (b)
• This sinc pulse X(F) is sampled(i.e. sampling
takes place in he frequency domain) at a
frequency Fs which is less than (1/2T) Hz
SAMPLING THEOREM IN FREQUENCY
DOMAIN
• The sampled version of X(F) is denoted by Xδ (F)
and each frequency sample is separated by
“Fs” wrt the adjacent frequency samples as
shown in fig 2.2.15(c) .The spectrum X(F) is
thus sampled a uniform interval less than
(1/2THz
• As seen from fig 2.2.15(c), the spectrum X(F)
has been sampled in the frequency domain.
We can represent CFS X(F) in terms of its
samples (encircled points in fig 2.2.15(c)
SAMPLING THEOREM IN FREQUENCY
DOMAIN
• The interpolation function i.e. sinc function as follows

• All the important concept discussed for the TDS are

applicable to the frequency domain sampling as well as For
eg. The replicas are now obtained in the time domain
• They are located at the time instants which are multiples of
Ts=1/Fs
• The problem of aliasing can occur in the frequency domain
sampling
• The problem of
aliasing can occur
in the frequency
domain sampling
Important points about
sampling Theorem
Fs > 2Fm
Showing the spectrum of g(t)
• Each term of the convolution is the original
spectrum shifted to a multiple of sampling
frequency
G(f)

G(f)
fs

fs 2fs
36
Recovering the original signal
• It is possible to recover the original spectrum
by lowpass filtering the sampled signal

G(f)
fs

W
fs 2fs
LPF

-W W
37
Fs = 2Fm
Nyquist sampling rate
• In order to cleanly extract baseband (original)
spectrum, we need sufficient separation with
the adjacent sidebands
• Min. separation can be found as follows
G(f)
fs fs-w>W

fs>2W
W fs

39
Fs < 2Fm
Sampling below Nyquist:
aliasing
• If signal is sampled below its Nyquist rate,
spectral folding, or aliasing, occurs.

Lowpass filtering will not recover

the baseband spectrum intact as a
result of spectral folding

fs<2W
41
 Nyquist rate
&
Harry Nyquist (1889-1976) Nyquist
Interval
• As an engineer at Bell Laboratories, Nyquist did
important work on thermal noise ("Johnson–
Nyquist noise"),
• The stability of feedback amplifiers,
telegraphy, facsimile,
• television, and other important communications
problems. With Herbert E. Ives,
• He helped to develop AT&T's first facsimile
machines that were made public in 1924. In
1932, he published a classical paper on stability
of feedback amplifiers.
• The Nyquist stability criterion can now be found
in all textbooks on feedback control theory.
Nyquist rate
• Nyquist rate – When sampling rate
becomes exactly equal to 2Fm samples
per second, then it is called Nyquist rate
• Nyquist rate is also called the minimum
sampling rate

It is given by
Fs = 2Fm
Nyquist interval
• Nyquist interval – Maximum
sampling interval is called Nyquist
interval
Nyquist interval
Ts = 1/2Fm seconds
 1.Example on Nyquist rate and
Interval
• EX. Determine the Nyquist rate for a continuous
time signal
X(t)= 6 cos 50πt +20 sin 300πt – 10 cos 100πt…..A
Solution:In a general form, any
continuous time signal may be expressed
as x(t)=A1cos ω1t+A2cos ω2t+A3cosω3t….B
And the given signal is
X(t)= 6 cos 50πt +20 sin 300πt – 10 cos 100πt
on comparing give signal B with standard
form of a signal A
 • We obtain the frequencies for the given
signal as

Thus, the highest frequency component of the given message signal will be

fmax = 150 Hz
Therefore, Nyquist rate=2 fmax = 2 X 150= 300Hz
Example-2
• Find the Nyquist rate and the Nyquist
interval for the signal
• X(t) 1/2π cos(4000 πt) cos(1000 πt)
Solution: Given signal is

Or

Or

{2cos A cos B = cos(A+B) +cos (A-B)}

Example-2

Let the two frequencies present in the signal be ω1 and

ω2 so that the new equation for the signal will be

……..2
Comparing equation 1 and 2, we have
ω1 t = 5000π t or
2πf1 t = 5000π t or 2f1 = 5000
f1 = 2500Hz
 Example-2
Similarly, for second factor
ω2 t = 3000π t or
2πf2 t = 3000π t or 2f2 = 3000
f2 = 1500Hz
Therefore, the maximum frequency present
in x(t) is f1 = 2500Hz
Nyquist rate given as
fs = 2fm or 2 X 2500 = 5kHz ……….Ans
Nyquist interval given as
Ts = 1/2fm or 1/2 X 2500 = 0.2 m sec……….Ans
 Example
1. In telephone system, the bandwidth of the voice signal
(message signal) is limited to W=3.1 kHz and the universal
sampling frequency is fs = 8kHz > 2W.
2. Determine the Nyquist rates and Nyquist intervals used to
sample the following signals:
x(t)=sinc(200t), y(t)=sinc2(200t).
To determine the sampling rate of a signal, we have to know
its frequency
spectrum. For x(t) and y(t), this may be done using Fourier
transform.
1  f 
X(f )  rect  ,
200  200 
 1  f 
 
 1  , f  200
Y ( f )   200  200 
0, f  200

©2000, John Wiley & Sons, Inc.
03/15/24 51
Haykin/Communication Systems, 4th Ed
 Example (Continued)
From X(f) and Y(f), it is clear that for x(t) the maximum frequency (bandwidth) W=100Hz and
for y(t) the maximum frequency (bandwidth) W=200Hz

X(f) Y(f)

1/200 1/200

f(Hz) f(Hz)
-100 0 100 -200 0 200

For x(t), fs=2W=200Hz and Ts=1/fs=5ms.

For y(t), fs=2W=400Hz and Ts=1/fs=2.5ms.

03/15/24 52
RECONSTRUCTION
FILTER(LPF)
Reconstruction Filter(LPF)
• The LPF is used to recover original signal
from its samples. This is also known as
interpolation filter

• A low-pass filter is a filter that passes low-

frequency up to a specified cut-off frequency
and rejects all other frequency above cut-off
frequency
• What are "up sampling" and
"interpolation"?
• "Up sampling" is the process of inserting
zero-valued samples between original
samples to increase the sampling rate. (This is
called "zero-stuffing".) Up sampling adds to
the original signal undesired spectral images
which are centered on multiples of the
original sampling rate.
Why interpolate?
• The primary reason to interpolate is simply to
increase the sampling rate at the output of
one system so that another system operating
at a higher sampling rate can input the signal.
 example
• suppose we have a table like this, which gives
some values of an unknown function f.
• Plot of the data points as given in the table.
• Interpolation provides a means of estimating
the function at intermediate points, such
as x = 2.5.
• There are many different interpolation
methods, How accurate is the method? How
expensive is it? How smooth is the
interpolant? How many data points are
needed?
x F(x)
0 0
1 0.8415
2 0.9093
3 0.1411
4 0.7568
5 0.9589
6 0.2794
Piecewise constant interpolation
• Piecewise constant
interpolation, or nearest-
neighbor interpolation.
• The simplest
interpolation method is
to locate the nearest
data value, and assign
the same value.
• In simple problems, this
method is unlikely to be
used, as linear
interpolation
Linear interpolation
• One of the simplest
methods is linear
interpolation
(sometimes known as
lerp).
• Consider the above
example of
estimating f(2.5). Since
2.5 is midway between 2
and 3, it is reasonable to
take f(2.5) midway
between f(2) = 0.9093
and f(3) = 0.1411, which
yields 0.5252.
 Linear interpolation
• Generally, linear
interpolation takes
two data points, say
(xa,ya) and (xb,yb),
and the interpolant
is given by:
At each point, the derivative of is the slope of
a line that is tangent to the curve. The line is
always tangent to the blue curve; its slope is
at the point (x,y) the derivative. Note derivative is positive where
green,negative where red, and zero where
Linear interpolation is quick and easy, but it is
black.
not very precise. Another disadvantage is that
the interpolant is not differentiable at the
pointxk.
 Polynomial interpolation

Polynomial interpolation is a
generalization of linear
interpolation.
Note that the linear
interpolant is a linear
function.
We now replace this
interpolant by a polynomial of
higher degree.
 Frequency response of Ideal LPF
From fig 3.3, it may be observed that in case of LPF,there is
sharp-change in response at cut-off frequency, that is
amplitude response becomes suddenly zero at cut-off frequency
which is not possible practically.
 this means that an ideal LPF is not physically realizable.
 In place of ideal LPF,we use practical filter.
 Frequency response of Practical LPF
 fig 3.4 shows the frequency response of practical LPF.
 From fig 3.4, it may be observed that in case of
practical filter.
 The amplitude response decreases slowly to become
zero.
 This means that there is a transition band in case of
practical filter.
Fig-3.5 shows the use of practical LPF in
reconstruction of original signal from its
sample
RECONSTRUCTION
USING
INTERPOLATION
FILTERS
Reconstruction using interpolation
filters

• The
Reconstruction using interpolation
filters
Reconstruction using interpolation
filters
Reconstruction using interpolation
filters
EFFECT OF UNDER
SAMPLING:
ALIASING
Aliasing Effect
 The aliasing phenomenon in which a high a high frequency
component in the frequency spectrum of the signal takes identity
of a lower frequency component in the spectrum of the sampled
signal
From fig.3.7- it is obvious that because of the overlap due to
aliasing phenomenon,
 it is not possible to recover original signal x(t) from sampled signal
g(t) by LP Filtering
Since the spectral components in the overlap regions add and
hence the signal is distorted
 Aliasing Effect
 In practice a message signal is not strictly band-limited.
 This means that the sampling rate fs=2W may result in an overlapping
between the different periods of G(f) as shown by Figure 3.7
 This overlapping is called aliasing effect.
 Aliasing effect can be eliminated by using an anti-aliasing filter prior to
sampling and using a sampling rate slightly higher than Nyquist rate
(fs=2W). This is shown in Figure

g(t) Anti-aliasing Sampler g(kTs)

Filter

74
Figure 3.3
(a) Spectrum of a signal. (b) Spectrum of an undersampled
version of the signal exhibiting the aliasing phenomenon.

03/15/24 75
Figure 3.4
(a) Anti-alias filtered
spectrum of an
information-bearing
signal. (b) Spectrum of
instantaneously
sampled version of the
signal, assuming the use
of a sampling
rate greater than the
Nyquist rate.
(c) Magnitude response
of reconstruction filter.

03/15/24 76
SAMPLING
TECHNIQUES
Types of Sampling Techniques
Sampling Techniques

Ideal Practical
Sampling sampling
Natural Flat-top
Instantaneous sampling
sampling
sampling
IDEAL
SAMPLING
Ideal or Instantaneous or
Impulse Sampling
Ideal or Instantaneous or
Impulse Sampling
Ideal or Instantaneous or
Impulse Sampling
Instantaneous sampling
spectrum
 Disadvantage of
Ideal sampling
1 Due to very narrow samples, the
transmitted power is very small &
the SNR is low. Thus the ideally
sampled pulses may get lost in the
background noise.

2 Ideal sampling is not possible to

achieve practically, because it is
practically impossible to have
pulses of width approaching
zero.
3 Therefore practically natural or
flat-top sampling is used. ideal
sampling was used only to
prove the sampling theorem
PRACTICAL
SAMPLING
Practical Aspects of sampling & signal
recovery
There are two popularly used practical
sampling techniques
Practical Sampling
Techniques
Sam ural
/Ch pling
/or pper
sam ary
Nat

of fi ples
din
o

dur nite
n
atio
NATURAL SAMPLING
/CHOPPER/ORDINARY
SAMPLES OF FINITE DURATION
Natural Sampling /Chopper/ordinary
samples of finite duration
• In instantaneous sampling results in the
samples whose width approaches zero.
• Due to this, the power content in the
instantaneous sampling pulse is negligible.
• Thus this method is not suitable for
transmission purpose
• Natural sampling is a practical method and
will be discussed in this section.
Natural sampling
Natural sampling
Natural sampling
Natural sampling
Natural sampling
Fig-3.14 105

Fig-3.1(f) 22
Fig-3.14 illustrates some arbitrary
spectrum of x(t) & corresponding
spectrum of F(f)

102
Merits and demerits of natural
sampling
• Generations is easy.
• We can use practical low pass filter for
reconstruction.
• The amplitudes of high frequency components
decrease therefore some distortion is
introduced.
• Increase SNR due to finite pulse width of the
sampling function
• for large values of “ ” there is a possibility of
crosstalk.
FLAT-TOP/
RECTANGULAR PULSE
SAMPLING
Flat-Top/Rectangular pulse Sampling
• Flat top sampling like natural sampling is also a
practically possible sampling method.
• But natural sampling is little complex where as
it is quite easy to get flat top samples.
• In flat top sampling or rectangular pulse
sampling, the top of the samples remains
constant & is equal to the instantaneous value of
the baseband signal x(t) at the start of sampling.
• The duration or width of each sample is &
sampling rate is equal to fs = 1/Ts
Flat Top sampling
Flat Top sampling w/f
Flat Top sampling spectrum
Merits and demerits of Flat Top
sampling
• Better SNR due to increased signal power. This is
due to the finite width “ ” of the pulses.
• Generation is easy
• Practical filters can be used for reconstruction
• Aperture effect introduces distortion.
APERTURE EFFECT
QUESTIONS?