0% found this document useful (0 votes)
119 views12 pages

Sampling of Signals: Sampling Is The Conversion of A Continuous Signal To A Discrete Signal, by Taking

1. Sampling converts a continuous signal to a discrete signal by taking samples at regular time intervals. The sampling frequency is the number of samples per second. 2. The Nyquist-Shannon sampling theorem states that a bandlimited signal with bandwidth B can be uniquely determined by its samples if the sampling frequency is greater than 2B. 3. For a signal to be reconstructed from its samples using ideal lowpass filtering, the sampling frequency must be at least twice the maximum frequency of the signal's spectrum. If the sampling frequency is lower, aliasing will occur.

Uploaded by

Ralu Damian
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
119 views12 pages

Sampling of Signals: Sampling Is The Conversion of A Continuous Signal To A Discrete Signal, by Taking

1. Sampling converts a continuous signal to a discrete signal by taking samples at regular time intervals. The sampling frequency is the number of samples per second. 2. The Nyquist-Shannon sampling theorem states that a bandlimited signal with bandwidth B can be uniquely determined by its samples if the sampling frequency is greater than 2B. 3. For a signal to be reconstructed from its samples using ideal lowpass filtering, the sampling frequency must be at least twice the maximum frequency of the signal's spectrum. If the sampling frequency is lower, aliasing will occur.

Uploaded by

Ralu Damian
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 12

3.

Sampling of signals
Sampling is the conversion of a continuous signal to a discrete signal, by taking
samples from a signal at given moments of times. Sampling is performed by measuring
the value of the continuous signal every Ts seconds, with a sampling frequency of fs=1/Ts.
The sampling frequency the number of samples obtained in one second and it is measured
in hertz or in samples per second. X()s
The ideal sampled signal is:

x ( t ) =

x ( kT ) ( t kT ) .
s

k =

Its spectrum is
1
X ( ) =
Ts

X ( k )

k =

The NyquistShannon Sampling theorem: A finite energy and band limited signal x(t),
with bandwidth B = M is uniquely determined by its samples x ( nTs ) n ] if

s 2M .
where

2
.
T
If we sample the signal with the sampling period Ts we obtain a sampled
signal x ( t ) with the spectrum X ( ) . The original signal is reconstructed from the
s =

spectrum of the sampled signal with lowpass filtering, with the cutoff frequency C , such
that,
M C s M 2M s or
fs 2 fM .

The limit case: The sampling frequency is equal with the bandwidth
s = 2M
and the cutoff frequency is the maximum frequency of the signals spectrum,
C = M
Unfavorable case: The sampling frequency is lower than the bandwidth,
s < 2M
when aliasing appears, the original signal cannot be reconstructed.
The spectrum of the ideal lowpass filter is:
H r ( ) = Ts pC ( ) .
Its impulse response is
hr ( t ) = Ts

sin ( C t )
t

The reconstructed signal is:


xr ( t ) = x ( t ) hr ( t ) =
At the limit we have: C = M =

s
2

k =

2C

x ( kTs )

sin (C ( t kTs ) )

C ( t kTs )

and the reconstructed signal is:

sin k

Ts

xr ( t ) = x ( kTs )
t

k =
k
Ts

Problems
1. Consider the system below

x1(t)

( )2

y(t)

x2(t)
x3(t)
The support of the Fourier transform for the signals xi(t), with i=1,2,3, are [-i, i].
a) Find the maximum sampling period T for which the signal y(t) can be reconstructed
from its samples using ideal low pass filtering.
3

b) Find T for 1 = , 2 = and 3 =


4
4
2
Solution.
a) The output signal and its spectrum are:

y ( t ) = x12 ( t ) x2 ( t ) + x3 ( t ) Y ( ) =

4 2
The support of the Fourier transform X 1 ( ) is:

X 1 ( ) X 1 ( ) X 2 ( ) + X 3 ( )

supp { X 1 ( )} = [ 1 , 1 ]

The support of the convolution of the two spectra is double:


supp { X 1 ( ) X 1 ( )} = [ 21 , 21 ]

The support of the convolution X 1 ( ) X 1 ( ) X 2 ( ) is:

supp { X 1 ( ) X 1 ( ) X 2 ( )} = [ 21 2 , 21 + 2 ]

This means that the support of the spectrum Y ( ) is

supp {Y ( )} = y , y , y = max ( 21 + 2 , 3 )

Applying the sampling theorem we obtain:

s 2 y ,

The sampling period

s y
Therefore, the maximum value for the sampling period is:

Ts =
max ( 21 + 2 , 3 )
Ts =

b)


supp { X 1 ( ) X 1 ( )} = [ 21 , 21 ] = ,
2 2
supp { X 1 ( ) X 1 ( ) X 2 ( )} = [ 21 2 , 21 + 2 ] = [ , ]
3
max ( 21 + 2 , 3 ) = max , =
4
supp {Y ( )} = [ , ]
2. Consider a system with the input signal x(t) and the output signal y(t) with the relation
output-input a polynomial:
N

y = P ( x) = ak x k .
k =0

If the signal x(t) is band limited to the interval [-M, M], find the maximum sampling
period for which the signal y(t) can be recovered from its samples using ideal lowpass
filtering.
Solution.
The Fourier transform of the signal y(t) is:
a
a
Y ( ) = a0 + a1 X ( ) + 2 X ( ) X ( ) + ... + N N X ( ) X ( ) ... X ( )
2
( 2 )  
N factors

supp (Y ( ) ) = [ N M , N M ]

Sampling theorem: s 2 N M =>

2 N M => T
.
T
N M

3. Consider the band limited signal z(t), in the interval [1, 2], with the spectrum
represented below:

Z()
1

0
Denote C =

2 + 1

; M =

2 1

2
2
a) Show the relation z (t ) = x1 ( t ) e jC t holds, where x1(t) is a bandlimited signal, in

the interval [-M, M].


b) Using the reconstruction theorem for the signal x1(t), find the maximum value of
T for which the relation is true:
t

sin n

j t nT
T
z ( t ) = z ( nT ) e s ( )
t

n =
n
T

Solution.
a) Assume there is a signal x1 ( t ) , a bandlimited signal, in the interval [-M, M], for

which we have

z (t ) = x1 ( t ) e jC t Z ( ) = X 1 ( C )

If the spectrum X 1 ( ) is shifted to the right with C we obtain the spectrum Z ( ) .

This implies that the support of Z ( ) is the same as the support of the shifted version of
X 1 ( ) :

supp {Z ( )} = supp { X 1 ( C )} = [1 , 2 ]

The support of Z ( ) is known:


We have supposed that:

supp {Z ( )} = [1 , 2 ]

supp { X 1 ( C )} = [ M + C , M + C ]

This is true if and only if


M + C = 1 and M + C = 2 (true)
b) The signal x1 (t ) = z ( t ) e jC t has the spectrum limited to [-M, M] =>

Sampling with the sampling frequency s 2M


=> The recovered signal
t

sin k

Ts

if C = M = s
x1 ( t ) = x1 ( kTs )
2
t

k =
k
Ts

=> z ( t ) = x1 ( t ) exp ( jC t ) =

=> z ( t ) =

exp ( j t ) x ( kT ) sinc T
C

k =

t
exp ( j ( t kT ) ) z ( kT ) sinc T

k =

k for C = M = s , Ts =

2
M

4. The signal z(t) with the spectrum sketched below is the analytical signal corresponding
to a real signal, x(t).

Z()
1

a) Sketch the spectrum of the real signal x ( t ) .

b) Use the relation between the signals x ( t ) , x ( t ) and z ( t ) , where x ( t ) is the Hilbert
transform of the signal x ( t ) , to find the maximum value of the sampling step T for which
we have:
x (t ) =

n =

x ( t ) =

n =

sin n

T
x ( nT ) cos C ( t nT ) x ( nT ) sin C ( t nT )
t

n
T

sin n

T
x ( nT ) cos C ( t nT ) + x ( nT ) sin C ( t nT )
t

n
T

c) Using the relations from b), systems that recover the original signals x ( t ) and x ( t )
from the sampled signals, xs ( t ) = x ( t ) T ( t ) and xs ( t ) = x ( t ) T ( t ) , respectively, are
represented below. Find the impulse response h1(t) and h2(t) and the frequency responses
H1() and H2() of these systems.
xs(t)

xs(t)

h1(t)

h1(t)

x(t)
xs ( t )

xs ( t )

h2(t)

x ( t )

h2(t)

Solution.
a) x(t) = Re(z(t)), x ( t ) = H ( x ( t ) ) =

1
x ( t ) X ( ) = j sgn X ( )
t

z ( t ) = x ( t ) + jx ( t )
Z ( ) = X ( ) + jX ( ) = X ( ) + j ( j sgn ) X ( ) = X ( ) + sgn X ( )
2 X ( ) ,
Z ( ) =
0,

>0
otherwise

=> < 0 : X ( ) =

Z ( )
2

, > 0 : X ( ) =

Z ( )
2

X()
1/2

-2

-1

b) We have to prove the relation:


x (t ) =

n =

x ( nT ) h1 ( t nT )

h1 ( t ) = cos C t sinc

h2 ( t ) = sin C t sinc

sinc 0t

x ( nT )h ( t nT )
2

n =

t H1 ( )
t H 2 ( )

p ( ) => for 0 = , sinc t Tp ( )


T
0
T
T
0

1
Tp ( ) ( C ) + ( + C )
T
2
1
H1 ( ) = T p ( C ) + p ( + C )

T
2 T
H1 ( ) =

Tp ( ) ( C ) ( + C )
T
2
j
1
H 2 ( ) = T p ( C ) p ( + C )
T

2 j T
H 2 ( ) =

H1()
T/2

-C-/T

-C -C+/T

C-/T C

C+/T

jH2()
T/2

-C
0

=> X ( ) H1 ( ) =

C-/T C

C+/T

T
T
X ( ) and jX ( ) jH 2 ( ) = sgn X ( ) jH 2 ( ) = X ( )
2
2

T
=> X ( ) H 2 ( ) = X ( )
2
X ( ) H1 ( ) X ( ) H 2 ( ) = TX ( )
After sampling of the signals x and x we obtain:
xs ( t ) = x ( nT ) ( t nT ) X s ( ) =

1
2

X k

T k =
T

1
2

xs ( t ) = x ( nT ) ( t nT ) X s ( ) = X k

T k =
T
X s ( ) = j sgn X s ( )

For the reconstruction of the signal x(t) from xs(t):

< 1 < 2 < C +

T
1
T 1
c) X s ( ) H1 ( ) = X ( ) = X ( )
T
2 2

1
X s ( ) H 2 ( ) = X ( )
2
Therefore, X s ( ) H1 ( ) X s ( ) H 2 ( ) = X ( )
In the time domain:

=>

>

2 1
2

x (t ) =

x ( nT ) cos ( t nT ) x ( nT ) sin ( t nT ) sinc T ( t nT )

k =

For the second relation:


j sgn X s ( ) H1 ( ) =
j sgn X s ( ) H 2 ( ) =

1
1
j sgn X ( ) => X s ( ) H1 ( ) = X ( )
2
2

1
1
j sgn X ( ) => X s ( ) H 2 ( ) = X ( )
2
2

Therefore, X s ( ) H 2 ( ) + X s ( ) H1 ( ) = X ( )
In the time domain:
x ( t ) =

x ( nT ) cos ( t nT ) + x ( nT ) sin ( t nT ) sinc T ( t nT )

k =

5. Consider the signal x(t) with the spectrum sketched below:

X()
1

-2

-1

We use the same relations from problem 3. The signal is ideally sampled with the
sampling frequency 2M .
a) Sketch the spectrum of the sampled signal if (4k+3)M = C
b) Sketch the spectrum of the sampled signal if (4k+1)M=C
c) Implement a system that permits the recovery of the original signal x(t) from the
sampled signal in case a) and in case b).
Solution.
a) The sampling pulsation is: s = 2 2M = 4M
The spectrum of the sampled signal:
1
2

X s ( ) = X k

T k =
T

If (4k+3)M = C k=1, C = 7M, 1 = 6M, 2 = 8M

Xs()
1/T

-2

-1

b) (4k+1)M = C k=1, C = 5M, 1 = 4M, 2 = 6M

Xs()
1/T

-2 -1

c) Reconstruction is made using bandpass filters

Hr(), case a)
2M/

-8M -6M -4M

4M

6M

8M

Hr(), case b)
2M/

-8M -6M -4M

4M

6M

8M

6. Consider the system represented below. x(t) is a bandlimited signal at the interval
[-M, M], and p(t) is a periodic signal of period T = M .

x(t)

xs(t)

a) The spectrum of the signal x(t) is X() and an are the


coefficients of the Fourier series for the signal p(t). Show the
spectrum of the sampled signal is:

p(t)
10

X s ( ) =

a X ( n )

n =

How much is s?
b) Assuming the continuous component of x(t) is not null a0 0 , show that x(t) can be
recovered from xs(t) using ideal lowpass filtering. Find the parameters for the lowpass
filter.
c) Implement a system for the reconstruction of the original signal x(t) from the sampled
signal xe(t) for a0=0.
d) For the spectrum of x(t) given below, sketch the magnitude spectrum for the signal
xs(t) in two cases: p(t)=T(t) and p(t)=T(t-).
X()
1

-M

Solution
a) p ( t ) =

P ( ) =

n =

an e

jn

2
t
T

2 a n T

n =

1
X ( ) P ( )
2
2

X s ( ) = an X n
,
T

n
X s ( ) =

2
= 2M
T
2

b) a0 0 => X s ( ) = a0 X ( ) + an X n

n0

s =

H()
1/a0

-M

M
0

11

c) a0 = 0, a1 0 => X s ( ) = a1 X ( ) +

a X n T

n 0,1

H1()
1/a1
xs(t)

x(t)

h1(t)

2/T-M
=M

2M

2/T+M
=3M

exp(-j2t/T)

2
2 ( )
T T
1
1
2

X s ( ) =
X ( ) P ( ) = X k

2
T
T

d) p(t)=T(t); P ( ) =

X()
1/T

0
p(t)=T(t-); P ( ) = e j
X s ( ) =
X s ( ) =

2
2 ( )
T T

1
1
2
X ( ) P ( ) =
X ( ) e j
2
2
T
1
2

X ( ) k
T
T

X s ( ) =

2M

-M

2
jk

k
k

2
T

1
2

X k

T k
T

1
2

X k

T k
T

12

T
2

jk T
e

You might also like