Module 3 : Sampling and Reconstruction

Problem Set 3

Problem 1

Shown in figure below is a system in which the sampling signal is an impulse train with alternating sign.

                      
The sampling signal p(t), the Fourier Transform of the input signal x(t) and the frequency response of the filter are shown below:

(a) For , sketch the Fourier transform of x p (t) and y(t) .

(b) For , determine a system that will recover x(t) from x p (t) and another that will recover x(t) from y(t).

 
(c) What is the maximum value of in relation to wM for which x(t) can be recovered from either x p ( t) or y(t)?

 
Solution 1

(a) As x p(t) = x(t) p(t)

So we first find the Fourier Transform of p(t) as follows :-

The Fourier Transform of a periodic function is an impulse train at intervals of .

Strength of impulse at being:

Thus we can also sketch and hence :

(b) To recover x(t) from x p (t):

Modulate x p (t) with .                                                                                                                    

has a spectrum with impluses of equal strength at . Thus the new signal will have copies of the original spectrum

(modulated by a constant of-course) at all even multiples of . Now an appropriate Low-pass filter can extract the original spectrum!

To recover x(t) from y(t):

Here too, notice from the figures that modulation with will do the job. Here too, the modulated signal will have copies of the

original spectrum at all even multiples of .

 
(c) So long as adjacent copies of the original spectrum do not overlap in , theoretically one can reconstruct the original signal.
Therefore the condition is:

 
Problem 2

The signal y(t) is obtained by convolving signals x 1 (t) ans x 2 (t) where:

Impulse train sampling is performed on y(t) to get .

Specify the range of values of T so that y(t) may be recovered from y p (t).

 
Solution 2

By the Convolution Theorem,

Thus from the Sampling Theorem, the sampling rate must exceed .

Thus T must be less than 10 -3 , i.e: 1millisecond.

In the figure below, we have a sampler, followed by an ideal low pass filter, for reconstruction of from its samples . From the

sampling theorem, we know that if is greater than twice the highest frequency present in and , then the

reconstructed signal will exactly equal . If this condition on the bandwidth of is violated, then will not equal . We

seek to show in this problem that if , then for any choice of T, and will always be equal at the sampling instants;

that is,

To obtain this result, consider the following equation which expresses in terms of the samples of :

With , this becomes

By considering the value of for which , show that, without any restrictions on , for any integer value

of k.

 
Solution 3

In order to show that and are equal at the sampling instants, consider
Thus,

Assuming the continuity of at ,

 
Problem 4

This problem deals with one procedure of bandpass sampling and reconstruction. This procedure, used when is real, consists of

multiplying by a complex exponential and then sampling the product. The sampling system is shown in fig. (a) below. With

real and with nonzero only for , the frequency is chosen to be , and the lowpass filter

has cutoff frequency .

(a) For shown in fig. (b) , sketch .

(b) Determine the maximum sampling period T such that is recoverable from .

(c) Determine a system to recover from .

  Fig. (b)

 
Solution 4

(a) Multiplication by the complex exponential in the time domain is equivalent to shifting left the Fourier transform by an
amount ' ' in the frequency domain. Therefore, the resultant transform looks as shown below:

After passing through the filter, the Fourier transform becomes:

enough to avoid overlapping between the copies):

(b) is recoverable from only if the copies of the Fourier Transform obtained by sampling do not overlap with each other. For

this to happen, the condition set down by the Shannon-Nyquist Sampling Theorem for a band-limited signal has to be satisfied, i.e. the
sampling frequency should be greater than twice the bandwidth of the original signal. Mathematically,

Hence, the maximum sampling period for to be recoverable from is .

 
(c) The system to recover from is outlined below :

 
Problem 5

The procedure for interpolation or upsampling by an integer factor N can be thought of as the cascade of two operations. The first
operation, involving system A, corresponds to inserting N-1 zero-sequence values between each sequence value of , so that

For exact band-limited interpolation, is an ideal lowpass filter.

(a) Determine whether or not system A is linear.

(b) Determine whether or not system A is time variant.

(c) For as sketched in the figure and with N = 3, sketch .

(d) For N = 3, as in the figure, and appropriately chosen for exact band-limited interpolation, sketch

 
Solution 5

(a) Let and be two inputs with corresponding outputs and respectively. Now, suppose the new input to this

system is . Then, the corresponding output is given by, if and 0

otherwise. From this it is clear that and hence the system is linear.

(b) The system A is not time invariant. We can see this from the following example. Let , for n = 0, 1 and otherwise. Also
assume that N=2. Thus, the output corresponding to x [ n ] will be if n = 0, 2 and otherwise. Now, let
if n = 1, 2 and 0 otherwise. The corresponding output is if n = 2, 4 and y 1[ n ] = 0 otherwise. Thus, and
hence this shows that the system is not time invariant.

(c) Refer to Fig. 1 and Fig. 2

Figure 1: Spectrum given in the problem.

Figure 2: Spectrum of the interpolated signal.

 
Problem 6

Shown in the figures is a system in which the sampling signal is an impulse train with alternating sign. The Fourier Transform of the input
signal is as indicated in the figures:

(i) For , sketch the Fourier transform of and .

(ii)  For , determine a system that will recover from .

(iii)  For , determine a system that will recover from .

(iv)  What is the maximum value of in relation to for which can be recovered from either or ?

 
Fig (a)

 
  
Fig (b)

Fig (c)

 
Fig (d)

 
Solution 6

(a) As , by dual of convolution theorem we have . So, we first find the Fourier

Transform of as follows:

The Fourier Transform of a periodic function is an impulse train at intervals of , each impulse being of magnitude:
Here w e see that the impulses on the axis vanish at even values of k .

Hence, the Fourier Transform of is as shown in figure (a). I n the frequency domain, the output signal Y can be found by

multiplying the input with the frequency response . Hence is as shown below in the figure (b).

 
Figure (a)

 
Figure (b)

(b) To recover from , we do the following two things:

1)     Modulate the signal by .

(c) To recover from , we do the following two things:

1)     Modulate the signal by .

2)     Apply a low pass filter of bandwidth .

(d) Maximum value for recoverability is as can be seen from the graphs.

 
Problem 7

A signal with Fourier transform undergoes impulse –train sampling to generate

where .

For each of the following set of constraints on and/or , does the sampling theorem guarantee that can be recovered
exactly from ?

(a)

(b)

(c)

(d) real and

(e) real and

(f)

(g)

 
Solution 7

From Sampling Theorem we know that if be a band-limited signal with

, then is uniquely determined by its samples

if

where

Now,
 

(a)

Here, obviously, .

Hence can be recovered exactly from .

(b)

Here, obviously ,

Hence can be recovered exactly from .

(c)

Real part of , but we can't say anything particular about imaginary part of the , thus not necessary that
for this particular range.

Hence cannot be recovered exactly from .

(d) real and

As is real we have

Taking mod on both sides

So, we get

Here , obviously ,

Hence x(t) can be recovered exactly from .

(e) real and

Proceeding as above we get

Here, obviously ,

Hence x(t) cannot be recovered exactly from .

(f)

When we convolve two functions with domain to and to then the domain of their convolution function varies

from to .

Here, and

Therefore,

Here, obviously ,

Hence x(t) can be recovered exactly from .

(g)

We cannot say anything about ,

Hence x(t) cannot be recovered exactly from .

 
Problem 8

Shown in the figures below is a system in which the input signal is multiplied by a periodic square wave. The period of is . The
input signal is band limited with

(a)   For , determine, in terms of , the maximum value of for which there is no aliasing among the replicas of in
.

 
(b) For , determine, in terms of , the maximum value of for which there is no aliasing among the replicas of in
.

Figure (a)

 
 Figure (b)

 
Solution 8

(a) We know that

is a periodic square wave of period .

 
With as shown in the figure.

Figure (a)

 
We calculate as follows: (FT of a periodic signal)

where,

Considering any one period , (say from 0 to ) and

Substituting 
which can never be 0.

Thus, is an impulse train situated at intervals of .

And has a maximum value of .

(Maximum value of T without aliasing).

(b) We know that

is a periodic square wave of period .

With as shown in the figure.

Figure (b)

We calculate as follows: (FT of a periodic signal)

where

Considering any one period , (say from 0 to T) and

Substituting 

for (i.e. k is even )

Thus, is an impulse train situated at intervals of .

And has a maximum value of .

(Maximum value of T without aliasing).

Figure I shows the overall system for filtering a continuous-time signal using a discrete time filter. If X c (jw) and H(exp(jw)) are as
shown in Figure II , with 1/T=20kHz, sketch X p (jw), X(exp(jw)), Y(exp(jw)), Y p (jw) and Y c (jw).

Figure ( I )

Figure ( II )

 
Solution 9 :

 
Problem 10 :

 Shown in figure below is a system in which the input signal is multiplied by a periodic square wave. The period of s (t) is T. The input
signal is band limited with | X(jw) |=0 for |w|>w m .

(a) For determine, in terms of wm , the maximum value of T for which there is no aliasing among the  replicas of X (jw) in
W(jw).

(b) For determine, in terms of wm , the maximum value of T for which there is no aliasing among the  replicas of X ( jw ) in W(
jw ).