Homework 2: DTFT and Discrete-Time Systems
Homework 2: DTFT and Discrete-Time Systems
Homework 2: DTFT and Discrete-Time Systems
Instructions: Both problems from Set-A has to be submitted by 5:00pm on 21/08/14 (Thursday). Please
follow the given instructions carefully.
If solutions from two or more students resemble or are copied from each other, all the concerned students
will get −5 marks. You must write your own solution in your own words.
Each question should be submitted on a different sheet. Write your name+roll no. on each page of your
submitted solution. Make a photocopy of your homework submission. Submit your Homework + photocopy
(homework submission and its copy should not be stapled together) in the box kept in EE-Office for EE603
Homework. The box will become available on Wednesday morning. If you have queries, then meet the instructor
or the TA during office hours.
Set-A
1. Consider a linear finite difference equation system with x[n] as input and y[n] as output. Their relationship
is given by
1
y[n] = {x[n] + 2x[n − 1] + 3x[n − 2] + 2x[n − 3] + x[n − 4]} .
9
Assume zero initial conditions. Sketch and label the plots of magnitude and phase responses correspond-
ing to the LTI system’s impulse response.
2. Let h[n] be the impulse response of the system y[n] = x[n] − y[n − 1].
(a) Find the impulse response h[n] of this system. Is it BIBO stable? Is it causal?
(b) Let the series cascade of M systems with impulse response h[n] as in Problem 2a be hM [n]. What
is the minimum value of M for which |hM [n]| → ∞, as n → ∞?
(c) The system is modified to
to improve stability. Let g[n] be the impulse response for this filter. As a designer, what values of
α will you choose to make the system BIBO stable?
Set-B
1. A causal LTI system’s transfer function is given by
1
Hd (ejω ) =
(2 + e−jω )(3 + e−jω )2
2. Consider the impulse response of a discrete-time ideal lowpass filter with cutoff frequency ωc = (π/2).
The N -point truncated version is defined as
The following MATLAB code implements the truncated version hN [n] for N = 10, 100, 1000.
1
clear all;
for i = 1:3
figure;
N = 10^i;
n = 1:1:2*N+1;
h = sin(pi.*(n-N-1)/2)./(pi.*(n-N-1));
h(N+1) = 1/2;
[H,w] = freqz(h, 1, 2048);
plot(w/pi, abs(H));
energy(i) = sum(h.*h);
end
figure;
plot(1:i, energy);
Let Hd,N (ejω ) be the corresponding frequency response of the truncated filter.
Rπ
(a) Show that −π |Hd,N (ejω ) − Hd (ejω )|2 dω decreases to zero as N increases to infinity. Do you observe
this by running the MATLAB code?
(b) Observe the output of your MATLAB code. Is it true that for N = 100, the truncated frequency
response HN,d (ejω ) converges (becomes closer than 5%) to Hd (ejω ) at every |ω| ≤ π? Will this hold
true if N is increased to 1000?
(c) Prove that if a signal g[n] is absolutely summable then its frequency response G(ejω ) is continuous
for −π ≤ ω ≤ π.
3. Consider the frequency response Hd (ejω ) of a discrete-time LTI system. Let h[n] be the corresponding
impulse response. Assume that h[n] satisfies the following five properties:
(a) Show that properties (i)–(iii) imply that h[n] is non-zero for only a finite duration.
(b) Find all possible discrete signals h[n] that satisfy properties (i)–(v).
1 − 2
Xd, (ejω ) = .
1 − 2 cos(ω0 − ω) + 2
Show that lim↑1 Xd, (ejω ) = 2πδ(ω − ω0 ). The frequencies ω and ω0 satisfy −π < ω, ω0 < π.
↓
5. Consider the signal x[n] = {−3, 1, −2, 0, 2, −1, 3} with DTFT Xd (ejω ). The arrow denotes the n = 0
location. The signal is zero for |n| ≥ 4. Compute the following quantities, without explicitly computing
Xd (ejω ):
(a) X ej0
(d) X ejπ
Rπ
(e) −π |X(ejω )|2 dω
2
6. A discrete-time LTI system has the impulse response
sin(πn/2)
h[n] = .
πn
Examine if this discrete-time system is BIBO stable or not.