Scribd
Upload
2 views

Homework_06_ECE437_Fall2024

The document outlines a homework assignment consisting of five problems related to Discrete Time Fourier Transform (DTFT) and filter design. It includes tasks such as deriving DTFT expressions, proving properties of even signals, sketching magnitude and phase responses, and designing low-pass filters. Additionally, it involves calculating DFTs, convolutions, and analyzing the effects of zero-padding on signals.

Uploaded by

nerdygamer69900
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
2 views

Homework_06_ECE437_Fall2024

The document outlines a homework assignment consisting of five problems related to Discrete Time Fourier Transform (DTFT) and filter design. It includes tasks such as deriving DTFT expressions, proving properties of even signals, sketching magnitude and phase responses, and designing low-pass filters. Additionally, it involves calculating DFTs, convolutions, and analyzing the effects of zero-padding on signals.

Uploaded by

nerdygamer69900
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

HOMEWORK #6 (Submit solutions through blackboard in PDF format)

Problem 1

a) For a signal h(n)= { -3, 1, 1, 1, -3 }

Find the expression for the DTFT of the signal, i.e. derive H(ω). What is the magnitude and what is the
phase of H(ω)? Hint: Simplify the expression before deriving the magnitude/phase.

b) Prove that the DTFT of an arbitrary even signal has a phase that is either 0 or π (i.e. that the
DTFT of such a signal is real). Hint: write DTFT as a sum of pairs of symmetric terms – then
simplify

Problem 2

Given is one period of DTFT function 𝑋(𝜔) and 𝑌(𝜔) for |𝜔| ≤ 𝜋
𝜋
1, |𝜔| ≤
𝑋(𝜔) = 16
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 − 𝜋, 𝜋

𝜋 𝜋 𝜋 𝜋
1, − ≤ |𝜔| ≤ +
𝑌(𝜔) = 2 16 2 16
0, 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 − 𝜋, 𝜋

a) Sketch the magnitude responses of 𝑋(𝜔) and 𝑌(𝜔) and the phase responses.
b) Find the x(n) that corresponds to 𝑋(𝜔). Is x(n) periodic? Is x(n) causal?
c) Find the y(n) that corresponds to 𝑌(𝜔). Is y(n) periodic? Is y(n) causal?
d) What is the convolution of x(n)*y(n) ?
e) Sketch the magnitude frequency response of a signal 𝑣(𝑛) = cos 𝑥(𝑛)

Hint: Try using the modulation property of the DTFT for evaluating y(n) and 𝑌(𝜔).

Problem 3

We want to design a (crude) stable, low-pass filter with the transfer function H(z) (i.e. 𝐻(𝜔)). The
general form of a transfer function is:

𝐵(𝑧) ∏ (𝑧 − 𝑧 )
𝐻(𝑧) = 𝐺 = 𝐺𝑧
𝐴(𝑧) ∏ (𝑧 − 𝑝 )
where G is a constant, real gain, and B(z) and A(z) are polynomials in z such that the coefficient next to
the highest power of z for those polynomials is 1.

a) Assume we want a 1st order all-pole system (i.e. system with 1 pole and with no zeros or
with a “trivial” zero in the transfer function – a trivial zero is a zero at z=0). Write H(z) that
would correspond to this case, and sketch the approximate pole-zero plot of such a system
to achieve a crude approximation of the desired low-pass frequency response
b) Additional requirements for the frequency response of the filter are: Magnitude of the freq.
response at ω=0 and ω=π/3 are: |𝐻(𝜔)| | = 1 and |𝐻(𝜔)|| / = 1/√3. Based on
this, calculate the exact parameters (i.e. gain G, and the position of the pole) of the filter
from part a). Justify the choice between multiple possibilities.
c) Assume you keep the same gain G and the pole position as obtained in part b), and that you
can also place 1 non-trivial zero in z-plane to improve the filter design to completely zero-
out the highest discrete frequncy. Sketch the position of the poles/zeros of the improved
H’(z) so that the frequency response 𝐻′(𝜔)| = 0 Write the corresponding H’(z). What
is the magnitude |𝐻′(𝜔)|| in this case?
d) Assuming a system with no zeros (or trivial zeros), would you be able to create a crude
band-pass filter with just 1 pole? Explain!

Hints: Magnitude squared of 𝐻(𝜔) can be obtained as: |𝐻(𝜔)| = 𝐻(𝑧)𝐻(𝑧 )

Problem 4

For two signals 𝑥 (𝑛) = {1 , 1} and 𝑥 (𝑛) = {-1 , 2},


↑ ↑

a) Calculate 2-point Discrete Fourier Transforms 𝑋 (𝑘) = 𝐷𝐹𝑇 {𝑥 (𝑛)} and 𝑋 (𝑘) = 𝐷𝐹𝑇 {𝑥 (𝑛)} of
𝑥 (𝑛) and 𝑥 (𝑛), respectively (k=0,1)
b) Using part a), find 𝑥 (𝑛) = 𝐼𝐷𝐹𝑇 {𝑋 (𝑘)} ,where 𝑋 (𝑘) = 𝑋 (𝑘)𝑋 (𝑘), 𝑘 = 0,1
c) Find linear convolution of 𝑥 (𝑛) and 𝑥 (𝑛) (without using DFT).
d) Explain briefly why are the results of b) and c) different. What is the minimum number of points DFT
required to make the result of b) equal to the result of c)?
e) Using the number of points N determined in d), calculate N-point DFT of 𝑥 (𝑛) and 𝑥 (𝑛) i.e.
𝑋 , (𝑘) , 𝑋 , (𝑘), 𝑘 = 0,1, . . , 𝑁 − 1
f) Using the number of points N determined in d), calculate 𝑥 (𝑛) = 𝐼𝐷𝐹𝑇 {𝑋 (𝑘)} ,where 𝑋 (𝑘) =
𝑋 , (𝑘)𝑋 , (𝑘), 𝑘 = 0,1, . . , 𝑁 − 1
Problem 5

Consider the signal

1 ,0 ≤ 𝑛 ≤ 3
𝑦(𝑛) =
0 , 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒

a) Compute the DTFT of the signal


b) Compute the 4 point DFT of the signal y(n). Explain how/why Y(k) resembles a delta impulse in
frequency domain (what does the periodic extension of y(n) look like? )
c) Compute the 8 point DFT of the signal y(n)
d) Compare the 8-point DFT with the 4-point DFT … are there any samples in frequency domain
that repeat? Provide an interpretation …
e) Is there a change in the underlying DTFT of the signal y(n) when it is zero-padded?

You might also like