DEPARTAMENTO DE ELÉCTRICA Y

ELECTRÓNICA

Carrera de Ingenierı́a en Electrónica y Telecomunicaciones

Carrera de Ingenierı́a en Electrónica, Automatización y Control

PROCESAMIENTO DIGITAL DE SEÑALES

DEBERES

Dr. Enrique V. Carrera

SANGOLQUÍ, ECUADOR
2018
Contents

1 Discrete Signals and Systems 1

1.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2 Discrete-time Signals and Systems 3

2.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Sampling 6
3.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

4 The Z-transform 8
4.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

5 Fourier Analysis 10
5.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
5.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

6 The Z and Fourier Transforms 12

6.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

7 Transform Analysis of Systems 14

7.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

8 The DFT 16
8.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
8.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

9 Discrete Transforms 18
9.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
9.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

10 The FFT 20
10.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
10.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

11 Digital Filters 21
11.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
11.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

i
12 Implementation of Discrete-Time Systems 24
12.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
12.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

13 FIR Filter Approximations 26

13.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
13.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

14 IIR Filter Approximations 28

14.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
14.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

15 Filter Design 30
15.1 Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
15.2 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

ii
Homework 1

Discrete Signals and Systems∗

1.1 Instructions
Solve the exercises listed in section 1.2 and write your report using LATEX† . After you generate
the PDF file, you must submit your report through the on-line learning platform.

1.2 Exercises
1. Find the period N of the sequence
 nπ   
2nπ  nπ   nπ 
x[n] = cos cos + sin sin
8 15 3 4

2. If x[n] = 0 for n < 0, and the odd part is xo [n] = n(0.5)|n| , find x[n] given that x[0] = 1.

3. Express the sequence (

(−1)n −2 ≤ n ≤ 2
x[n] =
0 otherwise
as a sum of scaled and shifted unit steps.

4. Listed below are several systems that relate the input x[n] to the output y[n]. For each,
determine whether the system is linear or nonlinear, shift-invariant or shift-varying, stable
or unstable, causal or non-causal, and invertible or non-invertible.

(a) y[n] = x[n] + x[−n]

(b) y[n] = nk=0 x[k]
P

(c) y[n] = n+n

P 0
k=n−n0 x[k]
(d) y[n] = log(x[n])
(e) y[n] = median(x[n − 1], x[n], x[n + 1])

5. Given below are the unit sample responses of several linear shift-invariant systems. For
each system, determine the conditions on the parameter a in order for the system to be
stable.
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).
†
There are many different TEX distributions, but I suggest to install TEXstudio (http://www.texstudio.org/)
for Windows, Linux or MacOS. However, if you do not want to worry about software installation, I strongly
recommend to use cloud applications like Overleaf (https://www.overleaf.com/).

1
 (a) h[n] = an µ[−n]
(b) h[n] = an (µ[n] − µ[n − 100])
(c) h[n] = a|n|

6. Consider the linear shift-invariant system described by the first-order linear constant co-
efficient difference equation
y[n] = ay[n − 1] + x[n]
Determine the conditions (if any) for which this system is stable.

7. If x[n] = ( 34 )n µ[n − 2] and h[n] = 2n µ[−n − 5], find the convolution y[n] = x[n] ∗ h[n].

8. The input to a linear shift-invariant system is the unit step, x[n] = µ[n], and the response
is y[n] = δ[n]. Find the unit sample response of this system.

9. Consider the linear shift-invariant system described by the LCCDE

1
y[n] = − y[n − 1] + 2x[n]
2
Find the response of this system to the input
(
2 n = 0, 2, 4, 6, . . .
x[n] =
0 otherwise

10. Consider the system described by the difference equation

y[n] − y[n − 1] + 0.25y[n − 2] = x[n] − 0.25x[n − 1]

(a) Find the unit sample response of the system.

(b) Find the response of the system to x[n] = (0.25)n µ[n].

2
Homework 2

Discrete-time Signals and Systems∗

2.1 Instructions
Solve the exercises listed in section 2.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

2.2 Exercises
1. Characterize the systems below as linear/non-linear, causal/non-causal and time invari-
ant/time varying:

(a) y[n] = (n + a)2 x[n + 4]

(b) y[n] = x[n + 1] + x3 [n − 1]
(c) y[n] = x[n] sin(ωn)
(d) y[n] = x[n] + sin(ωn)
(e) y[n] = x[n]/x[n + 3]
(f) y[n] = y[n − 1] + 8x[n + 3]
(g) y[n] = 2ny[n − 1] + 3x[n + 5]
(h) y[n] = n2 y[n + 1] + 5x[n + 2] + x[n + 4]
(i) y[n] = y[n − 1] + x[n + 5] + x[n + 5]
(j) y[n] = (2µ[n − 3] − 1)y[n − 1] + x[n] + x[n − 1]

2. For each of the discrete signals below, determine whether they are periodic or not. Calcu-
late the periods of those that are periodic.

(a) x[n] = cos2 (2πn/15)

(b) x[n] = cos(4πn/5 + π/4)
(c) x[n] = cos(πn/27 + 31)
(d) x[n] = sin(100n)
(e) x[n] = cos(11πn/12)
(f) x[n] = sin[(5π + 1)n]

3. Consider the system whose output y[m] is described as a function of the input x[m] by the
following difference equations:
∗
Exercises based on the book Digital Signal Processing: System Analysis and Design by Paulo S. R. Diniz et
al. (Cambridge University Press, 2nd Ed., 2010)

3
 P∞
(a) y[m] = n=−∞ x[n]δ[m − nN ]
x[m] ∞
P
(b) y[m] = n=−∞ δ[m − nN ]

Determine whether the systems are linear and/or time invariant.

4. Compute the convolution sum of the following pairs of sequences:

( (
1, 0 ≤ n ≤ 4 an , 0 ≤ n ≤ 7
a) x[n] = and h[n] =
0, otherwise 0, otherwise



1, 0≤n≤2 (

0, 3≤n≤6 n, 1≤n≤4
b) x[n] = and h[n] =


1, 7≤n≤8 0, otherwise

0, otherwise

1
(  2 , n = −1



a[n], n even 1, n = 0
c) x[n] = and h[n] = 1
0, n odd 

 2, n = 1

0, otherwise

5. Compute y[n] = x[n] ∗ x[n] ∗ x[n] ∗ x[n] for the sequence:

(
1, 0 ≤ n ≤ 1
x[n] =
0, otherwise

6. Supposing that all systems in next figure are linear and time invariant, compute y[n] as a
function of the input and the impulse responses of each system.

7. Determine the solutions of the difference equations below, supposing that the systems they
represent are initially relaxed:

(a) y[n] − √1 y[n

2
− 1] + y[n − 2] = 2−n sin(πn/4)µ[n]
√
(b) 4y[n] − 2 3y[n − 1] + y[n − 2] = cos(πn/6)µ[n]
(c) y[n] + 2y[n − 1] + y[n − 2] = 2n µ[−n]
(d) y[n] − y[n − 1] + y[n − 2] = (−1)n µ[n]
(e) y[n] + y[n − 3] = (−1)n µ[−n]

8. Compute the impulse responses of the systems below:

(a) y[n] = 5x[n] + 3x[n + 1] + 8x[n + 2] + 3x[n + 4]

4
 (b) y[n] + 13 y[n − 1] = x[n] + 21 x[n + 1]
(c) y[n] − 3y[n − 1] = x[n]
(d) y[n] + 2y[n − 1] + y[n − 2] = x[n]
(e) y[n] − y[n − 1] = x[n] − x[n + 5]

9. Determine the steady-state response for the input x[n] = sin(ωn)µ[n] of the filters de-
scribed by:

(a) 12y[n] − 7y[n − 1] + y[n − 2] = x[n]

(b) y[n] = x[n + 2] + x[n + 1] + x[n]
(c) y[n] − 21 y[n − 1] = x[n]
(d) y[n] = x[n + 2] + 2x[n + 1] + x[n]

10. Discuss the stability of the systems described by the impulse responses below:

(a) h[n] = 2−n µ[n]

(b) h[n] = 1.5n µ[n]
(c) h[n] = 0.1n
(d) h[n] = 2−n µ[−n]
(e) h[n] = 10n µ[n] − 10n µ[n − 10]
(f) h[n] = 0.5n µ[n] − 0.5n µ[4 − n]

5
Homework 3

Sampling∗

3.1 Instructions
Solve the exercises listed in section 3.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

3.2 Exercises
1. If the Nyquist rate for xa (t) is Ωs , find the Nyquist rate for (a) x2 (2t), (b) x(t/3), (c)
x(t) ∗ x(t).

2. A continuous-time signal xa (t) is known to be uniquely recoverable from its samples

xa (nTs ) when Ts = 1 ms. What is the highest frequency in Xa (f )?

3. Suppose that xa (t) is band-limited to 8 kHz (that is, Xa (f ) = 0 for |f | > 8000). (a) What
is the Nyquist rate for xa (t)? (b) What is the Nyquist rate for xa (t) cos(2π · 1000t)?

4. Let xa (t) = cos(650πt) + 2 sin(720πt). (a) What is the Nyquist rate for xa (t)? (b) If
xa (t) is sampled at twice the Nyquist rate, what are the frequencies of the sinusoids in the
sampled sequence?

5. If a continuous-time filter with an impulse response ha (t) is sampled with a sampling

frequency of fs , what happens to the cutoff frequency ωc of the discrete-time filter as fs
is increased?

6. A complex band-pass signal xa (t) with Xa (f ) nonzero for 10 kHz < f < 12 kHz is sampled
at a sampling rate of 2 kHz. The resulting sequence is x[n] = δ[n]. What is xa (t)?

7. If the highest frequency in xa (t) is f = 8 kHz, find the minimum sampling frequency for
the band-pass signal ya (t) = xa (t) cos(Ω0 t) if (a) Ω0 = 2π ·20·103 and (b) Ω0 = 2π ·24·103 .

8. The continuous-time signal xa (t) = 7.25 cos(2000πt) is sampled at a sampling frequency

of 8 kHz and quantized with a resolution ∆ = 0.02. How many bits are required in the
A/D converter to avoid clipping xa (t)?

9. Suppose that we want to sample the signal xa (t) with a 12-bit quantizer, where xa (t) is
assumed to be gaussian with a variance σx2 ?. What is the signal-to-quantization noise ratio
if we want the range of the quantizer to extend from −3σx to 3σx ?
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

6
10. A sequence x[n] corresponds to samples of a band-limited signal using a sampling frequency
of 10 kHz. However, the sequence should have been sampled using a sampling frequency
fs = 12 kHz. Design a system for digitally changing the sampling rate.

7
Homework 4

The Z-transform∗

4.1 Instructions
Solve the exercises listed in section 4.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

4.2 Exercises
1. Find the Z-transform of
1 |n|
(

2 |n| < 10
x[n] =
0 otherwise

2. How many different sequences have a z-transform given by

1 − 2z −1 + 3z −2
H(z) =
(1 − 18 z −1 + 14 z −2 )(1 + 13 z −1 )

3. A right-sided sequence x[n] has a Z-transform X(z) given by

2z −8 + z −6 − 2z −3 + 4
X(z) =
z −8 + 2z −5 − 4z −3
Find the values of x[n] for all n < 0.
4. Use the Z-transform to perform the convolution of the following two sequences:
x[n] = δ[n] − 2δ[n − 2]
h[n] = 2δ[n] − 2δ[n − 1] + 3δ(n − 2] + δ[n − 3]

5. A right-sided sequence has a Z-transform

z −7 + 3z −4 − 2z −3 + 1
X(z) =
3z −9 + 8z −7 − 4z −2
Find the index and the value of the first nonzero value of x[n].
6. Find the inverse Z-transform of
1
a) X(z) = |z| > 0.4
(1 − 0.4z −1 )2
1
b) X(z) = |z| > 0.2
1 − 0.2z −2
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

8
 7. Find the inverse Z-transform of
z5 − 3
X(z) = |z| > 1
1 − z −5

8. Use the residue theorem to find the value of x[n] at n = 10 when

1 − 3z −4
X(z) = |z| > 0.6
(1 − 0.2z −1 )(1 + 0.6z −1 )

9. The sequence y[n] is the solution to the LCCDE

3 1
y[n] − y[n − 1] + y[n − 2] = x[n] n≥0
2 2
with x[n] = δ[n]. Find a set of initial conditions on y[n] for n < 0 so that y[n] = 1 for
n ≥ 0.

10. Consider the following difference equation:

y[n] + y[n − 2] = x[n] + x[n − 1]

If x[n] = 10µ[n] and y[−2] = −10 and y[−1] = 0, find the output sequence y[n] for n ≥ 0.

9
Homework 5

Fourier Analysis∗

5.1 Instructions
Solve the exercises listed in section 5.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

5.2 Exercises
1. The input to a linear shift-invariant system is x[n] = cos(nπ/8) + cos(3nπ/4). Find the
output when the unit sample response is

(a) h[n] = 8 sin(7nπ/8)/(7nπ)

sin(nπ/2) sin(nπ/4) jπn/4
(b) h[n] = e
n2
2. The input to a linear shift-invariant system is x[n] = n(0.5)n µ[n] and the output is y[n] =
(1/3)n−2 µ[n − 2] − 0.5(1/3)n−3 µ[n − 3]. Find the frequency response H(ejω ) in terms of
both magnitude and phase.

3. If h[n] is the unit sample response of an ideal low-pass filter with a gain of one and a cutoff
frequency ωc = π/8, what is g[n] = cos(nπ/2)h[n]?

4. Find the group delay of the system that has a frequency response

1 − 0.5e−jω
H(ejω ) =
1 + 0.25e−jω

5. What is the magnitude of the frequency response of the cascade of the following two
systems?
e−jω − 0.5 sin(nπ/4)
H1 (ejω ) = h2 [n] = δ[n] −
1 − 0.5e−jω nπ

6. Find the inverse DTFT of X(ejω ) illustrated in figure 5.1.

7. Find the DTFT of

sin(nπ/2) sin(nπ/4)
x[n] =
nπ nπ
8. If the DTFT of x[n] = (0.2)n µ(n + 2) is X(ejω ), find the sequence that has a DTFT given
by Y (ejω ) = X(ej2ω ).
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

10
 Figure 5.1: Frequency response of a digital system

9. The DTFT of a sequence x[n] is X(ejω ) = cos3 (3ω). Evaluate the sum
∞
X
S= (−1)n x[n]
n=−∞

10. A causal linear shift-invariant system is defined by the difference equation 2y[n]−y[n−2] =
x[n − 1] + 3x[n − 2] + 2x[n − 3]. Find the frequency response, H(ejω ), in terms of both
magnitude and phase.

11
Homework 6

The Z and Fourier Transforms∗

6.1 Instructions
Solve the exercises listed in section 6.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

6.2 Exercises
1. Compute the Z transform of the following sequences, indicating their regions of conver-
gence:

(a) x[n] = sin(ωn + θ)µ[n]

(b) x[n] = cos(ωn)µ[n]
(c) x[n] = an µ[−n]
(d) x[n] = e−an µ[n]
(e) x[n] = e−an sin(ωn)µ[n]
(f) x[n] = n2 µ[n]

2. Suppose that the transfer function of a digital filter is given by

(z − 1)2
z 2 + (m1 − m2 )z + (1 − m1 − m2 )

Plot a graph specifying the region of the m2 × m1 plane in which the digital filter is stable.

3. Compute the impulse response of the system with transfer function

z2
H(z) = √
4z 2 − 2 2z + 1
supposing that the system is stable.

4. Compute the time response of the causal system described by the transfer function

(z − 1)2
H(z) =
z 2 − 0.32z + 0.8
when the input signal is the unit step.
∗
Exercises based on the book Digital Signal Processing: System Analysis and Design by Paulo S. R. Diniz et
al. (Cambridge University Press, 2nd Ed., 2010)

12
 5. Determine the inverse Z transform of the following functions of the complex variable z,
supposing that the systems are stable:
z
(a) z−0.8
z2
(b) z 2 −z+0.5
z 2 +2z+1
(c) z 2 −z+0.5
z2
(d) (z−a)(z−1)
1−z 2
(e) (2z 2 −1)(z−2)

6. Compute the inverse Z transform of the functions below. Suppose that the sequences are
right handed and one sided.
1
(a) X(z) = sin( )
z
r
z
(b) X(z) =
1+z
7. Compute the frequency response of a system with the following impulse response:
(
(−1)n , |n| < N − 1
h[n] =
0, otherwise

8. Compute and plot the magnitude and phase of the frequency response of the systems
described by the following difference equations:

(a) y[n] = x[n] + 2x[n − 1] + 3x[n − 2] + 2x[n − 3] + x[n − 4]

(b) y[n] = y[n − 1] + x[n]
(c) y[n] = x[n] + 3x[n − 1] + 2x[n − 2]

9. Plot the magnitude and phase of the frequency response of the digital filters characterized
by the following transfer functions:

(a) H(z) = z −4 + 2z −3 + 2z −1 + 1
z2 − 1
(b) H(z) =
z 2 − 1.2z + 0.95
10. If a digital filter has transfer function H(z), compute the steady-state response of this
system for an input of the type x[n] = sin(ωn)µ[n].

13
Homework 7

Transform Analysis of Systems∗

7.1 Instructions
Solve the exercises listed in section 7.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

7.2 Exercises
1. The system function of a causal filter is
1
H(z) =
1 + az −1 + 0.3z −2
For what values of a will this filter be stable?
2. A stable filter has a system function
(1 − 3z −1 )(1 − 13 z −1 )
H(z) =
1 − 0.2z −1 + 0.4z −2
Find a stable and causal system G(ejω ) such that |G(ejω )H(ejω )| = 1.
3. The system function of an FIR filter is H(z) = (1 + 0.2z −1 + 0.8z −2 )2 . Find a linear phase
system that has a frequency response with the same magnitude.
4. An FIR filter with generalized linear phase has the following properties:
(a) h[n] is real, and h[n] = 0 for n < 0 and for n > 5
P5 n
(b) n=0 (−1) h[n] = 0
(c) H(z) is equal to zero at z = 0.7ejπ/4
Rπ jω
(d) −π H(e )dω = 4π

Find H(z).
5. An FIR filter with a real-valued unit sample response has a group delay τ (ω) = 2. If the
system function has a zero at z = 21 j, and H(z)|z=1 = 1, find h[h].
6. Let x[n] be a sequence that is equal to zero for n < 0 and n > 5. If the Z-transform of
x[n] is
X(z) = 3(1 − 0.2z −1 )(1 + 0.5z −1 + 0.8z −2 )(1 + 0.4z −1 − 0.5z −2 )
how many other sequences are equal to zero for n < 0 and n > 5, have the same initial
value as x[n], and have the same phase?
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

14
 7. A causal and stable allpass filter has a unit sample response that is real. The system
function contains three poles, one of which is at z = 0.8. If H(z) has a zero at z = 2ejπ/4 ,
what is H(z)?

8. If a feedback system of the form G(z) = K is used to compensate the system

1
H(z) =
1+ 0.8z −1 + 1.6z −2
for what values of K will the closed-loop system be stable?

9. For the system

1
H(z) =
1 + 1.2z −1 + 1.5z −2
find a feedback system of the form G(z) = 1 + g(1)z −1 + g(2)z −2 that will move the poles
of H(z) to z = 0.5 and z = −0.5.

10. Find the closed-loop system function of a feedback network with

z −1
(a) H(z) = 1− 12 z −1
and G(z) = 2
3 − 16 z −1
z −1
(b) H(z) = 2
3 − 16 z −1 and G(z) = 1− 12 z −1

15
Homework 8

The DFT∗

8.1 Instructions
Solve the exercises listed in section 8.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

8.2 Exercises
1. Find the DFS coefficients for the sequence of period N = 8 whose first four values are
equal to 1 and the last four are equal to 0.

2. Find the 10-point DFT of each of the following sequences:

(a) x[n] = δ[n] + δ[n − 5]

(b) x[n] = µ[n] − µ[n − 6]

3. Find the 10-point inverse DFT of


3 k = 0

X[k] = 2 k = 3, 7

1 otherwise


4. If the 10-point DFTs of x[n] = δ[n] − δ[n − 1] and h[n] = µ[n] − µ[n − 10] are X[k] and
H[k], respectively, find the sequence w[n] that corresponds to the 10-point inverse DFT of
the product H[k]X[k].

5. If x[n] is real and x[n] = x[N − n], what can you say about the N -point DFT of x[n]?

6. If x[n] has an N -point DFT X[k], find the N -point DFT of y[n] = cos(2πn/N )x[n].

7. If X[k] is the N -point DFT of x[n], what is the N -point DFT of the sequence y[n] = X[n]?

8. How many DFTs and inverse DFTs of length N = 128 are necessary to linearly convolve
a sequence x[n] of length 1000 with a sequence h[n] of length 64 using the overlap-add
method? Repeat for the overlap-save method.

9. A sequence x[n] of length N1 = 100 is circularly convolved with a sequence h[n] of length
N2 = 64 using DFTs of length N = 128. For what values of n will the circular convolution
be equal to the linear convolution?
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

16
10. A continuous-time signal x(t) is sampled with a sampling frequency o fs = 2 kHz. If a
1000-point DFT of 1000 samples is computed, what is the spacing between the frequency
samples X[k] in terms of the analog frequency?

17
Homework 9

Discrete Transforms∗

9.1 Instructions
Solve the exercises listed in section 9.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

9.2 Exercises
1. One wishes to measure the frequency content of a fast pulse x(t) using a digital computer.
In order to do so, a fast data acquisition system detects the beginning of the pulse and
digitalizes it. Knowing that:

(a) The pulse duration is of approximately 1 ns.

(b) It has no significant frequency components beyond 5 GHz.
(c) One needs to discriminate frequency components spaced of 10 MHz.

One asks:

(a) Determine the smallest sampling rate at the A/D converter of the data acquisition
system that makes the desired measurement possible.
(b) Describe the measurement procedure, providing the relevant parameter values for the
minimum sampling frequency case.

2. Show that
N −1
(
X N, for k = 0, ±N, ±2N, . . .
WNnk =
n=0
0, otherwise

~ = [9, 1, 1, 9, 1, 1, 1, 1]T .
3. Given the DFT coefficients represented by the vector X

(a) Determine its length-8 IDFT.

(b) Determine the sequence whose length-8 DFT is given by Y [k] = W8−4k X[k].
~ = [4, 2, 2, 2, 2, 2, 2, 4]T .
4. Suppose the DFT X[k] of a sequence as represented by the vector X

(a) Compute the corresponding IDFT using the adequate DFT properties.
(b) Compute the sequence whose length-8 DFT is given by Y [k] = W83k X[k].

5. Consider the sequence x[n] = δ[n] + 2δ[n − 1] − δ[n − 2] + δ[n − 3].

∗
Exercises based on the book Digital Signal Processing: System Analysis and Design by Paulo S. R. Diniz et
al. (Cambridge University Press, 2nd Ed., 2010)

18
 Figure 9.1: Digital sequences from exercise 8.

Figure 9.2: Digital sequences from exercise 9.

(a) Compute its length-4 DFT.

(b) Compute the finite-length sequence y[n] of length-6 whose DFT is equal to the imag-
inary part of the DFT of x[n].
(c) Compute the finite-length sequence w[n] of length-4 whose DFT is equal to the imag-
inary part of the DFT of x[n].

6. Compute the coefficients of the Fourier series of the periodic sequences below using the
DFT.

(a) x0 [n] = sin(2πn/N ), for N = 20

(b) x0 [n] = (−1)n

7. Compute and plot the magnitude and phase of the DFT of the following finite-length
sequences:

(a) x[n] = 2 cos(πn/N ) + sin2 (πn/N ), for 0 ≤ n ≤ 10 and N = 11

(b) x[n] = e−2n , for 0 ≤ n ≤ 20
(c) x[n] = δ[n − 1], for 0 ≤ n ≤ 2
(d) x[n] = n, for 0 ≤ n ≤ 5

8. Compute the linear convolution of the sequences in figure 8 using the DFT.

9. Compute the linear convolution of the sequences in figure 9.2 using the DFTs with the
minimum possible lengths. Justify the solution, indicating the DFT properties employed.

10. We want to compute the linear convolution of a long sequence x[n], of length L, with
a short sequence h[n], of length K. If we use the overlap-and-save method to compute
the convolution, determine the block length that minimizes the number of arithmetic
operations involved in the convolution.

19
Homework 10

The FFT∗

10.1 Instructions
Solve the exercises listed in section 10.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

10.2 Exercises
1. Let x[n] be a sequence of length 1024 that is to be convolved with a sequence h[n] of
length L. For what values of L is it more efficient to perform the convolution directly than
it is to perform the convolution by taking the inverse DFT of the product X[k]H[k] and
evaluating the DFTs using a radix-2 FFT algorithm?

2. Suppose that we have a 1025-point data sequence (1 more than N = 210 ). Instead of
discarding the final value, we zero pad the sequence to make it of length N = 211 so that we
can use a radix-2 FFT algorithm. (i) How many multiplications and additions are required
to compute the DFT using a radix-2 FFT algorithm? (ii) How many multiplications and
additions would be required to compute a 1025-point DFT directly?

3. In a radix-3 decimation-in-time FFT, how is the input sequence indexed?

4. How many complex multiplications are necessary in a radix-3 decimation-in-frequency

FFT?

5. How many complex multiplications are required for a 12-point prime factor FFT with
N1 = 4 and N2 = 3 if we do not count multiplications by ±1 and ±j?

6. How many complex multiplications are required for a 15-point prime factor FFT if we do
not count multiplications by ±1?

7. Determine the basic cell of a radix-5 algorithm. Analyze the possible simplifications in the
graph of the cell.

8. Show that the number of complex multiplications of the FFT algorithm for generic N is
given by
M (N ) = N (N1 + N2 + . . . + Nl−1 + Nl − l)

∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

20
Homework 11

Digital Filters∗

11.1 Instructions
Solve the exercises listed in section 11.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

11.2 Exercises
1. Give two distinct realizations for the transfer functions below:

(a) H(z) = 0.0034 + 0.0106z −2 + 0.0025z −4 + 0.0149z −6

2
−1.349z+1 2−1.889z+1
(b) H(z) = ( z 2z−1.919z+0.923 )( z 2z−1.937z+0.952 )

Figure 11.1: Signal flowgraphs of three digital filters.

∗
Exercises based on the book Digital Signal Processing: System Analysis and Design by Paulo S. R. Diniz et
al. (Cambridge University Press, 2nd Ed., 2010)

21
 Figure 11.2: Digital filter structure.

Figure 11.3: Digital filter structure.

2. Write the equations that describe the networks in figure 11.1.

3. Determine the transfer functions of the digital filters in figure 11.1.

4. Describe the networks in figure 11.1 using state variables.

5. Implement the transfer function below using a parallel realization with the minimum num-
ber of multipliers.
z 3 + 3z 2 + 11
4 z+ 4
5
H(z) = 2 1
(z + 2 z + 12 )(z + 12 )

6. Given the realization depicted in figure 11.2:

(a) Show its state-space description.

(b) Determine its transfer function.
(c) Derive the expression for its magnitude response and interpret the result.

7. Consider the digital filter of figure 11.3:

(a) Determine its state-space description.

(b) Compute its transfer function and plot the magnitude response.

8. Determine the transfer function of the digital filter shown in figure 11.4 using the state-
space formulation.

22
 Figure 11.4: Second-order lattice structure.

Figure 11.5: Digital filter structure.

9. Determine the impulse response of the filter shown in figure 11.5.

10. Determine the impulse response of the filter shown in figure 11.6.

Figure 11.6: Signal flowgraph of a digital filter.

23
Homework 12

Implementation of Discrete-Time
Systems∗

12.1 Instructions
Solve the exercises listed in section 12.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

12.2 Exercises
1. Find the unit sample response for the following network:

2. What is the frequency response of the following network?

3. What is the minimum number of multiplications and additions and delays required to
implement a linear phase filter with h[n] = 0 for n < 0 and n > 63?
4. How many multiplies, adds, and delays are required to implement the filter
(
1
[1 + 12 cos( 2πn 1 4πn
64 ) + 4 cos( 64 ) 0 ≤ n ≤ 63
h[n] = 64
0 otherwise
∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

24
 5. Draw a frequency sampling structure for the FIR high-pass filter of length N = 32 with
(
1
k = 15, 16, 17
H[k] = 32
0 otherwise

6. Find the system function for the following network, where az −1 is a unit delay combined
with a multiplication by a:

7. Find the unit sample response of the following network:

8. Find the constant A so that the second-order all-pole filter

A
H(z) =
1 + α1 z −1 + α2 z −2
has unit gain at ω = 0, and find a structure that only requires two multiplications.

9. What is the system function corresponding to the following filter structure?

10. If the system function of a causal filter is

1
H(z) =
1 + 1.1z −1 + 0.9z −2 + 1.45z −3 + 0.5z −4
is this filter stable?

25
Homework 13

FIR Filter Approximations∗

13.1 Instructions
Solve the exercises listed in section 13.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

13.2 Exercises
1. Assume that a periodic signal has four sinusoidal components at frequencies ω0 , 2ω0 ,
4ω0 , 6ω0 . Design a nonrecursive filter, as simple as possible, that eliminates only the
components 2ω0 , 4ω0 , 6ω0 .

2. Given a lowpass FIR filter with transfer function H (z), describe what happens to the filter
frequency response when:

(a) z is replaced by −z.

(b) z is replaced by z −1 .
(c) z is replaced by z 2 .

3. Determine the relationship between L, M , and N which means that the overall filter in
figure 13.1 has linear phase.

Figure 13.1: Overall filter block diagram.

4. Design a highpass filter satisfying the following specification below using the frequency
sampling method: M = 12, Ωr = 1.0 rad/s, Ωp = 1.5 rad/s, Ωs = 5.0 rad/s.

5. Plot and compare the characteristics of the Hamming window and the corresponding
magnitude response for M = 5, 10, 15, 20.
∗
Exercises based on the book Digital Signal Processing: System Analysis and Design by Paulo S. R. Diniz et
al. (Cambridge University Press, 2nd Ed., 2010)

26
 Figure 13.2: Ideal magnitude response of a filter.

Figure 13.3: Ideal magnitude response of a filter.

6. Plot and compare the rectangular, triangular, Bartlett, Hamming, Hann, and Blackman
window functions and the corresponding magnitude responses for M = 20.

7. Determine the ideal impulse response associated with the magnitude response shown in
figure 13.2, and compute the corresponding practical filter of orders M = 8, 14, 20 using
the Hamming window.

8. For the magnitude response shown in figure 13.3, where ωs = 2π denotes the sampling
frequency:

(a) Determine the ideal impulse response associated with it.

(b) Design a fourth-order FIR filter using the triangular window with ωc = π/4.

9. For the magnitude response shown in figure 13.4:

(a) Determine the ideal impulse response associated with it.

(b) Design a fourth-order FIR filter using the Hann window.

Figure 13.4: Ideal magnitude response of a filter.

10. Design a bandpass filter satisfying the following specification using the Hamming, Hann,
and Blackman windows: M = 10, Ωc1 = 1.125 rad/s, Ωc2 = 2.5 rad/s, Ωs = 10 rad/s.

27
Homework 14

IIR Filter Approximations∗

14.1 Instructions
Solve the exercises listed in section 14.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

14.2 Exercises
1. Design an analog elliptic filter satisfying the following specifications: Ap = 1.0 dB, Ar = 40
dB, Ωp = 1000 Hz, Ωr = 1209 Hz.

2. Design a lowpass Butterworth filter satisfying the following specifications: Ap = 0.5 dB,
Ar = 40 dB, Ωp = 100 Hz, Ωr = 150 Hz, Ωs = 500 Hz.

3. Design a bandstop elliptic filter satisfying the following specifications: Ap = 0.5 dB, Ar =
60 dB, Ωp1 = 40 Hz, Ωr1 = 50 Hz, Ωr2 = 70 Hz, Ωp2 = 80 Hz, Ωs = 240 Hz.

4. Design highpass Butterworth, Chebyshev, and elliptic filters that satisfy the following
specifications: Ap = 1.0 dB, Ar = 40 dB, Ωr = 5912.5 rad/s, Ωp = 7539.8 rad/s, Ωs =
50265.5 rad/s.

5. Design three bandpass digital filters, one with a central frequency of 770 Hz, a second of
852 Hz, and a third of 941 Hz. For the first filter, the stopband edges are at frequencies
697 and 852 Hz. For the second filter, the stopband edges are 770 and 941 Hz. For the
third filter, the edges are at 852 and 1209 Hz. In all three filters, the minimum stopband
attenuation is 40 dB and use Ωs = 8 kHz.

6. Plot the zero-pole constellation for the three filters designed in previous exercise and
visualize the resulting magnitude response in each case.

7. The transfer function

κ
H(s) =
(s2 + 1.4256s + 1.23313)(s + 0.6265)

corresponds to a lowpass normalized Chebyshev filter with passband ripple Ap = 0.5 dB.

(a) Determine κ such that the filter gain at DC is 1.

(b) Design a highpass digital filter with cutoff frequency ωp = π/3 rad/s, sampling fre-
quency ωs = π rad/s, and passband ripple of 0.5 dB using the bilinear transformation.
∗
Exercises based on the book Digital Signal Processing: System Analysis and Design by Paulo S. R. Diniz et
al. (Cambridge University Press, 2nd Ed., 2010)

28
 (c) Suggest a possible realization for the resulting transfer function.

8. Given the analog transfer function

1
H(s) =
(s2 + 0.76722s + 1 − 33863)(s + 0.76722)

design transfer functions corresponding to discrete-time filters using both the impulse-
invariance method and the bilinear transformation. Choose Ωs = 12 rad/s. Compare the
resulting frequency responses with the one from the analog filter.

9. Determine the original analog transfer function corresponding to

4z z
H(z) = −0.4
−
z−e z − e−0.8
assuming that the following method was employed in the analog to discrete-time mapping,
with T = 4:

(a) Impulse invariance method.

(b) Bilinear transformation method.

10. Design a digital filter corresponding to the specifications: Ap = 1.0 dB, Ar = 40 dB,
Ωp = 1000 Hz, Ωr = 1209 Hz, Ωs = 8 kHz. Then transform the designed filter into
a highpass filter satisfying the specifications: Ap = 1.0 dB, Ar = 40 dB, Ωr = 5912.5
rad/s, Ωp = 7539.8 rad/s, Ωs = 50265.5 rad/s, using the frequency transformation in the
discrete-time domain.

29
Homework 15

Filter Design∗

15.1 Instructions
Solve the exercises listed in section 15.2 and write your report using LATEX. After you generate
the PDF file, you must submit the report using your account in https://engrade.com/.

15.2 Exercises
1. Use the bilinear transformation to design a first-order low-pass Butterworth filter that has
a 3-dB cutoff frequency ωc = 0.5π.

2. Use the bilinear transformation to design a second-order bandpass Butterworth filter that
has 3-dB cutoff frequencies ωl = 0.4π and ωu = 0.6π.

3. If the specifications for an analog low-pass filter are to have a 1-dB cutoff frequency of 1
kHz and a maximum stopband ripple δs = 0.01 for |f | > 5 kHz, determine the required
filter order for the following:

(a) Butterworth filter.

(b) Type I Chebyshev filter.
(c) Type II Chebyshev filter.
(d) Elliptic filter.

4. Let Ha (jΩ) be an analog filter with

Ha (jΩ)|Ω=0 = 1

(a) If a discrete-time filter is designed using the impulse invariance method, is it neces-
sarily true that
H(ejω )ω=0 = 1


(b) Repeat part (a) for the bilinear transformation.

5. Consider a causal and stable continuous-time filter that has a system function
s+1
Ha (s) =
(s + 2)2

If a discrete-time filter is designed using impulse invariance with Ts = 1, find H(z).

∗
Exercises based on the book Schaum’s Outline of Theory and Problems of Digital Signal Processing by Monson
H. Hayes (McGraw Hill, 2nd Ed., 2011).

30
 6. The system function of a digital filter is
2 1
H(z) = −1
−
1 − 0.5z 1 − 0.25z −1
(a) Assuming that this filter was designed using impulse invariance with Ts = 2, find the
system function of two different analog filters that could have been the analog filter
prototype.
(b) If this filter was designed using the bilinear transformation with Ts = 2, find the
analog filter that was used as the prototype.

7. The system function of an analog filter Ha (s) may be expressed as a parallel connection
of two lower-order systems
Ha (s) = Ha1 (s) + Ha2 (s)
If Ha (s), Ha1 (s), and Ha2 (s) are mapped into digital filters using the impulse invariance
technique, will it be true that

H(z) = H1 (z) + H2 (z)

What about with the bilinear transformation?

8. If an analog filter has an equiripple passband, will the digital filter designed using the im-
pulse invariance method have an equiripple passband? Will it have an equiripple passband
if the bilinear transformation is used?

9. Can an analog allpass filter be mapped to a digital allpass filter using the bilinear trans-
formation?

10. An IIR low-pass digital filter is to be designed to meet the following specifications: (i)
passband cutoff frequency of 0.22π with a passband ripple less than 0.01, (ii) stopband
cutoff frequency of 0.24π with a stopband attenuation greater than 40 dB.

(a) Determine the filter order required to meet these specifications if a digital Butterworth
filter is designed using the bilinear transformation.
(b) Repeat for a digital Chebyshev filter.
(c) Compare the number of multiplications required to compute each output value using
these filters, and compare them to an equiripple linear phase filter.

31