You are on page 1of 8

EECS 20N: Structure and Interpretation of Signals and Systems

Problem Set 2
Department of EECS
Issued: 13 September 2014
U NIVERSITY OF C ALIFORNIA B ERKELEY
Due: 19 September 2014, 6pm
Some people claim a lack of ability for science to justify failure and
discouragement. I enjoy laboratory work, they tell us, but am no good
at discovering things.
Certainly there are minds unsuited for experimental work, especially if
they have a short attention span and lack curiosity and admiration for the
work of nature.
But are the great majority of those professing incompetence really so?
Might they exaggerate how difficult the task will be, and underestimate
their own abilities?
I believe that this is often the case, and would even venture to suggest
that many people habitually confuse inability with the simple fact that
they learn and understand slowly, or perhaps are sometimes even lazy
or they dont have a secondary trait such as patience, thoroughness, or
determinationwhich may be acquired rapidly through hard work and the
satisfaction of success.a
a

y Cajal, transExcerpts from Advice for a Young Investigator, by Santiago Ramon


lated by Neely Swanson and Larry W. Swanson, The MIT Press, 1999, ISBN: 0-26268150-1.

Policy Statement
We encourage you to collaborate, but only in a group of up to five current
EECS 20N students.
On the solution document that you turn in for grading, you must write the
names of your collaborators below your own; each teammate must submit
for our evaluation a distinct, self-prepared solution document containing
original contributions to the collaborative effort.
Please write neatly and legibly, because if we cant read it, we cant grade it.
Unless we explicitly state otherwise, you will receive full credit only if you
explain your work succinctly, but clearly and convincingly.
Typically, we evaluate your solutions for only a subset of the assigned problems. A priori, you do not know which subset we will grade. It is to your
advantage to make a bona fide effort at tackling every assigned problem.
If you are asked to provide a sketch, it refers to a hand-drawn sketch, welllabeled to indicate all the salient featuresnot a plot generated by a computing device.
On occasion, a problem set contains one or more problems designated as optional. We do NOT grade such problems. Nevertheless, you are responsible
for learning the subject matter within their scope.
Overview, Subject Matter, and Reading
This problem set covers the convolution of discrete-time signals; the system properties of linearity and time invariance; and the impulse responses and frequency
responses of discrete-time LTI systems. It draws on material from the following
portions of the textbook (Lee & Varaiya):
(a) All of Ch. 2.
(b) All of Ch. 8.
(c) All of Ch. 9.

HW2.1 The impulse response of a discrete-time LTI system F

x
is the discrete-time doublet: f (n) = (n) (n 1).
(a) Determine the output of the system if the input is
(i) the unit-step function: x(n) = u(n).
(ii) a four-point discrete-time box function:

x(n) = (n) + (n 1) + (n 2) + (n 3).


(iii) a constant function: x(n) = 1.
(iv) the sign-alternating signal x(n) = (1)n .
Be sure to provide a well-labeled sketch of all the signals involved.
(b) Based on the results of part (a), explain why the system F may aptly be called
a one-dimensional edge detector. Does the systems response provide a clue as
to which direction (up or down) each edge in the input signal is?
(c) Without determining the frequency response of the filter F, provide a reasonable conjecture as to whether you expect the filter to favor low or high frequencies. That is, do you expect the filter to be a low-pass or a high-pass filter?
(d) Determine a reasonably simple expression for F (), where F is the frequency
response of the filter F.
(e) Provide well-labeled sketches of |F ()| and F (), the magnitude and phase
responses, respectively, of the filter. Is your magnitude response consistent
with your results in part (a) and your conjecture in part (b)?

HW2.2 (Frequency Response of a Discrete-Time FIR Filter) In this problem you


will explore, among other things, the relationship between time scaling and frequency scaling.
(a) A discrete-time LTI filter having impulse response g : Z R produces a twopoint difference of the input, as described below:
n Z,

1
g(n) = [(n) (n 1)].
2

Determine the frequency response G : R C of this filter. Sketch the magnitude response |G| and the phase response G over the frequency range
< . Is this filter low-pass, band-pass, or high-pass? Explain.
(b) A discrete-time LTI filter is described by the following linear, constant-coefficient
difference equation:
1
y(n) = [x(n) x(n 2)].
2
(i) Draw a delay-adder-gain block diagram representation (i.e., a signal flow
graph implementation) of this filter.
(ii) Determine the filters impulse response h : Z R.
(iii) Determine the frequency response H : R C of this filter filter. Sketch
the magnitude response |H| and the phase response H over the frequency range < . Is this filter low-pass, band-pass, or highpass? Explain.
(c) How are g(n) and h(n) related to each other? How are G() and H() related
to each other? Explain how the frequency-domain relationships tie in with the
time-domain relationships.

HW2.3 (System Properties) Consider each of the following systems Fi , i = 1, . . . , 10,


independently. For each system, x denotes the input, and y the corresponding output.1
F1
F2
F3
F4

:
:
:
:

F5 :
F6 :
F7 :
F8 :
F9 :

F10 :

[Z R] [Z R]; n Z, y(n) = x(n).


[R R] [R R]; t R, y(t) = x(e|t| ).
[Z R] [Z R]; n Z, y(n) = x(2n).
[R R] [R R]; t R, y(t) = x(t/2).
Z
1 t
[R R] [R R]; y(t) =
x( ) d, C > 0.
C
Z
1 t
[R R] [R R]; y(t) =
x( ) d, C > 0.
C 0
[R+ {0}] [R+ R]; t R+ , y(t) = cos(800t).
[R+ {0}] [R+ R]; t R+ , y(t) = cos(800(t + 1)).
(
0 if n is an odd integer
[Z R] [Z R]; n Z, y(n) =
1 if n is an even integer.
( n
n mod N = 0
x
N
[Z R] [Z R]; n Z, y(n) =
0
otherwise.
The system F10 is called an N -fold upsampler.

Provide succinct, but clear and convincing responses to the questions below.
(a) For each system, select the strongest true assertion from the list below.
(i) The system must be linear.
(ii) The system could be linear, but does not have to be.
(iii) The system cannot be linear.
(b) For each system, select the strongest true assertion from the list below.
(i) The system must be time-invariant.
(ii) The system could be time-invariant, but does not have to be.
(iii) The system cannot be time-invariant.
1

The definition of system F10 uses the mod function. For any n Z and N N, the expression
n mod N read n modulo N is the unique integer remainder k (0 k N 1) when n is
divided by N , e.g., 5 mod 3 = 2 and 4 mod 3 = 2. Note that N divides n k.

HW2.4 Consider a discrete-time system F : [Z R] [Z R]. It is known that


F is linear and time invariant. The following input-output signal pair (x, y) is a
behavior of the system. The signals x and y are zero outside the regions shown.

x(n)

y(n)
(1)

x
n

(2)
(1)

(1)

3 4 5

Provide succinct, but clear and convincing responses to the questions below.
(a) Can f be determined from the information given in this problem? If so, provide a well-labeled plot of the sample values f (n). If not, explain why it is not
possible to determine f based on what is known about the system F.
(b) Determine, and provide well-labeled, hand-drawn sketches of, the response of
the system to each of the following input signals:
(i) x(n) = 1 for all n Z.
(ii) x(n) = cos(n) for all n Z.
(iii) x(n) = u(n) for all n Z, where u is the unit-step function.

HW2.5 A continuous-time system F : [R R] [R R] is known to be linear.


Two exemplary input-output behaviors (xi , yi ), i = 1, 2, of the system are depicted
in the figure below (all signals are zero outside the regions shown):

x1(t)
0

(1)

x1

y2

x2

(1)

y2(t)
0

y1(t)

(1)

x2(t)

y1

(1)

(a) A continuous-time system G consists of a cascade interconnection of two identical systems F, as shown in the figure below; note that G = F F, and that
the input-output signal pair (q, r) is a behavior of G. Suppose q = x1 . Provide
a well-labeled sketch of the corresponding output signal r, or explain why r
cannot be determined from the information given in the problem.

q(t)
0

(1)

(b) Determine, and provide a well-labeled, hand-drawn sketch of, the response of
the system F to the input signal described below:
(
1 t/T if 0 < |t| T
t R, x(t) =
0
elsewhere.

HW2.6 (Polynomial Multiplication and Discrete-Time Convolution) The impulse


response of a real2 , discrete-time FIR filter A is described by
a(n) = a0 (n) + a1 (n 1) + + aN (n N ),

where N {1, 2, 3, . . .}.

(a) Show that the frequency response A() of the filter is a polynomial, in terms of
ei , whose coefficients have a simple relationship with the impulse response
values a0 , . . . , aN .
(b) Suppose the filter A is placed in a cascade (series) interconnection with another
discrete-time FIR filter B whose impulse response is described by
b(n) = b0 (n) + b1 (n 1) + + bM (n M ),

where M {1, 2, 3, . . .}.

The cascade structure, which we call C, is shown in the figure below.

Let c denote the impulse response of the cascade interconnection.


(i) Express c in terms of a and b.
(ii) Express the frequency response C() of the system C in terms of the frequency responses A() and B() of the cascaded systems A and B.
(iii) Explain why C must be an FIR filter.
(c) Consider two polynomials A(z) and B(z) described as follows:
A(z) = a0 + a1 z + + aN z N

B(z) = b0 + b1 z + + bM z M ,

where M and N and positive integers. Let C(z) = A(z) B(z), where
C(z) = c0 + c1 z + + cN +M z N +M .
Show that multiplying polynomials is tantamount toX
convolving their coefficients; in particular, explain how cn = (a b)n =
am bnm , where n =
m

0, 1, . . . , N + M .
2

A real filter is one that produces a real-valued output signal for every real-valued input signal.
An LTI filter is real if its impulse-response is real-valued.