Escolar Documentos
Profissional Documentos
Cultura Documentos
Spring 2006
Prof. Atlas
Final Exam
June 6, 2006
Your Name:
Solutions
Exam Instructions:
1. Open book and notes. But as listed in our course syllabus: No electronic
devices (calculators, laptops, Pilots, cell phones, beepers, etc.) are
allowed for exams.
2. Do not open this exam until 2:30!
3. Put all your answers in the appropriate space. Feel free to attach extra
worksheets if necessary.
4. Turn in your work and put your name at the top of all loose worksheets.
This work will be looked at for possible partial credit.
5. Justify all of your answers.
6. The exam will be collected promptly at 4:20! Continuing to work after
the bell will cause you to lose points.
7. This exam has a total of 9 pages (including this page).
8. The weight (out of 200) of each section of each problem is located to the
right of the problem in parentheses.
9. The total weight for this exam is 200 points.
After this exam is graded, your score will be recorded on the back of the last
page.
Page 1 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 1
Standard linear convolution within input x[n] , impulse response h[n] , and output y[ n] is represented
by y[n] = x[n] h[n] .
For this problem the input is x[n] = [n] + 2 [n 1] and for the impulse response h[n] we only know
that
h[n] = 7 .
n =
a) Must the system represented by this convolution be bounded-input bounded-output (BIBO) stable? Yes or
no? Please justify your answer.
(10 points)
The system is not necessarily stable. The necessary and sufficient condition for stability is absolute summability
which is that
h[n] is finite. There is no absolute value in the above problem statement. The infinite sum
n =
being finite does not necessarily imply that the infinite absolute sum is finite.
(10 points)
n =
So
n =
n =
n =
n =
n =
n =
m =
h[n] + 2 h[m] = 7 + 2 7 = 21
where the last infinite sum can be found via a simple change in variables
m=n-1where the limits remain infinite.
Page 2 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 2
Determine the N = 10 point inverse discrete Fourier transform for the following two sequences:
1, k = 0
X 1[k ] =
0, k = 1, 2,3,...,9
a)
(10 points)
1 N 1
1 9
1
X 1[k ]WN kn = [k ]W10 kn = W100 n
N k =0
10 k =0
10
1
= , n = 0,1, 2,...,9
10
x1[n] =
b)
X 2 [k ] = 1, k = 0,1, 2,...,9
(10 points)
W
k =0
(k p)n
N
N, k = p
=
0, k p
Thus
1 9 kn 10 /10, n = 0 1, n = 0
W10 = 0, n 0 = 0, n 0
10 k =0
x2 [n] = [n]
x2 [n] =
Page 3 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 3
A causal, discrete time linear system is described by the difference equation:
5 y [ n ] 4 y [ n 1] = x [ n ]
n
7
a) If x [ n ] = u [ n ] , determine y [ n ] .
5
(10 points)
5Y ( z ) 4 z 1Y ( z )= X ( z )
(5 4z ) Y ( z ) = X ( z )
1
Y (z) =
z
X ( z),
5 ( z 4 5)
Y (z) =
z
z
5 ( z 4 5) ( z 7 5)
X ( z) =
z
z
( 7 5)
(
)( z 7 5)
5 z 1Y ( z ) =
Y (z) =
y [ n] =
73
43
( z 7 5) ( z 4 5)
z
z
7
4
15 ( z 7 5 ) 15 ( z 4 5 )
n
7 7
4 4
u [ n] u [ n]
15 5
15 5
b) In your answer for part a), clearly label the forced response and the natural response. Very briefly justify
(10 points)
your answer.
n
7 7
u [ n ] Forced Response (it is an attenuated version of the input)
15 5
n
4 4
u [ n ] Natural Response (this is the system taking off on its own)
15 5
(10 points)
Yes, the system is stable. This can be deduced from the one pole belonging to H ( z ) at z = 4 5 ,
because it is inside the unit circle (the other pole belongs to the unbounded input, x [ n ] ). Stability can
also be determined from the difference equation. For a given bound on the input, x [ n ] M , we find
that y [ n ] 5M .
Page 4 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 4
x [ n ] = 5 [ n 1]+ [ n 4] arises by sampling some original continuous-time real function x ( t ) at
frequency f s = 5 MHz (5, 000, 000 Hz)
a) Determine the DTFT of this x[ n] .
X ( ) =
x[n]e
jn
n =
(10 points)
(5 [n 1]+ [n 4]) e
jn
n =
= 5e j + e j 4
b) Find the values of the DTFT at specific frequency points = 0, , 2 ,3 (your answers should be
(10 points)
simplified)
X (0) = X (2 ) = 6
X ( ) = X (3 ) = 4
c) For this x [ n ] , what is the minimum size (N) of the DFT needed to avoid creating errors in the inverse DFT
(i.e., recover x [ n ] exactly from the result of this size N DFT)?
(10 points)
d) For this problem, it is know that we can recover x ( t ) exactly from x [ n ] . What is the frequency range of the
original signal x ( t ) in Hertz?
(10 points)
Page 5 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 5
The pole-zero plot below corresponds to the z-transform X(z) of a causal sequence x[n].
Im z
z-plane
Re z
(10 points)
Since x[n] is causal (thus right-sided) and the ROC can contain no poles, the ROC must be outside the largest
magnitude pole. ROC x = z > 3 4
(10 points)
(10 points)
Y ( z) =
n =
y [n] z n =
n =
x [ n ] z n =
x [ n] ( )
1
n =
= X ( 1z )
Page 6 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 5 (continued)
d) Sketch the pole-zero plot of this Y(z) on the z-plane below, labeling the pole and zero locations. (10 points)
Im z
z-plane
Re z
Im
z-plane
Re
A zero at z = 0 becomes 1 z = 0 z = .
A pole at z = -3 4 becomes 1 z = -3 4 z = 4 3 .
A pole at z =
A pole at z =
1
2
1
2
e
e
j 4
becomes
becomes
z
1
=
z
1
2
1
2
j 4
z = 2e
j
j 4
z = 2e 4 .
(10 points)
1
ROC x
Page 7 of 9
EE 341
Spring 2006
Prof. Atlas
Problem 6
Given the following two sequences,
1,
x1 [n] =
0,
0n7
otherwise
1,
x2 [n] =
0,
0n4
otherwise
-2
-1
-2
-1
(10 points)
Page 8 of 9
EE 341
Spring 2006
Prof. Atlas
(10 points)
(10 points)
d) What is the minimum length circular convolution to prevent time aliasing of these two signals? (10 points)
L = M + N 1 = 8 + 5 1 = 12
Page 9 of 9