DSP Exam 1 Solutions and Key Information

EE538 Digital Signal Processing I Exam 1 Fall 2000
Examination Date: Session 11 (Live: 15 September 2000).

Test Duration: 50 minutes.
Open Book but Closed Notes. This test is composed of three problems.
All work should be done in the blue books provided.
There is no need to return the test sheet, just turn in the blue books.
Problem 1. [30 points]

Consider a very simplistic CDMA system with only two users assigned the following
length 4 orthogonal codes, respectively:
User 1’s code: c1 [n] = {1, −1, 1, −1}
User 2’s code: c2 [n] = {1, −1, −1, 1}

Consider transmitting a block of two bits for each of the two users,
User 1’s two info. bits: b1 [n] = {b1 [0], b1 [1]}
User 2’s two info. bits: b2 [n] = {b2 [0], b2 [1], }

where bk [n] is either a “+1” representing the binary bit “1” or “-1” representing the binary
bit “0” for any value of k or n. The transmitted code-division multiplexed block may be
mathematically expressed as
X
2 X
1
x[n] = bk [m]ck [n − 4m], n = 0, 1, ..., 7
k=1 m=0
Given that the received block has the following numerical values
0 , 0, 2, −2, 2, −2, 0, 0}
x[n] = {|{z}
↑
where the first entry above is the value of x[0], determine the numerical values of b1 [n], n =
0, 1 and b2 [n], n = 0, 1. Your answer should consist of 4 numerical values all together. Show
all work in arriving at your answer. Note that code 1 is orthogonal to code 2.
1

A realization of a causal IIR notch filter is the so-called coupled realization pictured
below.
x[n] + + y[n]
z -1
+ +
2 r cos(θ) 2 cos(θ)
z -1
r2
(a) Determine the overall transfer function H(z) = Y (z)/X(z) in terms of r and θ. Show
all work.
(b) Express the poles of the systems in terms of r and θ. What are the conditions on r
and θ in order that the system be BIBO stable?
(c) For this part of the problem, let r = .95 and θ = 90◦ .
(i) Plot the pole-zero diagram.

(ii) State and plot the region of convergence for H(z).
(iii) Determine the DTFT of h[n] and plot the magnitude |H(ω)| over the interval
−π < ω < π showing as much detail as possible. In particular, explicitly point
out if there are any values of ω for which |H(ω)| is exactly zero.
2
Problem 3. [30 points] Consider the transmission of a pulse amplitude-modulated signal

∞
X
x(t) = b[k]p(t − kTo )
k=−∞
where b[n] are the information-bearing symbols being transmitted which may be viewed as
a discrete-time sequence. In binary phase-shift keying, b[n] is either “+1” or “-1” for all n.
p(t) is the pulse symbol waveform and 1/To is the bit rate. Sampling p(t) at TWICE the bit
rate yields the discrete-time sequence

To sin( π2 n) cos( π4 n)
p̃[n] = p n = π , −∞ < n < ∞,
2 2
n 1 − n42
At the receiver, x(t) arrives by both a direct path and a multipath reflection having the same
strength as the direct path but at a delay of To and phase-shifted by θ. Denoting continuous
time convolution as ∗, the received signal, y(t), may be modeled as:
y(t) = x(t) ∗ g(t)
where g(t) is described below using δ(t) to denote the Dirac Delta function.
g(t) = δ(t) + ejθ δ (t − To ) (1)
Sampling y(t) at the bit rate, Fs = T1o , it is easily shown that the resulting sequence y[n] =
y(nTo ) may be modeled as having been generated by the following discrete-time system.
symbol y [n]= y (n To )
b[n] h [n]
sequence
P∞
Determine h[n] and plot the magnitude of its DTFT H(ω) = n=−∞ h[n]e−jωn over
−π ≤ ω ≤ π for each of the following three values of θ:
(i) θ = 0
π
(ii) θ = 2
(iii) θ = π
In each case, H(ω) will be exactly equal to zero for one specific value of ω. Determine this
frequency in each case.
3
EE538 Digital Signal Processing I 20 September 2001
Exam 1
Cover Sheet
Open Book but Closed Notes.
Calculators not allowed.
This test contains three problems.
You must show all work for each problem to receive full credit.
Do not return this test sheet, just return the blue books.
Prob. No. Topic(s) Points

1. Convolution and LTI Systems 30
2. Z-Transform (ZT); Relationship between ZT and DTFT 40
5. Digital Upsampling 30
1
h (n) y (n)
1 1
x(n)
h (n) y (n)
2 2
Consider the system above where the causal signal x(n) is simultaneously input to two
different causal FIR filters of length M = 2 with respective impulses responses h1 (n) and
h2 (n). That is, both h1 (n) and h2 (n) are only nonzero for n = 0 and n = 1. We don’t know
anything about the input signal x(n) except that it is causal. Yet, it is possible to determine
h1 (n) and h2 (n) (to within a scalar multiple) given the respective outputs y1 (n) and y2 (n).
(a) Show that y1 (n) ∗ h2 (n) = y2 (n) ∗ h1 (n), which may be alternatively expressed as
y1 (n) ∗ h2 (n) − y2 (n) ∗ h1 (n) = 0
(b) Exploiting this relationship and arbitrarily assigning h1 (0) = 1 without loss of gen-
erality, determine the numerical values of h1 (1), h2 (0), and h2 (1) given the following
output values.
y1 (0) = 1 y1 (1) = 4 y1 (2) = 7 y1 (3) = 10
y2 (0) = 3 y2 (1) = 5 y2 (2) = 7 y2 (3) = 9
Note: This is not all of the output values for both y1 (n) and y2 (n). However, it is
enough information to determine both h1 (n) and h2 (n) which, again, are only nonzero
for n = 0 and n = 1. Again, you are given h1 (0) = 1. SHOW ALL WORK.

Consider that we wish to filter a long stream of data with a filter having the following impulse
response
2π
h[n] = ej N ℓn {u[n] − u[n − N]}
where ℓ is an integer between 0 and N − 1. Convolution with this impulse response may be
effected via the following difference equation
N −1
2π
ej N ℓk x[n − k]
X
y[n] =
k=0
This implementation is seen to require N multiplications and N − 1 additions per output

point.
2
(a) It can be shown that we can achieve exactly the same input-output (I/O) relationship
via the following difference equation which requires only 1 multiplication and two
additions per output point.
y[n] = a1 y[n − 1] + x[n] − x[n − D]
Determine the values of a1 and D in terms of ` and N so that this IIR system has
exactly the same I/O relationship as the FIR filter above. SHOW ALL WORK – NO
CREDIT FOR ANSWERS THAT ARE NOT SUBSTANTIATED BY WORK.
(b) Consider the specific case of ` = 2 and N = 4:
(i) State the numerical values of a1 and D for this case so that the FIR filter and the
IIR filter have the same I/O relationship.
(ii) Plot the pole-zero diagram for the system. Show the region of convergence.
(iii) Is this a lowpass, bandpass, or highpass filter? Briefly explain why.
(iv) Plot the impulse response of the system (Stem plot).
x(n) Ideal LPF: ωc = π/2

w(n) y(n)
2 2 sin(ω cn)
h(n) =
π n
The DT signal x(n) is input to the system above to effect digital upsampling by a factor of
two, where (
x( n2 ), n even
w(n) =
0, n odd
sin( π2 n)
h(n) = π , −∞ < n < ∞,
2
n
Instead of having to actually run zeroes through the filter, we know that we can compute
the even and odd samples of the output signal y(n) above as shown below.
h [n] = h[2n] y [n] = y [2n]

1 1
x[n]
h [n] = h[2n+1] y [n] = y [1+2n]

2 2
3
(a) Let H1 (ω) and H2 (ω) denote the DTFT’s of h1 (n) and h2 (n), respectively. Plot the
magnitude and phase (separate plots) of H1 (ω) over −π < ω < π. Also plot the
magnitude and phase (separate plots) of H2 (ω) over −π < ω < π. Your answer to this
part of the question should be four plots.
(b) Repeat part (a) for a nonideal lowpass filter with impulse response
sin( π2 n) cos( π4 n)
hLP [n] = , −∞ < n < ∞,
π
2
n 1 − n42
having the raised-cosine spectrum plotted below for which HLP (ω) = 2 for |ω| ≤ π4 ,
HLP (ω) = 0 for 3π 4
≤ |ω| ≤ π, and HLP (ω) has a cosine roll-off from 2 at ωp = π4
to 0 at ωs = 3π4
. Let H1 (ω) and H2 (ω) denote the DTFT’s of h1 (n) = hLP [2n] and
h2 (n) = hLP [1 + 2n], respectively. Plot the magnitude and phase (separate plots) of
H1 (ω) over −π < ω < π. Also plot the magnitude and phase (separate plots) of H2 (ω)
over −π < ω < π. Your answer to this part of the question should also be four plots.
H (ω)
LP
2
ω
π 3π π π 3π π
4 4 4 4
4
EE538 Digital Signal Processing I Session 13
Exam 1 Live: Wed., Sept. 18, 2002
Cover Sheet
Coverage: Sessions 1-10.
Calculators not allowed.

1. DT Correlation 30
2. LTI Systems (Causal): 30
Properties, Frequency Response
Interconnection, Pole-Zero Diagram
3. Digital Upsampling 40
1
length 4 orthogonal complex-value codes, respectively, which are two rows of a 4-point DFT
matrix:
User 1’s code: c1 [n] = {1, −j, −1, j}
User 2’s code: c2 [n] = {1, j, −1, −j}
Consider transmitting a block of two QAM symbols for each of the two users,
User 1’s two info. symbols: b1 [n] = {b1 [0], b1 [1]}
User 2’s two info. symbols: b2 [n] = {b2 [0], b2 [1], }

where bk [n] can take on one of four different complex values (for each value of k and n):
1. 1+j representing the bit pair {1,1} ;
2. -1+j representing the bit pair {0,1} ;
3. -1-j representing the bit pair {0,0} ;
4. 1-j representing the bit pair {1,0} ;
The transmitted code-division multiplexed block may be mathematically expressed as
X
2 X
1
x[n] = bk [m]ck [n − 4m], n = 0, 1, ..., 7
k=1 m=0
x[n] = { 2j , −2j, −2j, 2j, −2j, 2j, 2j, −2j}

|{z}
↑
all work in arriving at your answer. Note that code 1 is orthogonal to code 2 and that the
two users are synchronized.
2

Consider the causal, second-order difference equation below where it is noted that .9025 =
2
.95 .
y[n] = −0.95 y[n − 1] − 0.9025 y[n − 2] + x[n] + x[n − 1] + x[n − 2]
Consider implementing this second-order difference equation as two first-order systems in
parallel (one pole each) as shown in the diagram.
y [n]
1
h [n]
1
x[n] + y[n]
y [n]
2
h [n]
2
(a) Determine and write the first-order difference equation for each of the two first-order
systems in parallel. The upper first-order system has impulse response h1 [n] and is
described by the difference equation
(1) (1) (1)
y1 [n] = a1 y1 [n − 1] + b0 x[n] + b1 x[n − 1]
The lower first-order system has impulse response h2 [n] and is described by the differ-
ence equation
(2) (2) (2)
y2 [n] = a1 y2 [n − 1] + b0 x[n] + b1 x[n − 1]
(i) (i) (i)
Determine the numerical values of a1 , b0 , and b1 , i = 1, 2 – six values total.
(b) For EACH of the two first-order systems, i = 1, 2, do the following:

(ii) State and plot the region of convergence for Hi (z).
(iii) Determine the DTFT of hi [n] and plot the magnitude |Hi(ω)| over the interval
−π < ω < π showing as much detail as possible. In particular, explicitly point
out if there are any values of ω for which |Hi (ω)| is exactly zero.
3
X (F) H (ω)
a LP
1/4W 2
F ω
W W
π 3π π π 3π π
4 4 4 4
x (t) x [n] w [n] Lowpass Filter h [n] y [n]

a Ideal A/D LP
= π
2
F = 4W
s ω 4
ω s = 3π
p 4
gain =2
Figure 1.
The analog signal xa (t) with CTFT Xa (F ) shown above is input to the system above,
where x[n] = xa (n/Fs ) with Fs = 4W , and
sin( π2 n) cos( π4 n)
hLP [n] = , −∞ < n < ∞,
π
2
n 1 − n2
4
such that HLP (ω) = 2 for |ω| ≤ π4 , HLP (ω) = 0 for 3π 4

≤ |ω| ≤ π, and HLP (ω) has a cosine
roll-off from 1 at ωp = π4 to 0 at ωs = 3π
4
. Finally, the zero inserts may be mathematically
described as (
x( n2 ), n even
w[n] =
0, n odd
(a) Plot the magnitude of the DTFT of the output y[n], Y (ω), over −π < ω < π.
(b) Given that

sin( π2 n)
x[n] = − ∞ < n < ∞,
πn
provide an analytical expression for y[n] for −∞ < n < ∞ (similar to the expression
for either x[n] or hLP [n] above, for example.)
THIS PROBLEM IS CONTINUED ON THE NEXT PAGE.
4
(c) The up-sampling by a factor of 2 in Figure 1 can be efficiently done via the block
diagram in Figure 2 below.
(i) Provide an analytical expression for h0 [n] = hLP [2n] for −∞ < n < ∞. Simplify
as much as possible.
(ii) Plot the magnitude of the DTFT of h0 [n], |H0 (ω)|, over −π < ω < π.
(iii) Provide an analytical expression for the output y0 [n] for −∞ < n < ∞.
(iv) Plot the magnitude of the DTFT of h1 [n], |H1 (ω)|, over −π < ω < π.
(v) Plot the phase of the DTFT of h1 [n] , 6 H1 (ω), over −π < ω < π.
(vi) Provide an analytical expression for the output y1 [n] for −∞ < n < ∞.
y [n]=y[2n]
h [n] = h [2n] 0
0 LP
y[n]
x [n] interleaver
y [n]=y[2n+1]
h [n] = h [2n+1] 1
1 LP
Figure 2.
5
Exam 1 Live: Fri., Sept. 26, 2002
Cover Sheet
Calculators allowed.

1. DT Correlation 30
2. Deriving DT Model from CT Signal/System 35
Digital Upsampling
Frequency Response of LTI System
3. Deriving DT Model from CT Signal/System 35
Digital Upsampling
Frequency Response of LTI System
1

length 4 orthogonal codes, respectively, which are two rows of a 4x4 Walsh-Hadamard matrix:
User 1’s code: c1 [n] = {1, −1, 1, −1}
User 2’s code: c2 [n] = {1, 1, −1, −1}

Consider transmitting a block of two PAM symbols for each of the two users,
User 1’s two info. symbols: b1 [n] = {b1 [0], b1 [1]}
User 2’s two info. symbols: b2 [n] = {b2 [0], b2 [1], }

where bk [n] can take on one of four different real values (for each value of k and n):
1. 3 representing the bit pair {1,1} ;
2. 1 representing the bit pair {0,1} ;
3. -1 representing the bit pair {0,0} ;
4. -3 representing the bit pair {1,0} ;
The transmitted code-division multiplexed block may be mathematically expressed as
X
2 X
1
x[n] = bk [m]ck [n − 4m], n = 0, 1, ..., 7
k=1 m=0
4 , −2, 2, −4, −4, −2, 2, 4}

x[n] = {|{z}
↑
all work in arriving at your answer. Note that code 1 is orthogonal to code 2 and that the
two users are synchronized.
2
The analog signal xa (t) is reconstructed from its samples x[n] = xa (nTs ) according to the
following equation

∞
X sin π Tts
xr (t) = x[k] g(t − kTs ) where: g(t) =
k=−∞ π Tts
Samples of the reconstructed signal at L times the original sampling rate may be obtained
via the following discrete-time system.
x [n] xr [n] = x n Ts ( )
r L
L h [n]
Figure 1.
Your primary task in this problem is to determine the appropriate filter impulse response
h[n] for different values of L so that the output of the system above is what you would have
obtained if you had sampled the reconstructed signal at L times the original sampling rate
as specified. NOTE: Correct answer for h[n] is different for each part (for each value of L.)
(a) For the case of L = 1, write a closed-form expression for the filter h[n]. Plot the
magnitude of the DTFT of h[n], H(ω), over −π < ω < π.
(b) For the case of L = 2, write a closed-form expression for the filter h[n]. Plot the
magnitude of the DTFT of h[n], H(ω), over −π < ω < π. The case of L = 2 in Figure
1 may be efficient implementated as in the block diagram in Figure 2 below.
h [n] = h [2n]
0
xr [n] = x n Ts
( )
r 2
x [n] interleaver
h [n] = h [2n+1]
1
Figure 2.
Using your answer for h[n] for this part (b), do the following:
(i) Write an expression for h0 [n] = h[2n]. Plot the magnitude of the DTFT of h0 [n],
|H0 (ω)|, over −π < ω < π.
(ii) Write an analytical expression for h1 [n] = h[2n + 1]. Plot the magnitude of the
DTFT of h1 [n], |H1 (ω)|, over −π < ω < π.
(c) For the case of L = 4, write a closed-form expression for the filter h[n]. Plot the
3

The analog signal xa (t) is reconstructed from its samples x[n] = xa (nTs ) according to the
following equation
∞ ( |t|
X t 1− for |t| < Ts
xr (t) = x[k] g(t − kTs ) where: g(t) = Λ = Ts
k=−∞ Ts 0 for |t| > Ts
Samples of the reconstructed signal at L times the original sampling rate may be obtained
via the following discrete-time system.
x [n] xr [n] = x n Ts ( )
r L
L h [n]
Figure 1.
Your primary task in this problem is to determine the appropriate filter impulse response
h[n] for different values of L so that the output of the system above is what you would have
obtained if you had sampled the reconstructed signal at L times the original sampling rate
as specified. NOTE: Correct answer for h[n] is different for each part (for each value of L.)
(a) For the case of L = 1, write a closed-form expression for the filter h[n]. Plot the
(b) For the case of L = 2, write a closed-form expression for the filter h[n]. Plot the
magnitude of the DTFT of h[n], H(ω), over −π < ω < π. The case of L = 2 in Figure
1 may be efficient implementated as in the block diagram in Figure 2 below.
h [n] = h [2n]
0
xr [n] = x n Ts
( )
r 2
x [n] interleaver
h [n] = h [2n+1]
1
Figure 2.
|H0 (ω)|, over −π < ω < π.
(c) For the case of L = 4, write a closed-form expression for the filter h[n]. Plot the
4
Exam 1 Live: Wed., Sept. 22, 2004
Cover Sheet
Calculators allowed.

1. LTI Systems: Properties, 35
Transfer Functions, Frequency Response
DT Autocorrelation, Cross-Correlation
2. Interconnection of LTI Systems: 35
3. Sampling Theory, CTFT-DTFT Relationship, 30
DT Frequency Selective Filtering
1

Consider a DT LTI system whose impulse response is
h[n] = (j)n u[n]
(a) Is the system BIBO stable? Substantiate your answer mathematically.
(b) Find a bounded input signal x[n] that produces an unbounded output from this system.
(c) Find the system transfer function H(z) of this system and draw the pole-zero diagram.
(d) Write the difference equation for the LTI system having the impulse response above.
(e) Plot a rough sketch of the magnitude of the DTFT of h[n], |H(ω)|, over −π < ω < π,
showing as much detail as possible.
(f) Consider the input signal below which is a sum of sinewaves “turned on” for all time.
x[n] = 1 + (−j)n + (−1)n
Write a closed-form expression for the corresponding output y[n]. ALSO, plot a rough
sketch of the magnitude of the DTFT of y[n], |Y (ω)|, over −π < ω < π, showing as
much detail as possible.
(g) Let y[n] denote the output obtained with the input signal below relative to the LTI
system with impulse response above.
x[n] = (0.5)n u[n]
Write a closed-form expression for the cross-correlation ryx [`] between the output y[n]
and the input x[n].
2

Consider the causal, second-order LTI system described by the difference equation below.
y[n] = 0.25 y[n − 2] + x[n] − x[n − 2]
(a) Find the system transfer function H(z) of this system and draw the pole-zero diagram.
(b) Plot the magnitude, |H(ω)|, of the DTFT of the impulse response of the system over
−π < ω < π, showing as much detail as possible. In particular, explicitly point out if
there are any values of ω for which |Hi(ω)| is exactly zero.
(c) Consider implementing the second-order difference equation above as two first-order
systems (one pole each) in parallel as shown in the diagram.
y [n]
1
h [n]
1
x[n] + y[n]
y [n]
2
h [n]
2
The upper first-order system has impulse response h1 [n] and is described by the differ-
ence equation
(1) (1) (1)
y1 [n] = a1 y1 [n − 1] + b0 x[n] + b1 x[n − 1]
The lower first-order system has impulse response h2 [n] and is described by the differ-
ence equation
(2) (2) (2)
y2 [n] = a1 y2 [n − 1] + b0 x[n] + b1 x[n − 1]
(i) (i) (i)
Determine the numerical values of a1 , b0 , and b1 , i = 1, 2 – six values total. NOTE:
Each of the two first-order systems has a single non-zero zero and a single non-zero
(1)
pole. In order to get a unique answer, you are given that b0 = 12 , and you must find
(1) (1) (2) (2) (2)
5 numerical values: a1 , b1 , a1 , b0 , and b1 .
(d) For EACH of the two first-order systems, i = 1, 2, do the following:

(ii) State and plot the region of convergence for Hi (z).
(iii) Determine the DTFT of hi [n] and plot the magnitude |Hi(ω)| over the interval
−π < ω < π showing as much detail as possible.
3
Problem 3. [30 points] ( )( )

sin(4t) sin(8t)
Consider the continuous-time signal x(t) = . A DT signal is obtained
πt πt
2π
by sampling x(t) according to x[n] = x(nTs ) for Ts = .
36
(i) Plot the magnitude of the DTFT of x[n] over −π < ω < π.
( )
n sin( π3 n)
(ii) x[n] is passed through a DT linear system with impulse response ha [n] = (−1)
πn
yielding the output ya [n]. Plot magnitude of the DTFT of ya [n].
( )
4π sin( 2π
9
n)
(iii) x[n] is passed through a DT linear system with impulse response hb [n] = 2 cos n
9 πn
yielding the output yb [n]. Plot magnitude of the DTFT of yb [n] over −π < ω < π.
sin( 2π
9
n)
(iv) x[n] is passed through a DT linear system with impulse response hc [n] =
πn
yielding the output yc [n]. Plot magnitude of the DTFT of yc [n] over −π < ω < π.
4
EE538 Digital Signal Processing I
Exam 1 Session 12. Live: Thurs, Sept. 28, 2006
Cover Sheet
Coverage: Chaps 1-5, 10.2,10.3
Calculators NOT allowed.
This test contains TWO problems.
1
(a) Consider the symmetric sequence below which is one for −2 ≤ m ≤ 2 and zero for
|m| > 2. Is this is a valid autocorrelation sequence? Justify your answer.
rxx [m] = u[m + 2] − u[m − 3]
(b) Consider the symmetric sequence below. Is this is a valid autocorrelation sequence?
Justify your answer.
rxx [m] = (3 − |m|)(u[m + 2] − u[m − 3])
(c) Let rxx [m] denote the autocorrelation sequence for the DT signal x[n]. Let y[n] =
x[n − no ], where no is an integer. Let ryy [m] denote the autocorrelation sequence for
the DT signal y[n]. Derive an expression relating ryy [m] and rxx [m]. That is, how is
ryy [m] related to rxx [m]?
(d) Let rxx [m] denote the autocorrelation sequence for the DT signal x[n]. Let y[n] =
ej(ω0 n+θ) x[n], where ωo is some frequency and θ is some phase value. Let ryy [m] denote
the autocorrelation sequence for the DT signal y[n]. Derive an expression relating
ryy [m] and rxx [m]. That is, how is ryy [m] related to rxx [m]?
(e) Consider that the signal x[n] below, where a = 12 , is the input to the LTI system
described by the difference equation in equation (2) below.
1
x[n] = an u[n] − an−1 u[n − 1] (1)
a
1
y[n] = y[n − 1] + x[n] (2)
4
(i) Determine a closed-form analytical expression for the auto-correlation rxx [m] for
x[n], when a = 12 . (Hint: examine the Z-Transform of x[n].)
(ii) Determine a closed-form analytical expression for the cross-correlation ryx [m] be-
tween the input and output.
(iii) Determine a closed-form analytical expression for the auto-correlation ryy [m] for
the output y[n].
2

Consider the transmission of a pulse amplitude-modulated signal
∞

s(t) = x[k]p(t − kTs )
k=−∞
where x[n] are the information-bearing symbols being transmitted which may be viewed as
a discrete-time sequence. In binary phase-shift keying, x[n] is either “+1” or “-1” for all n.
1/Ts is the symbol rate and p(t) is the pulse symbol waveform below

πt
p(t) = cos (u(t + Ts ) − u(t − Ts ))
2Ts
At the receiver, s(t) arrives by both a direct path and a multipath reflection having the
same strength and phase as the direct path but at a delay of Ts . Denoting continuous time
convolution as ∗, the received signal, xr (t), may be modeled as:
xr (t) = s(t) ∗ g(t)
where g(t) is described below using δ(t) to denote the Dirac Delta function.
g(t) = δ(t) + δ (t − Ts ) (3)
Samples of the received signal at L times the symbol rate rate may be obtained via the
following discrete-time system.
x [n] xr [n] = x n Ts
( )
r L
L h [n]
Figure 1.
Your primary task in this problem is to determine the appropriate filter impulse response
h[n] for different values of L so that the output of the system above is what you would have
obtained if you had sampled the received signal at L times the symbol rate as specified.
NOTE: Correct answer for h[n] is different for each part (for each value of L.)
(a) For the case of L = 1, write a closed-form expression for the filter h[n] (the nonzero
values). Plot the magnitude of the DTFT of h[n], H(ω), over −π < ω < π.
(b) For the case of L = 2, write a closed-form expression for the filter h[n] (the nonzero
values). Plot the magnitude of the DTFT of h[n], H(ω), over −π < ω < π. The case
of L = 2 in Figure 1 may be alternatively implemented as in the block diagram in
Figure 2 on the top of the next page.
3
h [n] = h [2n]
0
xr [n] = x n Ts
( )
r 2
x [n] interleaver
h [n] = h [2n+1]
1
Figure 2.
|H0 (ω)|, over −π < ω < π.
(c) For the case of L = 3, write a closed-form expression for the filter h[n].
4
EE538 Digital Signal Processing I Fall 2009
Exam 1 Tuesday, Oct. 6, 2009
Cover Sheet
Coverage: Chapters 1-5.
This test contains two problems.
All work should be done in blue books.
Do not return the exam itself; just your blue book.

2. DT Autocorrelation, Cross-Correlation 45
Correlation in terms of Convolution
1
Problem 1. [55 points] Consider a DT LTI system with the transfer function below, with
a single pole at z = a, where a is real-valued and satisfies 0 < a < 1, and a single zero at
z = 1/a. Consider computing the frequency response of this system graphically.
(z − a1 )
H(z) = (1)
(z − a)
(a) To this end, first find an expression for the length N(ω) of the vector that connects
the zero at z = 1/a to the point z = ejω on the unit circle, where 0 ≤ ω ≤ π.
(b) Next, find an expression for the length D(ω) of the vector that connects the pole at
z = a to the point z = ejω on the unit circle, where 0 ≤ ω ≤ π.
(c) Show that the ratio of the answer to (a) to the answer to (b), N(ω)/D(ω), is a constant
that does not depend on the frequency ω.
(d) Determine a closed-form expression for the impulse response h[n] of this system (in
terms of a).
(e) Determine a closed-form expression for the autocorrelation sequence rhh [ℓ] for the im-
pulse response h[n] of the system. Simplify as much as possible and show all work.
(f) Draw a block diagram for this system that uses only a single delay unit.
1
y[n] = ay[n − 1] + x[n] − x[n − 1] (2)
a
(g) The signal below is input to the system above. (i) Plot the energy density spectrum
for the output y[n] AND (ii) find the total energy in y[n] (in terms of a).
sin( 3π
( )
4
n)
x[n] = (3)
πn
2
(a) Let x[n] and y[n] be real-valued sequences both of which are even-symmetric: x[n] =
x[−n] and y[n] = y[−n]. Under these conditions, prove that rxy [ℓ] = ryx [ℓ] for all ℓ.
(b) Express the autocorrelation sequence rzz [ℓ] for the complex-valued signal z[n] = x[n] +
jy[n] where x[n] and y[n] are real-valued sequences, in terms of rxx [ℓ], rxy [ℓ], ryx [ℓ],
and ryy [ℓ].
(c) Determine a closed-form expression for the autocorrelation sequence rxx [ℓ] for the signal
x[n] below.
sin( π2 n)
( )
x[n] = (4)
πn
(d) Determine a closed-form expression for the autocorrelation sequence ryy [ℓ] for the signal
y[n] below.
π
( )
n n sin( 2 n)
y[n] = (−1) x[n] = (−1) (5)
πn
(e) Determine a closed-form expression for the autocorrelation sequence rzz [ℓ] for the
complex-valued signal z[n] formed with x[n] and y[n] defined above as the real and
imaginary parts, respectively, as defined below. You must show all work and simplify
as much as possible.
z[n] = x[n] + jy[n] (6)
(f) Plot rzz [ℓ].
3
Exam 1 Friday, Oct. 1, 2010
Cover Sheet
All work should be done in blue books.
Do not return the exam itself; just your blue book.

1
Problem 1. [55 points] Be sure to label each part carefully in your blue book.
(a) Determine and plot the autocorrelation sequence rxx [`] for x[n] below.
x[n] = {1, 1, −1, 1} (1)
(b) Determine and plot the autocorrelation sequence rzz [`] for z[n] defined in terms of x[n]
in Eqn. (1) as
z[n] = x[n − 4]
(c) Determine and plot the autocorrelation sequence rzz [`] for z[n] defined in terms of x[n]
in Eqn. (1) as
z[n] = (−1)n x[n]
(d) Determine the autocorrelation sequence rzz [`] for z[n] defined in terms of x[n] in Eqn.
(1) below. For this part, you need only write out the autocorrelation sequence; you
don’t need to plot it.
π π
z[n] = ej ( 2 n+ 3 ) x[n]
(e) Consider a simple radar example where there are two echoes such that the received
signal may be expressed in terms of x[n] in Eqn. (1) as
y[n] = x[n − 4] − x[n − 8]
Compute and plot the cross-correlation sequence ryx [`] given the input sequence x[n]
defined in Eqn. (1) above.
(f) Repeat for the case where the received signal may be expressed in terms of x[n] as
y[n] = x[n − 4] + x[n − 6]
2
Consider N = 4 information symbols, denoted {b0 , b1 , b2 , b3 }. A sequence of length N = 4
is created as a sum of sinewaves as indicated below.
2π
s0 [n] = ej0 4 n {u[n] − u[n − 4]} = {1, 1, 1, 1}
3 2π
X s1 [n] = ej1 4 n {u[n] − u[n − 4]} = {1, j, −1, −j}
x[n] = bk sk [n] where: 2π
k=0 s2 [n] = ej2 4 n {u[n] − u[n − 4]} = {1, −1, 1, −1}
2π
s3 [n] = ej3 4 n {u[n] − u[n − 4]} = {1, −j, −1, j}
The last L = 3 values of x[n] are appended to the beginning to form a sequence of length 7
as
x[n] = {x[−3], x[−2], x[−1], x[0], x[1], x[2], x[3]} (2)

= {x[+1], x[+2], x[+3], x[0], x[1], x[2], x[3]} (3)
Next, x[n] is input to a Linear Time-Invariant (LTI) system with an impulse response of
length L = 3 equal to
h[n] = {h[0], h[1], h[2]}

= {1, 0, j}
to produce an output sequence

y[n] = x[n] ∗ h[n]
of length equal to 7 + 3 − 1 = 9.
y[n] = {y[−3], y[−2], y[−1], y[0], y[1], y[2], y[3], y[4], y[5]} (4)
= {2, −2, 2 + 2j, 2 − 2j, 2 + 2j, −2 + 2j, 2 + 2j, −2j, 2j} (5)
(6)
Here’s your task: given h[n] = {h[0], h[1], h[2]} = {1, 0, j} and
y[n] y[-3] y[-2] y[-1] y[0] y[1] y[2] y[3] y[4] y[5]
= 2 -2 2+2j 2-2j 2+2j -2+2j 2+2j -2j 2j
determine the 4 values of the information symbols {b0 , b1 , b2 , b3 }. You MUST show all work
and explain how you got your answer concisely but with sufficient detail to receive full credit.
3
Exam 1 Monday, Oct. 3, 2011
Cover Sheet
Show your work in the space provided for each problem.
Always simplify your answers as much as possible.

1
(a) Consider x[n] to be a real-valued sequence with autocorrelation rxx [`]. Express the
autocorrelation sequence ryy [`] for y[n] = x[−n] in terms of rxx [`]. Be sure to show
clearly how you arrived at your answer.
2
(b) Determine and plot the autocorrelation sequence rxx [`] for x[n] defined below with
p = 12 . Note: This sequence is used in parts (c) thru (f).
1n o
x[n] = δ[n] + (p2 − 1)pn u[n] (1)
p
3
(c) Determine and plot the autocorrelation sequence rzz [`] for z[n] defined in terms of x[n]
in Eqn. (1) as below.
π
j 2
n+ √π
z[n] = e 2
x[n]
4
(d) Determine and plot the autocorrelation sequence rzz [`] for z[n] defined in terms of x[n]
in Eqn. (1) below.
z[n] = x[2 − n]
5
(e) Consider a simple radar example where there are two echoes such that the received
signal may be expressed in terms of x[n] in Eqn. (1) as
1
y[n] = x[n − 3] + x[n − 5] (2)
2
6
(f) Recall the simple radar example described by Eqn (2). Compute and plot the auto-
correlation sequence for the output ryy [`] given the input sequence x[n] defined in
Eqn. (1). Can echo delays be determined from the autocorrelation of the output ryy [`]?
7
Problem 2. [40 points] Consider the causal DT LTI system described the difference
equation below which is used for all four parts of this problem.
y[n] = j y[n − 1] + x[n] − x[n − 4]
(a) Determine the transfer function H(z) for this system and plot the pole-zero diagram.
Show the region of convergence.
8
(b) Determine the impulse response of the system h[n]. You can either list the value of
h[n] for each n or write a closed-form expression for h[n] that works for all n.
9
(c) Determine a closed-form expression for the frequency response of the system H(ω)
equal to the DTFT of h[n]. Plot the magnitude |H(ω)| over −π < ω < π. You
can provide a rough sketch, but you need to clearly indicate the frequencies where
|H(ω)| = 0 and the frequency where |H(ω)| reaches its peak value.
10
(d) Determine a closed-form expression for the output y[n] when the input is the following
sum of sinewaves turned-on forever:
x[n] = 1 + 2(−j)n + 3(j)n + 4(−1)n
11
Exam 1: Short Version to Keep When You Leave
(a) Note: System 1 and System 2 defined in parts (b) and (c), respectively, are NOT
in parallel, but they do have the same input equal to x[n] below with p = 43 .
1n o
x[n] = δ[n] + (p2 − 1)pn u[n] (1)
p
Determine & plot the autocorrelation sequence rxx [ℓ] for x[n] defined above with
p = 34 .
(b) Consider System 1 below:
System 1: y1 [n] = x[n] + x[n − 1] − x[n − 2] + x[n − 3]
x[n] y [n]
1
h [n]
1
(i) Determine the impulse response h1 [n]. You can write it in sequence form.
(ii) Determine the autocorrelation of the impulse response. Do a stem plot of
rh1 h1 [ℓ].
(iii) For the input signal in Equation (1) above with p = 34 , determine the
autocorrelation of the output y1 [n]. Do a stem plot of ry1 y1 [ℓ].
(c) Consider System 2 below:
System 2: y2 [n] = x[n] + x[n − 1] + x[n − 2] − x[n − 3]
y [n]
x[n] h [n] 2
2
(i) Determine the impulse response h2 [n]. You can write it in sequence form.
(ii) Determine the autocorrelation of the impulse response. Do a stem plot of
rh2 h2 [ℓ].
(iii) Determine the autocorrelation of the output y2 [n]. Do a stem plot of ry2 y2 [ℓ].
(d) Using your answers from parts (b) and (c), do a stem plot of r[ℓ] = ry1 y1 [ℓ] + ry2 y2 [ℓ].
2
Problem 2. [20 points] System 1 and System 2 are as defined in parts (b) and (c) of
Problem 1, respectively, but now they are connected in parallel, such that the overall
output is
y[n] = y1 [n] + y2 [n]
(a) Determine the overall response h[n] of the system. Do a stem plot of h[n].
(b) Plot the magnitude of the frequency response of the system |H(ω)| over −π < ω < π.
(c) Determine a closed-form expression for the output y[n] when the input is x[n] below.
x[n] = 1 + (j)n + (−j)n + (−1)n
(a) The autocorrelation of x[n] may be computed in terms of convolution as
rxx [ℓ] = x[ℓ] ∗ x∗ [−ℓ]
Which of the three main properties of convolution discussed in class is key to proving
that the autocorrelation of y[n] = x∗ [−n] is the same as the autocorrelation of x[n].
Write your one-word answer directly below.
(b) Is the convolution of two all-pass filters/sequences also an all-pass filter/sequence?

Prove your answer with a short mathematical analysis in the space provided on the
next few blank pages.
(c) Let x[n] be the input signal described below with p = 12 .
1n o
x[n] = δ[n] + (p2 − 1)pn u[n] (2)
p
This signal is input to the system below described the difference equation below
2 2
y[n] = y[n − 1] − x[n] + x[n − 1]
3 3
(i) Draw the pole-zero diagram for this system. Is the system stable?
(ii) Determine the autocorrelation of the output y[n]. Do a stem plot of ryy [ℓ].
(d) For the same input signal and same system described in part (c), determine the
cross-correlation ryx [ℓ] between the input and output. Just provide a closed-form
expression for ryx [ℓ]; you don’t need to plot it.
3
NAME:
Exam 1 Friday, Sept. 27, 2013
Cover Sheet
Write your name on this and every page

1. Frequency Response and Interconnection 50
of LTI Systems, Pole-Zero Diagrams
1
NAME:
Problem 1. [50 points] System 1 and System 2 defined in parts (a) and (b), respectively,
will eventually be connected in parallel. We first analyze them individually.
(a) Consider System 1 below:
System 1: y1 [n] = 0.9j y[n − 1] − 0.9j x[n] − x[n − 1]
(i) Determine the Transfer Function for System 1, denoted H1 (z). H1 (z) is the
Z-Transform of the impulse response, h1 [n], for System 1, although you can find
H1 (z) anyway you like. Identify the poles and zeros, and do a pole-zero plot.
(ii) Determine the frequency response of System 1, denoted H1 (ω). H1 (ω) is the
DTFT of the impulse response, h1 [n], for System 1, although you can find H1 (ω)
anyway you like. Plot the magnitude |H1 (ω)| over −π < ω < π.
(iii) Determine the autocorrelation of the impulse response h1 [n]. Do a stem plot of
rh1 h1 [ℓ].
(b) Consider System 2 below:
System 2: y1 [n] = −0.9j y[n − 1] + 0.9j x[n] − x[n − 1]
(i) Determine the Transfer Function for System 2, denoted H2 (z). H2 (z) is the
Z-Transform of the impulse response, h2 [n], for System 2, although you can find
H2 (z) anyway you like. Identify the poles and zeros, and do a pole-zero plot.
(ii) Determine the frequency response of System 2, denoted H2 (ω). H2 (ω) is the
DTFT of the impulse response, h2 [n], for System 2, although you can find H2 (ω)
anyway you like. Plot the magnitude |H2 (ω)| over −π < ω < π.
(iii) Determine the autocorrelation of the impulse response h2 [n]. Do a stem plot of
rh2 h2 [ℓ].
(c) The overall system is formed from connecting System 1 and System 2 in parallel.
(i) Determine the Transfer Function for the overall system, denoted H(z). H(z) is
the Z-Transform of the impulse response, h[n], for the parallel combination of
Systems 1 and 2. You can find H(z) anyway you like. Identify the poles and
zeros for the overall system, and do a pole-zero plot.
(ii) Determine the frequency response of the overall system, denoted H(ω). H(ω) is
the DTFT of the impulse response, h[n], for the overall system; you can find
H(ω) anyway you like. Plot the magnitude |H(ω)| over −π < ω < π. showing as
much detail as possible. Point out any frequencies for which H(ω) = 0.
(iii) Determine the output of the overall system, y[n], when the input is the DT
signal below. The overall system is the parallel combination of Systems 1 and 2.
π

x[n] = 3 + 2 cos n + (−1)n −∞<n<∞
2
2
NAME: Page intentionally blank for Problem 1 Work
3
4
5
NAME:
Problem 2. [50 points] Let x[n] and y[n] be DT signals with autocorrelations and
cross-correlations defined in terms of convolution as below.
rxx [ℓ] = x[ℓ]∗x∗ [−ℓ] ryy [ℓ] = y[ℓ]∗y ∗[−ℓ] rxy [ℓ] = x[ℓ]∗y ∗ [−ℓ] ryx [ℓ] = y[ℓ]∗x∗ [−ℓ] (1)
(a) Consider the case where x[n] and y[n] are both causal, finite-length signals of
duration N. That is, x[n] and y[n] are both only nonzero for n = 0, 1, ..., N − 1. A
concatenated signal of length 2N is formed as below:
z[n] = x[n] + y[n − N] (2)
Express the autocorrelation, rzz [ℓ], for z[n] = x[n] + y[n − N] in terms of rxx [ℓ], ryy [ℓ],
rxy [ℓ], and ryx [ℓ].
(b) Now, consider the specific case where x[n] and y[n] form a complementary pair of
+1’s and -1’s such that
rxx [ℓ] + ryy [ℓ] = 2Nδ[ℓ] (3)
Simplify your answer for rzz [ℓ] in part (a) for this special case.
(c) Specifically, consider the Barker codes of length 4
x[n] = {1, 1, 1, −1} y[n] = {1, 1, −1, 1} (4)
such that z[n] = x[n] + y[n − N] is:
z[n] = {1, 1, 1, −1, 1, 1, −1, 1} (5)
Determine the autocorrelation rzz [ℓ] for z[n] using the results that you derived above
(you can compare to a direct calculation of rzz [ℓ] to check your answer.)
(d) Now, we repeat the steps above for the case where the concatenated signal of length
2N is formed as below with a negative sign on the second term.
w[n] = x[n] − y[n − N] (6)
Express the autocorrelation, rww [ℓ], for w[n] = x[n] − y[n − N] in terms of rxx [ℓ],
ryy [ℓ], rxy [ℓ], and ryx [ℓ].
(e) Again, consider the specific case where x[n] and y[n] form a complementary pair of
+1’s and -1’s such that
rxx [ℓ] + ryy [ℓ] = 2Nδ[ℓ] (7)
Simplify your answer for rww [ℓ] in part (d) for this special case.
(f) Again, let x[n] = {1, 1, 1, −1} and y[n] = {1, 1, −1, 1} be Barker codes of length 4,
such that
w[n] = x[n] − y[n − N] = {1, 1, 1, −1, −1, −1, 1, −1} (8)
Determine the autocorrelation rww [ℓ] for z[n] using the results that you derived above
(g) Sum your answers to parts (b) and (d) to form rvv [ℓ] = rzz [ℓ] + rww [ℓ], Do a stem plot
of the sum rvv [ℓ]. Compare against the sum of your answers to (c) and (e).
6
7
8
9
NAME:
Cover Sheet

and their Related Properties
3. LTI Systems: Expressing Cross-Correlation 35
in terms of Convolution
1
NAME:
(a) Consider System 1 with impulse response h1 [n] below, where p1 = 0.8 (in fractional
form p1 = 45 ):
1 n o
System 1: h1 [n] = δ[n] + (p21 − 1)pn1 u[n]
p1
x[n] y [n]
1
h [n]
1
(i) Draw a pole-zero diagram for this system.

(ii) Plot the magnitude of the frequency response |H1 (ω)| over −π < ω < π.
(b) Consider System 2 with impulse response h2 [n] below, where p2 = −0.8 (in fractional
form p2 = − 54 ):
−1 n o
p2
y [n]
x[n] h [n] 2
2

(c) Consider Systems 1 and 2 to be connected in parallel. Plot the magnitude of the
frequency response |H(ω)| of the parallel combination of System 1 and System 2 over
−π < ω < π.
(d) Determine the overall output y[n] of the parallel combination of System 1 and
System 2 when the common input is:
x[n] = 3 + 2 cos(πn)
2
3
4
5
NAME:
(a) Determine the autocorrelation rx0 x0 [ℓ] of the length-4 sequence x0 [n] below, which is
written two different ways. Write your answer in sequence form indicating where the
value for ℓ = 0 is located.
x0 [n] = u[n] − u[n − 4] = {1, 1, 1, 1}
(b) Determine the autocorrelation rx1 x1 [ℓ] of the length-4 sequence x1 [n] below. Write
your answer in sequence form indicating where the value for ℓ = 0 is located.

π
j 2
n+ √π
x1 [n] = e 2
{u[n] − u[n − 4]}
(c) Determine the autocorrelation rx2 x2 [ℓ] of the length-4 sequence x2 [n] below. Write
x2 [n] = ejπ(n−2) {u[n − 2] − u[n − 6]}
(d) Determine the autocorrelation rx3 x3 [ℓ] of the length-4 sequence x3 [n] below. Write

π
−j 2
n+ √π
x3 [n] = e 2
{u[n] − u[n − 4]}
(e) Sum your answers to parts (a) thru (d) to form the sum below. Do a stem plot of
rxx [ℓ].
rxx [ℓ] = rx0 x0 [ℓ] + rx1 x1 [ℓ] + rx2 x2 [ℓ] + rx3 x3 [ℓ]
6
7
8
NAME:
x1 [n] = {1, −1, 1, 1} = δ[n] − δ[n − 1] + δ[n − 2] + δ[n − 3]
(b) Determine the autocorrelation rx1 x1 [ℓ] of the length-4 sequence x1 [n] below, which is
x2 [n] = {−1, 1, 1, 1} = −δ[n] + δ[n − 1] + δ[n − 2] + δ[n − 3]
(c) The sequence x1 [n] defined above is input to the system described by the simple
difference equation below. Do a stem plot of the cross-correlation ry1 x1 [ℓ] between the
output and input.
y1 [n] = x1 [n − 4] + x1 [n − 6]
(d) The sequence x2 [n] defined above is input to the same system described by the simple
output and input.
y2 [n] = x2 [n − 4] + x2 [n − 6]
(e) Sum your answers to parts (c) thru (d) to form the sum below. Do a stem plot of
ryx [ℓ].
ryx [ℓ] = ry1 x1 [ℓ] + ry2 x2 [ℓ]
9
10
11
12
NAME:
Exam 1 Monday, Sept. 28, 2015
Cover Sheet

2. LTI Systems: All-Pass Filters 20
1
NAME:
(a) Consider two LTI systems connected in SERIES. System 1 has impulse response h1 [n]
below, where p1 = 0.5 (in fractional form p1 = 21 )
1 n o
p1
System 2 has impulse response h2 [n] below, where p2 = −0.5
1 n o
p2
x[n] y [n]
h [n] h [n]
1 2
(a) Determine a closed-form expression for the impulse response, h[n], of the overall
system. Write how h[n] is related to h1 [n] and h2 [n], and then show all work in
determining your final answer. You may want to make use of the convolution result
below. Note: you can solve the rest of this problem (except the extra credit) without
the answer to part (a.)
α β
αn u[n] ∗ β n u[n] = αn u[n] + β n u[n]
α−β β−α
(b) Determine the Z-Transform of the overall system. Draw the pole-zero diagram.
(c) Plot the magnitude of the frequency response |H(ω)| of the overall system over
−π < ω < π. Explain your answer.
(d) Write the difference equation for the overall system.
(e) Determine the output y[n] of the overall system when the input is the sum of
sinewaves (turned on forever) below
x[n] = 4 + 3(j)n + 2(−j)n + (−1)n
(f) Extra Credit. Consider creating a new impulse response from your answer for part
(a) as
g[n] = h[2n + 1] where: h[n] is the overall impulse response of the system
That is, g[n] is formed by keeping only the values of h[n] for odd values of time, i.e.,
throwing the values of h[n] for even values of time. Compute the energy of g[n]:
∞
X
Eg = g 2[n]
n=−∞
2
3
4
5
Problem 2. (a)
(a) Consider h1 [n] and h2 [n] to be two distinct all-pass filters (p1 6= p2 ) with respective
impulse responses below. Is the sum h[n] = h1 [n] + h2 [n] an all-pass filter for any and
all values of p1 and p2 (p1 6= p2 ) ?? Explain your answer. Your explanation is much
more important than your answer.
1 n o
h1 [n] = δ[n] + (p21 − 1)pn1 u[n] (1)
p1
1 n o
h2 [n] = δ[n] + (p22 − 1)pn2 u[n] (2)
p2
6
Problem 2.(b)
(b) Consider h[n] to be an all-pass filter with respective impulse response below.
1n o
h[n] = δ[n] + (p2 − 1)pn u[n] (3)
p
Is the product
g[n] = ejωo n h[n]
an all-pass filter for any and all values of the frequency ωo ? Explain your answer.
Your explanation is much more important than your answer.
7
Problem 3. Note: this problem is different from a similar problem on last year’s exam in
that the two sequences, x1 [n] and x2 [n], are of different lengths.
x1 [n] = {1, 1, −1} = δ[n] + δ[n − 1] − δ[n − 2]
(b) Determine the autocorrelation rx2 x2 [ℓ] of the length-5 sequence x2 [n] below, which is
x2 [n] = {1, 1, 1, −1, 1} = δ[n] + δ[n − 1] + δ[n − 2] − δ[n − 3] + δ[n − 4]
(c) The sequence x1 [n] defined above is input to the system described by the simple
output and input.
y1 [n] = 4x1 [n − 3] + x1 [n − 4]
(d) The sequence x2 [n] defined above is input to the same system described by the simple
output and input.
y2 [n] = 4x2 [n − 3] + x2 [n − 4]
(e) Sum your answers to parts (c) and (d) to form the sum below. Do a stem plot of
ryx [ℓ].
ryx [ℓ] = ry1 x1 [ℓ] + ry2 x2 [ℓ]
8
9
10
11
NAME:
Cover Sheet

1
NAME:
(a) Consider System 1 characterized by the difference equation

9 9
System 1: y1 [n] = − y1 [n − 2] + x[n] + x[n − 2]
16 16
x[n] y [n]
1
h [n]
1

(b) Consider System 2 with impulse response h2 [n] below:

9
System 2: y2 [n] = − y2 [n − 2] + x[n] + x[n − 2]
16
y [n]
x[n] h [n] 2
2

(c) Determine the overall output y2 [n] of System 2 when the input is:
π

x[n] = 3 + 2 cos n + cos(πn)
2
2
3
4
5
NAME:
value for ℓ = 0 is located. Briefly indicate which properties of autocorrelation you use
to solve each part.
x0 [n] = {1, 2, 3, 4} = δ[n] + 2δ[n − 1] + 3δ[n − 2] + 4δ[n − 3]
(b) Determine the autocorrelation rx1 x1 [ℓ] of the sequence x1 [n] below defined in terms of
x0 [n] in part (a). Write your answer in sequence form indicating where the value for
ℓ = 0 is located.
x1 [n] = ej ( 2 (n−2)) x0 [n − 2]
π
(c) Determine the autocorrelation rx2 x2 [ℓ] of the sequence x2 [n] below defined in terms of
ℓ = 0 is located. √
x2 [n] = ej(πn+ π) x0 [n]
(d) Determine the autocorrelation rx3 x3 [ℓ] of the sequence x3 [n] below defined in terms of
ℓ = 0 is located.
π
x3 [n] = e−j 2 n x∗0 [−n]
(e) Sum your answers to parts (a) thru (d) to form the sum below. Do a stem plot of
rxx [ℓ].
rxx [ℓ] = rx0 x0 [ℓ] + rx1 x1 [ℓ] + rx2 x2 [ℓ] + rx3 x3 [ℓ]
6
7
8
NAME:
(a) Determine the autocorrelation rxx [ℓ] of the length-4 sequence x[n] below, which is
x[n] = {1, −1, 1, 1} = δ[n] − δ[n − 1] + δ[n − 2] + δ[n − 3]
(b) The sequence x[n] defined above is input to the system described by the simple
difference equation below. Determine and write an equation for the cross-correlation,
ryx [ℓ], between the output and input in terms of the autocorrelation of the input,
rxx [ℓ].
y[n] = 4x[n − 3] + x[n − 4]
(c) Compute the numerical values of the cross-correlation ryx [ℓ] between the output and
input and do a stem plot OR write out the values in sequence form clearly indicating
the point corresponding to ℓ = 0.
(d) What is the value of ryx [ℓ] at ℓ = 3? What is the value of ryx [ℓ] at ℓ = 4? Does the
cross-correlation ryx [ℓ] exhibit a peak at each of the two delays? Briefly explain.
9
10
11
12
NAME:
Cover Sheet

DT Autocorrelation, Cross-Correlation
1
NAME:
Problem 1. [55 points] Consider a DT System characterized by the difference equation
π
below where j = ej 2 .
System: y[n] = j y[n − 1] + x[n] − x[n − 4] (1)
(a) Draw a pole-zero diagram for this system.
(b) Plot the magnitude of the frequency response |H(ω)| over −π < ω < π.
(c) Determine the overall output y[n] of the System when the input is the sum of
infinite-length sine waves below
π √ π
x[n] = 3 + 2ej 2 n + 2e−j 2 n + ejπn (2)
(d) Determine the impulse response of the system, h[n]. Write the values out in sequence
form. Hint: the impulse response, h[n], is finite-length.
(e) For this same system, write an equivalent difference equation that only depends on
the current input and a finite number of past inputs.
(f) Determine the autocorrelation of the impulse response, rhh [ℓ] = h[ℓ] ∗ h∗ [−ℓ]. Write
the values out in sequence form. Hint: You can determine the autocorrelation any
way you want to, but it may be helpful to first write the impulse response in the form
h[n] = ejωo n {u[n] − u[n − N]} (determine ωo and N) and then use a property of
autocorrelation derived in class.
(g) Determine the cross-correlation, ryx [ℓ], between the output, y[n], and the input, x[n],
when the input is the signal below, where p = 12 .
1n o
x[n] = δ[n] + (p2 − 1)pn u[n] (3)
p
(h) Determine the output, y[n], only for the values n = 0, 1, 2, 3, when the input, x[n], is
the FINITE-length sinewave below, which has only 8 nonzero values (two periods.)
That is, x[n] is only nonzero over the range −4 ≤ n ≤ 3. VIP, you only need to
determine y[n] for four values of n. Write the output value for each these four DT
time instants, n = 0, 1, 2, 3, in sequence form OR you can write an expression.
π
x[n] = ej 2 n {u[n + 4] − u[n − 4]} (4)
2
3
4
5
NAME: Problem 2. [45 points] Consider a complex-valued sequence
sin( π2 n) sin( π2 n)
( ) ( )
z[n] = x[n] + jy[n] = + j (−1)n (5)
πn πn
where, as indicated, the real and imaginary parts are defined below.
sin( π2 n)
( )
x[n] = (6)
πn
sin( π2 n)
( )
n n
y[n] = (−1) x[n] = (−1) (7)
πn
(a) Determine the cross-correlation, ryx [ℓ], between y[n] and x[n]. Hint: Recommend
working in the frequency domain; same for part (b) below.
(b) Determine an expression for the autocorrelation sequence

rzz [ℓ] = z[ℓ] ∗ z ∗ [−ℓ] = rxx [ℓ] + ryy [ℓ] + j(ryx [ℓ] − rxy [ℓ]).
(c) Determine a closed-form expression for the autocorrelation sequence rww [ℓ] for the
signal w[n] below. Is your answer the same as your answer to part (b)? Explain.
π
w[n] = ej 2 n z[n] (8)
(d) Consider a DT system with impulse response below where the input is z[n] above.
Denote the output as ỹ[n], i.e., ỹ[n] = z[n] ∗ h[n]. Determine the cross-correlation,
rỹz [ℓ] = ỹ[n] ∗ z ∗ [−n], between the input and output for this system.
h[n] = 4δ[n − 3] − 3δ[n − 4] (9)
6
7
8

DSP Exam 1 Solutions and Key Information

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

DSP Exam 1 Solutions and Key Information

Transféré par

Droits d'auteur :

Formats disponibles

EE538 Digital Signal Processing I Exam 1 Fall 2000

Examination Date: Session 11 (Live: 15 September 2000).

Problem 1. [30 points]

User 1’s code: c1 [n] = {1, −1, 1, −1}

User 2’s code: c2 [n] = {1, −1, −1, 1}

User 1’s two info. bits: b1 [n] = {b1 [0], b1 [1]}

User 2’s two info. bits: b2 [n] = {b2 [0], b2 [1], }

Problem 2. [40 points]

(i) Plot the pole-zero diagram.

Problem 3. [30 points] Consider the transmission of a pulse amplitude-modulated signal

y(t) = x(t) ∗ g(t)

g(t) = δ(t) + ejθ δ (t − To ) (1)

Prob. No. Topic(s) Points

y1 (n) ∗ h2 (n) − y2 (n) ∗ h1 (n) = 0

Problem 2. [40 points]

This implementation is seen to require N multiplications and N − 1 additions per output

y[n] = a1 y[n − 1] + x[n] − x[n − D]

(b) Consider the specific case of ` = 2 and N = 4:

Problem 3. [30 points]

x(n) Ideal LPF: ωc = π/2

h [n] = h[2n] y [n] = y [2n]

h [n] = h[2n+1] y [n] = y [1+2n]

Prob. No. Topic(s) Points

User 1’s two info. symbols: b1 [n] = {b1 [0], b1 [1]}

User 2’s two info. symbols: b2 [n] = {b2 [0], b2 [1], }

1. 1+j representing the bit pair {1,1} ;

2. -1+j representing the bit pair {0,1} ;

3. -1-j representing the bit pair {0,0} ;

4. 1-j representing the bit pair {1,0} ;

The transmitted code-division multiplexed block may be mathematically expressed as

x[n] = { 2j , −2j, −2j, 2j, −2j, 2j, 2j, −2j}

Problem 2. [30 points]

(b) For EACH of the two first-order systems, i = 1, 2, do the following:

(i) Plot the pole-zero diagram.

Problem 3. [40 points]

x (t) x [n] w [n] Lowpass Filter h [n] y [n]

such that HLP (ω) = 2 for |ω| ≤ π4 , HLP (ω) = 0 for 3π 4

(b) Given that

Prob. No. Topic(s) Points

Problem 1. [30 points]

User 1’s code: c1 [n] = {1, −1, 1, −1}

User 2’s code: c2 [n] = {1, 1, −1, −1}

User 1’s two info. symbols: b1 [n] = {b1 [0], b1 [1]}

User 2’s two info. symbols: b2 [n] = {b2 [0], b2 [1], }

1. 3 representing the bit pair {1,1} ;

2. 1 representing the bit pair {0,1} ;

3. -1 representing the bit pair {0,0} ;

4. -3 representing the bit pair {1,0} ;

The transmitted code-division multiplexed block may be mathematically expressed as

4 , −2, 2, −4, −4, −2, 2, 4}

Problem 3. [35 points]

Prob. No. Topic(s) Points

Problem 1. [35 points]

h[n] = (j)n u[n]

(a) Is the system BIBO stable? Substantiate your answer mathematically.

x[n] = 1 + (−j)n + (−1)n

x[n] = (0.5)n u[n]

Problem 2. [35 points]

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

(d) For EACH of the two first-order systems, i = 1, 2, do the following:

(i) Plot the pole-zero diagram.

Problem 3. [30 points] ( )( )

Problem 1. [50 points]