SS202B 2018final Sol
SS202B 2018final Sol
1. (15 points) Determine whether or not each of the following statement is true or false. Justify your
answers.
(a) (5 points) For a causal & stable continuous-time LTI system, real parts of its poles are less than
zero.
Answer: True
For a causal system, region of convergence is the right half-plane from the location of poles.
To make it stable, the RoC should include Re{s} = 0 axis. Therefore, all poles should be in
the left half-plane, which means that real parts of poles are less than zero.
(b) (5 points) A system’s frequency response H(jω) can be uniquely determined, provided that poles,
zeros and Region of Convergence (RoC) are specified.
Answer: False
Only with poles, zeros and RoC, the system’s overall gain cannot be specified. For example,
the gain G of H(s) = G N (s)
D(s)
is unknown.
(c) (5 points) When a signal x(t) band-limited in ω ∈ [−ωM , ωM ] is sampled with Nyquist rate, the
original signal x(t) can be perfectly reconstructed from the sampled sequence x[n] using a single
zero-order-hold (ZOH) interpolation filter.
Answer: False
The frequency response of a ZOH interpolation filter is a sinc function, which has nonideal
stopband and passband characteristics as compared to the ideal low-pass filter.
2. (15 points) Let X(jω) be the continuous-time Fourier transform (CTFT) of a continuous-time domain
signal x(t).
(a) (5 points) Express the CTFT of the following signal y(t) in terms of X(jω).
Answer:
1
x3 (t) + 4 = x(t) × x(t) × x(t) + 4 ⇐⇒ X(jω) ∗ X(jω) ∗ X(jω) + 4 × 2πδ(ω)
(2π)2
(b) (5 points) Suppose that X(jω) is band-limited in ω ∈ [−ωM , ωM ]. What is the Nyquist rate to
sample the signal y(t) without aliasing?
Answer: 6ωM
After the three-times convolution of (a), the signal’s bandwidth is enlarged to ±3ωM . There-
fore, the Nyquist rate is 2 × 3ωM = 6ωM
(c) (5 points) A signal is given by x(t) = sin(t)/πt. When this signal x(t) is sampled with a sampling
rate ωs = 2 (rad/sec), evaluate the following sum of the sampled sequence x[n]:
∞
π
X
x[n]e−j 3 n =? (2)
n=−∞
Answer: π1
The Fourier transform of x(t) is
1, |ω| < W
x(t) = sin(W t)/πt ←→ X(jω) =
0, |ω| > W
In this case W=1. Since the CTFT is bounded in ω ∈ [−1, 1], no aliasing occurs when sampled
by ωs = 2.
ωs
After the sampling, the CTFT X(jω) is scaled by T1 = 2π = π1 , and DTFT is given by
1
Xd (ejω/T ) = X(jω)
T
The sum given in the problem is equal to the DTFT of the sampled sequence, and the DTFT
is uniform with magnitude π1 .
3. (15 points) For frequency responses of three causal LTI systems given below, answer to the following
questions.
A. H(jω) = (jω)12 +4
B. h(t) = e−t sin(t)u(t)
d2 y(t) d2 x(t)
C. H(s) : dt2
+ 3 dy(t)
dt
+ 2y(t) = dt2
(a) (3 points) Which one has the lowest damping? (A, B, C)
(b) (3 points) Which one has the highest break frequency? (A, B, C)
(c) (3 points) Which one is stable ? Multiple choices are allowed. (A, B, C)
(d) (3 points) Which one has a high-pass response? Multiple choices are allowed. (A, B, C)
(e) (3 points) Which one has a low-pass response? Multiple choices are allowed. (A, B, C)
EE202B Signals and Systems Final Exam - Page 3 of 7 June 11, 2018
1
A : H(jω) = → poles at ± 2j, ωn = 2, ζ = 0
(jω)2 + 22
1 (−1+j)t (−1−j)t
1 1 1
B : h(t) = e −e u(t) ⇐⇒ H(jω) = −
2j 2j jω + 1 − j jω + 1 + j
1 √ 1
H(jω) = 2
, poles at − 1 ± j, ωn = 2, ζ = √
(jω) + 2(jω) + 2 2
2
(jω)
C : H(jω) = 2
(jω) + 3(jω) + 2
√
(jω)2 √ 3 2
= , poles at − 2, −1, ωn = 2, ζ =
(jω + 2)(jω + 1) 4
4. (15 points) A discrete-time Fourier transform (DTFT) X(ejω ) of a sequence x[n] is shown below:
|X(ejω )|
1
ω
−4π −2π − π2 0 π
2
2π 4π
π
The spectrum X(ejω ) = 0 for 2
≤ |ω| ≤ π, and the phase response ]X(ejω ) = 0 for all ω.
(a) (5 points) Express the inverse Fourier transform y[n] of the following spectrum |Y (ejω )| in terms
of x[n]. ( ]Y (ejω ) = 0 for all frequency ω).
|Y (ejω )|
1
ω
−5π −3π −π 0 π 3π 5π
X(ej(ω−ω0 ) ) ⇐⇒ x[n]ejω0 n ,
X(ej(ω−π) ) ⇐⇒ x[n]ejπn = x[n](−1)n
(b) (10 points) Plot the magnitude of DTFT G(ejω ) of a sequence g[n] = x[2n]. Also indicate its
peak magnitude.
(Hint: This is a decimation process, which is equivalent to the sampling of the compressed CT
signal. Note that the compression - expansion property of CTFT does not directly hold for
DTFT.)
0.5
ω
−4π −2π 0 2π 4π with peak amplitude 0.5.
5. (10 points) Consider a linear phase discrete-time LTI system with frequency response H(ejω ) and real
impulse response h[n]. The group delay function for such a system is defined as
d
τ (ω) = − ]H(ejω ) (3)
dω
where ]H(ejω ) has no discontinuities. Suppose that, for this system,
π
π
|H(ej 3 )| = 4, ]H(ej0 ) = 0, and τ = 1. (4)
4
Determine the output y[n] of the system for the input x[n] = cos 11π
3
n .
π π
Answer: 4 cos 3
n − 3
The DT frequency response is periodic with a period of 2π.
1 j (12−1)π n −j
(12−1)π
π π
Therefore, x[n] = cos 11π
3
n = 2
e 3 + e 3
n
= 1
2
e−j 3 n + ej 3 n .
For a linear phase system, group delay is equivalent to the phase delay (= -φ/ω) and constant
across all frequency. The slope of phase is equal to one ( τ (π/4) = 1), so the phase of this system
π
is φ = −ω, and ]H(e±j 3 ) = ∓ π3 . From these magnitude and phase response of H, the output of
EE202B Signals and Systems Final Exam - Page 5 of 7 June 11, 2018
6. (25 points) A causal DT LTI system is described in terms of the difference equation
4y[n] − 4y[n − 1] + y[n − 2] = x[n] + 2x[n − 1] (5)
(a) (5 points) Derive the frequency response H(ejω ) of the system.
Y (ejω )
H(ejω ) =
X(ejω )
1 + 2e−jω
=
4(1 − e−jω + 14 e−2jω )
1 + 2e−jω
=
4(1 − 21 e−jω )2
(b) (5 points) Draw the pole-zero plot (Nyquist plot) of this system. Indicate the locations of poles
and zeros of this system. (Use } or xx to indicate duplicated zeros or poles.)
Answer:
Im{z}
Re{z}
-2 0 0.5
Solution:
The system response H(z) is given by
1 + 2z −1
H(z) =
4(1 − 21 z −1 )2
z+2
=z
4(z − 12 )2
(c) (5 points) Determine whether the system is stable or not. Justify your answer.
EE202B Signals and Systems Final Exam - Page 6 of 7 June 11, 2018
Answer: Stable
Solution: Since all poles of the system response H(z) is inside of unit circle (|zpole | < 1), the
RoC of the causal LTI system is the exterior region from |z| > zpole = 21 , which includes the
unit circle.
1 1 n 1 n
Answer: h[n] = 4 2
nu[n] + 2
(n − 1)u[n − 1]
Solution:
The frequency response can be rewritten as
1 + 2e−jω
H(ejω ) =
4(1 − 12 e−jω )2
1 e−jω
= +
4(1 − 12 e−jω )2 2(1 − 12 e−jω )2
1
X(ejω ) = −jω 2
⇐⇒ x[n] = (n + 1)an u[n]
(1 − ae )
X1 (ejω )X2 (ejω ) ⇐⇒ x1 [n] ∗ x2 [n]
e−jωn0 X(ejω ) ⇐⇒ x[n − n0 ],
n n−1
1 1 1 1
h[n] = (n + 1)u[n] + nu[n − 1]
4 2 2 2
n n
1 1 1
= (n + 1)u[n] + nu[n − 1],
4 2 2
7. (5 points) The Matlab code for filtering an audio signal is shown below. Which of the following
commands best suits for ( 1 ) & ( 2 )?
A. (1) y = dfft(b,a,fs); (2) y = dfilter(bd,ad,x);
B. (1) y = fft(b,a,fs); (2) y = filter(bd,ad,x);
C. (1) y = con2dis(b,a,fs); (2) y = dfilter(bd,ad,x);
D. (1) y = bilinear(b,a,fs); (2) y = filter(bd,ad,x);
[ x , f s ]= a u d i o r e a d ( ’ music . mp3 ’ ) ;
x = x (: ,1);
N = length ( x ) ;
t = ( 0 : N−1)/ f s ;
EE202B Signals and Systems Final Exam - Page 7 of 7 June 11, 2018
omega n = 100∗2∗ pi ;
zeta = 1;
b = [ 1 / omega n 1 0 ] ; % numerator c o e f f i c i e n t s
a = [ 1 / omega n 1 ] ; % denominator c o e f f i c i e n t s
% analog to d i g i t a l conversion
[ bd , ad ] = ( 1 ) ; % analog−to−d i g i t a l c o n v e r s i o n
y= ( 2 ) ; % filtering
%% p l a y a u d i o s i g n a l
ha= a u d i o p l a y e r ( y , f s ) ;
p l a y ( ha )
Answer: D. (CT filter coefficients are transformed to DT coefficients by ’bilinear’ function, and
then actual filtering occurs with ’filter’ command)