Exam II: Solutions

BME 171-02, Signals and Systems

Exam II: Solutions
100 points total
0. (5 pts.) Fourier transform tables.
1. (20 pts.) Determine the Fourier transforms of the following signals:
(a) x(t) = (cos(5t) + e−2t )u(t)
Solution:
 
1 1 1
cos(5t)u(t) ↔ πδ(ω + 5) + + πδ(ω − 5) +
2 j(ω + 5) j(ω − 5)
1
e−2t u(t) ↔
jω + 2
 
1 1 1 1
X(ω) = πδ(ω + 5) + + πδ(ω − 5) + +
2 j(ω + 5) j(ω − 5) jω + 2

(b) x(t) = (1 − t) p2 (t)

Solution:
ω 
p2 (t) ↔ 2 sinc
π
d ω 
tp2 (t) ↔ 2j sinc
dω  π  ω 
ω d
X(ω) = 2 sinc − j sinc
π dω π
Z t
(c) x(t) = e−3(t−λ) p1 (λ − 1/2)dλ
0
Solution: let x1 (t) = e−3t u(t) and x2 (t) = p1 (t − 1/2). Then x(t) = x1 (t) ⋆ x2 (t).
1
x1 (t) ↔
jω + 3
ω
x2 (t) ↔ sinc e−jω/2
2π
1 ω
X(ω) = X1 (ω)X2 (ω) = sinc e−jω/2
jω + 3 2π

 0,
 t<0
(d) x(t) = t, 0 ≤ t ≤ 2
 2, t > 2

Z t
Solution: let x1 (t) = p2 (t − 1) and note that x(t) = x1 (λ)dλ. Then
0
ω
x1 (t) ↔ 2 sinc e−jω
π
X1 (0) = 2
1 2 ω 
X(ω) = X1 (ω) + πX1 (0)δ(ω) = sinc e−jω + 2πδ(ω)
jω jω π
 Exam II: Solutions

2. (25 pts.) Write down the trigonometric Fourier series representation of the following signal:

x(t)
π

−π 0 π t

2π
Solution: we have T = 2π, ω0 = T = 1.

π 0 π  2 0 !
1 π2
Z   
1 1 1 t 3π
Z Z
2
a0 = x(t)dt = (−t)dt + πdt = − +π = + π2 =
2π −π 2π −π 0 2π 2 −π 2π 2 4

π  Z 0 π 
1 1
Z Z
ak = x(t) cos(kt)dt = − t cos(kt)dt + π cos(kt)dt
π −π π −π 0
0  !
π
1 0
 
1 1 1
Z
= − t sin(kt) + sin(kt)dt + π sin(kt)
π k −π k −π k 0
 0 !
1 1 1 cos(πk) − 1
= 0− cos(kt) +0 =
π k k −π πk 2

0, if k is even
=
− πk2 2 , if k is odd

π  Z 0 Z π 
1 1
Z
bk = x(t) sin(kt)dt = − t sin(kt)dt + π sin(kt)dt
π −π π −π 0
0 π !
1 0
 
1 1 1
Z
= t cos(kt) − cos(kt)dt − π cos(kt)
π k −π k −π k 0
  
1 1 1 1
= π cos(πk) + 0 − π cos(kπ) −
π k k k
1
=
k
Thus,
∞ ∞
3π X 2 X 1
x(t) = − 2
cos(kt) + sin(kt)
4 πk k
k=0 k=1
k odd

2
 Exam II: Solutions

3. (20 pts.) An LTI system generates the output

y(t) = (e−2t − e−3t )u(t)

in response to the input x(t) = e−2t u(t).

(a) Determine the unit impulse response h(t) of the system.
Solution: since the system is LTI, y(t) = x(t) ⋆ h(t). In the frequency domain,

Y (ω) = X(ω)H(ω);
1
X(ω) =
jω + 2
1 1 1
Y (ω) = − =
jω + 2 jω + 3 (jω + 2)(jω + 3)
Y (ω) jω + 2 1
H(ω) = = = .
X(ω) (jω + 2)(jω + 3) jω + 3
Therefore,
h(t) = e−3t u(t).
(b) Sketch the amplitude |H(ω)| and the phase 6 H(ω) in the sets of axes provided. Be sure
to mark the axes properly.

|Η(ω)| ∠Η(ω)
1/3
π/2
1/3√2

−3 0 3 ω 0 ω

−π/2

1
H(ω) =
jω + 3
3 ω
= −j ;
jω + 3 ω+3
1
|H(ω)| = √
ω2 + 9
 ω
6 H(ω) = tan−1 −
3
1
|H(0)| = ;
3
1 1
|H(±3)| = √ |H(0)| = √
2 3 2

3
 Exam II: Solutions

4. (20 pts.) Consider the discrete-time signal

x[n] = cos(3n)p[n],

where p[n] is the rectangular pulse

(
1, n = 0, 1, . . . , 6
p[n] =
0, otherwise

(a) Compute its discrete-time Fourier transform (DTFT) X(Ω).

Solution:
sin(7Ω/2) −j3Ω
P (Ω) = e
sin(Ω/2)
1
X(Ω) = [P (Ω + 3) + P (Ω − 3)]
2 
1 sin(7(Ω + 3)/2) −j3(Ω+3) sin(7(Ω − 3)/2) −j3(Ω−3)
= e + e
2 sin((Ω + 3)/2) sin((Ω − 3)/2)

(b) Express the 7-point discrete Fourier transform (DFT) of x[n] in terms of the DTFT X(Ω).
Solution: for k = 0, 1, . . . , 6,

Xk = X(2πk/7)
 
1 sin(7(2πk/7 + 3)/2) −j3(2πk/7+3) sin(7(2πk/7 − 3)/2) −j3(2πk/7−3)
= e + e
2 sin((2πk/7 + 3)/2) sin((2πk/7 − 3)/2)

4
 Exam II: Solutions

5. (10 pts.) You have two discrete-time signals, x[n] and ν[n], where x[n] = 0 for n < 0 and
n ≥ 1000 and ν[n] = 0 for n < 0 and n ≥ 1040. Explain how you would use the FFT
algorithm in order to efficiently compute the convolution x[n] ⋆ ν[n] and estimate the number
of (complex) multiplications you would need.
Solution: in general, the convolution of x[n] and ν[n] will have 1000 + 1040 = 2040 nonzero compo-
nents. The smallest power of 2 that is larger than 2040 is L = 2048 = 211 . Let us pad x[n] and ν[n]
with zeros so that

x[n] = 0, n = 1000, 1001, . . . , 2048

ν[n] = 0, n = 1040, 1041, . . . , 2048

Then to compute x[n] ⋆ ν[n], we would first use the FFT algorithm to compute the L-point DFT’s Xk
and Vk of x[n] and ν[n], and then use the FFT algorithm to compute the inverse L-point DFT of the
product of Xk and Vk .
We will need:

• On the order of (1/2)L log2 L = (1/2) · 2048 · 11 = 11264 multiplications to compute the L-point
DFT of x[n].
• On the order of (1/2)L log2 L = (1/2) · 2048 · 11 = 11264 multiplications to compute the L-point
DFT of ν[n].
• L = 2048 multiplications to compute the product of Xk and Vk .
• On the order of (1/2)L log2 L = (1/2) · 2048 · 11 = 11264 multiplications to compute the L-point
inverse DFT of the product of Xk and Vk .

Thus, the total number of multiplications is on the order of

11264 + 11264 + 2048 + 11264 = 35840.