Chapter 4 - Fourier Series and Fourier Transform of Discrete Time Signals 4.

64

4.10 Short Questions and Answers

+∞
Q4.1 Find Fourier coefficients of x(n) , where x(n) = ∑ δ (n − 3k) .
k = −∞

Solution
Given signal is a periodic impulse signal with impulses located at n = 3k, for integer values of k.
Let, one period of the given signal be x1(n).
Now, x1(n) = {1, 0, 0 }, with period N = 3, and with fundamental frequency, w 0=2p/3.
The Fourier coefficient ck is given by,

N− 1 − jk 2πn 2 − jk2πn
1 1 1 1
ck =
N
∑
n= 0
x1(n) e N =
3
∑
n=0
x1(n) e 3 =
3
x(0) + 0 + 0 =
3
; for all k.

Q4.2 Determine the discrete time Fourier series of x(n) = cos 2

FG π nIJ .
H6 K
Solution

Given that, x(n) = cos 2

FG π nIJ . Let us check, whether the given signal is periodic.
H6 K
bn + Ng = FGH cosFGH π6n + π6NIJK IJK
2
π
x(n + N) = cos 2
6
πN
Since cos (θ + 2πM) = cos θ, For periodicity, should be an integral multiple of 2π.
6
πN
Let, = M × 2π, where M and N are integers. ⇒ N = 12 M, Let M = 1, ∴ N = 12.
6
2π 2π π
∴ x(n) is periodic with fundamental period, N = 12 and fundamental frequency, ω 0 = = =
N 12 6
The Fourier series of x(n) can be obtained from Euler's formula as shown below.

F IJ LM FG IJ OP = LMM e OP LM e OP
jπn jπn 2 2
jπn jπn
π πn
2
6 + e
−
6 6 e
−
6 e jθ + e − jθ
x(n) = cos G 2 cosθ =
H K N H KQ M
6
n = cos
6 2 PP = M
MN 2
+
2 PP 2
N Q Q
j 2π n − j 2π n − j 2π n j 2π n
1 1 1 1 1 1
= e 6 + e 6 + = e 6 + + e 6
4 4 2 4 2 4
1 − j 2ω 0 n 1 1 j 2ω 0 n π
= e + + e ; where ω 0 =
4 2 4 6
Q4.3 Find the Fourier transform of x(n) = { 2, 1, 2 }.
Solution
By definition of Fourier transform,
+∞ 2
X(e jω ) = ∑ x(n) e − jωn = ∑ x(n) e − jωn = x(0) e 0 + x(1) e − jω + x(2) e − j2ω
n = −∞ n = 0

= 2 + e − jω + 2 e− j2ω = 2 e− jω (e jω + e − jω ) + e − jω e jθ + e − jθ
cosθ =
= 4 cos ω e − jω + e − jω = (1 + 4 cos ω ) e − jω 2
4. 65 Digital Signal Processing
Q4.4 Determine the Fourier transform of x(n) = u(n) – u(n–N).
Using finite geometric
Solution series sum formula
N – 1
x(n) can be expressed as, x(n) = 1 ; for n = 0 to N–1. 1– CN
By definition of Fourier transform,
∑
Cn =
1– C
n = 0

+∞ N−1 N− 1
− jω n 1 − e− jωN
X(e jω ) = ∑ x(n) e − jωn = ∑ 1× e − jωn
= ∑ de i =
1 − e− jω
n= −∞ n=0 n= 0

F − jωN − jωN
I − jωN
LMjωN − jωN
OP L ωN O LM sin ωN OP
1− GH e 2 e 2 JK e 2 e 2 _
MN
e 2
PQ = e − jω GFH N2 − 21JIK M sin 2 P − jω GFH N2−1JIK
=
F − jω − jω
I = − jω
LMe jω − jω
OP MM ω PP = e MM ω2 PP
1− G e
H
2 e 2
JK e 2
MN
2 − e 2
PQ N sin 2 Q N sin 2 Q
Q4.5 Find the Fourier transform of , x(n) = –an u(–n –1), where |a| < 1. Using finite geometric
Solution series sum formula
N – 1
1– CN
By definition of Fourier transform, when n = 0; a–n ejwn = 1 ∑
Cn =
1– C
n = 0

+∞ −1 ∞ ∞ ∞ n

X(e jω ) = ∑ x(n) e − jωn

= ∑ −a n
e− jωn = ∑ − a −n e jωn = 1 − ∑ a −n ejωn = 1− ∑ da −1 jω
e i
n = −∞ n= −∞ n=1 n=0 n=0

jω jω jω
1 a a− e −a −e e
= 1− = 1− = = = jω
1 − a −1e jω a − e jω a − e jω a − e jω e −a
n −n
Q4.6 Find the discrete time Fourier transform of the signal , x(n) = (0.2) u(n) + (0.2) u( − n − 1).
Solution
By definition of Fourier transform,
∞ ∞ ∞
X(e jω ) = ∑ x(n) e− jωn = ∑ ( 0.2)n u(n) e− jωn + ∑ (0.2) −n
u( −n − 1) e− jωn
n= −∞ n= −∞ n= −∞
∞ −1 ∞ ∞
= ∑ (0.2 e − jω n
) + ∑ ( 0.2 e jω )−n = ∑ (0.2 e − jω n
) + ∑ (0.2 e jω )n when n = 0; (0.2ejw )n =1
n=0 n = −∞ n=0 n=1

∞ ∞
1 1
= ∑ ( 0.2 e − jω n
) + ∑ (0.2 ejω )n − 1=
1 − 0.2 e − jω
+
1 − 0.2 e jω
−1
n=0 n =0
Using infinite geometric
1 − 0.2 e jω + 1 − 0.2 e− jω − (1 − 0.2 e− jω ) (1 − 0.2 e jω )
= series sum formula
(1 − 0.2 e − jω ) (1 − 0.2 e jω ) ∞
1
1 − 0.2 e jω + 1 − 0.2 e− jω − (1 − 0.2 ejω − 0.2 e − jω + 0.04)
∑ Cn =
1− C
n = 0
=
1 − 0.2 e jω − 0.2 e− jω + 0.04 when |C|< 1
1 − 0.04 0.96
= = e jθ + e − jθ
1 − 0.2 (e jω + e − jω ) + 0.04 104. − 0.4 cos ω cosθ =
2

Q4.7 Determine the energy density spectrum of a discrete time signal , x(n) = a n u(n) for − 1 < a < 1.
Solution
Using infinite geometric
By definition of Fourier transform, series sum formula
∞
1
∞ ∞ ∞
− jω n 1
∑ Cn =
1 − C
X(e jω ) = ∑ x(n) e − jωn = ∑ an e − jωn = ∑ da e i =
1 − a e − jω
n = 0

when |C| < 1

n=0 n=0 n=0
Chapter 4 - Fourier Series and Fourier Transform of Discrete Time Signals 4. 66
Now the energy density spectrum is, e jθ + e − jθ
cosθ =
2 1 1 2
X(e jω = X(ejω ) X* (e jω ) = ×
1 − a e− jω 1 − a e jω
1 1 1
= = =
1 − a e jω − a e− jω + a 2 1 − a (e jω + e− jω ) + a 2 1 − 2a cos ω + a 2

Q4.8 Find the inverse Fourier transform of the rectangular pulse spectrum defined as,
X(e jω ) = 1 ; |ω | ≤ W
= 0 ; W ≤ |ω | ≤ π
Solution
By definition inverse Fourier transform,
π W
1 e jθ + e− jθ
x(n) =
2π z
−π
X(e jω ) e jωn dω =
1
2π z
−W
e jωn dω sin θ =
2j
W sin θ
=
1 LM e OP
jωn
=
1 LM e
j Wn
−
e− jWn OP =
1 LM e j Wn
− e − jWn OP θ
= sinc θ
2π N jn Q −W
2π N jn jn Q πn N 2j Q
sinWn W sinWn W
= = = sinc Wn
πn π Wn π
Q4.9 Determine the inverse Fourier transform of X(e jω ) = 2 π δ ( ω − ω 0 ), ω0 ≤ π .

Solution
The inverse Fourier transform of X(ejw ) is,
π π
x(n) =
1
2π z
−π
X(ejω ) ejωn dω =
1
2π z
−π
2 π δ(ω − ω 0 ) e jωn dω
Note : Here the integral limit is -p to
π
+p, and in this range there is only one
= z
−π
δ(ω − ω 0 ) e jωn
dω = e jωn
ω =ω 0
=e jω 0n
impulse located at w 0 .

1 − 2 a z −1
Q4.10 A causal discrete time LTI system has a system function H(z) = . Here 'a' is real and
2 b + z −1
|a| < 1. Find the value of 'b' so that the frequency response H(ejw ) of the system satisfies the
condition |H(ejw )| = 1 for all w.
Solution
1 − 2a z −1
Given that, H(z) =
2b + z −1
The frequency response of the system can be obtained by putting, z = e jw in H(z).
1 − 2a e − jω
∴ H(e jω ) == H(z) z = e jω
=
2b + e − jω
1 − 2a e − jω
Here, H(e jω ) = 1 ; ∴ =1 ⇒ 1 − 2a e − jω = 2b + e − jω
2b + e − jω
∴ 1 − 2 a cos ω + j 2 a sin ω = 2 b + cos ω − j sin ω
e± jθ = cos θ ± j sin θ
2 2 2 2
c1 − 2a cos ωh + c2 a sin ωh = c2b + cos ωh + csin ωh
1 + 4 a 2 cos 2ω − 4 a cos ω + 4 a 2 sin2 ω = 4 b2 + 4 b cos ω + cos 2ω + sin2ω sin2 θ + cos 2 θ = 1

1 + 4 a 2 − 4 a cos ω = 4 b2 + 4 b cos ω + 1
The above equation is true, when b = - a .
jω
Hence to satisfy the condition H(e ) = 1 for all w , b = -a .
4. 67 Digital Signal Processing
Q4.11 Determine the sampling period for the signal X(j W) = U ( j W + j W 0) – U (j W – j W 0), to
sample without aliasing.
Solution
The frequency spectrum of the given signal can be plotted as shown in fig Q4.11.

U ( Ω + Ω0 ) U ( Ω − Ω0 ) X ( j Ω)

1
−

− Ω0 0 Ω 0 Ω0 Ω − Ω0 0 Ω0 Ω
F ig Q 4 .11.
From the frequency spectrum of fig Q4.11, it is observed that the maximum frequency, W max is,
Ω0
Ωmax = Ω0 ; ∴ 2 π Fmax = Ω0 ⇒ Fmax =
2π
1
∴ Sampling frequency, Fs ≥ 2 Fmax ⇒ Sampling period, T ≤
Fs
1 1 π
∴Minimum sampling period, T = = =
Fs 2Fmax Ω0
F i. e. , T < π I . π
∴ In order to avoid aliasing the sampling period T should be less than GH Ω JK Ω0 0

Q4.12 Determine the Nyquist sampling frequency and Nyquist interval for the signal, x(t) = M
L sin 200 π t OP .
N πt Q
Solution
2

x(t) =
LM sin 200 π t OP =
1
sin2(200 π t) =
1 1 − cos 2(200π t)
MN π t PQ π t 2 2
π t 2 2
2
1 − cos 2 θ
=
1
1 − cos 400π t =
1
−
cos 400πt sin2 θ =
2
2 π2 t 2 2 π2 t 2 2 π2 t2
On comparing the cosine component with standard cosine wave "A cosW t" we get,
W = 400p Þ 2pF = 400p Þ F = 200 Hz
From the above analysis it is observed that, the maximum frequency in the signal Fmax = 200 Hz.
∴ Nyquist rate = 2 Fmax = 2 × 200 = 400 Hz
1 1
Nyquist interval = = = 2.5 ms
Nyquist rate 400

Q4.13 A signal x(t) whose spectrum is shown in fig Q4.13.1

is sampled at a rate of 300 samples / sec. Sketch the X (jF )
spectrum of the sampled discrete time signal. 2

Solution 1
From the spectrum shown in fig Q4.13.1 it is observed
that the maximum frequency, Fm in the signal is 100 Hz.
Given that, Sampling frequency, Fs is 300 Hz, which is −100 −50 0 50 100 F (H z )
greater than 2 Fm , and so the signal is sampled without
F ig Q 4.13 .1.
aliasing.
Chapter 4 - Fourier Series and Fourier Transform of Discrete Time Signals 4. 68
Frequency "f" of sampled discrete time signal corresponding to any frequency "F" of continuous time
signal is given by, f = F / Fs .
The magnitude of the spectrum of discrete time signal will be scaled by 1/T, where T = 1/ Fs. The
frequency spectrum of a discrete time signal will be periodic with periodicity of - 0.5 to + 0.5. (Refer
Chapter-2, Section 2.3). Therefore the frequency spectrum of sampled discrete time signal will be as
shown in fig Q4.13.2.
X (e jf )
2
T
1
T

400 350 −1 250 200 − 1 − 100 50 0 50 100 1 200 250 1 350 400 f
− − − − − 300
300 300 300 300 2 300 300 300 300 2 300 300 300
F ig Q 4.1 3.2.

Q4.14 If the spectrum shown in fig Q4.13.1 is sampled at a rate of 100 samples / sec. Sketch the
spectrum of the sampled discrete time signal.
Since the sampling frequency is less than 2 Fm, the spectrum of the sampled signal will have aliasing as
shown in fig Q4.14.1.

X (e jf )
2
T

1
T

−1 50 1 0 50 1 1 f
− =− =
100 2 100 2
F ig Q 4.1 4.1.

Q4.15 Consider the sampling of the bandpass signal X (jF )

whose frequency spectrum is shown in fig
Q4.15. Determine the minimum sampling rate
Fs to avoid aliasing.
−106 −100 −94 0 94 100 106 F
F ig Q 4.1 5.
Solution
The given signal is a bandpass signal. The bandwidth, B = 106 – 94 = 12 Hz.
Here the upper cutoff frequency (106 Hz) is not an integer multiple of bandwidth, B. Hence the minimum
sampling rate should be 4B, in order to avoid aliasing.

∴ Minimum sampling rate = 4 × B = 4 × 12 = 48Hz