You are on page 1of 138

UNIT II

ANALYSIS OF DISCRETE TIME SIGNALS

Baseband signal Sampling – Discrete Fourier series - Fourier


Transform of discrete time signals (DTFT) & its properties - Z
Transform & its properties – Evaluation of the inverse z-
transform.

1
Transformation involves change in coordinates and domain of operations
A BB

𝐴∩𝐵

Set Theory – Venn Diagrams Probability


X X X X X X X X X X X X

……………………………………….

=
=

+ + + + + + + + + + + Maximum
X X X X X X X X X X X X

Beagle Pug
……………………………………….

=
=

+ + + + + + + + + + + < Maximum

Puggle
X X X X X X X X X X X X

Beagle Pug
……………………………………….

=
=

+ + + + + + + + + + + < Maximum

Puggle
X X X X X X X X X X X X

………………….

=
=

+ + + + + + + + + + +

= 0
How much of something is present in anything ?

Multiply in time and integrate in time

How much of is present in ?


Any signal which repeats over time
again and again

Periodicity
Fourier Series for periodic signals/functions
∞ ∞
1
𝑓 𝑥 = 𝑎𝑜 + ෍ 𝑎𝑛 cos 𝑛𝑥 + ෍ 𝑏𝑛 sin 𝑛𝑥
2
𝑛=1 𝑛=1

𝜋 𝜋
1 1
𝑎𝑜 = න 𝑓 𝑥 𝑑𝑥 𝑎𝑛 = න 𝑓 𝑥 cos 𝑛𝑥 𝑑𝑥
𝜋 𝜋
−𝜋 −𝜋

𝜋
1
𝑏𝑛 = න 𝑓 𝑥 sin 𝑛𝑥 𝑑𝑥 𝑛 = 1, 2, 3, 4, … … .
𝜋
−𝜋

Any periodic function/signal can be expressed


as weighted sum of sine and cosine
an , bn gives the spectral information

All sinusoids are the multiple of fundamental

Any periodic signal as a weighted


combination n = 1, 2, 3, 4, …….
of only certain frequencies
Fourier Series – In Complex form
1

𝜃 = 2𝜋𝐹𝑡

Substituting 2 in 1 & minimizing we get

Where

Spectral Amplitude {Un}

FOURIER SERIES gives the


Spectral Amplitude {Un}
Fourier Series – Square Wave (PERIODIC)
∞ ∞
1 ao= 0, an=0 and bn only exist
𝑓 𝑥 = 𝑎𝑜 + ෍ 𝑎𝑛 cos 𝑛𝑥 + ෍ 𝑏𝑛 sin 𝑛𝑥
2
𝑛=1 𝑛=1

Any periodic function/signal can


be expressed as weighted sum of Any periodic signal as a weighted
sine and cosine combination
of only certain frequencies
A periodic square wave contains infinite odd sine harmonics only
𝑥(𝑡) ∞ ∞
1 1
𝑓 𝑡 = 𝑎 + ෍ 𝑎𝑛 cos 𝑛𝑡 + ෍ 𝑏𝑛 sin 𝑛𝑡
2 𝑜
𝑛=1 𝑛=1

0 .01 .02 𝑛 = 1, 2, 3, 4, … … .

−1 0.02
𝑥(Ω) 1
𝑎𝑛 = න 𝑓 𝑡 cos 𝑛𝑡 𝑑𝑡
𝜋 0.02
1 0
𝑎𝑜 = න 𝑓 𝑡 𝑑𝑡
𝜋
−𝜋 𝑎𝑛 = 0
0.02
1 Discrete spectrum No Cosine component is present
𝑎𝑜 = න 𝑓(𝑡) 𝑑𝑡
0.02 0.02
0 1
𝑏𝑛 = න 𝑓 𝑡 sin 𝑛𝑡 𝑑𝑡
0.01 0.02 0.02
1 0
𝑎𝑜 = න 1 𝑑𝑡 + න −1 𝑑𝑡
0.02 1
0 0.01 𝑏𝑛 = 1 − (−1)𝑛
𝑛𝜋
2
𝑎𝑜 = 0 𝑏𝑛 = 0, when n is even 𝑏𝑛 = 𝑛𝜋, when n is odd
No dc component is present
Square Wave – weighted sum of
sinusoids – ODD Function
Fourier Series – Saw tooth Wave

Any periodic signal as a


weighted combination
of only certain frequencies

𝑛 = 1, 2, 3, 4, … … .
Fourier Series Discrete Fourier Series
Input is continuous x(t) Input is discretised x[n]
Continuous Periodic signal Discrete Periodic signal

FS DFS

𝑁−1
𝑗2𝜋𝑛𝑘
𝑥(𝑡)
𝑥𝑛 =෍ 𝐶𝑘 𝑒 𝑁
𝑘=0
𝑁−1
𝑥(𝑡) 1 −𝑗2𝜋𝑛𝑘
𝐶𝑘 = ෍ 𝑥 𝑛 𝑒 𝑁
𝑁
Discrete Non periodic spectrum 𝑘=0

Discrete Periodic Spectrum


No of frequency component Ck required
to describe the signal = Number of
samples of x[n] in the signal
Fourier series (FS) and Discrete Fourier series (DFS)

𝑥(Ω)
x(t)
FS

tp
Continuous Periodic signal Discrete Non Periodic spectrum

x[n]
DFS

Ts

Discrete Periodic signal Discrete Periodic spectrum


Non repeating sequence as a signal with one period
extending from –∞ to + ∞

(i.e.) T = ∞ & f = 0
∑ → ∫

ω
Spectral Amplitude {Un} → Spectral density {F(ω)}
F → 0

Any aperiodic signal as a weighted combination of all possible frequency.


Any aperiodic signal as a weighted combination of all possible frequency.

A(ω)
A(t)

t ω

f(t) FT [ f(t) ] = f(ω) f(ω)


Integration in Time & variation
in frequency

FT [ f(t) ] = f(ω)
δ(t) = 1 t=0 δ(t-τ) = 1 t= τ
δ(t) = 0 t≠0 δ(t-τ) = 0 t≠ τ
h(t)

System
Impulse response
δ(n)

n ω
Impulse spectrum

All frequency components are present with unit


magnitude in the spectrum of Impulse

Impulse the only all frequency signal


δ(n)

n
=

All frequency components are present with


unit magnitude in the spectrum of Impulse
System

=
=

=
System

System

System

System

System
.
.
.

System
x(n) System y(n) δ(n) System h(n)
T[] T[]
∞ ∞

𝒚 𝒏 = ෍ 𝒉 𝒌 𝒙(𝒏 − 𝒌) 𝒚 𝒏 = ෍ 𝒙 𝒌 𝒉(𝒏 − 𝒌)
𝒌=−∞ 𝒌=−∞
CONVOLUTION SUM CONVOLUTION SUM

𝑰𝒇 𝒙(𝒏) = 𝒆𝒋𝝎𝒏
𝑫𝒊𝒔𝒄𝒓𝒆𝒕𝒆 𝒕𝒊𝒎𝒆 𝑭𝒐𝒖𝒓𝒊𝒆𝒓 𝑻𝒓𝒂𝒏𝒔𝒇𝒐𝒓𝒎
∞ ∞
𝒚 𝒏 = ෍ 𝒉 𝒌 𝒆𝒋𝝎(𝒏−𝒌) 𝑯 𝒆𝒋𝝎 = ෍ 𝒉 𝒌 𝒆−𝒋𝝎𝒌
𝒌=−∞ 𝒌=−∞

𝒚 𝒏 = 𝒆𝒋𝝎𝒏 ෍ 𝒉 𝒌 𝒆−𝒋𝝎𝒌
𝒌=−∞ Frequency response is a
function of continuous
variable ω

𝑯(𝒆𝒋𝝎 )
𝒚 𝒏 = 𝒆𝒋𝝎𝒏 𝑯(𝒆𝒋𝝎 ) Frequency response of the system
System

Fourier Transform
is a Signal Prism
Fourier Transform Discrete time
Input is continuous x(t)
Fourier Transform
Continuous Non Periodic signal
Input is discretised x[n]
FT

𝑥(𝝮) = න 𝑥(𝑡)𝑒 −𝑗𝝮𝑡 𝑑𝑡 Discrete Non Periodic signal
−∞
∞ DTFT
𝑥(𝑡) = න 𝑥(𝝮)𝑒 𝑗𝝮𝑡 𝑑𝑡 ∞
−∞
𝑋(𝑒 𝑗𝜔 ) = ෍ 𝑥[𝑛]𝑒 −𝑗𝜔𝑛
Continuous Non Periodic spectrum 𝑘=−∞
2𝜋
1
𝑥𝑛 = න 𝑋(𝑒 𝑗𝜔 )𝑒 𝑗𝜔𝑛 𝑑𝜔
2𝜋 0

Continuous Periodic spectrum


Fourier Transform (FT) and Discrete Time Fourier Transform (DTFT)

x(t) 𝑥(𝝮)

FT

Continuous Non Periodic signal Continuous Non Periodic spectrum

x[n] 𝑋(𝑒 𝑗𝜔 )
DTFT

Discrete Non Periodic signal Continuous Periodic spectrum


Periodic Signal Non Periodic Signal

Power Signal Energy Signal

Power Finite Energy Finite

𝑁−1 ∞
1 −𝑗2𝜋𝑛𝑘
𝐶𝑘 = ෍ 𝑥 𝑛 𝑒 𝑁 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 < ∞
𝑁 𝒏=−∞
𝑘=0
Spectral Amplitude {Un} Spectral density {F(ω)}
What if power and energy are infinite?
What if Fourier transform diverges to infinity ?
Power and Energy are infinite

When will this converge ?
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 < ∞
1 𝒏=−∞

𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 < ∞ 𝒆−𝒋𝝎𝒏 = 𝑐𝑜𝑠 2 𝜔𝑛 + 𝑠𝑖𝑛2 𝜔𝑛 = 1


𝒏=−∞


Fourier Transform will converge only when
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 <∞ x(n) is absolutely sumable.
𝒏=−∞

A system is said to be stable only if it satisfies the condition


෍ 𝒙 𝒏 <∞
𝒏=−∞

Therefore we can conclude that FOURIER TRANSFORM can only be applied for
STABLE SYSTEMS
𝒏
𝒙 𝒏 = 𝟐𝒏 u n 𝟏
𝒙 𝒏 = ൗ𝟐 u n
FT Diverge ∞ ∞ FT Converge
Unstable ෍ 𝒙 𝒏 =∞ Stable ෍ 𝒙 𝒏 <∞
𝒏=−∞ 𝒏=−∞
Laplace transform is used to make an unstable system stable
and determine at which point the system leads to instability

Find exactly at which point the system diverges or


converges.
How can this be achieved?

Diverging Signal Fast rate Converging signal


Decaying signal
∞ ∞
𝑥(𝝮) = න 𝑥(𝑡)𝑒 −𝑗𝝮𝑡 𝑑𝑡 𝒆−𝝈𝒕 𝑥(𝝮) = න 𝑥(𝑡)𝒆−𝝈𝒕 𝑒 −𝑗𝝮𝑡 𝑑𝑡
−∞ −∞
∞ ∞
𝑋(𝑆) = න 𝑥(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 𝑋(𝑆) = න 𝑥(𝑡)𝑒 −(𝜎+𝑗𝝮)𝑡 𝑑𝑡
−∞ −∞
Laplace Transform
Neither Energy nor Power signal Unstable system

Find exactly at which point the system diverges or converges.

What is Pole, Zero and Region of convergence?

𝒙 𝒕 = 𝒆−𝒕 𝒖(𝒕) Let us assume different values of σ


∞ ∞ ∞
𝑋(𝑆) = න 𝑒 −𝑡 𝑒 − 𝜎+𝑗𝝮 𝑡 𝑢 𝑡 𝑑𝑡 𝜎 = 0, =න 𝑒 −(0+1)𝑡
𝑑𝑡 = න 𝑒 −𝑡 𝑑𝑡
−∞ 0 0
∞ LT converges
𝑋(𝑆) = න 𝑒 −𝑡 𝑒 − 𝜎+𝑗𝝮 𝑡
𝑑𝑡 ∞ ∞
0 𝜎 = 1, = න 𝑒 −(1+1)𝑡 𝑑𝑡 = න 𝑒 −2𝑡 𝑑𝑡
∞ 0 0
𝑋(𝑆) = න 𝑒 −(𝜎+1)𝑡 𝑒 −𝑗Ω𝑡 𝑑𝑡 LT converges faster
0 ∞ ∞
For σ > 1 Converges 𝜎 = −1, = න 𝑒 −(−1+1)𝑡 𝑑𝑡 = න 𝑒 −0𝑡 𝑑𝑡
σ = -1 Undefined 0 0
𝑹𝑶𝑪 > −𝟏 LT Undefined
σ < -1 Diverges
∞ ∞
The value of σ for which X(s) is undefined is
𝜎 = −2, = න 𝑒 −(−2+1)𝑡 𝑑𝑡 = න 𝑒 𝑡 𝑑𝑡
called the POLE of the system or the 0 0
LT Diverges
boundary which marks the ROC 𝑹𝑶𝑪 is the area between Pole and zero of a system
𝒙 𝒕 = −𝒆−𝒕 𝒖(−𝒕)

𝑋 𝑆 = − න 𝑒 −𝑡 𝑒 − 𝜎+𝑗𝝮 𝑡 𝑢 𝑡 𝑑𝑡
−∞ 0 ∞
0 𝜎 = 1, = න 𝑒 −(1+1)𝑡 𝑑𝑡 = න 𝑒 −2𝑡 𝑑𝑡
𝑋 𝑆 = − න 𝑒 −𝑡 𝑒 − 𝜎+𝑗𝝮 𝑡 𝑑𝑡 −∞ 0
−∞ LT Diverge
0 0 ∞
𝑋 𝑆 = − න 𝑒 −(𝜎+1)𝑡 𝑒 −𝑗Ω𝑡 𝑑𝑡 𝜎 = −1, = න 𝑒 −(−1+1)𝑡 𝑑𝑡 = න 𝑒 −0𝑡 𝑑𝑡
−∞ −∞ 0
LT Undefined
Let us assume different values of σ 0 ∞

0 0
𝜎 = −2, = න 𝑒 −(−2+1)𝑡 𝑑𝑡 = න 𝑒 𝑡 𝑑𝑡
−∞ 0
𝜎 = 0, =න 𝑒 −(1)𝑡 𝑑𝑡 = න 𝑒 −𝑡 𝑑𝑡 LT Converge
−∞ −∞ 0 ∞
LT Diverge 𝜎 = −3, = න 𝑒 −(−3+1)𝑡 𝑑𝑡 = න 𝑒 2𝑡 𝑑𝑡
−∞ 0
Since the limit is from -∞ to 0
LT Converge
The boundary is still -1, but the ROC is < -1
For σ > 1 Diverge
Though the LT of 𝒙 𝒕 = −𝒆−𝒕 𝒖 −𝒕 𝒂𝒏𝒅 𝒙 𝒕 = 𝒆−𝒕 𝒖(𝒕) happens to be σ = -1 Undefined
σ < -1 Converge
same, boundary lone or pole are same, ROC is entirely different.
S Plane

Laplace expressed a signal in terms of


exponents and sinusoids

𝑋(𝑆) = න 𝑥(𝑡)𝑒 −(𝜎+𝑗𝝮)𝑡 𝑑𝑡 σ=0 σ<0 σ>0
−∞
Diverging Signal Fast rate Converging signal
Decaying signal

∞ ∞

𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 𝒓−𝒏 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒓−𝒏 𝒆−𝒋𝝎𝒏


𝒏=−∞ 𝒏=−∞

Let us relate DTFT and Z transform


DTFT 𝑿 𝒆𝒋𝝎 = 𝑿 𝒁 │𝑟 = 1 or Z = 𝒆
𝒋𝝎
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 (𝒓𝒆𝒋𝝎 )−𝒏
𝒏=−∞
r=1

-1 1 𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏
𝒏=−∞

Z = 𝒓𝒆𝒋𝝎 New complex variable


Z transform evaluated on the unit circle
corresponds to FT r- magnitude, angle ω
Fourier Transform DTFT
∞ ∞
𝑥(𝝮) = න 𝑥(𝑡)𝑒 −𝑗𝝮𝑡 𝑑𝑡 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏
−∞
𝒏=−∞

Laplace Transform Z Transform


∞ ∞
Discrete LT
𝑥(𝑆) = න 𝑥(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏
−∞ 𝒏=−∞

Discrete DTFT
DTFT DFT
∞ ∞
𝟐𝝅𝒌𝒏
−𝒋 𝑵
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 𝑿 𝒌 = ෍ 𝒙 𝒏 𝒆
𝒏=−∞ DTFT requires more 𝒏=−∞
𝑋(𝑒 𝑗𝜔 ) memory location for 𝑋(𝑘)
implementation due to
continuous frequency ω

Continuous Periodic spectrum Discrete Periodic spectrum


SAMPLING
Conversion of a continuous time signal to discrete time signal is known as Sampling

The most fundamental operation in signal processing

Sampling theorem - Most important for digital communication systems


SAMPLING THEOREM
1. A sufficient number of samples of the signal must be taken so as to represent the
original signal in its samples completely.
2. It should be possible to recover or reconstruct the original signal completely from its
samples. The number of samples to be taken depends on the maximum signal
frequency fm present in the signal.

1. If a signal x(t) contains no frequency component for frequencies above fm


then it is completely described by instantaneous sample values uniformly
1
spaced in time with period 𝑇𝑠 ≤ 𝑓𝑚 .
2

2. If a signal x(t) contains no frequency components for frequencies above fm


the signal can be exactly reconstructed/recovered from the knowledge of
its samples taken at the rate of 2fm samples / sec.
SAMPLING THEOREM
A continuous time signal x(t) may be completely represented in its samples
and recovered back if the sampling frequency 𝒇𝒔 > 𝟐𝒇𝒎
𝑓𝑠 - Sampling Frequency
𝑓𝑚 - maximum frequency present in the signal

Proof
Consider a signal x(t) whose spectrum is band limited to fm

𝑿 𝝎 = 𝟎 𝒇𝒐𝒓 𝝎 > 𝝎𝒎

𝑥(𝑡) 𝑿(𝝎)

FT

0 𝑡 −𝝎𝒎 𝟎 𝝎𝒎
Analog Signal Spectrum of analog Signal
The analog signal is multiplied with impulse train to obtain discrete time signal

𝑥(𝑡) 𝑐(𝑡)

.... ....
0 𝑡 -4Ts -3Ts -2Ts -Ts 0 Ts 2Ts 3Ts 4Ts 𝑡
Analog Signal Impulse Train

𝑐 𝑡 consist of unit impulses repeating periodically every Ts .



1 𝟐𝝅
𝑇𝑠 = 𝝎𝒔 = = 𝟐𝝅𝑓𝑠 𝑐 𝑡 = ෍ 𝛿(𝑡 − 𝑛𝑇𝑠 )
𝑓𝑠 𝑻𝒔
𝑛=−∞

𝑐(𝑡) is an impulse train which samples at a rate of fs


𝑦[𝑡]
Achieved by multiplying x(t) by 𝑐(𝑡) 𝑦(𝑡) = 𝑥 𝑡 𝑐(𝑡)

Frequency Convolution Theorem – Multiplication


of two functions in time domain is equivalent to
convolution of their spectra in Frequency domain. 0 𝑛𝑇
Sampled Signal
1
𝑦(𝑡) = 𝑥 𝑡 𝑐(𝑡) Frequency Convolution Theorem 𝑌𝜔 = 𝑋 𝜔 ∗ 𝐶(𝜔)
2𝜋
𝑭𝑻 𝒙(𝒕) = 𝑿 𝝎

1
𝑭𝑻 𝒄(𝒕) = 𝑪 𝝎 𝑌𝜔 = 𝑋 𝜔 ∗ 𝝎𝒔 ෍ 𝜹(𝝎 − 𝒏𝝎𝒔 )
2𝜋
𝒏=−∞
Fourier transform of a periodic signal

∞ 𝝎𝒔
𝑌𝜔 = ෍ 𝑋 𝜔 ∗ 𝜹(𝝎 − 𝒏𝝎𝒔 )
𝑋(𝜔) = 2𝜋 ෍ 𝑐𝑛 𝛿(𝜔 − 𝑛𝜔𝑜 ) 2𝜋
𝒏=−∞
𝑛=−∞

𝑐𝑛 fourier coefficient of a periodic signal


Using the property
𝑋(𝑛𝜔𝑜 ) 1 𝑭𝑻 𝜹(𝒕) = 𝟏 𝑥 𝑡 ∗ 𝛿 𝑡 − 𝑡1 = 𝑥(𝑡 − 𝑡1 )
𝑐𝑛 = =
𝑇𝑜 𝑇𝑜 𝛿(𝑡) ∞
𝟏
∞ 1 𝑌𝜔 = ෍ 𝑿(𝝎 − 𝒏𝝎𝒔 )
1 𝑻𝒔
𝑋(𝜔) = 2𝜋 ෍ 𝛿(𝜔 − 𝑛𝜔𝑜 ) 𝒏=−∞
𝑇𝑜
𝑛=−∞

0
2𝜋 𝟏
𝑋(𝜔) = ෍ 𝛿(𝜔 − 𝑛𝜔𝑜 ) 𝒀𝝎 = 𝑿 𝝎 + 𝑿 𝝎 − 𝝎𝒔 + 𝑿 𝝎 + 𝝎𝒔 + ⋯ .
𝑇𝑜 𝑻𝒔
𝑛=−∞
∞ n=0 n=1
𝑿(𝝎) = 𝝎𝒐 ෍ 𝜹(𝝎 − 𝒏𝝎𝒐 )
𝒏=−∞
𝟏
𝒀𝝎 = 𝑿 𝝎 + 𝑿 𝝎 − 𝝎𝒔 + 𝑿 𝝎 + 𝝎𝒔 + ⋯ .
𝑻𝒔
n=0 n=1

𝑥(𝑡) 𝑿(𝝎)

0 𝑡 −𝝎𝒎 𝟎 𝝎𝒎
Analog Signal Spectrum of analog Signal

𝒀(𝝎)

−𝟐𝝎𝒔 −𝝎𝒔 −𝝎𝒎 𝟎 𝝎𝒎𝜔𝑠 − 𝜔𝑚 𝝎𝒔 𝜔𝑠 + 𝜔𝑚 𝟐𝝎𝒔

𝝎𝒔 − 𝝎𝒎 > 𝝎𝒎 𝝎𝒔 > 𝟐𝝎𝒎 Perfect reconstruction of the original signal


The process of reconstructing a continuous time signal from its samples - INTERPOLATION

The LPF used to recover original from its samples - INTERPOLATION FILTER
When the sampling rate 𝒇𝒔 = 𝟐 𝒇𝒎 , it is called Nyquist rate or Minimum sampling rate

1
Maximum sampling interval is called Nyquist interval 𝑇𝑠 = 𝑓
𝑠

When the continuous time band limited signal is sampled at nyquist rate, the sampled
spectrum contains non-overlapping successive cycles repeating periodically. But they
touch each other.
𝒀(𝝎)
𝝎𝒔 − 𝝎𝒎 = 𝝎𝒎
𝝎𝒔 = 𝟐𝝎𝒎

−𝟐𝝎𝒔 −𝝎𝒔 −𝝎𝒎 𝟎 𝝎𝒎 𝝎𝒔 𝟐𝝎𝒔


Perfect reconstruction of the original signal

The original spectrum X(ω) can be recovered from the sampled spectrum Y(ω) by using
LPF with a sharp cut –off frequency fm
When the continuous time band limited signal is sampled at a rate lower than nyquist
rate 𝒇𝒔 < 𝟐 𝒇𝒎 , the spectrum Y(ω) of the sampled signal y(t) overlap with each other.

𝝎𝒔 − 𝝎𝒎 < 𝝎𝒎 𝝎𝒔 < 𝟐𝝎𝒎


𝒀(𝝎) Aliasing

−𝟐𝝎𝒔 −𝝎𝒔 𝟎 𝝎𝒔 𝟐𝝎𝒔

𝝎𝒔 − 𝝎𝒎 𝝎𝒎
Phenomenon in which a high frequency component in the frequency spectrum of signal
takes identity of a lower frequency component in the spectrum of the sampled
spectrum.
Effects of aliasing
✓ Since high and low frequencies interfere with each other distortion is generated.

✓ The date is lost and it cannot be recovered.


How to avoid aliasing
1. Sampling rate 𝒇𝒔 ≥ 𝟐𝒇𝒎
2. Strictly bandlimit the signal to ‘𝑓𝑚 ’: when 𝑓𝑠 = 2𝑓𝑚 there should be no aliasing
(ideal). If there are few components higher than 2𝑓𝑚 . low pass filter or Pre alias
filter must be used to limit the band of frequencies of the signal to fm
x(t)
x(t) - bandlimited 𝑋𝛿 (𝑓)
Band limiting low
Sampler
pass filter

44.1Khz (CD) and 48Khz (DVD) are the most common sampling rates for digital audio.

It is necessary to capture audio signal covering human hearing range.


Human beings can hear frequencies in the range of 20Hz to 20KHz.
Sampling theorem 𝑓𝑠 > 2𝑓𝑚 , So, in order to preserve the quality of sound sensed by the human ear,
roughly 40Khz sampling rate is required.
When the signal is a band pass signal, then a different criteria must be used to sample the signal

The band pass signal x(t), whose maximum bandwidth is 2fm can be completely represented and
recovered from its samples if it is sampled at the minimum rate of twice the bandwidth.

fm is the maximum frequency component present in the signal


Bandwidth is 2fm
Minimum sampling rate for band pass signal must be 4fm

X(f)
The center frequency fc > fm

The band pass signal x(t) is represented


in terms of in phase and quadrature
components.
-fc 0 fc -fm fc fc + f m
𝑥 𝑡 = 𝑥𝐼 (𝑡) cos 𝜔𝑐 𝑡 + 𝑥𝑄 (𝑡) sin 𝜔𝑐 𝑡
2fm 2fm
Suppress the sum frequencies using LPF
Spectrum of an arbitrary band pass signal
The in phase and quadrature components contain only low frequency components

The spectrum of these components is limited between –fm to fm

X(f) X(f)

-fc 0 fc -fm fc fc + fm -fm 0 fm

2fm Spectrum of xI(t) and xQ(t) components


2fm
Spectrum of an arbitrary band pass signal

For band pass signals of Bandwidth 2fm ,


Minimum sampling rate = 2 x BW = 4fm samples/sec
There are three types of sampling techniques:
1. Ideal / Instantaneous or Impulse sampling.
2. Natural sampling.
3. Flat Top sampling. Ideal / Instantaneous or Impulse sampling.
y(t)

x(t)

𝛿 𝑇𝑠 𝑡 = ෍ 𝛿(𝑡 − 𝑛𝑇𝑠 )
𝑛=−∞
Sampled signal
Sampling Function 𝑦(𝑡) = 𝑥 𝑡 𝛿𝑇𝑠 (𝑡)

It is impossible to have a pulse with pulse width
𝑦 𝑓 = 𝑓𝑠 ෍ 𝑋(𝑓 − 𝑛𝑓𝑠 )
approaching to zero. 𝑛=−∞
Natural sampling
Most practical way of accomplishing sampling of a band limited signal x(t) is to
multiply x(t) with a pulse train c(t).

τ
x(t) y(t)

𝑦(𝑡) = 𝑥 𝑡 𝑐(𝑡)

𝑦 𝑓 = τ 𝐴 𝑓𝑠 ෍ 𝑠𝑖𝑛𝑐 𝑛𝑓𝑠 τ 𝑋(𝑓 − 𝑛𝑓𝑠 )


𝑛=−∞

Sampling is termed natural as the profile of each sample retains the shape of analog
segment during the pulse interval.
Flat top or Rectangular sampling.
The top of the sample remain constant which is equal to the instantaneous value of the
base band signal x(t) at the beginning of sampling.

x(t) y(t)

𝑦(𝑡) = 𝑥 𝑡 𝑐(𝑡)

𝐻 𝑓 spectrum of rectangular
pulse (Sinc function)

𝑦 𝑓 = 𝐻(𝑓) 𝑓𝑠 ෍ 𝑋(𝑓 − 𝑛𝑓𝑠 )


𝑛=−∞
The primary effect of flat top sampling is an attenuation of high frequency components.
This effect is called aperture effect.
The width of the pulse in flat top and natural sampling is increased to the maximum
extent to reduce the transmission bandwidth.
Comparison of various sampling methods

Natural Flat top


S. No Parameter Ideal Sampling
Sampling Sampling

Multiplication by an Chopping Sample and Hold


1 Sampling principle
impulse function process Circuit

Most practically
2 Feasibility Practically not possible Practically used
used

3 Sampling rate Tends to infinity Nyquist rate Nyquist rate

4 Noise Interference Nominal Maximum Minimum


Consider the analog signal 𝑥𝑎 𝑡 = cos 100𝜋𝑡 + 2 sin 50𝜋𝑡 − 8 cos 150𝜋𝑡

𝝎 What is the Nyquist Rate? 𝝮𝐬 ≥𝟐𝝮𝐍 F𝐬 ≥𝟐F𝐍


𝑭=
𝟐𝝅
𝝎𝟏 𝟏𝟎𝟎𝝅 F 𝐍 is the maximum frequency in the signal xa(t)
𝑭𝟏 = = = 𝟓𝟎
𝟐𝝅 𝟐𝝅
𝑵𝒚𝒒𝒖𝒊𝒔𝒕 𝒓𝒂𝒕𝒆 Fs ≥ 2FN
𝝎𝟐 𝟓𝟎𝝅
𝑭𝟐 = = = 𝟐𝟓 𝑵𝒚𝒒𝒖𝒊𝒔𝒕 𝒓𝒂𝒕𝒆 Fs ≥ 2*FN ≥ 2*75 ≥ 150Hz
𝟐𝝅 𝟐𝝅

𝝎𝟑 𝟏𝟓𝟎𝝅
𝑭𝟑 = = = 𝟕𝟓
𝟐𝝅 𝟐𝝅
Calculate the minimum sampling frequency required for x(t) = 0.5 sin 50𝜋t + 0.25 sin 25𝜋t
so as to avoid aliasing
𝝎𝟏 𝟓𝟎𝝅
F𝐬 ≥𝟐F𝐍
𝑭𝟏 = = = 𝟐𝟓
𝟐𝝅 𝟐𝝅 minimum sampling frequency required
𝝎𝟐 𝟐𝟓𝝅 to avoid aliasing
𝑭𝟐 = = = 𝟏𝟐. 𝟓
𝟐𝝅 𝟐𝝅
Fs = 2*FN = 2*25 = 50Hz
Continuous time domain Discrete time domain
Fourier Series (FS) Discrete Fourier Series (DFS)
Fourier Transform (FT) Discrete Time Fourier Transform (DTFT)
Laplace Transform (LT) Z Transform

When DTFT does not converge for a sequence DTFT [ x(n) ] = ∞

An alternate tool for analysis – Z Transform



When will DTFT converge ?
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 < ∞
1 𝒏=−∞

𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 < ∞ 𝒆−𝒋𝝎𝒏 = 𝑐𝑜𝑠 2 𝜔𝑛 + 𝑠𝑖𝑛2 𝜔𝑛 = 1


𝒏=−∞


Fourier Transform will converge only when
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 <∞ x(n) is absolutely summable of finite
𝒏=−∞

A system is said to be stable only if it satisfies the condition


෍ 𝒙 𝒏 <∞
𝒏=−∞

Therefore we can conclude that FOURIER TRANSFORM can only be applied for
STABLE SYSTEMS 𝒏
𝒙 𝒏 = 𝟐𝒏 u n 𝒙 𝒏 = 𝟏ൗ𝟐 u n
FT does not converge for FT converge for decaying
growing sequence sequence
∞ ∞
Unstable ෍ 𝒙 𝒏 =∞ Stable ෍ 𝒙 𝒏 <∞
𝒏=−∞ 𝒏=−∞
Need for Z - Transform

Diverging Signal Fast rate Converging signal


Decaying signal

∞ ∞

𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 𝒓−𝒏 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒓−𝒏 𝒆−𝒋𝝎𝒏


𝒏=−∞ 𝒏=−∞

DTFT does not converge for all sequence 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 (𝒓𝒆𝒋𝝎 )−𝒏
𝒏=−∞
Hence move on to an alternative tool for
signal analysis ∞

𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏
Laplace Transform 𝒏=−∞

Z = 𝒓𝒆𝒋𝝎 New complex variable
𝑥(𝑆) = න 𝑥(𝑡)𝑒 −𝑠𝑡 𝑑𝑡
−∞ r- magnitude, angle ω
Lets Relate DTFT and Z Transform
∞ ∞

𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏 <∞ 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 (𝒓𝒆𝒋𝝎 )−𝒏


𝒏=−∞ 𝒏=−∞

𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏
𝒏=−∞

DTFT 𝑿 𝒆𝒋𝝎 = 𝑿 𝒁 │r = 1 or z= 𝒆 𝒋𝝎


𝑟 =1

-1 1 σ

Z transform evaluated on the unit circle corresponds to FT


Z transform and DTFT of various functions
𝒙 𝒏 =𝜹 𝒏 𝜹 𝒏 = 𝟏, 𝒏 = 𝟎
= 0, n≠ 0

∞ ∞

𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏


𝒏=−∞ 𝒏=−∞
∞ ∞

𝑿 𝒁 = ෍ 𝜹 𝒏 𝒁−𝒏 𝑿 𝒆𝒋𝝎 = ෍ 𝜹 𝒏 𝒆−𝒋𝝎𝒏


𝒏=−∞ 𝒏=−∞

𝑿 𝒁 = 𝛿 𝟎 𝒁−𝟎 𝑿 𝒆𝒋𝝎 = 𝛿 𝟎 𝒆−𝒋𝝎𝟎

𝑿 𝒁 =1 𝑿 𝒆𝒋𝝎 = 1

Z transform and DTFT of 𝜹 𝒏 = 1


Z transform and DTFT of various functions
𝒖 𝒏 = 𝟏, 𝒏 ≥ 𝟎

𝒙 𝒏 =𝒖 𝒏 = 0, n < 0
𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏 ∞
𝒏=−∞ 𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏

𝒏=−∞
𝑿 𝒁 = ෍𝒖 𝒏 𝒁−𝒏 ∞
𝒏=𝟎
𝑿 𝒆𝒋𝝎 = ෍ 𝒖 𝒏 𝒆−𝒋𝝎𝒏
𝑿 𝒁 = 1 + 𝒁−𝟏 + 𝒁−𝟐 + 𝒁−𝟑 + ⋯ + 𝒁−∞ 𝑨 𝒏=𝟎

Multiply both sides by 𝒁−𝟏 𝑿 𝒆𝒋𝝎 = 1 + 𝒆−𝒋𝝎 + 𝒆−𝟐𝒋𝝎 + 𝒆−𝟑𝒋𝝎 + ⋯ + 𝒆−𝒋𝝎∞ 𝑪

𝒁−𝟏 𝑿 𝒁 = 𝒁−𝟏 + 𝒁−𝟐 + 𝒁−𝟑 + ⋯+ 𝒁−∞ 𝑩 Multiply both sides by 𝒆−𝒋𝝎

𝒆−𝒋𝝎 𝑿 𝒆𝒋𝝎 = 𝒆−𝒋𝝎 + 𝒆−𝟐𝒋𝝎 + 𝒆−𝟑𝒋𝝎 + ⋯ + 𝒆−𝒋𝝎∞ 𝑫


Subtract A from B
Subtract C from D
𝑿 𝒁 − 𝒁−𝟏 𝑿 𝒁 = 𝟏 𝑿 𝒆𝒋𝝎 − 𝒆−𝒋𝝎 𝑿(𝒆𝒋𝝎 ) = 𝟏
𝑿 𝒁 (𝟏 − 𝒁−𝟏 ) = 𝟏 𝑿 𝒆𝒋𝝎 (𝟏 − 𝒆−𝒋𝝎 ) = 𝟏
𝟏 𝟏
𝑿 𝒁 = 𝑿 𝒆𝒋𝝎 =
(𝟏−𝒁−𝟏 ) (𝟏 − 𝒆−𝒋𝝎 )
Z transform and DTFT of various functions
𝒖 𝒏 = 𝟏, 𝒏 ≥ 𝟎
∞ 𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏
= 0, n < 0
𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏

𝒏=−∞
𝑿 𝒆𝒋𝝎 = ෍ 𝒙 𝒏 𝒆−𝒋𝝎𝒏

𝒏=−∞
𝑿 𝒁 =෍ 𝒂𝒏 𝒖 𝒏 𝒁−𝒏 ∞
𝒏=𝟎 𝑿 𝒆𝒋𝝎 = ෍ 𝒖 𝒏 𝒆−𝒋𝝎𝒏
𝑿 𝒁 = 1 + 𝒂𝟏 𝒁−𝟏 + 𝒂𝟐 𝒁−𝟐 + 𝒂𝟑 𝒁−𝟑 + ⋯ + 𝒂∞ 𝒁−∞ 𝑨 𝒏=𝟎

Multiply both sides by 𝒂𝒁−𝟏

𝒂𝒁−𝟏 𝑿 𝒁 = 𝒂𝟏 𝒁−𝟏 + 𝒂𝟐 𝒁−𝟐 + 𝒂𝟑 𝒁−𝟑 + ⋯ + 𝒂∞ 𝒁−∞ 𝑩

Subtract A from B

𝑿 𝒁 − 𝒂𝒁−𝟏 𝑿 𝒁 = 𝟏
𝑿 𝒁 (𝟏 − 𝒂𝒁−𝟏 ) = 𝟏

𝟏 𝟏
𝑿 𝒁 = 𝑿 𝒆𝒋𝝎 =
(𝟏−𝒂𝒁−𝟏 ) (𝟏 − 𝒂𝒆−𝒋𝝎 )
Region of Convergence, Causality and Stability
Region of convergence (ROC) – the area where ZT converges or where ZT has finite values

Stability and Causality of the system can be determined based on ROC


𝒏 𝟐
𝟏
𝟑 ∞
A geometric series ෍ 𝒒 = 𝟏 + 𝒒 + 𝒒 + 𝒒 + ⋯+ 𝒒 =
𝟏−𝒒
𝒏=𝟎
Converges if and only if 𝑞 < 1

In general, geometric Progression

𝒂(𝟏 − 𝒓𝒏 )
𝒂 + 𝒂𝒓 + 𝒂𝒓𝟐 + 𝒂𝒓𝟑 + ⋯+ 𝒂𝒓∞ =
𝒓−𝟏

𝒂
The above series converges to Since r is small
𝒓−𝟏 &
when n=∞, rn =0
Converges if and only if 𝐫 < 𝟏
Find the ZT and ROC of the causal sequence defined by 𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏 , 𝒂 <𝟏


𝒂(𝟏 − 𝒓𝒏 )
𝑿 𝒁 = ෍ 𝒙 𝒏 𝒁−𝒏 𝒂 + 𝒂𝒓 + 𝒂𝒓𝟐 + 𝒂𝒓𝟑 + ⋯+ 𝒂𝒓∞ =
𝒓−𝟏
𝒏=−∞
∞ Converges if and only if 𝐫 < 𝟏
𝑿 𝒁 = ෍ 𝒂𝒏 𝒖 𝒏 𝒁−𝒏
𝒂
The above series converges to
𝒏=𝟎 𝒓−𝟏
𝑿 𝒁 = 1 + 𝒂𝟏 𝒁−𝟏 + 𝒂𝟐 𝒁−𝟐 + 𝒂𝟑 𝒁−𝟑 + ⋯ + 𝒂∞ 𝒁−∞

a = 1 & r = 𝒂𝒁−𝟏
𝒂 𝟏 𝒁
= −𝟏
=
𝒓 − 𝟏 𝒂𝒁 − 𝟏 𝒂 − 𝒁
Converges if and only if 𝐫 < 𝟏
𝒂
𝒂𝒁−𝟏 < 𝟏 <𝟏
𝒁

ROC 𝒁 >𝒂 Area outside the circle of radius a is the ROC


𝒏
0<n<∞ 𝟏 𝒖 𝒏 = 𝟏, 𝒏 ≥ 𝟎
x[n] is a right sided sequence 𝒙 𝒏 = 𝒖 𝒏 = 0, n < 0
𝟐
∞ 𝒏
𝟏 𝟏
𝑿 𝒁 =෍ 𝒁−𝒏 Converges if and only if <𝟏
𝟐 𝟐𝒁
𝒏=𝟎
∞ 𝒏
𝟏 ROC 𝒁 >𝟐
𝟏
𝑿 𝒁 =෍
𝟐𝒁
𝒏=𝟎

𝟏 𝟏 𝟏 𝟏
𝑿 𝒁 = 1 + 𝟐𝒁 + (𝟐𝒁)𝟐 + (𝟐𝒁)𝟑 + ⋯ + (𝟐𝒁)∞
𝟏
a=1&r=
𝟐𝒁
-1 1
x1/2
𝒂 𝟏 𝟐𝒁
= =
𝒓−𝟏 𝟏
− 𝟏 𝟏 − 𝟐𝒁
𝟐𝒁

Zero of the system lies at Z = 0


If ROC includes the Unit Circle,
Pole of the system lies at Z = 1/2
Then the system is STABLE and DTFT also converges
𝒏
-∞ < n < -1 𝟏 𝒖 𝒏 = 𝟏, 𝒏 ≥ 𝟎
x[n] is a Left sided sequence 𝒙 𝒏 =− 𝒖 −𝒏 − 𝟏 = 0, n < 0
𝟐
−∞ 𝒏
𝟏 𝑿 𝒁 = 2𝑧 + (2𝑧)2 +(2𝑧)3 + … … .
𝑿 𝒁 =− ෍ 𝒁−𝒏
𝟐
𝒏=−𝟏 a = 2z & r = 𝟐𝒛
−𝟏 𝒏
𝟏 𝒂 𝟐𝒛
𝑿 𝒁 =− ෍ =
𝟐𝒁 𝒓 − 𝟏 𝟐𝒛 − 𝟏
𝒏=−∞
∞ −𝒏
Converges if and only if 𝟐𝒛 < 𝟏
𝟏
𝑿 𝒁 =−෍ ROC 𝒁 <𝟐
𝟏
𝟐𝒁
𝒏=𝟏

𝑿 𝒁 = − ෍ 𝟐𝒁 𝒏

𝒏=𝟏 x 1
-1 1/2
𝟐𝒁
𝑿 𝒁 =−
(𝟏 − 𝟐𝒁)
𝟐𝒁 𝒛
𝑿 𝒁 = = If ROC does not include the Unit Circle,
(𝟐𝒁 − 𝟏) 𝒛 − 𝟏ൗ𝟐
Hence the system is UNSTABLE and DTFT diverges
Sum of 2 exponentials sequences
𝒏 𝒏
𝟏 −𝟏
𝒙 𝒏 = 𝒖 𝒏 + 𝒖 𝒏
𝟐 𝟑
x(n) contains 2 right sided sequences
𝒏 𝒏
𝟏 𝒏 −𝟏
𝒙𝟏 𝒏 = 𝒖 𝒏 𝒙 𝒏 =𝒂 𝒖 𝒏 𝒙𝟐 𝒏 = 𝒖 𝒏
𝟐 𝟏
𝟑
𝑿 𝒁 =
∞ (𝟏−𝒂𝒁−𝟏 ) ∞

𝑿𝟏 𝒁 = ෍ 𝒙 𝒏 𝒁
−𝒏 𝑿𝟐 𝒁 = ෍ 𝒙 𝒏 𝒁
−𝒏

𝒏=−∞
𝒏=−∞
∞ 𝒏 ∞ 𝒏
𝟏 −𝟏
𝑿𝟏 𝒁 =෍ 𝒁
−𝒏
𝑿𝟐 𝒁 =෍ 𝒁
−𝒏

𝒏=𝟎
𝟐 𝒏=𝟎
𝟑

𝟏 𝟏
𝑿𝟏 𝒛 = 𝑿𝟐 𝒛 = 𝟏
(𝟏−
𝟏
𝒁−𝟏 ) (𝟏+ 𝒁−𝟏 )
𝟑
𝟐

𝟏 𝟏
𝑿𝟏 𝒛 = 𝟏 𝑿𝟐 𝒛 = 𝟏
(𝟏− ) (𝟏+ )
𝟐𝒛 𝟑𝒛
𝟏 𝟏
𝑿𝟏 𝒛 = 𝟏 𝑿𝟐 𝒛 = −𝟏
(𝟏− ) (𝟏−( )
𝟐𝒛 𝟑𝒛

𝟐𝒁 𝟑𝒁
𝑿𝟏 𝒛 = 𝑿𝟐 𝒛 =
(𝟐𝒁−𝟏) (𝟑𝒁+𝟏)

𝒁 𝒁
𝑿𝟏 𝒛 = 𝑿𝟐 𝒛 =
(𝒁− 𝟏ൗ𝟐) (𝒁+ 𝟏ൗ𝟑)
𝟏 −𝟏
ROC <𝟏 ROC <𝟏
𝟐𝒁 𝟑𝒁
𝟏
ROC 𝟐𝒁 > 𝟏 ROC <𝟏
𝟑𝒁
ROC 𝒁 > 𝟏Τ𝟐 ROC 𝟑𝒁 > 𝟏
Z transform of x(n) is X(z) =𝑋1 𝑧 + 𝑋2 𝑧 ROC 𝒁 > 𝟏Τ𝟑
𝒁 𝒁
𝑿 𝒛 = 𝑿𝟏 𝒁 + 𝑿𝟐 𝒛 = +
(𝒁 − ൗ𝟐) (𝒁 + 𝟏ൗ𝟑)
𝟏
𝒁 𝒁
𝑿 𝒛 = +
(𝒁 − ൗ𝟐) (𝒁 + 𝟏ൗ𝟑)
𝟏

𝟐𝒁 𝟑𝒁
= +
(𝟐𝒁 − 𝟏) (𝟑𝒁 + 𝟏) 1
2
𝟐𝒁(𝟑𝒁 + 𝟏) 𝟑𝒁(𝟐𝒁 − 𝟏) x x
= + -1 −1 1 1
(𝟐𝒁 − 𝟏) (𝟑𝒁 + 𝟏) 3 12

𝟔𝒛𝟐 + 𝟐𝒛 + 𝟔𝒛𝟐 − 𝟑𝒛
=
(𝟐𝒁 − 𝟏)(𝟑𝒁 + 𝟏)

𝟏𝟐𝒛𝟐 − 𝒛 ROC is the intersection of X1(z) and X2(z)


=
(𝟐𝒁 − 𝟏)(𝟑𝒁 + 𝟏)
ROC 𝒁 > 𝟏Τ𝟐 ROC 𝒁 > 𝟏Τ𝟑
𝒛(𝟏𝟐𝒛 − 𝟏)
= As ROC includes the Unit Circle,
(𝟐𝒁 − 𝟏)(𝟑𝒁 + 𝟏)
The system is STABLE and DTFT Converges
Zero of the system lies at Z = 0, 1/12

Pole of the system lies at Z = 1/2 and -1/3


Sum of 2 exponentials sequences
𝒏 𝒏
𝟏 −𝟏
𝒙 𝒏 =− 𝒖 −𝒏 − 𝟏 + 𝒖 𝒏
𝟐 𝟑
x(n) contains 1 right sided and 1 left sided sequences
𝒏 𝒏
𝟏 −𝟏
𝒙𝟏 𝒏 = − 𝒖 −𝒏 − 𝟏 𝒙𝟐 𝒏 = 𝒖 𝒏
𝟐 𝟑
𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏
𝟏 ∞ 𝒏
−∞ 𝒏 𝑿 𝒁 = −𝟏
𝑿 𝒁 =− ෍
𝟏
𝒁 −𝒏
(𝟏−𝒂𝒁−𝟏 )
𝑿𝟐 𝒁 =෍ 𝒁
−𝒏

𝟐 𝒏=𝟎
𝟑
𝒏=−𝟏

𝟐𝒁 𝒛 𝟏
𝑿 𝒁 = = 𝑿𝟐 𝒛 = 𝟏
(𝟏+ 𝒁−𝟏 )
(𝟐𝒁 − 𝟏) 𝒛 − 𝟏ൗ𝟐 𝟑

𝟏
𝑿𝟐 𝒛 = 𝟏
(𝟏+ )
𝟑𝒛
𝟏 𝟏
𝑿𝟏 𝒛 = 𝟏 𝑿𝟐 𝒛 = −𝟏
(𝟏− ) (𝟏−( )
𝟐𝒛 𝟑𝒛

𝟐𝒁 𝟑𝒁
𝑿𝟏 𝒛 = 𝑿𝟐 𝒛 =
(𝟐𝒁−𝟏) (𝟑𝒁+𝟏)

𝒁 𝒁
𝑿𝟏 𝒛 = 𝑿𝟐 𝒛 =
(𝒁− 𝟏ൗ𝟐) (𝒁+ 𝟏ൗ𝟑)
−𝟏
ROC 𝟐𝒁 < 𝟏 ROC <𝟏
𝟑𝒁

ROC 𝒁 < 𝟏Τ𝟐 𝟏


ROC <𝟏
𝟑𝒁

ROC 𝟑𝒁 > 𝟏
Z transform of x(n) is X(z) =𝑋1 𝑧 + 𝑋2 𝑧 ROC 𝒁 > 𝟏Τ𝟑
𝒁 𝒁
𝑿 𝒛 = 𝑿𝟏 𝒁 + 𝑿𝟐 𝒛 = +
(𝒁 − ൗ𝟐) (𝒁 + 𝟏ൗ𝟑)
𝟏
𝒁 𝒁
𝑿 𝒛 = +
(𝒁 − ൗ𝟐) (𝒁 + 𝟏ൗ𝟑)
𝟏

𝟐𝒁 𝟑𝒁
= +
(𝟐𝒁 − 𝟏) (𝟑𝒁 + 𝟏)
1
𝟐𝒁(𝟑𝒁 + 𝟏) 𝟑𝒁(𝟐𝒁 − 𝟏)
x x 2
-1 −1 1 1
= +
(𝟐𝒁 − 𝟏) (𝟑𝒁 + 𝟏) 3 12

𝟔𝒛𝟐 + 𝟐𝒛 + 𝟔𝒛𝟐 − 𝟑𝒛
=
(𝟐𝒁 − 𝟏)(𝟑𝒁 + 𝟏)

𝟏𝟐𝒛𝟐 − 𝒛 ROC is the intersection of X1(z) and X2(z)


=
(𝟐𝒁 − 𝟏)(𝟑𝒁 + 𝟏)
ROC 𝒁 < 𝟏Τ𝟐 ROC 𝒁 > 𝟏Τ𝟑
𝒛(𝟏𝟐𝒛 − 𝟏)
=
(𝟐𝒁 − 𝟏)(𝟑𝒁 + 𝟏)

Zero of the system lies at Z = 0, 1/12

Pole of the system lies at Z = 1/2 and -1/3


0<n<∞
𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏 , 𝒂 <𝟏 Right sided sequence ROC 𝒁 >𝒂

𝒏 0<n<∞
𝟏 𝟏
𝒙 𝒏 = 𝒖 𝒏 Right sided sequence ROC 𝒁 >𝟐
𝟐

𝒏 -∞ < n < -1
𝟏 𝟏
𝒙 𝒏 =− 𝒖 −𝒏 − 𝟏 Left sided sequence ROC 𝒁 <𝟐
𝟐
Right sided sequence ROC is the intersection of X1(z) and X2(z)
0<n<∞
𝒏 𝒏
𝟏 −𝟏 ROC 𝒁 > 𝟏Τ𝟐 ROC 𝒁 > 𝟏Τ𝟑
𝒙 𝒏 = 𝒖 𝒏 + 𝒖 𝒏
𝟐 𝟑
Right and left sided sequence ROC is the intersection of X1(z) and X2(z)

𝒏 𝒏
-∞ < n < -1 0<n<∞
𝟏 −𝟏
𝒙 𝒏 =− 𝒖 −𝒏 − 𝟏 + 𝒖 𝒏 ROC 𝒁 < 𝟏Τ𝟐 ROC 𝒁 > 𝟏Τ𝟑
𝟐 𝟑
Observation from the problems
Properties of ROC
1. ROC is a ring in the Z plane centered at origin.

2. FT of x(n) converges absolutely if and only if the ROC of the Z transform includes
the unit circle.

3. The ROC cannot contain any poles.

4. For finite duration sequence, ROC is entire Z plane except Z = 0 or Z = ∞.

5. If x(n) is a right sided sequence (i.e) n ≥ 0. ROC extends outward from the outermost
pole in X(z)

6. If x(n) is a left sided sequence (i.e) n < 0. ROC extends inward from the innermost pole

7. If x(n) is a two sided sequence , ROC is a ring in the Z plane bounded on the
interior and exterior by a pole and not containing any pole in the ROC.
Common Z transform pairs
Z transform ROC
δ(n) 1 All Z
𝟏
U(n) 𝐳 >𝟏
𝟏 − 𝒛−𝟏
𝟏 𝐳 <𝟏
-U(-n-1)
𝟏 − 𝒛−𝟏
All z except z=0,
δ(n-m) 𝒛−𝒎
If m>0 &if m<0
𝟏 𝐳 > 𝐚
𝐚𝐧 𝐮(𝐧)
𝟏 − 𝒂𝒛−𝟏
𝟏
-𝐚𝐧 𝐮(−𝐧 − 𝟏) 𝐳 < 𝐚
𝟏 − 𝒂𝒛−𝟏
𝒂𝒛−𝟏
n𝐚𝐧 𝐮(𝐧) 𝐳 > 𝐚
(𝟏 − 𝒂𝒛−𝟏 )𝟐
𝐧 𝒂𝒛−𝟏
-n𝐚 𝐮(−𝐧 − 𝟏) 𝐳 < 𝐚
(𝟏 − 𝒂𝒛−𝟏 )𝟐
Properties of ZT and DTFT
Linearity of Z transform

Let x1(n) and x2(n) Let X1(z) and X2(z) 𝑍 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = 𝑎𝑋1 𝑧 + 𝑏𝑋2 (𝑧)

𝑍 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = ෍ 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 𝑧 −𝑛


𝑛=−∞
∞ ∞

𝑍 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = ෍ 𝑎𝑥1 𝑛 𝑧 −𝑛 + ෍ 𝑏𝑥2 (𝑛)𝑧 −𝑛


𝑛=−∞ 𝑛=−∞

𝑍 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = aX1(z) + bX2(z)

Linearity of DTFT
Let x1(n) and x2(n) Let X1(ω) and X2(ω) 𝐷𝑇𝐹𝑇 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = 𝑎𝑋1 ω + 𝑏𝑋2 (ω)

𝐷𝑇𝐹𝑇 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = ෍ 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 𝑒 −𝑗𝜔𝑛


𝑛=−∞
∞ ∞

𝐷𝑇𝐹𝑇 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = ෍ 𝑎𝑥1 𝑛 𝑒 −𝑗𝜔𝑛 + ෍ 𝑏𝑥2 𝑛 𝑒 −𝑗𝜔𝑛


𝑛=−∞ 𝑛=−∞

𝐷𝑇𝐹𝑇 𝑎𝑥1 𝑛 + 𝑏𝑥2 𝑛 = 𝑎𝑋1 ω + 𝑏𝑋2 (ω)


Time shifting in Z transform Time shifting in DTFT
Let x(n) be delayed by k units x(n-k) Let x(n) be delayed by k units x(n-k)

𝑍 𝑥 𝑛−𝑘 = 𝑋 𝑧 𝑧 −𝑘 𝐷𝑇𝐹𝑇 𝑥 𝑛 − 𝑘 = 𝑋 𝜔 𝑒 −𝑗𝜔𝑘


∞ ∞

𝑍 𝑥 𝑛−𝑘 = ෍ 𝑥 𝑛 − 𝑘 𝑧 −𝑛 𝐷𝑇𝐹𝑇 𝑥 𝑛 − 𝑘 = ෍ 𝑥 𝑛 − 𝑘 𝑒 −𝑗𝜔𝑛


𝑛=−∞ 𝑛=−∞

Assume 𝑛 − 𝑘 = 𝑚 and 𝑛 = 𝑘 + 𝑚 Assume 𝑛 − 𝑘 = 𝑚 and 𝑛 = 𝑘 + 𝑚


∞ ∞
𝑍 𝑥 𝑛−𝑘 = ෍ 𝑥 𝑚 𝑧 −(𝑘+𝑚) 𝐷𝑇𝐹𝑇 𝑥 𝑛 − 𝑘 = ෍ 𝑥 𝑚 𝑒 −𝑗𝜔(𝑘+𝑚)
𝑛=−∞ 𝑛=−∞
∞ ∞
𝑍 𝑥 𝑛−𝑘 = ෍ 𝑥 𝑚 𝑧 −𝑚−𝑘 𝐷𝑇𝐹𝑇 𝑥 𝑛 − 𝑘 = ෍ 𝑥 𝑚 𝑒 −𝑗𝜔𝑚 𝑒 −𝑗𝜔𝑘
𝑛=−∞ 𝑛=−∞

𝑍 𝑥 𝑛−𝑘 = 𝑋(𝑍)𝑍 −𝐾 𝐷𝑇𝐹𝑇 𝑥 𝑛 − 𝑘 = 𝑋 𝜔 𝑒 −𝑗𝜔𝑘

Find the Z transform and DTFT of 𝒙 𝒏 = 𝒂𝒏 𝒖(𝒏 − 𝟓)

𝒁−𝟓 𝒆−𝒋𝝎𝟓
𝒁 𝒂𝒏 𝒖 𝒏−𝟓 = 𝑫𝑻𝑭𝑻 𝒂𝒏 𝒖 𝒏−𝟓 =
𝟏 − 𝒂𝒛−𝟏 𝟏 − 𝒂𝒆−𝒋𝝎
Time scaling in Z transform Time scaling in DTFT
Let y(n) =x(an) Let y(n) =x(an)
1ൗ 1ൗ
𝑍 𝑥 𝑎𝑛 =𝑋 𝑧 𝑧 𝑎 𝐷𝑇𝐹𝑇 𝑥 𝑎𝑛 =𝑋 𝜔 𝑒 𝑎

∞ ∞

𝑍 𝑥 𝑎𝑛 = ෍ 𝑥 𝑎𝑛 𝑧 −𝑛 𝐷𝑇𝐹𝑇 𝑥 𝑎𝑛 = ෍ 𝑥 𝑎𝑛 𝑒 −𝑗𝜔𝑛
𝑛=−∞ 𝑛=−∞

Assume a𝑛 = 𝑚 and 𝑛 = 𝑚Τ𝑎 Assume a𝑛 = 𝑚 and 𝑛 = 𝑚Τ𝑎


∞ ∞
𝑚
𝑍 𝑥 𝑎𝑛 = ෍ 𝑥 𝑚 𝑧 −( Τ𝑎) 𝐷𝑇𝐹𝑇 𝑥 𝑎𝑛 = ෍ 𝑥 𝑚 𝑒 −𝑗𝜔(
𝑚Τ )
𝑎
𝑛=−∞ 𝑛=−∞
∞ ∞
1ൗ 1ൗ
𝑍 𝑥 𝑎𝑛 = ෍ 𝑥 𝑚 𝑧 −𝑚 𝑧 𝑎 𝐷𝑇𝐹𝑇 𝑥 𝑛 − 𝑘 = ෍ 𝑥 𝑚 𝑒 −𝑗𝜔𝑚 𝑒 𝑎

𝑛=−∞ 𝑛=−∞

1ൗ 1ൗ
𝑍 𝑥 𝑎𝑛 = 𝑋(𝑍)𝑧 𝑎 𝐷𝑇𝐹𝑇 𝑥 𝑎𝑛 =𝑋 𝜔 𝑒 𝑎

Find the Z transform and DTFT of 𝒙 𝒏 = 𝒂𝒏 𝒖(𝟓𝒏)


𝟏 𝟏
𝒁 ൗ𝟓 𝒆 ൗ𝟓
𝒁 𝒂𝒏 𝒖 𝒏 − 𝟓 = 𝑫𝑻𝑭𝑻 𝒂𝒏 𝒖 𝒏 − 𝟓 =
𝟏 − 𝒂𝒛−𝟏 𝟏 − 𝒂𝒆−𝒋𝝎
Time Reversal in Z transform Time Reversal in DTFT

If 𝑍 𝑥(𝑛) = 𝑋 𝑧 If 𝐹 𝑥(𝑛) = 𝑋 𝑒 𝑗𝜔 = 𝑋(𝜔)

Then 𝑍 𝑥 −𝑛 = 𝑋 𝑧 −1 Then 𝐹 𝑥 −𝑛 = 𝑋 𝑒 −𝑗𝜔 = 𝑋(−𝜔)



Proof: 𝐹 𝑥(𝑛) = ෍ 𝑥(𝑛) 𝑒 −𝑗𝜔𝑛

𝑛=−∞
𝑍 𝑥(−𝑛) = ෍ 𝑥(−𝑛) 𝑧 −𝑛 ∞
𝑛=−∞ 𝐹 𝑥(−𝑛) = ෍ 𝑥(−𝑛) 𝑒 −𝑗𝜔(−𝑛)
Assume −𝑛 = 𝑙 𝑛=−∞

∞ Assume −𝑛 = 𝑚
𝑍 𝑥(−𝑛) = ෍ 𝑥(𝑙) 𝑧 𝑙 ∞
𝑙=−∞ 𝐹 𝑥(−𝑛) = ෍ 𝑥(𝑚) 𝑒 𝑗𝜔𝑚

𝑚=−∞
𝑍 𝑥(−𝑛) = ෍ 𝑥(𝑙) 𝑧 −1 −𝑙 ∞
𝑙=−∞ = ෍ 𝑥(𝑚) 𝑒 −𝑗(−𝜔)𝑛
𝑚=−∞
= 𝑋(𝑧 −1 )
= 𝑋 𝑒 −𝑗𝜔 𝑜𝑟 𝑋(−𝜔)
5
Frequency shifting Property of DTFT

If 𝐹 𝑥(𝑛) = 𝑋 𝜔
Then 𝐹 𝑥 𝑛 𝑒 −𝑗𝜔0 𝑚 = 𝑋 𝜔 − 𝜔0

Proof:

𝐹 𝑥(𝑛)𝑒 𝑗𝜔0𝑚 = ෍ 𝑥(𝑛)𝑒 𝑗𝜔0𝑚 𝑒 −𝑗𝜔𝑛


𝑛=−∞

𝐹 𝑥(𝑛)𝑒 𝑗𝜔0 𝑚 = ෍ 𝑥(𝑛) 𝑒 −𝑗 𝜔−𝜔0 𝑛

𝑛=−∞

= 𝑋(𝜔 − 𝜔0 )

This property is the dual of the time shifting property.

6
Symmetry Property of DTFT
The DTFT 𝑋 𝜔 is a complex function of 𝜔 and can be expressed as

𝑋 𝜔 = 𝑋𝑅 𝜔 + 𝑗𝑋𝐼 (𝜔)

Where 𝑋𝑅 𝜔 is real part & 𝑋𝐼 𝜔 is imaginary part of 𝑋 𝜔 respectively. We have


∞ ∞ ∞

𝑋 𝜔 = ෍ 𝑥 𝑛 𝑒 −𝑗𝜔𝑛 = ෍ 𝑥 𝑛 cos 𝜔𝑛 − 𝑗 ෍ 𝑥 𝑛 sin 𝜔𝑛


𝑛=−∞ 𝑛=−∞ 𝑛=−∞
∞ ∞

𝑋𝑅 𝜔 + 𝑗𝑋𝐼 𝜔 = ෍ 𝑥 𝑛 cos 𝜔𝑛 − 𝑗 ෍ 𝑥 𝑛 sin 𝜔𝑛


∞ 𝑛=−∞ 𝑛=−∞ ∞

𝑋𝑅 𝜔 = ෍ 𝑥 𝑛 cos 𝜔𝑛 𝑋𝐼 𝜔 = − ෍ 𝑥 𝑛 sin 𝜔𝑛
𝑛=−∞ 𝑛=−∞
Since cos −𝜔 𝑛 = cos 𝜔𝑛 and sin −𝜔 𝑛 = − sin 𝜔𝑛
∞ ∞

𝑋𝑅 −𝜔 = ෍ 𝑥 𝑛 cos(−𝜔)𝑛 𝑋𝐼 −𝜔 = − ෍ 𝑥 𝑛 sin −𝜔 𝑛
𝑛=−∞ 𝑛=−∞
∞ ∞

= ෍ 𝑥 𝑛 cos 𝜔𝑛 = 𝑋𝑅 𝜔 = ෍ 𝑥 𝑛 sin 𝜔 𝑛 = −𝑋𝐼 𝜔


𝑛=−∞ 𝑛=−∞

𝑋𝑅 −𝜔 = 𝑋𝑅 𝜔 (Even Symmetry) 𝑋𝐼 −𝜔 = −𝑋𝐼 𝜔 (Odd Symmetry)


7
Multiplication by a time factor n in ZT Multiplication by a time factor n in DTFT
Let y(n) =n x(n) Let y(n) =n x(n)
𝑑𝑋 𝑧 𝑑𝑋 𝜔
𝑍 𝑛𝑥 𝑛 = −𝑧 𝐷𝑇𝐹𝑇 𝑛𝑥 𝑛 =𝑗
𝑑𝑧 𝑑𝜔
∞ ∞

𝑍𝑥 𝑛 = ෍ 𝑥 𝑛 𝑧 −𝑛 𝐷𝑇𝐹𝑇 𝑥 𝑛 = ෍ 𝑥 𝑛 𝑒 −𝑗𝜔𝑛
𝑛=−∞ 𝑛=−∞
Differentiate w.r.t Z Differentiate w.r.t ω

𝑑𝑋(𝑧) ∞
= ෍ 𝑥 𝑛 (−𝑛)𝑧 −𝑛−1) 𝑑𝑋(𝜔)
𝑑𝑧 = ෍ 𝑥 𝑛 (−𝑗𝑛)𝑒 −𝑗𝜔𝑛
𝑛=−∞ 𝑑𝜔
𝑛=−∞
Multiply both sides by -z Multiply both sides by j
∞ ∞
−𝑧𝑑𝑋(𝑧) 𝑑𝑋(𝜔)
= ෍ 𝑥 𝑛 𝑛 𝑧 𝑧 −𝑛−1) 𝑗 = ෍ 𝑛 𝑥 𝑛 𝑒 −𝑗𝜔𝑛
𝑑𝑧 𝑑𝜔
𝑛=−∞ 𝑛=−∞


−𝑧𝑑𝑋(𝑧) 𝑑𝑋 𝜔
= ෍ 𝑥 𝑛 𝑛 𝑧 −𝑛 𝐷𝑇𝐹𝑇 𝑛𝑥 𝑛 =𝑗
𝑑𝑧 𝑑𝜔
𝑛=−∞

𝑑𝑋 𝑧
𝑍 𝑛𝑥 𝑛 = −𝑧
𝑑𝑧
Parseval’s Theorem - DTFT
∞ 𝜋
1
If 𝐹 𝑥(𝑛) = 𝑋 𝜔 Then 𝐸 = ෍ 𝑥(𝑛) 2
= න 𝑋(𝜔) 2 𝑑𝜔
2𝜋
𝑛=−∞ −𝜋

Proof:
𝜋 ∗
∞ ∞ ∞
2
1
𝐸 = ෍ 𝑥(𝑛) = ෍ 𝑥 𝑛 𝑥∗ 𝑛 = ෍ 𝑥 𝑛 න 𝑋 𝜔 𝑒 𝑗𝜔𝑛 𝑑𝜔
2𝜋
𝑛=−∞ 𝑛=−∞ 𝑛=−∞ −𝜋
∞ 𝜋
1
= ෍ 𝑥 𝑛 න 𝑋 ∗ (𝜔)𝑒 −𝑗𝜔𝑛 𝑑𝜔
2𝜋
𝑛=−∞ −𝜋

Interchanging the order of summation and integration, we get

𝜋 ∞
1
𝐸= න 𝑋 ∗ (𝜔) ෍ 𝑥 𝑛 𝑒 −𝑗𝜔𝑛 𝑑𝜔
2𝜋
−𝜋 𝑛=−∞
𝜋
1
= න 𝑋 ∗ (𝜔) 𝑋(𝜔) 𝑑𝜔
2𝜋
−𝜋
𝜋
1
= න 𝑋(𝜔) 2 𝑑𝜔
2𝜋 9
−𝜋
Find the Z transform and DTFT of 𝒙 𝒏 = 𝒏𝒖(𝒏)

𝟏 𝟏
𝒁𝒖 𝒏 = 𝑫𝑻𝑭𝑻 𝒖 𝒏 =
𝟏 − 𝒛−𝟏 𝟏 − 𝒆−𝒋𝝎

𝒛 Differentiate w.r.t ω
𝒁𝒖 𝒏 =
𝒛−𝟏
Differentiate w.r.t Z 𝑑𝑋(𝜔) −𝑗𝑒 −𝑗𝜔
=
𝑑𝑋(𝑧) 𝑧−1−𝑧 −1 𝑑𝜔 (1 − 𝑒 −𝑗𝜔 )2
= =
𝑑𝑧 (𝑧 − 1)2 (𝑧 − 1)2 Multiply both sides by j

Multiply both sides by -z


𝒅𝑿(𝝎) 𝒆−𝒋𝝎
𝒋 =
𝒅𝝎 (𝟏 − 𝒆−𝒋𝝎 )𝟐
−𝒛𝒅𝑿(𝒛) 𝒛
=
𝒅𝒛 (𝒛 − 𝟏)𝟐

−𝒛𝒅𝑿(𝒛) 𝒛−𝟏
=
𝒅𝒛 (𝟏 − 𝒛−𝟏 )𝟐
Convolution in ZT Convolution in DTFT

𝑦 𝑛 = 𝑥 𝑛 ⊗ ℎ 𝑛 = ෍ 𝑥 𝑘 ℎ(𝑛 − 𝑘)
𝑘=−∞
∞ ∞

𝑍 𝑦(𝑛) = ෍ ෍ 𝑥 𝑘 ℎ(𝑛 − 𝑘) 𝑧 −𝑛
𝑛=−∞ 𝑘=−∞

Assume 𝑛 − 𝑘 = 𝑚 and 𝑛 = 𝑘 + 𝑚
∞ ∞

𝑍 𝑦(𝑛) = ෍ ෍ 𝑥 𝑘 ℎ(𝑚) 𝑧 −(𝑚+𝑘)


𝑚=−∞ 𝑘=−∞
∞ ∞

𝑍 𝑦(𝑛) = ෍ ℎ 𝑚 𝑧 −𝑚 ෍ 𝑥 𝑘 𝑧 −𝑘
𝑚=−∞ 𝑘=−∞

𝒀(𝒛) = 𝑿 𝒛 𝑯(𝒛) 𝒀(𝝎) = 𝑿 𝝎 𝑯(𝝎)

Convolution becomes Multiplication


Inverse Z Transform
The basic methods employed to recover the original sequence x[n] from X(z)

1. Using Properties of Z Transform

2. Convolution Theorem Method

3. Partial Fraction Method

4. Long Division Method

5. Contour Integration Method


1. Using Properties of Z Transform

𝟏 IZT
𝑿 𝒁 =
(𝟏−𝒂𝒁−𝟏 )
ROC 𝒁 > 𝒂 𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏
IZT
𝑿 𝒁 =
𝟏
(𝟏−𝒂𝒁−𝟏 )
ROC 𝒁 < 𝒂 𝒙 𝒏 = −𝒂𝒏 𝒖 −𝒏 − 𝟏

2. Convolution Theorem Method :


If the given Z transform X(z) of x[n] can be split into a product of two Z transforms
whose inverses are known, the convolution summation of the inverses of these two Z
transforms will yield the desired inverse Z transform.

𝒊𝒇 𝑿(𝒛) = 𝑿𝟏 𝒛 𝑿𝟐 (𝒛), Then 𝒙 𝒏 = 𝒁−𝟏 [𝑿(𝒛)]


𝒙 𝒏 = 𝒁−𝟏 [𝑿𝟏 𝒛 𝑿𝟐 (𝒛)]

𝒙 𝒏 = ෍ 𝒙𝟏 𝒌 𝒙𝟐 (𝒏 − 𝒌)
𝒌=−∞

𝒙 𝒏 = 𝒙𝟏 𝒏 ∗ 𝒙𝟐 (𝒏)
Using Convolution Theorem, find the inverse Z transform.

𝑧
𝑋(𝑧) =
(𝑧 − 1)2
Split into a product of two Z transforms 𝑿(𝒛) = 𝑿𝟏 𝒛 𝑿𝟐 (𝒛)

𝑧 1
𝒙 𝒏 =𝒖 𝒏 𝑿𝟏 𝒛 = 𝑿𝟐 𝒛 = 𝒙 𝒏 =𝒖 𝒏−𝟏
𝑧−1 𝑧−1
𝟏
𝒙𝟏 𝒏 = 𝒖 𝒏 𝒁−𝟏
𝑿 𝒁 = 𝒙𝟐 𝒏 = 𝒖 𝒏 − 𝟏 𝑿 𝒁 =
(𝟏−𝒁−𝟏 ) (𝟏−𝒁−𝟏 )

𝒛 ∞ 𝟏
𝑿 𝒁 = 𝑿 𝒁 =
𝒛−𝟏 𝒙 𝒏 = ෍ 𝒙𝟏 𝒌 𝒙𝟐 (𝒏 − 𝒌) 𝒛−𝟏
𝒌=−∞
𝒏

𝒙 𝒏 = ෍ 𝒖 𝒌 𝒖(𝒏 − 𝟏 − 𝒌)
𝒌=𝟎

𝒙 𝒏 = 𝒖 𝟎 𝒖 𝒏 − 𝟏 + 𝒖 𝟏 𝒖 𝒏 − 𝟐 + 𝒖 𝟑 𝒖 𝒏 − 𝟑 + ⋯ … … … . + 𝒖 𝒏 𝒖 −𝟏

𝒙 𝒏 = 𝟏 + 𝟏 + 𝟏 + ⋯……..+ 𝟏 + 𝟎 = 𝒏

𝒙 𝒏 = 𝒏 𝒖(𝒏)
Using Properties of Z Transform find x[n] for
𝐗(𝐳) = 𝐥𝐨𝐠 𝟏 + 𝒂𝒛−𝟏 , 𝒛 >𝒂
Differentiate X(z) w.r.t Z 𝟏
𝑿 𝒁 = (𝟏−𝒂𝒁−𝟏 )
𝟏
𝑿 𝒁 =
(𝟏+𝒂𝒁−𝟏 )
dX(z) −𝑎𝑧 −2
= 𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏 𝒙 𝒏 = −𝒂 𝒏 𝒖 𝒏
𝑑𝑧 1 + 𝑎𝑧 −1
𝒂
Multiply both sides by Z 𝑿 𝒁 = (𝟏+𝒂𝒁−𝟏 )

dX(z) 𝑎𝑧 −2 × 𝑧 𝒙 𝒏 = 𝒂 −𝒂 𝒏 𝒖 𝒏
−𝑧 =
𝑑𝑧 1 + 𝑎𝑧 −1
Time shifting property
dX(z) 𝑎𝑧 −1
−𝑧 = 𝒂𝒁−𝟏
𝑿 𝒁 = (𝟏+𝒂𝒁−𝟏 )
𝑑𝑧 1 + 𝑎𝑧 −1
Taking Inverse Z transform on both sides 𝒏−𝟏 𝒖
𝒙 𝒏 = 𝒂 −𝒂 𝒏−𝟏
By multiplication property
𝒏−𝟏 𝒖(𝒏 − 𝟏)
𝒏𝒙 𝒏 = 𝒂 −𝒂
𝐝𝑿(𝒛)
𝒛−𝟏 −𝒛 = 𝒏𝒙(𝒏) 𝟏 𝒏
𝟏
𝒅𝒛 𝒙 𝒏 = 𝒂 −𝒂 𝒖(𝒏 − 𝟏)
𝒏 −𝒂

𝒂𝒛−𝟏 − −𝒂 𝒏
𝒛−𝟏 = 𝒂 −𝒂 𝒏−𝟏 𝒖(𝒏 − 𝟏) 𝒙 𝒏 = 𝒖(𝒏 − 𝟏)
𝟏 + 𝒂𝒛−𝟏 𝒏
3. Partial Fraction Method
1
𝑋(𝑧) = ROC 𝒁 > 𝟏
1 + 𝑧 −1 (1 − 𝑧 −1 )2

1. Convert X(z) into positive powers of Z


z3
X(z) =
z + 1 (z − 1)2

X(z) 𝑧2
=
z 𝑧 + 1 (𝑧 − 1)2
2. Apply partial fraction to determine the constants

X(z) 𝐴1 𝐴2 𝐴3
= + +
z 𝑧+1 𝑧−1 (𝑧 − 1)2
X(z) 1ൗ 3ൗ 1ൗ
𝐴1 = (𝑧 + 1) × | = 1/4 X(z) 4 + 4 + 2
z z=-1 =
𝑑 X(z) z 𝑧+1 𝑧−1 (𝑧 − 1)2
𝐴2 = {(𝑧 − 1)2 × }|z=1 = 3/4
𝑑𝑧 z
X(z)
𝐴3 = (𝑧 − 1) × | = 1/2
z z=1
𝒁 𝟏
1ൗ 3ൗ 1ൗ 𝑿 𝒁 = 𝑿 𝒁 =
X(z) (𝒁−𝒂) (𝟏−𝒂𝒁−𝟏 )
= 4 + 4 + 2
z 𝑧+1 𝑧−1 (𝑧 − 1)2 𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏
1ൗ 𝑧 3ൗ 𝑧 1ൗ 𝑧 𝒁 𝟏
X(z) = 4 + 4 + 2 𝑿 𝒁 = 𝑿 𝒁 =
(𝒁+𝒂) (𝟏+𝒂𝒁−𝟏 )
𝑧+1 𝑧−1 (𝑧 − 1)2
𝒙 𝒏 = −𝒂𝒏 𝒖 𝒏
1ൗ 3ൗ 1ൗ 𝑧 −1
X(z) = 4 + 4 + 2
1 + 𝑧 −1 1 − 𝑧 −1 (1 − 𝑧 −1 )2 𝑿 𝒁 = 𝒁 𝒛−𝟏
(𝒁+𝒂)𝟐
𝑿 𝒁 =
(𝟏−𝒂𝒁−𝟏 )𝟐

3. Take Inverse ZT based on ROC ROC 𝒁 > 𝟏 𝒙 𝒏 = 𝒏𝒂𝒏 𝒖 𝒏

𝟏 𝟑 𝟏
𝒙 𝒏 = (−𝟏) 𝒖 𝒏 + (𝟏) 𝒖 𝒏 + 𝒏(𝟏)𝒏 𝒖 𝒏
𝒏 𝒏
𝟒 𝟒 𝟐

𝟏 𝟑 𝟏
𝒙 𝒏 = (−𝟏) + (𝟏) + 𝒏(𝟏)𝒏 𝒖 𝒏
𝒏 𝒏
𝟒 𝟒 𝟐
Determine x(n) for (i) ROC 𝒛 > 𝟏 (ii) ROC 𝒛 < 𝟎. 𝟓 (iii) ROC 0.5 < 𝒛 < 𝟏
1
𝑋(𝑧) =
1 − 1.5𝑧 −1 + 0.5𝑧 −2
1. Convert X(z) into positive powers of Z
z2
X(z) =
z − 1 (z − 0.5)
X(z) 𝑧 x x
= 0.5 1
z z − 1 (z − 0.5)
2. Apply partial fraction to determine the constants

X(z) 𝐴1 𝐴2 2𝑧 𝑧
= + X(z) = −
z 𝑧−1 𝑧 − 0.5 𝑧−1 𝑧 − 0.5
X(z)
𝐴1 = (𝑧 − 1) × | =2 Poles are located at 1 and 0.5
z z=1
X(z)
𝐴2 = (𝑧 − 0.5) × | = -1 2 1
z z=0.5 X(z) = −
1−𝑧 −1 1 − 0.5𝑧 −1
X(z) 2 1
= −
z 𝑧−1 𝑧 − 0.5
2 1
X(z) = − ROC 𝒛 > 𝟏
1 − 𝑧 −1 1 − 0.5𝑧 −1
(i) If ROC 𝑧 > 1, then it is a right sided sequence
Poles are located at 1 and 0.5
𝟏 x x
𝑿 𝒁 = 0.5 1
(𝟏−𝒂𝒁−𝟏 )

𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏
𝒙 𝒏 = 𝟐(𝟏)𝒏 𝒖 𝒏 − (𝟎. 𝟓)𝒏 𝒖 𝒏

𝒙 𝒏 = {𝟐(𝟏)𝒏 − 𝟎. 𝟓 𝒏 }𝒖 𝒏

(ii) If ROC 𝑧 < 0.5, then it is a left sided sequence


𝟏
𝑿 𝒁 =
(𝟏−𝒂𝒁−𝟏 )

𝒙 𝒏 = −𝒂𝒏 𝒖 −𝒏 − 𝟏 x x
0.5 1

𝒙 𝒏 = 𝟐(−𝟏)𝒏 𝒖 −𝒏 − 𝟏 − {−𝟎. 𝟓𝒏 𝒖(−𝒏 − 𝟏)}

𝒙 𝒏 = {𝟐 + 𝟎. 𝟓𝒏 }𝒖(−𝒏 − 𝟏)}
(iii) If ROC 0.5 < 𝑧 < 1, then it is a right and left sided sequence

ROC 𝒛 > 𝟎. 𝟓 ROC 𝒛 < 𝟏

𝟏 𝟏
𝑿 𝒁 = 𝑿 𝒁 =
(𝟏−𝒂𝒁−𝟏 ) (𝟏−𝒂𝒁−𝟏 )

𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏 𝒙 𝒏 = −𝒂𝒏 𝒖 −𝒏 − 𝟏
x x
𝒏 𝒖(𝒏) 0.5 1
𝒙 𝒏 = −𝟐𝒖 −𝒏 − 𝟏 − 𝟎. 𝟓
4. Long Division Method (Power Series Expansion)

In this method, the given X(z) is to be expanded into a power series of the form,

which converges in the given ROC.
𝑿 𝒛 = ෍ 𝑪𝒏 𝒛−𝒏
𝒏=−∞

Then by uniqueness of the Z transform, x(n) = Cn for all n.

Prior to performing the long division, it is usually most convenient to arrange both
the numerator and denominator powers of z - 1 or z.

For a right sided or causal sequence ( ROC is 𝒛 > 𝒓𝒐 ), the numerator and
denominator both will be put in ascending powers of z - 1 (or descending powers of z)

For a left sided or Anti causal sequence ( ROC is 𝒛 < 𝒓𝒐 ), the numerator and
denominator both will be put in descending powers of z - 1 (or ascending powers of z)

The z transform of a two sided sequence 𝑿(𝒛) = ෍ 𝒙(𝒏)𝒛−𝒏
𝒏=−∞

X(z) has both positive and negative powers of z

It is not possible to obtain the two sided sequence by long division

If the sequence x[n] is { causal , right sided, ROC 𝒛 > 𝜶 }


𝑿 𝒛 = ෍ 𝒙 𝒏 𝒛−𝒏 = 𝒙 𝟎 𝒛−𝟎 + 𝒙 𝟏 𝒛−𝟏 + 𝒙 𝟐 𝒛−𝟐 + ⋯ … .


𝒏=𝟎
X(z) will have only negative powers of z

If the sequence x[n] is { Anti causal , Left sided, ROC 𝒛 < 𝜶 }


𝟎

𝑿 𝒛 = ෍ 𝒙 𝒏 𝒛−𝒏 = + ⋯ 𝒙 −𝟑 𝒛𝟑 + 𝒙 −𝟐 𝒛𝟐 + 𝒙 −𝟏 𝒛𝟏 + 𝒙 𝟎 𝒛𝟎
𝒏=−∞
X(z) will have only positive powers of z

The inverse Z transform of X(z) is only determination of x(n),


𝒙 𝟎 , 𝒙 𝟏 , 𝒙 𝟐 , … … . If it is causal
𝒙 𝟎 , 𝒙 −𝟏 , 𝒙 −𝟐 , … … . If it is Anti causal
4. Long Division Method (Power Series Expansion)

1
Obtain the causal or right sided sequence 𝑋(𝑧) =
1 + 𝑧 −1 + 𝑧 −2
𝟏 − 𝒛−𝟏 + 𝒛−𝟑 − 𝒛−𝟒 + 𝒛−𝟔 For a right sided the numerator and
𝟏 + 𝒛−𝟏 + 𝒛−𝟐 𝟏 denominator both will be put in
- + −1
- ascending powers of z - 1
1−𝑧 + 𝑧 −2
𝑧 −1 − 𝑧 −2
- −1 + −2
+
𝑧 −𝑧 − 𝑧 −3
𝑧 −3
- - -
𝑧 −3 + 𝑧 −4 + 𝑧 −5
−𝑧 −4 − 𝑧 −5
+ + +
−𝑧 −4 − 𝑧 −5 − 𝑧 −6
𝑿 𝒛 =𝟏− 𝒛−𝟏 + 𝒛−𝟑 − 𝒛−𝟒 + 𝒛−𝟔
𝑧 −6
𝒙 𝒏 = 𝟏, −𝟏, 𝟎, 𝟏, −𝟏, 𝟎, 𝟏, …
𝑧 𝟏
𝑋 𝑧 = 2 1
, 𝒛 <
2𝑧 − 3𝑧 + 1 𝟐
𝟏
As ( ROC is 𝒛 < 𝟐 ), the numerator and denominator both are arranged in descending

powers of z - 1 (or ascending powers of z)


𝒛 + 𝟑𝒛𝟐 + 𝟕𝒛𝟑 + 𝟏𝟓𝒛𝟒 + ⋯
𝟏 − 𝟑𝒛 + 𝟐𝒛𝟐 𝒛
- + 2-
𝑧 − 3𝑧 + 2𝑧 3
3𝑧 2 + 2𝑧 3
-3𝑧 2 +− 9𝑧 3++ 6𝑧 4
7𝑧 3 − 6𝑧 4
- + -
7𝑧 3 − 21𝑧 4 + 14𝑧 5
15𝑧 4 + ⋯

𝑿 𝒛 = 𝒛 + 𝟑𝒛𝟐 + 𝟕𝒛𝟑 + 𝟏𝟓𝒛𝟒 + ⋯ 𝒙 𝒏 = … … 𝟏𝟓, 𝟕, 𝟑, 𝟏, 𝟎


𝑧2 + 2
𝑋 𝑧 = 3 2
, 𝒛 >𝟏
𝑧 − 3𝑧 + 3𝑧 − 1
For a right sided sequence ( ROC is 𝒛 > 𝒓𝒐 ), the numerator and denominator both
will be put in ascending powers of z - 1 (or descending powers of z)
𝒛−𝟏 + 𝟑𝒛−𝟐 + 𝟖𝒛−𝟑 + 𝟏𝟔𝒛−𝟒 + ⋯
𝑧 3 − 3𝑧 2 + 3𝑧 − 1 𝑧 2 + 2
-2 + - +
𝑧 − 3𝑧 + 3 − 𝑧 −1
3𝑧 − 1 + 𝑧 −1
-3𝑧 −+ 9 +- 9𝑧 −1 +− 3𝑧 −2
8 − 8𝑧 −1 + 3𝑧 −2
- + - +
8− 24𝑧 −1 + 24𝑧 −2 − 8𝑧 −3
16𝑧 −1 − 21𝑧 −2 + 8𝑧 −3
-16𝑧 −1 +− 48𝑧 −2 +- 48𝑧 −3 +− 16𝑧 −4
𝑿 𝒛 = 𝒛−𝟏 + 𝟑𝒛−𝟐 + 𝟖𝒛−𝟑 + 𝟏𝟔𝒛−𝟒 + ⋯
27𝑧 −2 − 40𝑧 −3 + 16𝑧 −4
𝒙 𝒏 = 𝟎, 𝟏, 𝟑, 𝟖, 𝟏𝟔, …
Stability and Causality of Systems in terms of ZT

LSI system is causal if and only if the ROC of the system


function H(z) is exterior of a circle of radius r < ∞

If the system has more than one pole, then system is causal if and only if the
ROC of the system function H(z) is exterior of the largest pole.

If the system has more than one pole, then system is anti causal if and only if
the ROC of the system function H(z) is interior of the smallest pole.

LSI system is stable if and only if the ROC of the system


function H(z) includes the unit circle.
3 − 4𝑧 −1
The system function of the LSI system is given as 𝐻(𝑧) =
1 − 3.5𝑧 −1 + 1.5𝑧 −2
Specify the ROC of H(z) and determine unit sample response h(n) for the following cond
1. Stable System, 2. Causal System, 3. Anti causal System

Solving for A and B


3 − 4𝑧 −1
𝐻(𝑧) =
1 − 3.5𝑧 −1 + 1.5𝑧 −2 𝐴 = 2&𝐵 =1
Rewriting in positive powers of Z 𝐻(𝑧) 2 1
= +
𝑧(3𝑧 − 4) 𝑧 𝑧 − 3 𝑧 − 0.5
𝐻(𝑧) = 2 2𝑧 𝑧
𝑧 − 3.5𝑧 + 1.5 𝐻(𝑧) = +
𝐻(𝑧) (3𝑧 − 4) 𝑧 − 3 𝑧 − 0.5
= 2 2 1
𝑧 𝑧 − 3.5𝑧 + 1.5
𝐻(𝑧) = −1
+
Apply partial fractions 1 − 3𝑧 1 − 0.5𝑧 −1

𝐻(𝑧) 𝐴 𝐵
= +
𝑧 𝑧 − 3 𝑧 − 0.5
𝟐 𝟏
𝑯(𝒛) = −𝟏
+
𝟏 − 𝟑𝒛 𝟏 − 𝟎. 𝟓𝒛−𝟏
(i) Determine ROC & h(n) for stable system.
Since the system is stable, ROC must include unit circle.
To include unit circle, ROC must be within 0.5 & 3

ROC 𝒛 > 𝟎. 𝟓 ROC 𝒛 < 3 x 1


x
0.5 3
𝒉 𝒏 = (𝟎. 𝟓)𝒏 𝒖 𝒏 + 𝟐(−𝟑)𝒏 𝒖 −𝒏 − 𝟏

𝒉 𝒏 = (𝟎. 𝟓)𝒏 𝒖 𝒏 − 𝟐(𝟑)𝒏 𝒖 −𝒏 − 𝟏

(ii) Causal system.


ROC of the system function H(z) is exterior of the largest pole.

ROC 𝒛 > 𝟎. 𝟓 ROC 𝒛 > 𝟑

𝒉 𝒏 = (𝟎. 𝟓)𝒏 𝒖 𝒏 + 𝟐(𝟑)𝒏 𝒖 𝒏


x x
0.5 1 3
𝒉 𝒏 = [(𝟎. 𝟓)𝒏 +𝟐(𝟑)𝒏 ]𝒖 𝒏
(ii) Anti Causal System.
ROC of the system function H(z) is interior of the smallest pole.

ROC 𝒛 < 0.5 ROC 𝒛 < 3


x x
0.5 1 3

𝒉 𝒏 = (−𝟎. 𝟓)𝒏 𝒖 −𝒏 − 𝟏 + 𝟐(−𝟑)𝒏 𝒖 −𝒏 − 𝟏

State the causality & stability of the system defined by the following eqs.
𝟏 𝟏 𝟏
𝑿 𝒛 = −𝟏
+ −𝟏
+
𝟏 − 𝟐𝒛 𝟏 − 𝟓𝒛 𝟏 − 𝟎. 𝟓𝒛−𝟏
ROC 𝒛 > 0.5 ROC 𝒛 > 2 ROC 𝒛 > 5
Inverse Discrete Time Fourier Transform
𝟏
Determine the sequence from the given FT 𝑿(𝒆𝒋𝝎 ) =
(𝟏 − 𝒂𝒆−𝒋𝝎 )(𝟏 − 𝒃𝒆−𝒋𝝎 )
Apply partial fraction
𝑨 𝑩
𝑿 𝒆𝒋𝝎 = +
𝟏 − 𝒂𝒆−𝒋𝝎 𝟏 − 𝒃𝒆−𝒋𝝎

𝑨 𝑩
𝒙=𝒆 −𝒋𝝎
𝑯 𝒆𝒋𝝎 = +
𝟏 − 𝒂𝒙 𝟏 − 𝒃𝒙

Solve for A and B

𝟏 𝒂
𝑨= × 𝟏 − 𝒂𝒙 𝑨=
𝟏 − 𝒂𝒙 𝟏 − 𝒃𝒙 𝒙 = 𝟏ൗ𝒂 (𝒂 − 𝒃)

𝟏
𝑩= × 𝟏 − 𝒃𝒙 −𝒃
𝟏 − 𝒂𝒙 𝟏 − 𝒃𝒙 𝒙 = 𝟏ൗ𝒃 𝑩=
(𝒂 − 𝒃)

𝒂ൗ 𝒃ൗ
𝒂−𝒃 𝒂−𝒃 𝟏 𝒂 𝒃
𝑿 𝒆𝒋𝝎 = − = −𝒋𝝎

𝟏 − 𝒂𝒙 𝟏 − 𝒃𝒙 𝒂 − 𝒃 𝟏 − 𝒂𝒆 𝟏 − 𝒃𝒆−𝒋𝝎
𝒂ൗ 𝒃ൗ
𝒂−𝒃 𝒂−𝒃 𝟏 𝒂 𝒃
𝑿 𝒆𝒋𝝎 = − = −
𝟏 − 𝒂𝒙 𝟏 − 𝒃𝒙 𝒂 − 𝒃 𝟏 − 𝒂𝒆−𝒋𝝎 𝟏 − 𝒃𝒆−𝒋𝝎

𝟏
Using property 𝒂𝒏 𝒖 𝒏 = 𝑿 𝒆𝒋𝝎 =
(𝟏−𝒂𝒆−𝒋𝝎 )

𝟏
𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏 − 𝒃𝒏 𝒖 𝒏
𝒂−𝒃
𝒂 𝒂
𝒙 𝒏 = 𝒂𝒏 𝒖 𝒏 − 𝒃𝒏 𝒖 𝒏
𝒂−𝒃 𝒂−𝒃
𝒂𝒏+𝟏 𝒃𝒏+𝟏
𝒙 𝒏 = − 𝒖 𝒏
𝒂−𝒃 𝒂−𝒃

𝟏
𝒙 𝒏 = 𝒂𝒏+𝟏 − 𝒃𝒏+𝟏 𝒖 𝒏
𝒂−𝒃
Determine the sequence x(n) for the given FT
𝒋𝝎
−𝟐
𝑿 𝒆𝒋𝝎 =𝒆 𝒇𝒐𝒓 − 𝝅 ≤ 𝝎 ≤ 𝝅
𝟏 𝝅
𝒙 𝒏 = න 𝑿(𝒆𝒋𝝎 ) 𝒆𝒋𝝎𝒏 𝒅𝝎
𝟐𝝅 −𝝅

𝟏 𝝅 −𝒋𝝎/𝟐 𝒋𝝎𝒏
𝒙 𝒏 = න 𝒆 𝒆 𝒅𝝎
𝟐𝝅 −𝝅

𝟏 𝝅 𝒋𝝎(𝒏−𝟏)
𝒙 𝒏 = න 𝒆 𝟐 𝒅𝝎
𝟐𝝅 −𝝅
𝝅
𝟏 𝟏
𝒋𝝎(𝒏−𝟐)
𝒙 𝒏 = 𝒆
𝟏
𝟐𝝅𝒋(𝒏 − )
𝟐 −𝝅

𝟏 𝟏
𝒋𝝅(𝒏− )
𝟏
−𝒋𝝅(𝒏− )
𝒙 𝒏 = 𝒆 𝟐 − 𝒆 𝟐
𝟏
𝟐𝝅𝒋(𝒏 − )
𝟐 𝟏
sin 𝝅(𝒏 − )
𝒙 𝒏 = 𝟐
𝒆𝒋𝜽 − 𝒆−𝒋𝜽 𝟏
= sin 𝜽 𝝅(𝒏 − )
𝟐𝒋 𝟐
Discrete Fourier series (DFS)
Any periodic function/signals can be expressed in Fourier series (FS)
X[Ꞷ] Discrete
x(t) Continuous
Periodic Aperiodic
FS spectrum
signal

Discrete Discrete
X[k]
Periodic Periodic
x[n] spectrum
signal
DFS

The Fourier series representation of a periodic discrete-time sequence is called discrete Fourier serious (DFS)

1
Discrete Fourier series (DFS)
Consider a periodic discrete time signal x[n] with period N
i.e. 𝑥[𝑛] = 𝑥[𝑛 + 𝑘𝑁] for any integer value of k.

Periodic function x[n] can be synthesized/expressed as

1. Sum of sine & cosine sequences –Trigonometric form of Fourier series

2. Linear combination of complex exponentials – Exponential form of Fourier series

Whose frequencies are multiples of fundamental frequency 𝜔0 = 2𝜋Τ𝑁


Exponential form of Discrete Fourier series
The discrete time periodic signals with period N have 𝑥 𝑁 + 𝑛 = 𝑥(𝑛)

The exponential form of the Fourier series for discrete time signal x[n] is given by

𝑁−1
1 𝑗2𝜋𝑘𝑛ൗ
𝑥[𝑛] = ෍ 𝑋(𝑘)𝑒 𝑁
𝑁
𝑘=0

Here 𝑿(𝒌) , 𝒌 = 𝟎, 𝟏, … … . . 𝑵 − 𝟏 are the coefficients of series expansion. They are also called Fourier coefficients.

These coefficients are derived from x(n) as,

𝑁−1 𝑁−1 𝑁−1


−𝑗2𝜋𝑘𝑛ൗ
−𝑗 2𝜋ൗ𝑁 𝑛𝑘
𝑋 𝑘 =෍ 𝑥(𝑛)𝑒 𝑁 =෍ 𝑥(𝑛)𝑒 = ෍ 𝑥(𝑛)𝑊𝑁𝑛𝑘
𝑛=0 𝑛=0 𝑛=0

−𝑗 2𝜋Τ
where 𝑊𝑁 = 𝑒 𝑁

The equation for x[n] & X(k) are called DFS synthesis and analysis pair. Hence, X(k) & x[n] are periodic sequences.
Trigonometric form of Discrete Fourier series
∞ ∞

𝑥 𝑡 = 𝑎𝑜 + ෍ 𝑎𝑛 cos 𝑛𝜔0 𝑡 + ෍ 𝑏𝑛 sin 𝑛𝜔0 𝑡


𝑛=1 𝑛=1
𝑇 𝑇 𝑇
1 2 2
𝑎𝑜 = න 𝑥 𝑡 𝑑𝑡 𝑎𝑛 = න 𝑥 𝑡 cos 𝑛𝜔0 𝑡 𝑑𝑡 𝑏𝑛 = න 𝑥 𝑡 sin 𝑛𝜔0 𝑡 𝑑𝑡
𝑇 𝑇 𝑇
0 0 0
n = 1, 2, 3, 4, …….

Alternate form DFS for Even N


𝑁ൗ −1 𝑁ൗ −1
2 2
2𝜋 2𝜋 𝑁
𝑥 𝑛 =𝐴 0 + ෍ 𝐴(𝑘) cos 𝑘 𝑛 + ෍ 𝐵(𝑘) sin 𝑘 𝑛 +𝐴 cos 𝜋𝑛
𝑁 𝑁 2
𝑘=1 𝑘=1
𝑁−1 𝑁−1
𝑛 1 2 2𝜋 𝑁
cos 𝑛𝜋 = −1 𝐴 0 = ෍ 𝑥(𝑛) 𝐴 𝑘 = ෍ 𝑥(𝑛) cos 𝑘 𝑛 , 𝑘 = 1,2, … . . … , − 1
𝑁 𝑁 𝑁 2
𝑛=0 𝑛=0

𝑁−1 𝑁−1
2 2𝜋 𝑁 𝑁 1
𝐵 𝑘 = ෍ 𝑥(𝑛) sin 𝑘 𝑛 , 𝑘 = 1,2, … . . … , −1 𝐴 = ෍ 𝑥(𝑛) cos 𝜋𝑛
𝑁 𝑁 2 2 𝑁
𝑛=0 𝑛=0
Trigonometric form of DFs
𝑁−1 /2 𝑁−1 /2
2𝜋 2𝜋 𝑁
𝑥 𝑛 =𝐴 0 + ෍ 𝐴(𝑘) cos 𝑘 𝑛 + ෍ 𝐵(𝑘) sin 𝑘 𝑛 +𝐴 cos 𝜋𝑛
𝑁 𝑁 2
𝑘=1 𝑘=1

DFS for N is odd

𝑁−1 𝑁−1
1 2 2𝜋 𝑁−1
𝐴 0 = ෍ 𝑥(𝑛) 𝐴 𝑘 = ෍ 𝑥(𝑛) cos 𝑘 𝑛 , 𝑘 = 1,2, … . . … ,
𝑁 𝑁 𝑁 2
𝑛=0 𝑛=0

𝑁−1 𝑁−1
2 2𝜋 𝑁−1 𝑁 1
𝐵 𝑘 = ෍ 𝑥(𝑛) sin 𝑘 𝑛 , 𝑘 = 1,2, … . . … , 𝐴 = ෍ 𝑥(𝑛) cos 𝜋𝑛
𝑁 𝑁 2 2 𝑁
𝑛=0 𝑛=0

𝑵−𝟏
If N is odd, A(0) remains same as in the case of N is even, A(k) & B(k) are good for 𝒌 = 𝟏, 𝟐, … . . … , ,
𝟐

There will not be A(N/2) coefficient


Relationship between the Exponential & Trigonometric forms of DFs
For N is even, relationship between X(k) & A(k) and B(k)

𝑋(0) 𝑋 𝑘 + 𝑋(𝑁 − 𝑘) 𝑁
𝐴 0 = 𝐴 𝑘 = , 𝑘 = 1,2, … , − 1
𝑁 𝑁 2

𝑗 𝑋 𝑘 − 𝑋(𝑁 − 𝑘) 𝑁
𝐵 𝑘 = , 𝑘 = 1,2, … , − 1
𝑁 2
For N is odd, relationship between X(k) & A(k) and B(k)

𝑋(0)
𝐴 0 =
𝑁
𝑋 𝑘 + 𝑋(𝑁 − 𝑘) 𝑁−1
𝐴 𝑘 = , 𝑘 = 1,2, … , Note:
𝑁 2
A(0) – remains same
𝑗 𝑋 𝑘 − 𝑋(𝑁 − 𝑘) 𝑁−1 𝑁−1
𝐵 𝑘 = , 𝑘 = 1,2, … , A(k) & B(k) will be for 𝑘 = 1,2, … ,
2
𝑁 2
Find both the exponential and trigonometric forms of the DFS representation of x(n) for the figure shown
x(n) 3
2
1
The exponential form of DFS is given by

𝑁−1 𝑁−1
-3 -2 -1 0 1 2 3 4 5 6 7 8 9
1 𝑗2𝜋𝑘𝑛ൗ 1
𝑥 𝑛 = ෍ 𝑋(𝑘)𝑒 𝑁 = ෍ 𝑋(𝑘)𝑊𝑁−𝑛𝑘 for all n
𝑁 𝑁 −𝑗 2𝜋ൗ4 ×0
𝑘=0 𝑘=0 𝑊40 = 𝑒 =1
𝑁−1 𝑁−1
−𝑗2𝜋𝑘𝑛ൗ −𝑗 2𝜋ൗ4 ×1 −𝑗 𝜋ൗ2
𝑋 𝑘 =෍ 𝑥(𝑛)𝑒 𝑁 =෍ 𝑥(𝑛)𝑊𝑁𝑛𝑘 for all k 𝑊41 = 𝑒 = 𝑒
𝑛=0 𝑛=0
𝜋 𝜋
= cos − 𝑗 sin = −𝑗
To determine the exponential form of the DFS, calculate 2 2
𝑊𝑁𝑘 for different value of k. −𝑗 2𝜋ൗ4 ×2
𝑊42 = 𝑒 = 𝑒 −𝑗 𝜋
= −1
−𝑗 2𝜋ൗ𝑁 𝑘 −𝑗 2𝜋ൗ4 ×3 −𝑗 3𝜋ൗ2
𝑊𝑁𝑘 = 𝑒 From the problem, N = 4 𝑊43 = 𝑒 = 𝑒 =𝑗
2𝜋ൗ ×4
K varies from 0 to 3 (N-1) 𝑊44 = 𝑒 −𝑗 4 =1
𝑁−1 𝑁−1
1 𝑗2𝜋𝑘𝑛ൗ 1
The exponential form of DFS is given by 𝑥 𝑛 = ෍ 𝑋(𝑘)𝑒 𝑁 = ෍ 𝑋(𝑘)𝑊𝑁−𝑛𝑘 for all n
𝑁 𝑁
𝑘=0 𝑘=0
𝑁−1 𝑁−1
where 𝑋 𝑘 =෍ 𝑥(𝑛)𝑒
−𝑗2𝜋𝑘𝑛ൗ
𝑁 = ෍ 𝑥(𝑛)𝑊𝑁𝑛𝑘 for all k
𝑛=0 𝑛=0
4−1
(0)𝑛
For k=0, 𝑋 0 = ෍ 𝑥(𝑛)𝑊4 = 𝑥 0 + 𝑥 1 + 𝑥(2) + 𝑥 3 = 0 + 1 + 2 + 3 = 𝟔
𝑛=0
3
(1)𝑛
For k=1, 𝑋 1 = ෍ 𝑥(𝑛)𝑊4 = 𝑥 0 𝑊40 + 𝑥 1 𝑊41 + 𝑥 2 𝑊42 + 𝑥 3 𝑊43
𝑛=0
= 0 1 + 1 −𝑗 + 2 −1 + 3 𝑗 = −𝟐 + 𝒋𝟐
3
(2)𝑛
For k=2, 𝑋 2 = ෍ 𝑥(𝑛)𝑊4 = 𝑥 0 𝑊40 + 𝑥 1 𝑊42 + 𝑥 2 𝑊44 + 𝑥 3 𝑊46
𝑛=0
= 0 + 1 −1 + 2 1 + 3 −1 = −𝟐
3
For k=3, 𝑋 3 = ෍ 𝑥(𝑛)𝑊4(3)𝑛 = 𝑥 0 𝑊40 + 𝑥 1 𝑊43 + 𝑥 2 𝑊46 + 𝑥 3 𝑊49
𝑛=0
= 0 + 1 𝑗 + 2 −1 + 3 −𝑗 = −𝟐 − 𝒋𝟐
𝑁−1 𝑁−1
−𝑗2𝜋𝑘𝑛ൗ
𝑋 𝑘 =෍ 𝑥(𝑛)𝑒 𝑁 = ෍ 𝑥(𝑛)𝑊𝑁𝑛𝑘 for all k
𝑛=0 𝑛=0

𝑁−1 𝑁−1
1 𝑗2𝜋𝑘𝑛ൗ 1
𝑥 𝑛 = ෍ 𝑋(𝑘)𝑒 𝑁 = ෍ 𝑋(𝑘)𝑊𝑁−𝑛𝑘 for all n
𝑁 𝑁
𝑘=0 𝑘=0

4−1
1 −𝑛𝑘
1
𝑥 𝑛 = ෍ 𝑋(𝑘)𝑊4 = 𝑋 0 + 𝑋 1 𝑊4−𝑛 + 𝑋 2 𝑊4−2𝑛 + 𝑋(3)𝑊4−3𝑛
4 4
𝑘=0

1 2𝜋ൗ 𝑛 𝑗 2𝜋ൗ4 2𝑛 𝑗 2𝜋ൗ4 3𝑛


= 6 + −2 + 𝑗2 𝑒 𝑗 4 + −2 𝑒 + −2 − 𝑗2 𝑒
4

1 𝜋ൗ 𝑛 3𝜋ൗ 𝑛
= 3 + −1 + 𝑗1 𝑒 𝑗 2 − 𝑒 𝑗𝜋𝑛 − 1 + 𝑗 𝑒 𝑗 2 Answer
2
𝑁ൗ −1 𝑁ൗ −1
2 2
2𝜋 2𝜋 𝑁
𝑥 𝑛 = 𝐴 0 + ෍ 𝐴(𝑘) cos 𝑘 𝑛 + ෍ 𝐵(𝑘) sin 𝑘 𝑛 +𝐴 cos 𝜋𝑛
𝑁 𝑁 2
𝑘=1 𝑘=1

For N is even, relationship between X(k) & A(k) and B(k)

𝑋(0)
𝐴 0 =
𝑁
𝑋 𝑘 + 𝑋(𝑁 − 𝑘) 𝑁 𝑗 𝑋 𝑘 − 𝑋(𝑁 − 𝑘) 𝑁
𝐴 𝑘 = , 𝑘 = 1,2, … , − 1 𝐵 𝑘 = , 𝑘 = 1,2, … , − 1
𝑁 2 𝑁 2
To determine trigonometric form of DFS , A(0), A(1), B(1) and A(2) are determined as follow

4 𝑋(0) 6 3 𝑋 1 + 𝑋(4 − 1) −2 + 𝑗2 + −2 − 𝑗2
𝑘 = 1 𝑡𝑜 1 −1 =1 𝐴 0 = = = 𝐴 1 = = = −1
2 𝑁 4 2 4 4

𝑋 1 + 𝑋(4 − 1) −2 + 𝑗2 − −2 − 𝑗2 4 𝑋(4ൗ2) 1
𝐵 1 =𝑗 =𝑗 = −1 𝐴 = =−
4 4 2 4 2
1 1
2𝜋 2𝜋
𝑘 = 1, 𝑎𝑛𝑑 𝑁 = 4 𝑥 𝑛 = 𝐴 0 + ෍ 𝐴(𝑘) cos 𝑘 𝑛 + ෍ 𝐵(𝑘) sin 𝑘 𝑛 + 𝐴 2 cos 𝜋𝑛
4 4
𝑘=1 𝑘=1

𝜋 𝜋 3 𝜋 𝜋 1
𝑥 𝑛 = 𝐴 0 + 𝐴 1 cos 𝑛 + 𝐵(1) sin 𝑛 + 𝐴 2 cos 𝜋𝑛 𝑥 𝑛 = − 𝑐𝑜𝑠 𝑛 − 𝑠𝑖𝑛 𝑛 − 𝑐𝑜𝑠 𝜋𝑛
2 2 2 2 2 2
Find both the exponential and trigonometric forms of the DFS representation of x(n) for the figure shown

x(n) 3
2
1

-3 -2 -1 0 1 2 3 4 5 6 7 8 9

The exponential form of DFS is given by 1 𝜋ൗ 𝑛 3𝜋ൗ 𝑛


= 3 + −1 + 𝑗1 𝑒 𝑗 2 − 𝑒 𝑗𝜋𝑛 − 1 + 𝑗 𝑒 𝑗 2
2

3 𝜋 𝜋 1
The trignometric form of DFS is given by 𝑥 𝑛 = − 𝑐𝑜𝑠 𝑛 − 𝑠𝑖𝑛 𝑛 − 𝑐𝑜𝑠 𝜋𝑛
2 2 2 2
Find the DFS of a periodic impulse train.

1, 𝑛 = 𝑟𝑁, 𝑟 − 𝑎𝑛𝑦 𝑖𝑛𝑡𝑒𝑟𝑔𝑒𝑟
We Consider the periodic impulse train 𝑥 𝑛 = ෍ 𝛿 𝑛 − 𝑟𝑁 = ቊ
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
𝑟=−∞

𝑥 𝑛 = 𝛿 𝑛 𝑓𝑜𝑟 0 ≤ 𝑛 ≤ 𝑁 − 1, 𝑓𝑖𝑛𝑑 𝑡ℎ𝑒 𝐷𝐹𝑆 𝑐𝑜𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑡𝑠 𝑋(𝑘)

𝑁−1 𝑁−1
−𝑗2𝜋𝑘𝑛ൗ
𝑋 𝑘 =෍ 𝛿[𝑛]𝑒 𝑁 = ෍ 𝑥(𝑛)𝑊𝑁𝑛𝑘 = 𝑊𝑁0 = 1 for all k
𝑛=0 𝑛=0

In this case, X[k] is the same for all k.


∞ 𝑁−1 𝑁−1
1 1 2𝜋ൗ 𝑘𝑛
𝑥 𝑛 = ෍ 𝛿 𝑛 − 𝑟𝑁 = ෍ 𝑊𝑁 = ෍ 𝑒 𝑗
−𝑘𝑛 𝑁
𝑁 𝑁
𝑟=−∞ 𝑘=0 𝑘=0

DFS result produced is a useful representation of a periodic impulses train in terms of sum of complex
exponentials, where all the complex exponentials have the same magnitude and phase and add to unity at
integer multiples of N and to zero for all other integers(Duality).
Periodicity of Fourier Coefficients
The Fourier coefficient X(k) represent x(n) in frequency domain. These coefficients are periodic with period ‘N’
𝑁−1
1 −𝑗2𝜋𝑘𝑛ൗ
W.K.T 𝑋 𝑘 = ෍ 𝑥(𝑛) 𝑒 𝑁 (1)
𝑁
𝑛=0
𝑁−1
1 −𝑗2𝜋𝑘𝑛ൗ
× 𝑒 −𝑗2𝜋𝑛
Consider X(k + N), 𝑋 𝑘 + 𝑁 = ෍ 𝑥(𝑛) 𝑒 𝑁 (2)
𝑁
𝑛=0
Here kn – is an integer as k & n are integer.

𝑒 −𝑗2𝜋𝑛 = cos 2𝜋𝑛 − 𝑗 sin 2𝜋𝑛 = 1 (𝑎𝑙𝑤𝑎𝑦𝑠)


𝑁−1
1 −𝑗2𝜋𝑘𝑛ൗ
Hence Eq(2) becomes 𝑋 𝑘 + 𝑁 = ෍ 𝑥(𝑛) 𝑒 𝑁 (3)
𝑁
𝑛=0
The RHS of the above equation is X(k) from Eq (1), Hence, 𝑋 𝑘 + 𝑁 = 𝑋(𝑘) (4)
Fourier coefficients are periodic.
It satisfy the properties like linearity, duality, shifting, symmetry properties etc.
Properties of DFS
Property Periodic sequence (Period N) DFS coefficients
1. Linearity 𝑎𝑥1 𝑛 + 𝑏𝑥2 (𝑛) 𝑎𝑋1 𝑘 + 𝑏𝑋2 (𝑘)
𝑥 𝑛−𝑚 𝑊𝑁𝑘𝑚 𝑋 𝑘
2. Time shifting
𝑥(𝑛 + 𝑚) 𝑊𝑁−𝑘𝑚 𝑋(𝑘)
3. Frequency shifting 𝑊𝑁𝑙𝑛 𝑋 𝑘 𝑋 𝑘+𝑙
𝑁−1
4. Periodic convolution ෍ 𝑥1 𝑚 𝑥2 𝑛 − 𝑚 𝑋1 (𝑘)𝑋2 (𝑘)
𝑚=0
𝑁−1
1
5. Multiplication 𝑥1 𝑛 𝑥2 (𝑛) ෍ 𝑋1 (𝑙)𝑋2 (𝑘 − 𝑙)
𝑁
𝑙=0

𝑥∗ 𝑛 𝑋 ∗ −𝑘
𝑥 ∗ −𝑛 𝑋∗ 𝑘
𝑅𝑒[𝑥 𝑛 ] 1
6. Symmetry Property 𝑋𝑒 𝑘 = 𝑋 𝑘 + 𝑋 ∗ (−𝑘)
𝑗𝐼𝑚[𝑥 𝑛 ] 2
1 1
𝑥𝑒 𝑛 = 𝑥 𝑛 + 𝑥 ∗ (−𝑛) 𝑋𝑜 𝑘 = 𝑋 𝑘 − 𝑋 ∗ (−𝑘)
2 2
1 𝑅𝑒[𝑋 𝑘 ]
𝑥𝑜 𝑛 = 𝑥 𝑛 − 𝑥 ∗ (−𝑛) 𝑗𝐼𝑚[𝑋 𝑘 ]
2
Tutorial on Z and DTFT
Tutorial – Z Transform
1 − 𝑧 −1 cos 𝜔𝑜
𝑥(𝑛) = cos 𝜔𝑜 𝑛 𝑢(𝑛) 𝑋(𝑧) =
1 − 2𝑧 −1 cos 𝜔𝑜 + 𝑧 −2

𝑧 −1 sin 𝜔𝑜
𝑥(𝑛) = sin 𝜔𝑜 𝑛 𝑢(𝑛) 𝑋(𝑧) =
1 − 2𝑧 −1 cos 𝜔𝑜 + 𝑧 −2

1 − 𝑎𝑧 −1 cos 𝜔𝑜
𝑥(𝑛) = 𝑎𝑛 cos 𝜔𝑜 𝑛 𝑢(𝑛) 𝑋(𝑧) =
1 − 2𝑎𝑧 −1 cos 𝜔𝑜 + 𝑎2 𝑧 −2
𝑑 2𝑋 𝑧 𝑑𝑋(𝑧)
𝑦(𝑛) = 𝑛2 𝑎𝑛 𝑥(𝑛) 𝑋 𝑧 =𝑧 2 +𝑧
𝑑𝑧 2 𝑑𝑧
sin[(𝑛 + 1)𝜔 1
𝑥(𝑛) = 𝑟 𝑛 𝑢(𝑛), 0 < 𝑟 < 1 𝑋(𝑧) =
𝑠𝑖𝑛𝜔 1 − 2𝑟𝑧 −1 cos 𝜔 + 𝑟 2 𝑧 −2

2𝑧 2
𝑥(𝑛) = 𝑛 𝑛 + 1 𝑢(𝑛) 𝑋(𝑧) =
(𝑧 − 1)3
Tutorial – DTFT
𝑥 𝑛 = 1, −1,1, −1 𝑿 𝒆𝒋𝝎 = (𝟏 − 𝒆−𝒋𝟐𝝎 )(𝟏 − 𝒆−𝒋𝝎 )

𝑥 𝑛 = 1, 0<𝑛<𝑁 1 − 𝒆−𝒋𝝎(𝑵+𝟏)
= 0 elsewhere 𝑿 𝒆𝒋𝝎 =
1 − 𝒆−𝒋𝝎
𝑥 𝑛 = 𝛿 𝑛 − 𝛿(𝑛 − 1) 𝑿 𝒆𝒋𝝎 = (𝟏 − 𝒆−𝒋𝝎 )

𝑥 𝑛 = 𝑢 𝑛 − 𝑢(𝑛 − 1) 𝑿 𝒆𝒋𝝎 = 𝟏

1 − 𝑎7 𝒆−𝒋𝟕𝝎
𝑥 𝑛 = 𝑎𝑛 [𝑢 𝑛 − 𝑢 𝑛 − 8 ], 𝑎 < 1 𝑿 𝒆𝒋𝝎 =
1 − 𝒂𝒆−𝒋𝝎

𝒋𝝎
1
𝑦(𝑛) = (1Τ2)𝑛 𝑢(−𝑛) 𝑿 𝒆 =
1 − 𝟐𝒆−𝒋𝝎
1 − 𝑎2
𝑥(𝑛) = (𝑎) 𝑛 𝑥(𝑛) = (1ൗ2) 𝑛 𝑿 𝒆𝒋𝝎 =
1 − 𝟐𝒂 cos 𝝎 + 𝒂𝟐
𝑛𝜋
Find the DTFT of the following sequences (a) 𝑥 𝑛 = sin 𝑢(𝑛)
2

𝑛𝜋
𝐷𝑇𝐹𝑇 𝑥 𝑛 =𝑋 𝜔 = ෍ 𝑥 𝑛 𝑒 −𝑗𝜔𝑛 (b) 𝑥 𝑛 = cos 𝑢(𝑛)
3
𝑛=−∞

𝒏𝝅
(a) 𝒙 𝒏 = 𝒔𝒊𝒏 𝒖(𝒏)
𝟐


𝑛𝜋 1 1 1
𝑋 𝜔 = ෍ sin 𝑢(𝑛) 𝑒 −𝑗𝜔𝑛 = −
2 2𝑗 1 − 𝑒 𝑗 𝜋ൗ2 −𝜔 𝜋ൗ +𝜔
𝑛=−∞ 1 − 𝑒 −𝑗 2


𝑛𝜋 −𝑗𝜔𝑛
= ෍ sin 𝑒 𝜋 𝜋
2 1 1 − 𝑒 −𝑗 ൗ2 𝑒 −𝑗𝜔 − 1 + 𝑒 𝑗 ൗ2 𝑒 −𝑗𝜔
𝑛=0 =
∞ ∞
2𝑗 1 + 𝑒 −𝑗2𝜔 − 𝑒 −𝑗𝜔 𝑒 𝑗 𝜋ൗ2 + 𝑒 −𝑗 𝜋ൗ2
1 𝜋ൗ −𝜔 𝑛 𝜋ൗ +𝜔 𝑛
= ෍ 𝑒𝑗 2 − ෍ 𝑒 −𝑗 2
2𝑗
𝑛=0 𝑛=0
𝑒 −𝑗𝜔 sin 𝜋ൗ2
=
∞ 𝑛𝜋ൗ 𝑛𝜋ൗ 1 + 𝑒 −𝑗2𝜔 − 𝑒 −𝑗𝜔 cos 𝜋ൗ2
𝑒𝑗 2 − 𝑒 −𝑗 2
=෍ 𝑒 −𝑗𝜔𝑛
2𝑗
𝑛=0 𝑒 −𝑗𝜔
=
1 + 𝑒 −𝑗2𝜔
1 𝑛 𝑛𝜋
Find the DTFT of the following sequences (a) 𝑥 𝑛 = sin 𝑢(𝑛)
2 4

1 𝑛−2
(b) 𝑥 𝑛 = 𝑢(𝑛 − 2) c) cos 𝜔0 𝑛 𝑢(𝑛) d) sin 𝜔0 𝑛 𝑢(𝑛)
2
1 𝑛 𝑛𝜋
(a) 𝑥 𝑛 = sin 𝑢(𝑛)
2 4
∞ 𝑛
1 𝑛𝜋
𝑋 𝜔 = ෍ sin 𝑢(𝑛) 𝑒 −𝑗𝜔𝑛 1 1 1
2 4 = −
𝑛=−∞ 𝜋ൗ −𝜔 𝜋ൗ +𝜔
2𝑗 1 − 1ൗ 𝑒 𝑗 4 1 − 1ൗ2 𝑒 −𝑗 4
∞ 𝑛 2
1 𝑛𝜋 −𝑗𝜔𝑛
=෍ sin 𝑒
2 4 1ൗ 𝑒 −𝑗𝜔 sin 𝜋ൗ
𝑛=0
= 2 4
∞ 𝑛𝜋ൗ 𝑛𝜋ൗ
1 + 1ൗ4 𝑒 −𝑗2𝜔 − 𝑒 −𝑗𝜔 cos 𝜋ൗ4
𝑛
1 𝑒𝑗 4 − 𝑒 −𝑗 4
=෍ 𝑒 −𝑗𝜔𝑛 1ൗ 𝑒 −𝑗𝜔
2 2𝑗 2 2
𝑛=0 =
1 − 1ൗ 𝑒 −𝑗𝜔 + 1ൗ4 𝑒 −𝑗2𝜔
∞ 𝑛 ∞ 𝑛 2
1 1 𝜋 1 𝜋ൗ +𝜔 𝑛
= ෍ 𝑒 𝑗 ൗ4 −𝜔 𝑛 −෍ 𝑒 −𝑗 4
2𝑗 2 2
𝑛=0 𝑛=0

∞ 𝑛 ∞ 𝑛
1 1 𝑗 𝜋ൗ −𝜔 1 −𝑗 𝜋ൗ +𝜔
= ෍ 𝑒 4 −෍ 𝑒 4
2𝑗 2 2
𝑛=0 𝑛=0
𝑛−2
1
𝑥 𝑛 = 𝑢(𝑛 − 2)
2

𝐷𝑇𝐹𝑇 𝑥 𝑛 = 𝑋 𝜔 = ෍ 𝑥 𝑛 𝑒 −𝑗𝜔𝑛
𝑛=−∞

∞ 𝑛−2
1
𝑋 𝜔 = ෍ 𝑢(𝑛 − 2) 𝑒 −𝑗𝜔𝑛
2
𝑛=−∞
∞ 𝑛−2
1
=෍ 𝑒 −𝑗𝜔𝑛
2
𝑛=2
2
−𝑗2𝜔
1 −𝑗3𝜔 1
=𝑒 + 𝑒 + 𝑒 −𝑗4𝜔 + ⋯
2 2

2
1 −𝑗𝜔 1
= 𝑒 −𝑗2𝜔 1+ 𝑒 + 𝑒 −𝑗2𝜔 + ⋯
2 2

𝑒 −𝑗2𝜔
=
1 − 1ൗ2 𝑒 −𝑗𝜔
𝑥 𝑛 = cos 𝜔0 𝑛 𝑢(𝑛)

𝐷𝑇𝐹𝑇 𝑥 𝑛 = 𝑋 𝜔 = ෍ cos 𝜔0 𝑛 𝑢(𝑛) 𝑒 −𝑗𝜔𝑛


𝑛=−∞


𝑒 𝑗𝜔0 𝑛 + 𝑒 −𝑗𝜔0 𝑛 −𝑗𝜔𝑛
𝑋 𝜔 = ෍ 𝑒
2
𝑛=−∞

∞ ∞
1 𝜔0 −𝜔 𝑛 𝜔0 +𝜔 𝑛
= ෍ 𝑒𝑗 + ෍ 𝑒 −𝑗
2
𝑛=0 𝑛=0

1 1 1
= +
2 1 − 𝑒 𝑗 𝜔0−𝜔 1 − 𝑒 −𝑗 𝜔0+𝜔

1 1 − 𝑒 −𝑗 𝜔0+𝜔 + 1 − 𝑒 𝑗 𝜔0−𝜔
=
2 1 + 𝑒 −𝑗2𝜔 − 𝑒 −𝑗𝜔 𝑒 𝑗𝜔0 + 𝑒 −𝑗𝜔0

1 − 𝑒 −𝑗𝜔 cos 𝜔0
=
1 − 2𝑒 −𝑗𝜔 cos 𝜔0 + 𝑒 −𝑗2𝜔

You might also like