Digital Signal Processing

Dr. Abhishek Kumar

Assistant Professor
ICT, SOT-PDEU Gandhinagar
E-Mail: [email protected] , [email protected]
Mob:+91-9015532511

1
2
3
4
5
Syllabus

6
  Morris Mano, “Digital Design”, PHI.
 J. Millman, C. Halkias and C. Parikh, “Integrated Electronics”, Tata McGraw Hill.
Internal Assessment
 Course Assignments: Share and align assignment topics, due dates, and grading criteria.

Theory IA Marks distribution:

Type Marks Time

Internal Assisment-1 10 Marks
Multiple Choice Quiz Before Mid Semester
Internal Assisment-2 Assignments from Unit-1, 2, 3, 15 Marks
Throughout semester
and 4

Lab Marks distribution:

 Lab Work (LAB IA): Continuous performance (15), journal (20), viva (15)
 Lab Viva (LAB Exam): Write-up (15), Performance (15), Viva (20)
**5-Marks may vary based on your class behavior**
 Course Lab Practical: Coordinate lab schedules, experiments, and safety protocols.
Sinusoid:

8
Sinusoid:

9
Sinusoid:

10
Let’s Start

11
Digital Signal Processing
Processing: Signal:
Modify A function that carries some information {f(t)}
Extract
A variation of the dependent variable against the independent
Detect
variable

Continuous-time signals Discrete-time signals

Digital Signals
Continuous signals Discrete Signal
Analog signals

12
Digital Signal Processing
Processing: Signal:
Modify A function that carries some information {f(t)}
Extract
A variation of the dependent variable against the independent
Detect
variable

Continuous-time signals Discrete-time signals

Continuous signals Discrete Signal Digital Signals
Analog signals Sequence

13
Digital Signal Processing
Processing: Signal:
Modify A function that carries some information {f(t)}
Extract
A variation of the dependent variable against the independent
Detect
variable

Continuous-time signals Discrete-time signals

Digital Signals
Continuous signals Discrete Signal
Analog signals

14
Digital signal processing
Advantages:
→ Noise is easy to control after initial quantization
→ Highly linear (within limited dynamic range)
→ Complex algorithms t into a single chip
→ Flexibility, parameters can easily be varied in software
→ Digital processing is insensitive to component tolerances, aging, environmental conditions, electromagnetic
interference

But:
→ Discrete-time processing artifacts (aliasing)
→ Can require signi cantly more power (battery, cooling)
→ Digital clock and switching cause interference

15
fi
fi
Typical DSP applications
experimental
communication aviation
music physics
systems
modulation/demodulation, sensor-data evaluation
synthetic instruments, audio radar, radio navigation
channel equalization, echo
cancellation effects, noise reduction

medical
consumer engineering diagnostics security
electronics
control systems, feature ultrasonic imaging, computer tomography, digital watermarking, biometric
extraction for pattern ECG, EEG, MEG, AED, audiology identi cation, surveillance
recognition systems, signals intelligence
perceptual coding of audio and video on
DVDs, speech synthesis, speech
recognition

geophysics

seismology, oil exploration

16
fi
Sampling and Reconstruction
∞

∑
s(t) = δ(t − nTs)
n=−∞

17
Sampling and Reconstruction
∞

∑
s(t) = δ(t − nTs)
n=−∞

18
Contd.
xs(t) = x(t) . s(t)
∞ ∞ ∞
s(t) =
∑
δ(t − nTs) = a0 +
∑ ( an cos(nωst) + bn sin(nωst)) = 1/Ts +
∑ ( 2/Ts cos(nωst))
n=−∞ n=1 n=1

Ts/2

∫−T /2
a0 = 1/Ts δ(t)dt = 1/Ts
s

Ts/2

∫−T /2
an = 2/Ts δ(t)cos(nωst)dt = 2/Ts cos(nωs0) = 2/Ts
s

Ts/2

∫−T /2
bn = 2/Ts δ(t)sin(nωst)dt = 2/Ts sin(nωs0) = 0
s

19
Contd.
xs(t) = x(t) . s(t)
∞

{ }
= x(t) 1/Ts +
∑ ( 2/Ts cos(nωst))
n=1

= 2/Ts {x(t)/2 + x(t)cos(ωst) + x(t)cos(2ωst) + x(t)cos(3ωst) + ⋯}

F.T
Xs(ω) = 1/Ts {X(ω) + X(ω − ωs) + X(ω + ωs) + X(ω − 2ωs) + X(ω + 2ωs) + ⋯}
∞

∑
Xs(ω) = 1/Ts X(ω + nωs)
n=−∞ 20
A little more understanding of sampling in time domain

21
Reconstruction

22
Reconstruction in Frequency Domain:

23
Reconstruction in Time Domain:

24
Reconstruction:
The ideal low pass filter for eliminating any frequency content above fs /2 and its
Fourier transform is given by;

{1
1 if | f | < fs /2
H( f ) = = rect( f/fs)
if | f | > fs /2
Thus,
πt
sin( t ) 1
h(t) = fs πt s
= sinc (t/ts)
ts
ts

Hence,
̂ = h(t) * xs(t)
x(t)
25
Nyquist Theorem:
A continuous time signal can be represented in its samples and can be
recovered back when sampling frequency fs is greater than or equal to
the twice the highest frequency component of message signal. i. e.

fs ≥ 2fm

26
Aliasing
When: fs < 2fm

Is it always bad ? magnitude FFT of sine

1

0.9

Case 1: 0.8

0.7

0.6
magnitude

0.5

0.4

0.3

0.2

0.1

0
-6000 -4000 -2000 0 2000 4000 6000
Frequency (Hz)
27
28
Digital Signal Processing System:

29
Digital Signal Processing System:

30
Discrete-Time Systems

31
Discrete-Time Systems

32
Discrete Signal Representation:
A. Functional representation: ⎧ 1 for n = 1,3
⎪
x(n) = ⎨ 6 for n = 0,7
⎪0 elsewhere
33
⎩

B. Graphical x ( n)
representation

33
n
 Discrete Signal Representation:
C. Tabular representation:
n … -2 -1 0 1 2

x(n) … 0.12 2.01 1.78 5.23 0.12

x[n] = {1, 0.5, 2, − 0.7, 0.5, − 1, 0, 0.5}

D. Sequence representation:
2

-1
-4 -2 0 2 4
34
Some Elementary Discrete Time Signal
1. Unit Sample Sequence: δ[n]
2. Unit Step Signal: u[n]
3. Unit Ramp Signal: r[n]
n
4. Exponential Signal: x[n] = a ∀ n

{0, for n ≠ 0
1, for n = 0
δ[n] =

35
Some Elementary Discrete Time Signal
1. Unit Sample Sequence: δ[n]
2. Unit Step Signal: u[n]
3. Unit Ramp Signal: r[n]
n
4. Exponential Signal: x[n] = a ∀ n

36
Some Elementary Discrete Time Signal
1. Unit Sample Sequence: δ[n]
2. Unit Step Signal: u[n]
3. Unit Ramp Signal: r[n]
n
4. Exponential Signal: x[n] = a ∀ n

37
Exponential Signal:
n
x[n] = a ∀ n

If a is real; then x[n] is real signal;

n jθn
If a is complex valued; then x[n] =r e ;
n
xR[n] = r cos θn
n
xI[n] = r sin θn
n
| x[n] | = r = A(n) Amplitude Function
∠x[n] = θn = ϕ(n) Phase Function
38
1200 60

1000 50

800

600
0<a<1
40

30
a>1
400 20
X 6.25
Y 12.6058

200 10

0 0
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10

1200 60

50

a<−1
1000

−1 < a < 0
800 40

30
600
20
400
10
200
0
0
-10
-200 -20
-400 -30

-600 -40
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10

39
Some important classification:

1. Energy signals and power signals

2. Periodic signals and aperiodic signals
3. Symmetric (even) and antisymmetric (odd) signals

N
∞ 1 2
∑
2 P = lim | x(n) |
∑
E= | x(n) |
N→∞ 2N + 1
n=−∞ n=−N

40
Some important classification:

1. Energy signals and power signals

2. Periodic signals and aperiodic signals
3. Symmetric (even) and antisymmetric (odd) signals

x(n + N) = x(n) ∀ n

41
Some important classification:

1. Energy signals and power signals

2. Periodic signals and aperiodic signals
3. Symmetric (even) and antisymmetric (odd) signals

x(−n) = x(n) x(−n) = − x(n)

xe(n) = 1/2[x(n) + x(−n)] xo(n) = 1/2[x(n) − x(−n)]

x(n) = xe(n) + xo(n)

42
 Input-Output Model of Discrete-Time System
(input-output relationship description)

x ( n) discrete-time y ( n)
system
input signal H []
. output signal
excitation response
y (n) ≡ H [x(n) ]
H
x(n) ⎯⎯
→ y ( n)
43
Discrete-Time Systems. Definition
A discrete-time system is a device or algorithm that operates on a discrete-time
signal called the input or excitation (e.g. x(n)), according to some rule (e.g.
H[.]) to produce another discrete-time signal called the output or response (e.g.
y(n)).
y (n) ≡ H [x(n) ]

This expression denotes also the transformation H[.] (also called operator
or mapping) or processing performed by the system on x(n) to produce y(n).

44
Discrete-Time Systems. Definition
A discrete-time system is a device or algorithm that operates on a discrete-time
signal called the input or excitation (e.g. x(n)), according to some rule (e.g.
H[.]) to produce another discrete-time signal called the output or response (e.g.
y(n)).
y (n) ≡ H [x(n) ]

This expression denotes also the transformation H[.] (also called operator
or mapping) or processing performed by the system on x(n) to produce y(n).

45
Example
Determine the response of the following system to the input signal

{0
|n| − 3 ≤ n ≤ 3
x[n] =
otherwise
1. y[n] = x[n] (identity system)
2. y[n] = x[n − 1] (unit delay system)
3. y[n] = x[n + 1] (Unit advance system)
1
4. y[n] = [x(n + 1) + x(n) + x[n − 1]] (Moving average lter)
3
5. y[n] = median(x[n + 1], x[n], x[n − 1]) (Median system)
n

∑
6. y[n] = x[k] (Accumulator)
k=−∞
46
fi
 systems. For this purpose we need to define some basic building blocks that
interconnected to form complex systems.
Building Block Diagram Representation of Discrete-Time System
An adder. Figure 2.2.2 illustrates a system (adder) that performs the addi
two signal sequences to form another (the sum) sequence, which wedenote a
Note that it isnot necessary to store either one of the sequences in order to p
the addition. In other words, the addition operation is memoryless.

x1 (n)

y(n) = x(n) + xz(n)

An Adder Figure 2.2.2

Graphical representation of
an adder. Xy(n)

47
 Building Block Diagram Representation of Discrete-Time System

Systems
Constant Multiplier
and
Signas
Discrete-Time
isdepicted by
58 Chapter 2 D operation Note that
This (n). Fig.2.2.3 an
this

ov(n)p=eralioNaix(ny
multiplier. theinputx
constant factoron
A ascale
applying rn)

Figure2.2.3
Graphicalrepresentationof
the,

multiplicaon
nconstant multiplier. illustrates in
2.2.4 denoted
Figure sequence,
multipliler. product) multiplication the figureoft,
Asignal (the
another canvicwthe operation as
toform we 48
58 Chapter 2 D This (n). Note that this
Fig.2.2.3
ov(n)p=eralioNaix(ny
multiplier. theinputx
Buildingconstant
Block factor
Diagram on
Representation of Discrete-Time System
an
A scale
applyinga rn)

Figure2.2.3
Graphicalrepresentationof
the,

multiplicaon
nconstant multiplier. illustrates in
2.2.4 denoted
Signal Multiplier
multipliler.
Figure sequence,
multiplication the figure oft,
Asignal (the
product)
another canvicwthe operation as
form
to wo cases,we
ceding
y(n) = X|(n)xz(n)
me n
xz(n)
Figure2.2.4
Graphicalrepresentationof 49
 Thus
signal paing
memoryattimex(n),
n-1 output is sample
anditisrecalled from memory at time nto iorn
the
input signal is y(n) = x(n-1)
x(n)
Building Block Diagram Representation of Discrete-Time System
Figure 2.2.5 1)
(n-
Graphical representation of y(n) =
element.
the unit delay
the unit of delaybuilding
will block
becomne requires memory.
contrast
apparent to
when the
we
The use of1 a unit
unit
discuss delay,
the the symbol ad
z-transtormin
Unit Adelay element
basic element. In +
Thus this
unit advance
sample in time to yield x(n 1). Figure 2
(n) ahead by one to denote the unit adva
input x operator z being used y(n) = x(n-1)
operation,
Figure 2.2.5 with the x(n)
impossible in real time, since, i
Graphical advance
representation physically
is
of we store
that any such of the signal. On the other hand, if
looking
unit into
delay the future
element.
at any time. In
the sample
unit delay, a unit
Unit advance element
memory of
advance the computer,
element.
A application, it is possible
unit
we
In can
to
in
recall
contrast to
advance
time to
any
the
the
yield signal
x(n + x(n)
1). in
Figure advance
time.
2.2.6ilig
time ahead by one sample denote the unit
input x (n) z being used to advance.
1)
Figure 2.2.6with the operator y(n) =
time,x(n+
since, in fac
operation, physically x(n)
impossible in real s
Graphical
that any such advance is
representation of
the other hand, if we store the
future of
element. the signal. On
the unit advance
the 50
In sucha
Draw a basic building block of the following system:

1 1 1
y[n] = y[n − 1] + x[n] + x[n − 1]
4 2 2

51
Draw a basic building block of the following system:

1 1 1
y[n] = y[n − 1] + x[n] + x[n − 1]
4 2 2

52
Draw a basic building block of the following system:
1 1 1
y[n] = y[n − 1] + x[n] + x[n − 1]
4 2 2

53
Find
y[n] = ?

54
Classi cation of Discrete-Time Systems

Static vs. Dynamic Systems

Time-Invariant vs. Time-Variable Systems
Linear vs. Non-linear Systems
Causal vs. Non-causal Systems
Stable vs. Unstable of Systems
Recursive vs. Non-recursive Systems

55
fi
Static vs. Dynamic Systems. Definition

A discrete-time system is called static or memoryless if its output at any time instant n
depends on the input sample at the same time, but not on the past or future samples of the
input. In the other case, the system is said to be dynamic or to have memory.
If the output of a system at time n is completely determined by the input samples in the
interval from n-N to n ( N ≥ 0 ), the system is said to have memory of duration N.
If N = 0, the system is static or memoryless.
If 0 < N < ∞, the system is said to have finite memory.
If N → ∞ , the system is said to have infinite memory.

56
Examples:
The static (memoryless) systems:
3
y (n) = nx(n) + bx (n)

The dynamic systems with finite memory:

N
y ( n) = ∑ h( k ) x ( n − k )
k =0

The dynamic system with infinite memory:

∞
y ( n) = ∑ h( k ) x ( n − k )
k =0
57
Time-Invariant vs. Time-Variable Systems

A discrete-time system is called time-invariant if its input-output characteristics do not change

with time. In the other case, the system is called time-variable.
Definition. A relaxed system H [.] is time- or shift-invariant if only if
y (n) ≡ H [x(n) ] x(n) ⎯⎯ H
→ y ( n)
implies that
y (n − k ) ≡ H [x(n − k ) ] x(n − k ) ⎯⎯ H
→ y (n − k )
for every input signal x ( n) and every time shift k .

58
Examples:
The time-invariant systems:
3
y (n) = x(n) + bx (n)
N
y ( n) = ∑ h( k ) x ( n − k )
k =0

The time-variable systems:

3
y (n) = nx(n) + bx (n − 1)
N
N −n
y ( n) = ∑ h (k ) x(n − k )
k =0

59
Linear vs. Non-linear Systems
A discrete-time system is called linear if only if it satisfies the linear
superposition principle. In the other case, the system is called non-linear.
Definition. A relaxed system H [.] is linear if only if

H [a1 x1 (n) + a2 x2 (n) ] = a1H [x1 (n) ]+ a2 H [x2 (n) ]

for any arbitrary input sequences x1 ( n) and x2 ( n) , and any arbitrary constants a1 and a2.
The multiplicative (scaling) property of a linear system:
H [a1 x1 (n) ] = a1H [x1 (n) ]
The additivity property of a linear system:

H [x1 (n) + x2 (n) ] = H [x1 (n) ]+ H [x2 (n) ]

60
Examples:
The linear systems:

N
2
y ( n) = ∑ h( k ) x ( n − k ) y (n) = x(n ) + bx(n − k )
k =0

The non-linear systems:

N
3
y (n) = nx(n) + bx (n − 1) y (n) = ∑ h(k ) x(n − k ) x(n − k + 1)
k =0

61
Causal vs. Non-causal Systems. Definition
Definition. A system is said to be causal if the output of the system at any time n (i.e., y(n))
depends only on present and past inputs (i.e., x(n), x(n-1), x(n-2), … ). In mathematical terms,
the output of a causal system satisfies an equation of the form

] ]
y (n) = F [x(n), x(n − 1), x(n − 2), !
where F [.] is some arbitrary function. If a system does not satisfy this definition, it is called
non-causal.

62
Examples:
The causal system:

N
2
y ( n) = ∑ h( k ) x ( n − k ) y (n) = x (n) + bx(n − k )
k =0

The non-causal system:

10
3
y (n) = nx(n + 1) + bx (n − 1) y (n) = ∑ h( k ) x ( n − k )
k =−10

63
Stable vs. Unstable of Systems. Definitions
An arbitrary relaxed system is said to be bounded input - bounded output (BIBO) stable if
and only if every bounded input produces the bounded output. It means, that there exist some
finite numbers say M and M y , such that
x

x ( n) ≤ M x < ∞ ⇒ y ( n) ≤ M y < ∞

for all n. If for some bounded input sequence x(n) , the output y(n) is unbounded (infinite),
the system is classified as unstable.

64
Examples:
The stable systems:

N
2
y ( n) = ∑ h( k ) x ( n − k ) y ( n) = x ( n ) + 3 x ( n − k )
k =0

The unstable system:

n 3
y (n) = 3 x (n − 1)

65
Recursive vs. Non-recursive Systems. Definitions
A system whose output y(n) at time n depends on any number of the past outputs values ( e.g.
y(n-1), y(n-2), … ), is called a recursive system. Then, the output of a causal recursive system
can be expressed in general as

y (ny)(=n)F=[Fy ([ny−(n1),− 1),

y (ny−(n2), ⋯,
− 2),
! ,! ,⋯,
y (, ny−(nN−),Nx),(nx),(nx),(nx−(n1),− 1),! ,nx−(x(n
x(! ]−)] M)]
nM− )M

where F[.] is some arbitrary function. In contrast, if y(n) at time n depends only on the
present and past inputs

yy((nn))==FF[[xx((nn),),xx((nn−−1), ⋯,
1),!
!, ,xx((nn−−M
M))]]

then such a system is called non-recursive.

66
Examples:
The nonrecursive system:

N
y ( n) = ∑ h( k ) x ( n − k )
k =0

The recursive system:

N N
y ( n) = ∑ b( k ) x ( n − k ) − ∑ a ( k ) y ( n − k )
k =0 k =1

67
Z -Transform
Definition: The Z – transform of a discrete-time signal x(n) is defined as the
power series:
∞
−n
k
X ( z) = ∑ x ( n) z X ( z ) = Z [ x(n)]
n k =−∞

where z is a complex variable. The above given relations are sometimes

called the direct Z - transform because they transform the time-domain
signal x(n) into its complex-plane representation X(z).
Since Z – transform is an infinite power series, it exists only for those values
of z for which this series converges. The region of convergence of X(z) is
the set of all values of z for which X(z) attains a finite value.
68
The procedure for transforming from z – domain to the time-domain is called
the inverse Z – transform. It can be shown that the inverse Z – transform is
given by

1
2πj ∫c
n−1
x(n) = X(z)Z dz
−1
x ( n) = Z [X ( z ) ]

where C denotes the closed contour in the region of convergence of X(z) that
encircles the origin.

69
Determine the Z-transform of the following signals

70
Geometric series formullas required for z-transform calculation

71
Find z-transform of following signal:

n
x[n] = a u[n]
n
x[n] = − a u[−n − 1]
n n
x[n] = a u[n] + b u[−n − 1]
72
Properties of the z-Transform
TABLE33.2 Properties of the z-lansform
Property Time Domain
z-Domain ROC
Notation (n) X(z)
ROC: < |z| < n
(n)
XË(z) ROC;
2(n)
Xz(2) ROC,
Lincarit aNj(n)+
a)N2 (n)
ajX1(z) +t a, X(z) At least the intersection of
ROC, and ROC,
Time shifting x(n-k)
zX() That of X(z), except z =0 if
k>0 and z = 0 if k<0
Scaling in the a"xn)
domain
X(a-'z) |aln < |z <lalr
Time reversal x(-n) X(1)
2
Conjugation x*(n) X*(¢*) ROC
Realpart Re{x(n)) [X(2) + X*(z)] Includes ROC
Imaginary part Im{x(n)) 73
Includes ROC
 Scaling in the a"xn)
Properties
domain of the z-Transform
X(a-'z) |aln < |z <lalr
Time reversal x(-n) X(1)
2
Conjugation x*(n) X*(¢*) ROC
Realpart Re{x(n)) [X(2) + X*(z)] Includes ROC
Imaginary part Im{x(n)) i[X()-X(z")) Includes ROC
Differentiation in nx(n) dX(3)
the ¿-domain dz r2 <|z <1
Convolution x1(n) * x2(n) X1(2)X2(2) At least,the intersection of
ROC] and ROC;
Correlation Rsy (2) = XË(2)X2(z) Atleast, the intersection of
x1l) *x2(-)
ROCof X(2) and X(z)
Initialvalue If x (n) causal x (0) = lim X(z)
theorem Z’00

Multiplication x(n)x2(n) Xi(o)X, ($) dv At least, å < lz| < u2u

Parseval's relation X n)x(n) =
n=-o Xi(0)X;(1/v*)v'dv74
 k>0 and z = 0 if k<0
Properties of the z-Transform
Scaling in the
domain
a"xn)
X(a-'z) |aln < |z <lalr
Time reversal x(-n) X(1)
2
Conjugation x*(n) X*(¢*) ROC
Realpart Re{x(n)) [X(2) + X*(z)] Includes ROC
Imaginary part Im{x(n)) i[X()-X(z")) Includes ROC
Differentiation in nx(n) dX(3)
the ¿-domain dz r2 <|z <1
Convolution x1(n) * x2(n) X1(2)X2(2) At least,the intersection of
ROC] and ROC;
Correlation Rsy (2) = XË(2)X2(z) Atleast, the intersection of
x1l) *x2(-)
ROCof X(2) and X(z)
Initialvalue If x (n) causal x (0) = lim X(z)
theorem Z’00

Multiplication x(n)x2(n) Xi(o)X, ($) dv At least, å < lz| < u2u

Parseval's relation X n)x(n) =
n=-o Xi(0)X;(1/v*)v'dv
75
ROC Properties:

76
ROC Properties:

77
ROC Properties:

78
ROC Properties:

79
ROC Properties:

80
ROC Properties:

81
ROC Properties:

82
ROC Properties:

83
 We will study these properties
after pole zero
ROC Properties:

84
Important Z-tranforms

85
Poles and Zeros in Rational Z-Transform
M −k
−1
B(z) b0 + b1Z + b2Z + ⋯ + bM Z −2 −M ∑k=0 bk Z
X(Z) = = = M
A(z) a0 + a1Z + a2Z + ⋯ + qM Z
−1 −2 −M
∑k=0 ak Z k
−

Zeros: z values for X(z) =0

Poles: z values for X(z) =∞

86
nverse z-transforms 2j 

c

Inverse Z-Transform As we know, we will not evaluate the comp

that the equations that define z-transforms are
we will use one of the following technique

• Partial fraction expansion
–n
X z =  x  n z
• Long division
n = –
• Taylor series expansion
1 n–1
x  n  = --------  X  z z dz
2j
Partial fraction expansion
c As is discussed briefly in the videotape, pa
enables us to obtain the results that we wo
know, we will not evaluate the complex contour
of contour integral
integration over for the inve
the complex z-
ll use one of the following techniques: 87
 • Taylor series expansion
2. all the poles of the z-transform are at different locations in the z-p
The simplest case:

If
Partial fraction
18-491 expansion
Z-transform properties and inverses -7-
then we can write for the example we have been considering S
18-491 Z-transform
18-491 Z-transformproperties
propertiesand
andinverses
inverses -7--7- Spring,
Spring, 20192019
As is discussed briefly in the videotape, 3 –
fraction 1
partial
–1 expansion is a computational ha
3 – --
- z
1. the order of the numerator of the polynomial in z is less than the order of its denominator A (as itAis
in this case), and 4 1 2
enables
inthis
in us to
thiscase), obtain
case),and
and the results H 
that
z  we
= would have obtained
------------------------------------------------
- =if we had gone
----------------------
- + through
-------------------------
- the
of contour2.integration over  1 – 1  1 – 1  1 – 1  1 – 1
all the poles of the the complex
z-transform are 1z-plane.
–at---different
z 1 +locations
--- z in the1 –z-plane
--- z (as they 1 + are --- z in this
2. all the poles of the z-transform
2. all the poles of the z-transform are at different 
are at different
Copyright 2
locations
locations
2019, Richard
 in
M. Stern
 the 4
z-plane  (as 
they 2
are
in the z-plane (as they are in this case), in 
this 
case), 4 
The then
partial we can
fraction write for
method the example
of obtaining we have been
inverse considering
z-transforms builds on the fact that we
then
then we
wecan
canwrite
writefor forthe examplewe
theexample wehave havebeen been considering
considering
where the as-yet3 undetermined –1 coefficients are referred to as the residues
n 3 3– 1–z1 3 – --- z 1 A A
a u Hnterm 

3
z 3


– –--
=
- z
--
-----------
usedz - in = -------------------
complex
4 A Acalculus. -
4 4z------------------------------------------------ -
for=
the A ROC
A 1
---------------------- - +
z  a2 and that
-------------------------
-
– a
HHzz == ------------------------------------------------
-
------------------------------------------------
1 -=1 = – az
1– 1
1
----------------------
- +
----------------------
1 - +
2 2
-------------------------
-
-------------------------
1 – 1 -  1
 1 – 1 – 1–1  1
– 1 
1–1---–z1–1 1 + -- – 1 
1- z1–1–1  1 –1--- z–11– 1 1 + --
- z
– 1
1
In  1 1+ +--- z--2- zthe residues
z- z general,
-----  1–14 ---–z--- z A corresponding
1+1 ---+2z --- z    to4a pole  at location z = z ca
1 –
  2    4     2  i   4   i
n 2 4 1 2 4
– a u mula – n – 1   -------------------- for the ROC z  a
whereundetermined
the as-yet undetermined coefficients– 1 are referred to as the residues of the z-transform, fol
where the as-yet
where the as-yet undetermined coefficients coefficients 1 –are
are azreferred to as the residues of the z-transform, following the
referred to as the residues of the z-transform, following the
term used in complex calculus. i
term used in complex calculus.
term used in complex calculus. Ai = H  z   z – z 
z=z
The simplest
In general, In
case:
thegeneral,
residues the
A residues A corresponding
corresponding to a pole
to a pole at location
i
z =atzlocation z = zobtained
can be easily can beusing
easilythe
obtained
for- us
i
In general, the residues Ai i corresponding i obtained using the for-
to a pole at location z =i z i can be easily
mulaIf mula For our example this becomes:
 3 –1
our example this becomes:
For our example this becomes:
3 – --
- z
 1 – 1 4
3 –1 3 –1 3 A 2 = H  z  1 + --- z = -----------------------
3 – --- z3 – --- z --- 3  4  1 –1
1 4 2 --
- 1 1 + – --
- z
 z  =1H– --z-z 1 – 1--- z – 1 = -------------------
– 1
A 1 = HA - 4 = --- = 21 z = – --
- 2
1 2  2  =1 -------------------
– 1 1 – 1- 3 = --3- = 1 4
1 1 + --- z --- ---
z = --- z = 1---
2
4 1 + --4- z 1 2 2
2 z = --- 1
2 z = --2-
Hence
3 –1 3 –1
3 – --- z3 – --- z
 1 – 1 1 – 1 4 4 6 6 1 2
A 2 = HA2z = 1H+ z---z 1 + --- z ==--z-2= =2
= ----------------------- = --- H
= ----------------------- ----------------------- + --------------------------
 4  4  1 –1 1 –1 3 3
1 1 1+ – ---1z+ – --- z  1 – 1--- z –1  1 + 1--- z – 1
z = – --- z = – --- 2 2  2  
4 4 1 1
z = – ---z = – --4-
4 
4

nce Hence and if we are told that the system is causal, then the corr
1 2
1= ----------------------- +2 -------------------------- n n
H  z 
H  z  = ----------------------- +1 -------------------------   1
 – 1
 – 1  - 1 – 1 h  n  = --- + 2 ------ un
 1 – 1--- z –11– --- z 1 + 1--- z1– 1+ --- z  2   4
   2  4
2 4
and if we are told that the system is causal, thenNote that the firstinverse
the corresponding samples of thisisfunction starting with
z-transform
if we are told that the system is causal, then the corresponding inverse z-transform is
18-491 Z-transform properties and inverses -8- Spring, 2019

1 1 n – 1 n
1. If the ROC is z  --- , h  n  =  --- + 2  ------ u  n  , as noted above. This system is causal and
2  2  4
stable.

1 1 1 n – 1 n
 
2. If the ROC is ---  z  --- , h  n  = – --- u  – n – 1  + 2 ------ u  n  . This system is neither causal
4 2  2  4
nor stable.

1 1 n – 1 n
 
3. If the ROC is z  --- , h  n  = – --- u  – n – 1  + ------ u  – n – 1  . This system is also neither
4  2  4
causal nor stable.

Partial fractions with numerator order greater than or equal to denominator order:

If the order of the numerator is too large, we can reduce it via long division. For example, if we have the
transform
 Partial fractions with1 –numerator 3z
–1 order
transform greater than or equal to denominator order:
l fractions with numerator order greater than or equal to denominator order:
–3 –2 –1
If the order of the numerator is too large, we can 3z +it4zvia+long
reduce z +division. 5 For example, iforder:
we
Using MPartial fractions
ATLAB notation and withsettingnumeratorin plain text toorder H  z  =greater
control positioning:) than or equal
--------------------------------------------------- , we to
candenominator
apply long division as have the
follows:
order of the
transform numerator is too large, we can reduce it via long 1 – 3z division.
–1 For example, if we have the
rm If the order of the numerator is too large, we can reduce it via long division. For example, if we
–3 –2 –1
-z^(-2)-(5/3)z^(-1)-(8/9)
– 3 3z – 2 + 4z
– 1 (Using
+ z M + 5
ATLAB notation and setting in plain text to control positioning:)
transform  z  =+ 4z
H3z ---------------------------------------------------
+ z + 5 , we can apply long division as follows:
-3z^(-1)+1 )3z^(-3)+4z^(-2)+ –
H  z  = --------------------------------------------------- 1, we can applyz^(-1)+
long 5
division as follows:
1
–1 –3 – 3z – 2 – 1
1 – 3z 3z z^(-2)
3z^(-3)- + 4z + z + 5 -z^(-2)-(5/3)z^(-1)-(8/9)
H  z  = --------------------------------------------------- , we can apply long division as follows:
(Using MATLAB notation and 5z^(-2)+ setting in –1
plain
-3z^(-1)+1 text to
z^(-1) control
)3z^(-3)+4z^(-2)+ positioning:) z^(-1)+ 5
MATLAB notation and setting in 1
plain – 3z
text to control positioning:)
5z^(-2)-(5/3)z^(-1)3z^(-3)- z^(-2)
(Using MATLAB notation and setting in plain
(8/3)z^(-1)+ text
-z^(-2)-(5/3)z^(-1)-(8/9) to5 control 5z^(-2)+ positioning:) z^(-1)
-z^(-2)-(5/3)z^(-1)-(8/9)
(8/3)z^(-1)-(8/9) 5z^(-2)-(5/3)z^(-1)
-3z^(-1)+1 )3z^(-3)+4z^(-2)+ z^(-1)+ 5
-3z^(-1)+1 )3z^(-3)+4z^(-2)+ -z^(-2)-(5/3)z^(-1)-(8/9) z^(-1)+ 5 53/9 (8/3)z^(-1)+ 5
3z^(-3)- z^(-2) (8/3)z^(-1)-(8/9)
other words, the result3z^(-3)-of this divisionz^(-2) operation is that
-3z^(-1)+1 )3z^(-3)+4z^(-2)+ 5z^(-2)+ z^(-1) z^(-1)+ 5 53/9
–3 –2 – 15z^(-2)+
3z + 4z + z + 5 – 2 5 – 1 z^(-1)
8 53  9
H  z  = ---------------------------------------------------
3z^(-3)-
In =
other zz^(-2)
5z^(-2)-(5/3)z^(-1)
z – --- the
– words, – --- + of
result -------------------
this -
division operation is that
1 – 3z
–1
5z^(-2)-(5/3)z^(-1) 3 9 1 – 3z
– 1
5z^(-2)+ (8/3)z^(-1)+
–3
3z + 4z + z + 5
–2 – 1z^(-1) 5 –2 5 –1 8 53  9
H  z  = ---------------------------------------------------
(8/3)z^(-1)+ 5 = – z – --- z – --- + --------------------
5
(8/3)z^(-1)-(8/9) –81 3 9 –1
d the corresponding inverse z-transform equals –   n5z^(-2)-(5/3)z^(-1) – 2  – ---   n1––13z  – ---   n  plus the inverse z-trans-1 – 3z
 – z – --- z – --- + -------------------
--------------------------------------------------- =(8/3)z^(-1)-(8/9) 53/9 -
1 – 3z
–1 3 9 1 – 3z
–1
other words, the result of this division operation is that 53/9
and the corresponding inverse z-transform equals –   n
–3 –2 –1
sult of this division
3z operation
+ 4z + z is that
+ 5 – 2 5 – 1 8 53  9
H  z  = --------------------------------------------------- = – z – --- z 5– --- + --------------------853  9
g inverse z-transform 1 – 3z equals
–1 –   n – 2 
3 – --
- 9 n –
form
1 – 13z 
of -   n 
––1-------------------
-- - ,plus the
which inverse
now has a z-trans-
denominator of hig
–3 –2
3z + 4z + z + 5
–1
–2 5 –1 8 53  9 3 9
1 – 3z
– 1
--------------------------------------------------- = – z – --- z – --- + -------------------- n n
1 – 3z
–1 3 9 1 – 3z
– 15
either  538  9   3  u  n  or –  53  9   3  u  – n – 1  , de
 
hich now has a denominator of higher order than the numerator, and which would be
d the corresponding inverse z-transform equals – n – 2  – --
-   n – 1  – --
-   n  plus the inverse z-trans-
3 9
53  9
m of -------------------- , which now has a denominator of5higher order8than the numerator, and which would be
g inverse
1 – 3z
z-transform
–1 equals
n –   n – 2  – --
-   n – 1  – --
-   n  plus the inverse z-trans-
n or –  53  9   3  u  – n – 1  , depending 3 on9whether the ROC is z  3 or Copyright z  3 , 2019, Ri
n n
her  53  9   3  u  n  or –  53  9   3  u  – n – 1  , depending on whether the ROC is z  3 or z  3 ,
hich now has a denominator of higher order than the numerator, and which would be

n
n or –  53  9   3  Copyright
u  – n – 1  ,Copyright
depending
2019, 2019,on
Richard whether
M. Stern
Richard the ROC is z  3 or z  3 ,
M. Stern

Copyright 2019, Richard M. Stern

 M–N N s
Let us consider a A
simple example
– r same location, the situation
If we have multiple poles in the k with multiple poles:
m
is more difficult. As– 1described in the text, If we
Partial
have
Hz =
s polesfractions d with
at location the contribution
Br z
multiple
in the z-plane,
---------------------
+
poles
-
– 1in
+
of the multiple
the 1
C
same m  1 – d z
location:
pole i 
to the partial-fraction of the z-
r=0 H
k = 1 k  i1z –
 =d k z
------------------------------------------------------
m=1
transform is 1 –1 2
 1 – 1--- z 1 + --- z
– 1 
 2   4 
If we have multiple poles in the same location, the situation is more difficult.
s
here d i is the location C ofmthe pole of multiplicity s. If there is more than one location with multipl
 ---------------------------
have s poles at1 –location –1 m
-
d in the z-plane, the contribution of the multiple pole
dditional seriesmof= 1termsdzsimilar  Thistotransfer function,
the latter of course,
term above will has a single pole at z = – 1  4 and a do
be obtained.
transform is
et uswhere
consider a simple example Using with multiple
partial poles:
fractions, we would like to rewrite this equation in the form
s
1 C
1 m  d
s–m A
s 1 –1  C 1 C
–1 2
1
H  zCm == ------------------------------------------------------
---------------------------
----------------------------------------
– 1 2 s–m 1 –
1s –– m---!z  –d   1–+1dw
- 
--
- m
z
- H
s 1– m


-----------------z 
1 =
– dw

 H
1 + --- z
 w
1 – 1  

-------------------------
- +
 ------------------------
evaluated
1 – --- z
1 – 1 
at- +
w ---------------------------
=
1
d
–
1 – 1
--
- z
-
2

m =properties
18-491 Z-transform 1 2  1 – dz
and inverses 4  
-10- 4   2   2 
Spring, 2019
–1
While the use of w = z in the equation above may seem needlessly confusing and arbitrary, it actually
his where Thehas
transfer function, of course, residue A 1 is
a single obtained
pole 1  4 and a double
at z =in–conventional – 1 at z = 1  2 .
fashion:pole
is helpful because the derivatives
 1 –are
A = H  z  1 + --- z
1 easier to evaluate 1 in
= ----------------------------
terms of 1w than1 in terms of z
= ----- = ---
.
1  41  9 2 2
1  1 – --- z 3 – 1
singInpartial fractions, we would likez to
= – rewrite
--
-
the general case, with some isolated poles,  this
one set equation
 s – m in the form of
2 of multiple poles in the same location,and a polyno-
4 1 d z = – 1--- s –1
mial of numeratorCorderA
m 1 = ----------------------------------------
M and denominator C order N, -  -----------------
theC z-transform
4  1 – dw
would 
be H
of  w
the 
form  evaluated
1 2 s – m s – m
H  z  = -------------------------- + ------------------------- + ----------------------------
 1 1 – --- z
 1 – 1 z = – ---  12 
4= --------------------------- 1 1
A 1 = H  z  1 + --- z - =1 ----- = ---
or the residue C we  will
4 solve the equation
1 2 z = – --- 2 9
1  1 – --
- z
– 1 4 3
For the residue C 1 we
z = –will
1
---
4  solve
2 the
 equation
1
z = – ---
For the residue C 1 we will solve the equation s– m 4
s –1  1  d s – m s – 1 
-  1 – dw  HC w = -----------------------------------------1 ----------------- 1d– dw  H  w s 
m For the residuemC 1 we will solve the equation s s––m m
 C m1 = ----------------------------------------
 -
s – sm -----------------  1 – dw  
H  w
– 1

 d
 s – m–1! s––md- -----------------
C m = ----------------------------------------  1dw
– sdw
–  –1 
mH  w s– m  
w = d
 s – m !1 – d   ddw  s –
 m
s – !
s
m
– 
m – d   dw  w=d
– 1
s –1  –1 –1
C m = -----------------------------------------  -----------------  1 – dw  H  w   w=d w=d
s–m s–m
 s – m !  – d   dw 
d = 1  2 . This produces–1 ––11 – 1
–1
w=d
r sfor=s =2 2, ,mm = = 11, w , w= z= ,zw , =wz , and–=1d z=, 1–and 1 2 . This d =produces 1  2 . This produces
for s = 2 , m –=1 1–,1 w = z , w = z , and d = 1  2 . This produces
for s = 2 , m = 1 , w = z , w  = z , and d = 1  2 . This produces 

2 1 1  d
1
 1 – 1 2  4 1 2  – 1  4 2
w ----------------------------------------------
C = - 
------------------------------  =
---------
- – 12 – -------------------
1
--
- w  - = --
-
----------------------------------------------
 -  = – 2 -------------------
- = --
- 
 2 1
11 ! 1–1  21ddw 1d  11– 1
1 1 1  2  2 1 11  2 29 1   2 1  2 9 – 1  4 2
1 – 1 --- w C 
C 1 1==

------------------------------
2
---------
1 - 1 – --
d- w – 1 
----------------------------------------------
1 4
1 -
1 2

= – 2 -------------------
- – =
1  --
4-
 2  1  14! C
+ --
- w
------------------------------
 =  ---------
 - 1 – --
- w 1 + --
- --
- w
 1 + --
----------------------------------------------

1 1 2 1  21 ---------
1 ------------------------------ 2  
2 - 11 –4--
-
- w

- 
w 
 = – 2 1 +
-------------------
 - 
--
2 2 9
-
= --
-
---------------------------------------------- - = – 2 -------------------
-
– 11 2   dw  1 –--- w2  1 + --- w  1 w = 2 2 1 + 1 ---1   1 2 9
 1 ! w– 1= 2 2   dw  1 2  1  4 1 2– ---w  1 1+ --- 2w2   1   1 + --- 1 2
 1 ! – 1  2  dw  w2= 2 1– --- w4  1 + --- w  
2  1 + ---
For the residue C 2 we solve the same equation butwith m = 2 . This produces  2   w 4= 2  2
n but with m = 2 . This produces
For the residue C we solve the same equation but with m = 2 . This produces w=2
2
 0 
or the residue C we 1 solve  the
d  same 1 
2
equation 1 but with  m = 21. This 2 produces
For the
C 2 = 2
residue  C 
------------------------------ we 0 - solve
---------
0 d 0  1 2 1 – --
- w2the same equation 
----------------------------------------------
 -  but with
= m
-----------
- = =
--
- 2 . This produces
2 1 2 1 1 2 1 1 2 2 3
 1C2 = ------------------------------
 0 !  – 1  20  ---------  dw - 1 – 1--- w ----------------------------------------------
 
 21 – --
- w   1 + -- w
--  = -----------1- += ------
w ---------------------------------------------- -    =0 -----------
2 - =  --- 1 2   1    2 43
 2  0 !  – 1  2  dw 2  3 2  
4 1 + ---
1 + ---0 
1 1 1 – --
- w 1 + --
- w
 1 – --- w  1 + --- w    2  
 10 2 4  w=2 4

 2   4  1  d4 w=2
1  1 2
+ 1 a u  n   --------------------------- for z  a
n
–1 2 1
 n + 1 a u  n   --------------------------- for z  a

 
–1 2
Combining the terms, we obtain
19
1 –
29
az  1 – az 
23
H  z  = poles
Note that higher-order would - +yield
------------------- components- + ---------------------------
------------------- -
of the corresponding inverse z-transforms that
1 –1 1 –1  1 – 1 
2
1 + --- z by polynomials
include exponentials multiplied 1 – --- z of a 1higher – --
- z order, with the order of the polynomial equal to
4
the multiplicity of the pole at a given location minus 2  1. 2 

r-order poles would yield components of the corres

If the ROC of the z-transform in question is z  1  2 , the inverse transform can be obtained by combining
You can verify that this is the correct partial-fraction expansion by multiplying the terms back together or
all of the inverse transforms:
ntials multiplied by polynomials of a higher order, w
hrough the use of the MATLAB command residuez.
n n
 1---  – 1--- +  2---  1--- +  2   1
---  n + 1  ---
n 19 29 23
u  n   -------------------------- + ------------------------- + ----------------------------
 9  4  9  2  3  2
As noted above, we can obtain the inverse z-transform for the final term through the use of the differentia-
1 1 1 2
 1 + --- z   1 – --- z   1 – --- z 
of the for pole at a given location minus 1.
– 1 – 1 – 1
ion property z-transforms:  4   2   2 

dX  z 
nx  n   – z --------------
dz
 
Inverse z-transforms by long division
e z-transform in question is z 1 2 , the inverse
Applying this property to the case of the decaying exponential signal produces the transform pairs
t
Long division can also be used to obtain inverse z-transforms numerically. For example, consider again the
transform
e transforms: n
na u  n   3 az– 1
–1
3 –-------------------
--- z
4 –1
- for the ROC z  a and
1 – az
H  z  = -------------------------------------
1 –1 1 –2
 Inverse z-transforms by long division
Long division can also be
Inverse z-transforms by used to obtain inverse z-transforms numerically. For example, consider aga
long division
Inverse
Long z-transforms
division
transform
can also be usedbyto long
obtain division
inverse z-transforms numerically. For example, consider again the
Long division can also be used to obtain inverse z-transforms numerically. For example, consider again the
transform
transform
Long division can also33be –31used – 1 to obtain inverse z-transforms numerically. For example, conside
3 – ---– z 3--- –z1
transformH 34– ---4z
H zz ==-------------------------------------
-------------------------------------
4
H  z  = -------------------------------------
1 –11 – 11 –21 – 2
11 - z– ---1---zz+– 1---+z+1--- z---–2z
– 1--–
4 4 8 8 3 –1
4 3 – 8--- z
4
Arranging
H
the terms
 z  =
in order
-------------------------------------
of increasing
––11
Arranging the terms in order of – 1 1powers
1 increasing – 2 ofofz z and
powers anddividing
dividing
– 1
we obtain
we obtain
Arranging the terms in order 1 – --of - z increasing + --- z powers of z and dividing we obtain
4 8
3+ 0z^(-1)+(3/8)z^(-2)+
0z^(-1)+(3/8)z^(-2)+. .
1-(1/4)z^(-1)-(1/8)z^(-2)) )3-(3/4)z^(-1)+
1-(1/4)z^(-1)-(1/8)z^(-2) – 1 0z^(-3)+
3-(3/4)z^(-1)+ 0z^(-2)+
0z^(-2)+ 0z^(-3)+0z^(-4)+
0z^(-4)+...
...
Arranging the terms in order of increasing powers of z and dividing we obtain
3+ 0z^(-1)+(3/8)z^(-2)+
3-(3/4)z^(-1)-(3/8)z^(-2)
3-(3/4)z^(-1)-(3/8)z^(-2)
1-(1/4)z^(-1)-(1/8)z^(-2) ) 3-(3/4)z^(-1)+ 0z^(-2)+
0z^(-1)+(3/8)z^(-2)+ 0z^(-3) 0z^(-3)+ 0z^(-4)+
0z^(-1)+(3/8)z^(-2)+ 0z^(-3)
3-(3/4)z^(-1)-(3/8)z^(-2)
0z^(-1) - 0z^(-2) + 0 z^(-3)
0z^(-1) - 0z^(-2) + 0 z^(-3)3+ 0z^(-1)+(3/8)z^(-
(3/8)z^(-2)+0 z^(-3)+ 0z^(-4)
0z^(-1)+(3/8)z^(-2)+ 0z^(-3)
(3/8)z^(-2)+0 z^(-3)+ 0z^(-4)
1-(1/4)z^(-1)-(1/8)z^(-2) ) 3-(3/4)z^(-1)+ 0z^(-2)+ 0z^(-3)+ 0z^(-
Hence, the first several terms of the quotient will be 0z^(-1) - 0z^(-2) + 0 z^(-3)
Hence, the first several terms of the quotient will be 3-(3/4)z^(-1)-(3/8)z^(-2)
(3/8)z^(-2)+0 z^(-3)+ 0z^(-4)
0z^(-1)+(3/8)z^(-2)+ 0z^(-3)
-491 Z-transform properties and inverses -12- Sp
18-491 Z-transform properties and inverses -12-
3 –2
–1
H  z  = 3 + 0z + --- z + 
–1 3 –2
H  z  = 3 + 0z + --- z +  8
8 –1 3 –2
H  z  = 3 + 0z + --- z + 
8
hich,if causal,
hich, if causal,has byhas by inspection
inspection the inverse
the inverse transform of transform of
which, if causal, has by inspection 3the inverse transform of
h  n  = 3  n  + 0  n – 1  + ---   n – 2  + 3
h  n  = 3  n  + 08 n – 31  + ---   n – 2  + 
h  n  = 3  n  + 0  n – 1  + ---   n – 82  + 
8
omment:
omment:
Comment:
• A left-sided inverse z-transform could be obtained in similar fashion by arranging the term
opposite order. This technique cannot be used to obtain both-sided inverse
• A left-sided inverse z-transform could be obtained in similar fashion by arrangingz-transforms.
• A left-sided inverse z-transform could be obtained in similar fashi
opposite order. This technique cannot be used to obtain both-sided inverse z-trans
opposite order. This technique cannot be used to obtain both-sided
verse z-transforms by Taylor series expansion
 1 + z – 1
For example, the z-transform X  z  = ln  ----------------- with ROC z  1 can be obtained
Inverse z-transforms by Taylor series expansion –1
1 – z 
sion:
Occasionally we are asked to obtain the inverse z-transform of a function that is not a r
–1
in z or z. Sometimeswe can – 1use Taylor series
– 3 expansion
– 5 –to
7 accomplish this.
1+z  –1 z z z
X  z  = ln  ----------------- = 2 z + ------- + ------- + ------- + 
1 – z  –1 3– 1
5 7
1 + z 
For example, the z-transform X  z  = ln  ----------------- with ROC z  1 can be obtained u
–1
1 – z 
sion:
The corresponding inverse is obtained easily by inspection as in the case of the result
1 + z  – 1 – 3 – 5 – 7
–1 z z z
X  z  = ln  ----------------- = 2 z + ------- + ------- + ------- + 
1 – z  –1 3 5 7
An Introduction to Linear-Discrete Time Time-Invariant
Systems (LTI Systems)

Time-Domain Representation
Frequency-Domain Representation

99
Time Domain: Impulse Response and Convolution

δ ( n) h(n) = H [δ (n) ]
LTI system

unit impulse impulse response

H []
.

LTI system description by convolution (convolution sum):

∞ ∞
y ( n) = ∑ h( k ) x ( n − k ) = ∑ x ( k ) h( n − k ) = h( n) * x ( n) = x ( n) * h( n)
k =−∞ k =−∞

Viewed mathematically, the convolution operation satisfies the

commutative law. 100
Step Response

u ( n) LTI system g (n) = H [u (n) ]

unit step step response
H []
. unit-step
response
∞ n
g ( n) = ∑ h(k )u (n − k ) = ∑ h(k )
k =−∞ k =−∞

These expressions relate the impulse response to the step response of the system.

101
Impulse Response Property and Classification of LTI Systems
Causal LTI Systems
A relaxed LTI system is causal if and only if its impulse response is zero for negative
values of n , i.e.

h(n) = 0 for n < 0

Then, the two equivalent forms of the convolution formula can be obtained for the causal
LTI system:
∞ n
y ( n) = ∑ h( k ) x ( n − k ) = ∑ x ( k ) h( n − k )
k =0 k =−∞

102
Stable LTI Systems
A LTI system is stable if its impulse response is absolutely summable, i.e.

∞
2
∑ h( k ) < ∞
k =−∞

103
Finite Impulse Response (FIR) LTI Systems and Infinite Impulse
Response (IIR) LTI Systems

Causal FIR LTI systems:

N
y ( n) = ∑ h( k ) x ( n − k )
k =0

IIR LTI systems:

∞
y ( n) = ∑ h( k ) x ( n − k )
k =0

104
Transfer Function

The LTI system can be described by means of a constant coefficient linear

difference equation as follows
N M
y ( n) = ∑ b( k ) x ( n − k ) − ∑ a ( k ) y ( n − k )
k =0 k =1

Application of the Z-transform to this equation under zero initial conditions

leads to the notion of a transfer function.

105
 input signal output signal
LTI System y ( n)
x ( n)
h( n) H ( z)
X ( z ) = Z [x(n) ] Y ( z ) = Z [y ( n) ]

H ( z ) = Z [h(n) ]

Transfer function: the ratio of the Z - transform of the output signal

and the Z - transform of the input signal of the LTI system:

Y ( z ) Z [ y (n)]
H ( z) = =
X ( z ) Z [ x(n)] 106
LTI system: the Z-transform of the constant coefficient linear difference equation
under zero initial conditions: N M
y ( n) = ∑ b( k ) x ( n − k ) − ∑ a ( k ) y ( n − k )
k =0 k =1

N M
−k −k
Y ( z ) = ∑ b( k ) z X ( z ) − ∑ a ( k ) z Y ( z )
k =0 k =1

The transfer function of the LTI system:

N
−k
Y ( z) ∑ b( k ) z
H ( z) = = k =0
M
X ( z) −k
1 + ∑ a(k ) z
k =1

H(z): may be viewed as a rational function of a complex variable

107
z (z -1).
Poles, Zeros, Pole-Zero Plot
Let us assume that H(z) has been expressed in its irreducible or so-called
factorized form: N N
−k
∑ b( k ) z ∏ ( z − z k )
b0 N − M
H ( z) = k =0
M
= z k =1
M
−k a0
1 + ∑ a(k ) z ∏(z − p ) k
k =1 k =1

Zeros of H(z): the set {zk} of z-plane for which H(zk)=0

Poles of H(z): the set {pk} of z -plane for which H ( pk ) → ∞

Pole-zero plot: the plot of the zeros and the poles of H(z) in the z-plane
represents a strong tool for LTI system description.
108
Recursive and Non-recursive LTI Systems

N
Causal non-recursive LTI: y ( n) = ∑ h( k ) x ( n − k )
k =0

Causal recursive LTI:

N M
y ( n) = ∑ b( k ) x ( n − k ) − ∑ a ( k ) y ( n − k )
k =0 k =1
LTI systems:
characterized by constant-coefficient difference equations

109
We study the frequency domain
representation after DTFT

110
.4
Determine the system function and the unit sample response of
em function
system is
function and
the system describedthe
by unit
the disample
erenceresponse
equationof the system described
tion Y(z) 2
H(z) =
X(¢)
1 1- -i
y(n) =y(n- 1) +2x(n)
Using Table 3.3 we obtainthe in
pole at z= and a zero at the origin.
s a
computing the z-transform of the difference equation, we obtain

?Y(z) + 2X (z)
1
h(n) =2(;'u(n)
1
Y(z) =
t samnle response of thesystem.

m function is 111
ff
 .4 Y(z) = Y(z) + 2X
Determine the system function and the unit sample response of
system function and
the system describedthe
by unit
the disample
erenceresponse
equationof the system described
tion function is
system
Y(z) 2
1
H(z) =
X(¢)
y(n) =y(n- 1) +1-
2x(n)-i
origin. Using Table 3.3 we obtainth
z= and a zero at the
has a pole atthe z-transform of the difference equation, we obtain
mcomputing
1
h(n)
1 =2(;'u(n)
Y(z) = Y(z) + 2X (z)
t samnle response of the system.

m function is 112
ff
 ElinXeaArMPLA3.5E.2 all
the
(c) (b) (Sa) pecify
time-invaTheriant The
The poles
systemsystemsystem the time-
ROC
is is is of
anticausal. causal.stable. o f
H(z)
H(Z)
system and are
determine H(2)=
inside chara
H(2)= cterized is
chara 1-3.5z-l+1.5z-2
1 1
--3
}
113
h(n) 2z-1 1
the+ 4z-31
for
3.5z-l+1.5z-2 the unit
1-3-}2z-1 1
following -lvaran2t
+
4z3-1 circle.
by f
r conthdeitions:
e system
system
2

function is
BIBO
A linear time-invariant system is characterized by the system function

s: 3 stable
Frequency-Domain Representation of Discrete Signals and LTI
Systems

x ( n) = e jω n
LTI system y ( n)
complex-valued h( n) LTI system output
exponencial
signal

impulse response
∞
y ( n) = ∑ h( k ) x ( n − k )
k =−∞

114
Comments on Fourier Transform of Discrete Signals and Frequency-Domain Description of L
Systems

jω
x(n), X (e ) LTI system jω
y (n), Y (e )
input signal H (e ) h ( n )
jω
output signal

frequency response impulse response

115
The input signal x(n) and the spectrum of x(n):
π
∞ 1 jω jωn
jω
X (e ) = ∑ x ( k )e − jω k
x ( n) = ∫ X (e )e dω
k =−∞
2π −π

The output signal y(n) and the spectrum of y(n):

π
∞ 1 jω jωn
jω
Y (e ) = − jω k y ( n) = ∫ Y ( e ) e dω
∑ y ( k )e
k =−∞
2π −π

The impulse response h(n) and the spectrum of h(n):

π
jω
∞
− jω k 1 jω jωn
H (e ) = ∑ h ( k )e h( n) = ∫ H (e )e dω
k =−∞ 2π −π

Frequency-domain description of LTI system:

jω jω jω
Y (e ) = H (e ) X (e ) 116
LTI system output:

∞ ∞
jω ( n − k )
y ( n) = ∑ h ( k ) x ( n − k ) = ∑ h ( k )e =
k =−∞ k =−∞
∞ ∞
− jω k jω n jω n − jω k
= ∑ h ( k )e
k =−∞
e =e ∑ h ( k )e
k =−∞

jω n jω
y ( n) = e H (e )

∞
jω − jω k
Frequency response: H (e ) = ∑ h ( k )e
k =−∞

117
 jω jω jφ (ω )
H (e ) = H (e ) e

jω jω jω
H (e ) = Re ⎡⎣ H (e ) ⎤⎦ + j Im ⎡⎣ H (e ) ⎤⎦
∞ ∞
jω ⎡ ⎤
H (e ) = ∑ h(k )cos ωk + j ⎢ − ∑ h(k )sin ωk ⎥
k =−∞ ⎣ k =−∞ ⎦
∞
jω
Re ⎡⎣ H (e ) ⎤⎦ = ∑ h(k )cosωk
k =−∞

∞
jω
Im ⎡⎣ H (e ) ⎤⎦ = − ∑ h(k )sin ωk
k =−∞

118
 Magnitude response:

jω jω 2 jω 2
H (e ) = Re ⎡⎣ H (e ) ⎤⎦ + Im ⎡⎣ H (e ) ⎤⎦

Phase response:
jω
jω
Im ⎡
⎣ H ( e ) ⎤
⎦
φ (ω ) = arg ⎡⎣ H (e ) ⎤⎦ = arctg jω
Re ⎡⎣ H (e ) ⎤⎦

Group delay function:

dφ (ω )
τ (ω ) = −
dω
119
Thank You!

120