D I GI T A L SI GN A L PRO CESSI N G

U ni t 1












U N I T I O V ERV I EW O F SI GN A LS A N D SY ST EM S 9
--Basic elements of digital signal Processing
Concept of frequency in continuous time and discrete time signals
Discrete time signals. Discrete time systems
Analysis of Linear time invariant systems
Z transform
Convolution and correlation.

I nt r oduct i on

W hat i s si gnal pr ocessi ng ?

Signal processing is the action of changing one or more features (parameters)
of a signal according to a predetermined requirement. The parameters that is to
be changed may be , amplitude, frequency, phase etc... of the signal . The signal
which undergoes such a process is known as input signal. And the entity which
performs this processing is known as "Signal processing system" or simply
system. ie; the input signal is fed to the system and the appropriately processed
signal is coming out of the system. This signal is known as the "Output signal".
This process is illustrated below.


Some of the signal processing tasks are Amplification, Attenuation, Filtering,
Modulation, etc... . To accomplish the required task, the "system" is designed with
appropriate behavior and features. ie; for an amplifier, the "system" is designed
such that any input signal going to the "system" will be amplified. In actual circuit,
such a behavior can be realized using transistors , FET or Op-Amps. This sort of
signal processing is known as analog signal processing. Well ! , then what is Digital
Signal Processing ....?

A nal og si gnal pr ocessi ng V s D i gi t al Si gnal Pr oce ssi ng
In analog processing system the information is represented
as an analog quantity. ie; something that varies continuously with time. This
"something" can be voltage , current etc.. . for example, the signal that comes out
of a microphone is analog in nature. The varying voltage from the microphone is
processed using an analog signal processing system known as "amplifier" , and
reproduced through a speaker.

Where in Digital Signal Processing system , the information is represented in
digital format. The signals that are to be processed is converted into numerical
form before any processing. This conversion is known as sampling. The
information contained in an analog signal is first converted to digital samples which
are equally spaced in time. The figure below shows an analog signal and its
sampled version.
Each of these samples are converted in to a numerical value and stored in the
computer' s memory. Processing is then done on these samples. The sampling in a
Digital Signal Processing system is governed by a theorem known as sampling
theorem.


Cl assi f icat i on of Signal s: Cont inuous- T i me ver se s D iscr et e- T ime Signal s

Continuous time or analog signals are signals that are defined for every value of a <
t < b, where ( a, b) can be ( , ) , i.e., x ( t ) = e
- | t |
or x(t ) cos( t ). .

Discrete-time signals are defined at discrete- time instants and between the two
discrete time instants are undefined but are not zero. They can be obtained either
by sampling analog signals or they can be discrete in nature like discrete
measurement signals.


A discrete-time signal having a set of discrete values is called a digital signal. Note
that sampling an analog signal produces a discrete-time signal. Then quantization
of its values produces a digital signal.



D et er mi ni st ic ver sus Random Signal s

Any signal that can be uniquely described by an explicit mathematical expression or
a well-defined rule is called deterministic . The past, present and future of a
deterministic signal are known with certainty. Otherwise, it is called Random
and its properties is explained by statistical techniques.


Review of Si nusoids in Cont i nuous and D iscr et e T i me

x
a
t Acost t

A cos2Ft 2F

xa t Tp xa t , Tp
1
: fundamental period. Increasing F means
increasing
F

oscillation in time domain. F = 0 corresponds to Tp = .
1
DIGITAL SIGNAL PROCESSING
Lecture 1 - Chapter 1
Classification of Signals: Continuous-Time verses Discrete-Time Signals
Continuous time or analog signals are signals that are defined for every value of a < t <
b, where (a, b) can be ( + , ), i.e., x (t) = e
-|t|
or ). cos( ) ( t t x = .
Discrete-time signals are defined at discrete-time instants and between the two discrete
time instants are undefined but are not zero. They can be obtained either by sampling
analog signals or they can be discrete in nature like discrete measurement signals.
A discrete-time signal having a set of discrete values is called a digital signal. Note that
sampling an analog signal produces a discrete-time signal. Then quantization of its
values produces a digital signal.
Deterministic versus Random Signals
Any signal that can be uniquely described by an explicit mathematical expression or a
well-defined rule is called deterministic. The past, present and future of a deterministic
signal are known with certainty. Otherwise, it is called Random and its properties is
explained by statistical techniques.
Review of Sinusoids in Continuous and Discrete Time

( ) ( )
( ) F 2 Ft 2 A
t t A t x
a


= + =
< < + =
cos
cos
( ) ( )
F
T t x T t x
p a p a
1
, = = + : fundamental period. Increasing F means increasing
oscillation in time domain. F = 0 corresponds to Tp = .
2
Also, for complex exponential signals, ( )
( ) +
=
t j
a
Ae t x . A sinusoidal signal then can
also be expressed as
( ) ( )
( ) ( )

+ +
+ = + =
t j t j
a
Ae
2
1
Ae
2
1
t A t x cos
Discrete-Time Sinusoid Signals

( ) ( )
( ) ( ) < < + =
= + =
n n f A n x
f n A n x


0
0
2 cos
2 cos
A few important differences between continuous sinusoid and discrete sinusoids:
1) A discrete-time sinusoid is periodic only if its frequency is a rational number.
By default, ( ) ( ) n x N n x = + for all n if x(n) is periodic. The smallest N is called
Fundamental Period.
( ) ( ) ( ) ( ) + = + + = + n f N n f A N n x
0 0
2 cos 2 cos
This relationship is true if and only if K N f 2 2
0
=
N
K
f =
0
: a rational number.
To determine the period N of a periodic discrete time sinusoid, we express f as two
relatively prime numbers. Observe that a small change in frequency can result in a
large change in period. For example, ( ) ( ) + =

= n A n A n x cos
2
1
2 cos
1
its
period is
60
30
2
1
1
= = =
f
K
N .
Now consider ( ) 60 517 . 0
60
31
,
60
31
2 cos
2 2 2
= = =

= N f n A n x
2. An analog ( ) + , F maps to
2
1
f
2
1
or equivalently to or in
other words, the highest rate of oscillation occurs at
2
1
f + = or + = . To see
what happens for 2 consider
1
=
0
and
2
= 2 -
0
. When
1
varies
between to 2, then
2
varies between and 0. Now
( ) n A n A n x
0 1 1
cos cos = =
3
( ) ( ) ( ) ( ) n x n A n A n A n A n x
o o 1
*
0 2 2
cos cos 2 cos cos ) ( = = = = =
* this is only true because n is an integer value, i.e., x(n) is a discrete signal.
Hence, ( ) ( ) n x n x
1 2
= is an alias because + < < F maps only to
2
1
f
2
1

by sampling.
Analog to Digital Conversion (A/D)
- sampling (sampling rate)
- quantization
Sampling:
( ) ( )
( )
nTs t a
s a
t x
n nT x n x
=
=
< < = where ,
s
s
F
n
nT t = = , F
s
= Sampling rate (Frequency) (Hz)
Consider an analog sinusoid: ( ) ( ) + = Ft 2 A t x
a
cos
( ) ( ) ( )

+ = + = = n
F
F
2 A nT F 2 A nT x n x
s
s s a
cos cos
Now recall that < < F maps to
2
1
f
2
1

2
1
F
F
f
2
1
s
=

max
2F F
s
Nyquist rate
Example 1:
( )
( ) Hz 50 F t 100 t x
Hz 10 F t 20 t x
2 2 a
1 1 a
= =
= =

cos
cos
If we digitize both of these signals with Fs = 40Hz, then
( ) ( ) n
2
n
40
20
40
n
x nT x n x
1 a s 1 a 1

cos cos = =

= =

( )
( ) ! ! cos
cos cos cos
n x n
2
n
2
2 n
2
5
n
40
100
40
n
x n x
1
2 a 2
= =

+ = = =

Therefore, by this sampling rate x
2
(n) has become same as x
1
(n), which is an aliasing
error. Equivalently, in this case, the frequency of 50 Hz is an alias of 10 Hz by sampling
max
2F F
s

4
rate of 40 Hz. Furthermore, all frequencies (F
1
+ 40K) are aliases of F
1
. Hence, do not
use the Nyquist rate blindly.
Example 2: ( )
43 42 1 4 43 4 42 1 43 42 1
3 2 1
x x x
a
t 100 t 300 10 t 50 3 t x cos sin cos + =
Nyquist rate? f
1
= 25, f
2
= 150, f
3
= 50 Hz
F
max
= 150 Hz F
s
= 300 Hz?!
Problem: with F
x
= 300 Hz, ( ) 0 n 10
F
n
x n x
s
2 a 2
= =

= sin all the time!

If it had a phase shift n o < < , then it would have been fine, but it is best to choose a
higher sampling rate.
Sampling
( ) ( ) ( ) t p t x t x
a p
=
( ) ( )

+

=
s
nT t t p
( ) ( ) ( ) ( )
( ) ( ) ( ) [ ]
( ) ( )
( ) ( )
s a
s
p
Ks
s
s
a p
s s nT t p
k X
T
X
k
T
P
P X X
nT t nT x t x n x
x

=
=
=
= =

+

+

+

=
1
2
*
2
1
|
Therefore, X
p
() is a periodic function of shifted the X
a
().
x
P(t)
x
a
(t)
x(n)
T
s
x(n)
( ) t x
a
5
( ) X
Obviously, if
M s
2 there is no
aliasing and the signal can be
reconstructed accurately.
In practice however, generating very narrow impulse is very difficult. Therefore, the
practical way for sampling is zero-order hold. Such a system samples x
a
(t) at a given
sampling instant and holds that value until the succeeding sampling instant.
Reconstruction of x
r
(t) from the output of this system requires a cascade of low-pass
filters or a non-constant gain of LPF.

X
p
()

M

s
x
p
(t)
It is as if ( ) t x
a
x
P(t) H
0
(t)
( ) t x
o
( )
( )
r
r
H
t h
x
r
(t)
|H
r
()|
-
s
/2
s
/2
w

s
/2
-
s
/2 2

<Hr()
6
Lecture 2 (Chapter 2) Discrete-Time Signal and Systems
Classification of Signals
1. Finite duration x(n) =0 n > N
Infinite duration
2. Right-Left sided
( )
1
0 N n n x < = right-sided
( )
2
0 N n n x > = left-sided
Some Elementary Discrete-Time Signals
- unit sample sequence ( )

=
=
0 0
0 1
n
n
n
- unit step sequence ( )

<

=
0 0
0 1
n
n
n u
- unit ramp ( )

<

=
0 0
0
n
n n
n u
r

- exponential ( )
n
a n x = for all n
If a is complex, then ( )
n j n j
e r n x re a

= =
( ) n j n r
n
sin cos + =











n
for real a
< a< 1
a > 1
-1 < a < o a < -1
r is a damping factor
7
For complex a, ( ) ( ) ( ) n jx n x n x
I R
+ = , where ( ) ) cos( n r n x
n
R
=
if r < 1 then
( )
( )

=
=
n n x
r n x
n

Energy and Power Signals
Energy is defined as ( )
2
n x E

+

= if E is finite, i.e., < < E o , then x(n) is called
Energy Signal. However, many signals that have an infinite energy, have a finite average
power. Average power is defined as
( )
2
1 2
1
lim

+
=
N
N
N
ave
n x
N
P .
If we define the signal energy of x(n) over the interval (-N, N) as
( )
2
N
N
N
n x E

= then
n
N
E E

= lim
and therefore,
N
N
ave
E
N
P
1 2
1
lim
+
=

clearly if E is finite, then P
ave
= 0.

Example Unit Step Sequence




Obviously, it is not an energy signal but it is a power signal.
n
u(n)
8

( ) ( )
=
+
+
=
+
=
+
=

=


2
1
1 2
1
1
1 2
1
1 2
1
2
0
2
N
N
im
N
im n x
N
im p
N
N
n
N
N
N
l
l l

Periodic and Aperiodic Signals
( ) ( ) = + n x N n x signal is periodic
Energy of periodic signals is infinite but it might be finite over a period. On the other
hand, the average power at the periodic signal is finite and is equal to the P
ave
over a
single period. Hence, periodic signals are power signals.

Symmetric (even) and odd Signals
( ) ( ) n x n x = even
( ) ( ) n x n x = odd
Any signal can be written as: ( ) ( ) ( ) n x n x n x
o e
+ =
Where
( ) ( ) ( ) [ ]
( ) ( ) ( ) [ ]

=
+ =
n x n x
2
1
n x
n x n x
2
1
n x
o
e

Read Section 2.1.8
Classification of Discrete-Time Systems
Static versus Dynamic Systems
Static Systems memory less the output doesnt depend on past or future values of
the input.
Dynamic Systems having either infinite or finite memory.
Example: ( ) ( ) ( )
2
n x n x 2 n y + = Static
( ) ( )

=
=
N
k
k n x n y
0
Dynamic-finite
( ) ( )

=
=
0 k
k n x n y Dynamic-infinite

it is a power signal!
9
Time invariant versus Time-Invariant Systems
A relaxed system is time-invariant if
( ) ( ) n y n x
( ) ( ) k n y k n x K ,
Example: 1) y(n) = x(n) x (n 1)
y(n k) = x(n k) x(n k 1) time invariant
2) y(n) = nx(n)
y(n k) = (n k) x(n k) = nx(n k) kx(n k)
but x(n k) nx(n k) y(n k)
time variant
Causality
A system is causal if the output at any time depends only on present and past values of
the inputs and not on future values of the input. y(n) = x(-n) is non-casual because y (-1)
= x(1)!

Stable versus Unstable Systems
A system is Stable if any bounded input produces bounded output (BIBO).
Otherwise, it is unstable.
Linearity
A system is linear if
( ) ( ) ( ) ( ) [ ] ( ) [ ] n x T a n x T a n x a n x a T
2 2 1 1 2 2 1 1
+ = +
two systems can be connected to each other in two ways:








T
1
T
2

x(n)
y(n) T = T
1
T
2
= T
2
T
1


= for LTI systems only
T= T
1
+ T
2

T
1

T
2

+
y
x
10
LTI Systems
LTI systems are the most important class of systems because the behavior of the system
is known by knowing its response to unit sample input (x(n)).
Then
( ) ( ) { } ( ) ( )
( ) ( ) [ ]
( )
( ) ( )
( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) ( ) K n x K h K n h K x
n x n h n h n x n y
K n h K x K n F K x
K n K x F n x F n y
K K
K
K n h
K
K
= =
= =
= =

= =

+
=

+
=
+
=
* *
43 42 1

Causality of LTI System
[ ] [ ] ... ) 1 ( ) 1 ( ) ( ) 0 ( ) 1 ( ) 1 ( ) 2 ( ) 2 ( ...
) ( ) ( ) ( ) ( ) ( ) ( ) (
0
1
+ + + + + + + =
+ = =

n x h n x h n x h n x h
k n x k h k n x k h k n x k h n y
4 4 4 4 4 4 3 4 4 4 4 4 4 2 1

this depends on future values of x(n)
Hence, for a system to be casual, its h(n) must be zero for n < 0.

Stability of LTI Systems
The BIBO system means:

( ) ( )
( ) ( ) ( )
( ) ( ) ( ) ( ) ( ) ( )
x
k k k
k
y x
M k h k n x k h k n x k h n y
k n x k h n y
M n y M n x

=
=
< < <
+
=
+
=
+
=

Therefore if ( ) <

k
k h meaning h(k) is absolutely summable, then ( ) < | | n y the
system is stable. This condition is not only sufficient for stability of the system, but it is
also necessary. Indeed we should show that if ( ) , =

k
k h then a bounded input can
produce an unbounded output. Let
11
( )
( )
( )
( )
( )

=
0 n h 0
0 n h
n h
n h
n x
*
*
then ( ) ( ) ( ) k n x k h n y
k
=

and
( ) ( ) ( )
( )
( )
( ) = = = =

+

k h
k h
k h
k x k h o y
k
2
( ) = 0 y unbounded.
The condition ( ) <

k
k h also implies that h(n) goes to zero as n approaches the
output of the system goes to zero as n if the input is set to zero beyond n > n
o
. In
other words, a finite excitation to the system creates an output that eventually dies out
(transient response).
Let's prove it. Suppose ( )
x
M n x < and also x(n) = 0 for n > n
o
. Then at any
n = n
o
+ N , the system output is:
( ) ( ) ( ) ( ) ( ) k N n x k h k N n x k h N n y
o
N k
o
N
k
o
+ + + = +


=
=

=
4 4 4 3 4 4 4 2 1
0
1

( ) 0 = n x Q for
o
n n
Hence, ( ) ( ) ( ) ( ) ( )


=

=
+ = +
N k
x
N K N k
o o
k h M xk k h k N n x k h N n y
Now as ( ) ( ) 0 0 = + =

=


N n y im k h im N
o
N
N n
N
l l Therefore, the transient response
goes to zero and hence the system is stable.
Example: h(n) = a
n
u(n) determine the range that h(n) is stable.
1) System is causal for h(n) = 0 for n < 0
2) ( ) ... 1
2
0
+ + + = =


=

=
a a a k h
k
k
k

Now if |a| <1 this converges to
a 1
1

and the system would be stable but if |a| > 1 the

system is unstable.

If h(n) = 0 for n > M and n < 0, then the system is called FIR (Finite duration Impulse
Response); otherwise the system if called IIR.
12
FIR system has a finite memory and it only looks at the input within a window.

Analyzing the FIR and IIR System
One approach is to use convolution sum but if the system is IIR, its practical
implementation is impossible because it needs an infinite sum. Then to realize an IIR
system, the solution is to use different equations.
There is a sub-class of recursive and non-recursive systems. Consider a commulative
average:

( ) ( )
( ) ( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) n x
n
n y
n
n
n y
n x n ny n x k x k x n y n
n k k x
n
n y
n
k
n
k
n
k
1
1
1
1
1 1
0
1
1
1
0 0
0
+
+
+
=
+ = + = = +

+
=

= =
=







Is this system LTI? No, as it is time-variant because of multiplying by n.

2.4.2 LTI Systems Characterized by a Constant Coefficient Difference Equation
Consider y(n) = a y(n 1) + x(n)
y(o) = ay(0) + x(1) = a
2
y(-1) +ax(0) + x(1)

y(n) = a
n+1
y(-1) + a
n
x(0) + a
n-1
x(1) +
or ( ) ( ) ( ) ( ) ( ) n y n y K n x a 1 y a n y
zs zi
response forced or state zero
n
0 K
K
response input zero
1 n
+ = + =

+

4 4 3 4 4 2 1
43 42 1
**
The linearity applies to each of these responses separately. This system is linear and
time-invariant.
z
-1
x
x
y(n)
x(n)
1
n+1
n
A Recursive System
z
-1

+
x(n)
y(n)
a
13
Impulse Response h(n) is simply equal to the zero-state response of the system.
( ) ( ) ( ) ( ) 0 then ) ( ) ( Let
0 0
= = = =

= =
n a k n a n h n n x k n x a n y
n
n
k
k
n
k
k
zs


Stability Example: is y(n) = a y(n-1) + x(n) stable?
Given a bounded input: ( ) <
x
M n x for all 0 n , from (**) we have
( ) ( ) ( ) ( )
y
n
x
n
n
k
k n
M
a
a
M y a k n x a y a n y =

+ +
+
+
=
+

1
1
1 1
1
1
0
1

if n is finite, M
y
is finite but if n , M
y
is bounded only if |a| < 1.
15
Lecture 4
Properties of ROC
1) ROC is a ring < <
2 1
r z r o
2) The Fourier Transform of x(n) converges if and only if ROC of X(z) includes the unit circle.
(Remember, z = re
j
and if |z| = 1 then r = 1 and then ( ) ( ) ( )
n j j
e n x e X z X

+

= = =
Fourier Transform of x(n). So, if X(z) convergence region includes the unit circle, then
X(e
j
)=X() exists.)
3) ROC cannot contain any poles.
4) If x(n) is finite, then ROC is the entire plane except z = o/
5) If x(n) is right-sided (i.e., x(n) = o for n < N
1
, < ) ROC is the exterior of the largest pole.
6) If x(n) is the left-sided (i.e. x(n) = o for n > N
2
> -) then ROC is the innermost ring of the
smallest pole.
7) If x(n) is two-sided, ROC consists of a ring in z plane, bounded on the interior and exterior by
a pole and not containing any pole.
8) ROC must be a connected region.

Properties of the Z-Transform
Time Shifting: x(n k) Z z
-k
X(z)
Linearity: ax
1
(n) + bx
2
(n) Z aX
1
(z) + bX
2
(z)
But we cannot say ROC = ROC
1
+ ROC
2

Example 1:
( )

=
else
N n
n x
1 0
0
1


Direct Method:

( ) ( )

= =
=

1
1
1
1
1
1
0
1
0
z if
z if
z
z
N
z
z n x z X
N
N
n
n
N

N -1
x(n)
1
0
16
The function
( ) 1
1
1
1
1 1

z z
z
z
z
N
N n
has two poles at 0 and 1, but z = 1 is not a pole for X(z)
because it is defined to be 1 at z = 1. Therefore, ROC the entire plane except z = 0.
Now using Z-Transform properties:
x(n) = u(n) u(n N)
( ) ( ) ( ) ( )
1
1
1
1 1

= =
z
z z U z z X
N N

and ROC of this one is |z| > 1 while that is different from ROC found earlier.
So, if the linear combination of several signals has finite duration, the ROC of its z-transform is
exclusively dictated by the finite duration of this signal and not by the ROC of the individual
transforms.
Scaling ( )
|
.
|

\
|

a
z
X n x a
n
ROC: |a| r
1
< |z| < |a| r
2

Time-Reversal ( ) ( )
1
z X n x ROC:
1 2
r
1
z
r
1
< <
Differentiation ( )
( )
dz
z dX
z n nx same ROC
Example 2:
Determine x(n) if X(z) = log (1+az
-1
) and |z| > || .
( ) ( )
( )
(

=
+
=
+

1
1
1
1
1
2
1
1
1 1 az
az
az
az
dz
z dX
z
az
az
dz
z dX

( ) ( )
( ) ( )
( ) ( ) ( ) n nx n u a a
az
az
InvZ
az
n u a
n n
=
)
`

1
1 1
1
1
1
1
1
Q
( ) ( ) ( ) 1 1
1
=

n u
n
a
n x
n
n

Convolution
( ) ( ) n x n x
2 1
* ( ) ( ) z X z X
2 1

ROC is at least the intersection of that for X
1
(z) and X
2
(z).
Correlation
( ) l
2 1x x
r ( ) ( )
1
2 1

z X z X remember that ( ) ( ) ( ) l l l =
2 1 2 1
* x x r
x x

x*(n) x*(z*)
Z
Z
Z
17
Time Multiplication
( ) ( ) n x n x
2 1
Z ?
( ) ( ) ( ) ( ) ( )
( ) ( )
( ) dv v
v
z
X v X
j
dv v z v n x v X
j
z n x dv v v X
j
z n x n x z X
n n
n
c
n
n
n n
n
1
2 1
1
2 1
2
1
1 2 1
2
1
2
1
2
1

+
+
=

+
=

+
=
|
.
|

\
|
=
(

=
(

= =

Parseval's Theorem
( ) ( ) ( ) dv v
v
X v X
j
n x n x
1 *
2 1
*
2 1
*
1
2
1

|
.
|

\
|
=

It is like evaluating ( ) ( ) { } n x n x Z
*
2 1
at z = 1 circle.
Initial Value Theorem
If x(n) is causal, ) ( lim ) 0 ( z X x
z
=
Proof: ( ) ( ) ( ) ( ) ( ) ... 2 1
2 1
+ + + = =

z x z x o x z n x z X
o
n

if z z
-n
o therefore, ) ( lim ) 0 ( z X x
z
= .
Example:
Using Z-transform properties, find X(z) of the following signal.
( ) ( ) ( ) ( ) 2 2
3
cos 5 . 0 2
2

=

n u n n n x
n


( ) ( ) ( )
( ) ( )
(

)
`

=
)
`

n u
n
Z
dz
d
z z
n u
n
n Z z z X
n
n
3
cos 5 . 0
3
cos 5 . 0
2
2

( ) ( )
5 . 0 ;
25 . 0 25 . 0 1
25 . 0 1
25 . 0
3
cos 5 . 0 2 1
3
cos 5 . 0 1
3
cos 5 . 0
2 1
1
2 1
1
>
+

=
+
|
.
|

\
|

|
.
|

\
|

=
)
`

z
z z
z
z z
z
n u
n
Z
n

From Table on page 174

18
( )

+

=

2 1
1
1
25 . 0 5 . 0 1
5 . 0 1
z z
z
dz
d
z z X
( )
4 3 2 1
5 4 3
0625 . 0 25 . 0 75 . 0 1
0625 . 0 5 . 0 25 . 0


+ +
+
=
z z z z
z z z
z X ; |z| > 0.5
Now, lets use MATLAB to see if weve computed correctly.
b = [0, 0, 0, 0.25, - 0.5, 0.0625];
a = [1, -1, 0.75, -0.25, 0.1625];
n = 0: 20 % checking the fist 21 samples of x(n)
delta = [n =0]; % creating (n)
x = filter (b, a, delta),
plot (n, x), hold
x = [zeros (1, 2) n.* (0.5.

n) * cos(pi * n/3)]; % creating the original signal

n1 = 0:22;
plot (n1, x, r)

Rational Z-Transforms
( )

=
N
k
k
k
M
k
k
k
z a
z b
z X
0
0
if a
o
and b
o
0, then we can rewrite it as: ( )
( )
( )
k
N
k
k
M
k N M
p z
z z
z
a
b
z X


=
=
= +
1
1
0
0

It has M finite zeros at z
1
, z
2
, , z
M
and N finite poles at p
1
, p
2
, ., p
N
as well as N M zeros or
M N poles at origin and a possible zero/pole at . Depending on the location of the poles, the
signal has different behaviors. Read Section 3.3.2.

The System Function of a LTI System
Y(z) = H(z) X(z) H(z) is called the system-function. A system in general can be presented
by a difference equation:

( ) ( ) ( )
( ) ( ) ( ) z X z b z Y z a z Y
k n x b k n y a n y
k
N
k
k
k
N
k
k
M
k
k
N
k
k

=
= =


+ =
+ =
0 1
0 1

19
( )
( )
( )

=
+
= =
N
k
k
k
M
k
k
k
N
k
k
k
M
o k
k
k
z a
z b
z a
z b
z X
z Y
z H
0
0
1
1
, where a
0
= 1
Special Cases:
If a
k
= 0 for 1 < k < N, then ( )
k M
M
k
k M
k
M
k
k
z b
z
z b z H

=

=

= =
0 0
1
, which is an all-zeros system.
The system has M trivial poles at the origin. Such a system has a finite duration impulse response
and therefore is called FIR system.
On the other hand, if b
k
= 0 for M k 1 then ( ) 1 ,
1
0
0
0
1
0
= =
+
=

=

a
z a
z b
z a
b
z H
N
K
k N
k
N
N
k
k
k
.
This system is an all-pole system (has N trivial zeros at origin) and therefore, has an infinite
duration impulse response and thus is called IIR system. A pole-zero system is still IIR because
of the poles.

The Inverse of Z Transform
( ) ( )
k
k
z k x z X

+
=

=
By multiplying both sides of the above formula by z
n-1
and integrating both sides over a closed
contour within ROC of X(z), which encloses the origin, we have:
( ) ( ) dz z n x dz z z X
k n
c
k
n
c

+
=

=
1 1

Since the series converges on this contour, we can interchange and

. Then
( )
43 42 1

=
=

=
k n
k n j
c
k n
k
c
n
dz z k x dz z z X
0
2
1 1
) (

Cauchy Integral Theorem

( ) ( ) dz z z x
j
n x
n
c
1
2
1

One of Caushy Theorems states ( )

=

=

1
1
2 n
n
j
o
dz z z
n
c
o

20
Let z
o
= o. Then, f(z) = z
n
. If n is positive the antiderivitive
1
1
+
+
n
z
n
is analytic every where and
therefore, its contour integral is zero. But only for f(z) = z
-1
it doesnt have an antiderivitive even
in a punctured plane. For 2 n , it is analytic in a punctured plane with origin deleted.
Remember that if f is analytic in a simply connected domain, D, and is any loop (close
contour) in D, then 0 ) ( =

dz z f because in a simply connected domain any loop can be shrunk

to a point. Therefore, the integral of a continuous function over a shrinking loop converges to
zero.

21
Lecture 5
Calculating the Inverse Z-Transform
( ) ( ) dz z z X
j 2
1
n x
1 n

Three methods to calculate it:

1) Direct Method by contour integration
2) Expansion into a series of terms z/z
-1

3) Partial Fraction expansion and look-up table.

Cauchy-Residue Theorem
Let f(z) be a function of the complex variable z and C be a closed path in the z-plane. If the
derivative ( ) z f
dz
d
exists on and inside the contour C and if f(z) has no poles at z = z
0
, then
( )
( )
( )

else
C inside is z if z f
dz
z z
z f
j
c
0 0
0
0 2
1

More generally:
( )
( )
( )
( )

else
C inside z if
dz
z f d
k
dz
z z
z f
j
z z k
k
c
k
0 1
1
0
0
0
! 1
1
2
1

RHS of the above equation is called residue of the pole z
o
.
Now suppose the function can be written as
( )
( ) z g
z f
where f(z) has no poles inside C and g(z) is a
polynomial with simple roots z
1
, z
2
, , z
n
inside C. Then,
( ) ( ) ( )
( )
i
n
i
i
i
i
n
i
c
n
i i
i
c
z A dz
z z
z A
j
dz
z z
z A
j
dz
z g
z f
j

= = =
=

=
(

=
1 1 1
2
1
2
1
) ( 2
1

,

where ( ) ( )
( )
( ) z g
z f
z z z A
i i
= .
( ) ( ) ( ) ( )
i
z z
n
N
i
i
n
z z X z z dz z z x
j
n x
=

= =
1
1
1
2
1

= sum at residue of x(z)z

n-1
at z = z
i
and N =
number of poles.
Example: Problem 3.56 (C)
22
( )
a
z
az
a z
z X
1
1
>

=
( ) dz z
a z
a z
a j
dz z
az
a z
j
n x
n
c
n
c
1 1
1
1
2
1
1 2
1

=



a
r z
1
> = This is the contour C.
Three cases:
1) For 1 n then ( ) ( )
1
1

=
n
z a z
a
z f and the only pole inside c is
a
1
. Therefore, for 1 n

( ) ( )
1 1
1
1
1
1 1
1 1 1 1
+

|
.
|

\
|

|
.
|

\
|
=
|
.
|

\
|
|
.
|

\
|

=
n n
n
a
z
n
a a
a
a
a a
z a z
a
n x

2) For ( )
( )
dz
az z
a z
j
n x o n

= =
1 2
1

then, ( ) ( ) a z
a
z f

=
1
and two poles (0 and
a
1
) inside C. Therefore,
( )
( ) ( )
a z
a z
a z
a z
a
n x
a
z
o z
a
1 1 1
1
1

=

+

=
=
=

3) For n < 0, ( )
( )
dz
z a z
a z
a j
n x
n
c
1
1
1
1
2
1

Therefore, it has a pole at zero with order of (n1) and a pole at 1/a. Since
ROC ( ) n x
a
1
z > is right-sided and therefore, it is enough to find where it reaches the zero
on the left side.
( )
( )
0
1
1
1
1
2
1
1
1 2 2
=
(
(

|
|
.
|

\
|

+

=

=
=
=

o z
a
z
a z
a z
dz
d
z
a z
a
dz
a z z
a z
a j
x

With the same method, we can prove that x(-2)=x(-3)==0 .


23
The Inverse z-Transform by Power Series Expansion
The method is to use long division. How to divide.
Example:
( )
( )( ) 2 3 2 3 1
1
2 1 1
1
2
2
2 1 1 1
+
=
+
=

=

z z
z
z z z z
z X
It has two poles: z = 1 and z = 2. Therefore, if the signal is casual, then ROC: |z| > 2 and if the
signal is non-casual, ROC: |z| < 1.
Case 1:
ROC: |z| > 2 signal is casual and therefore ( ) ( )
n
o n
z n x z X

= has terms with negative power

for z. Therefore, we divide as:

... 317 15 7 3 1
30 32
30 45 15
14 15 0
14 21 7
6 7 0
6 9 3
2 3 0
2 3 1
1
2 3 1
4 3 2 1
5 4
5 4 3
4 3
4 3 2
3 2
3 2 1
2 1
2 1
2 1
+ + + + +
+
+
+
+

+
+
+
+

z z z
z z
z z z
z z
z z z
z z
z z z
z z
z z
z z
x(n) = {1, 3, 7, 15, 31,}

Case 2: ROC: |z| < 1 signal is non-casual ( ) ( )
n
z n x z X

=
0
has terms with positive
powers of z. Therefore, we divide in a way to get z with positive powers.
24

...
8
7
4
3
2
1
4
3
4
7
0
4
3
4
9
2
3
2
1
2
3
0
2
1
2
3
1
1
1 3 2
4 3 2
3 2
3 2
2
2 1
1 2
+ + +
+
+
+
+
+


z z z
z z
z z z
z z
z z
z z
( )

= 0 , 0 ,
2
1
,
4
3
,
8
7
..., n x

Question: How would you use this method for a case like ROC: 1 < |z| < 2?

The Inverse z-Transform by Partial- Fraction Expansion
( )
( )
( ) z D
z N
z X = The goal is to write it in the form
( )
( ) ( ) ( ) ( )
1 1
3
3
1
2
2
1
1
1
1
...
1 1 1

+ +

=
z p
A
z p
A
z p
A
z p
A
z X
N
N

Let ( )
N
N
M
M
z a z a z a
z b z b b
z X


+ + + +
+ + +
=
... 1
....
2
2
1
1
1
1 0
. If it is an improper rational function (M > N) then the
first step is to write it as:
( )
( )
( )
( ) z D
z N
z c z c c z X
N M
N M
1 1
1 0
..... + + + + =

, where now the degree of N

1
(z) is less than N.
The inverse z-transformation of c
0
+ c
1
z
-1
+ . is very straight-forward. Next step is to write
( )
( ) z D
z N
1
or in general a proper function
( )
( ) z D
z N
, where M < N, in terms of positive powers of z,
factorize denominator and then write
( )
z
z X
in terms of partial fractions. In general, lets assume
an X(z) has N simple poles and L multiple poles at k = j. Then:

( )
( ) ( ) ( )
N
N
j
j
j
j
j
j
P
A
p
A
p
A
p
A
p z
A
p z
A
z
z X

+ +

+ +

=
1
...
1
....
1 1
....
2
2 1
2
2
1
1
l
l

25
Then
( ) ( )
k
p z
k
k
z
z X p z
A
=

= k = 1, 2, ., n excluding k = j and
( ) ( )
(

z
z X p z
z d
d
A
k
i
k
k
jk l
l
, k = 1, 2, ..,l
Then

( ) ( )
( ) ( )

<
>
=
)
`

k
N
k
k
n
k
k
p z ROC n u p
p z ROC n u p
z p
Z
: 1
:
1
1
1
1


Example
( )
( )( )
2
1
1
5 . 0 5 . 1 1
1
2
2 1

=
+
=

z z
z
z z
z X

( )
( )( )
2
1
1
2
1
1
2 1

=

=
z
A
z
A
z z
z
z
z X

( )
( )
2
2
1
1
1
1 1
= = =
= z
z
z X
z A
( )
( )
1
2
1
2
1
2
1
2
1 2
=

= =
= z
z
z X
z A
( )
1
1
2
1
1
1
1
2
2
1
1
1
2

=
(
(

=
z
z
z
z
z z X

Now depending on ROC, we get different x(n).

1) ROC: |z| > 1 casual
( ) ( ) ( ) ( ) ( ) ( ) ( ) n u n u n u n x
n
n
n
5 . 0 2
2
1
1 2 = =
2) ROC: |z| < non-casual and left-sided

( ) ( ) ( ) ( )
( ) ( ) 1 5 . 0 2
1 5 . 0 1 1 2
+ =
+ =
n u
n u n u n x
n
n n

3) 1
2
1
< < z non-casual and two-sided
26
( ) ( ) ( ) ( ) ( ) n u n u n x
n n
5 . 0 1 1 2 =

One-Sided Z Transform
One-sided Z Transform is defined as ( ) ( ) ( ) ( )
n n
o n
z n u n x z n x z X

+

=
+

= = . It does not contain
information about x(n) for n < 0. It is unique only for the causal signals that are zero for n < 0. It
is useful to solve difference equations of the systems that are not relaxed initially, but the input is
not necessarily zero before applying to the system.
Properties
If ( ) ( ) z X n x
Z +

+
then ( ) ( ) ( ) o k z n x z X z k n x
k
n
n k Z
>
(

+

=
+
+
,
1

Proof: ( ) { } ( )
n
o n
z k n x k n x Z

=
+

= . Let n k = l then n = 0 l = -k and also -n = -l k.

( )
{
( )
( )
( ) ( )
(

+ =
(
(
(
(

+
|
|
.
|

\
|
=
=
=
+

=
+
K
1 n
n K
1
K
z x
o
n
K
K
K
z X z n x z z x z x z
z z x
l l
l l
l
l
43 42 1
l l
l

Time Advance: ( ) ( ) ( ) 0 ,
1
>
(

=
+
+
k z n x z X z k n x
k
o n
n k Z

Final Value Theorem: ( ) ( ) z X z im n x im
z n
+

=
1
) 1 ( l l
This limit exists if the ROC of (z-1)X
+
(z) includes the unit circle.
This can be proved by the following analogy. If the limit ) ( lim n x
n
exists, then the function
x(n) can be written as 0 ) ( lim and ) ( lim where ), ( ) ( = = + =

n f c n x n f c n x
n n
. Then take the one
sided Z-Transform from both sides and prove the theorem.

Analysis of the Systems in Z-Domain
Without too much restriction, lets assume
( )
( )
( ) z Q
z N
z x = and ( )
( )
( ) z A
z B
z H =
27
( )
( ) ( )
( ) ( ) z A z Q
z N z B
z Y

= . If the system is relaxed, the initial condition = o

If the system is not relaxed, then
( ) ( ) ( )
( )
( )
( )
( )
3 2 1
43 42 1
z Y
o
z Y
zi
zs
z A
z N
z x z H z Y
+
+ = where ( ) ( )
n
k
n
k
N
k
k
z n y z a z N

=

=

=
1 1
0

H(z) has p
1
, ., p
N
poles and X(z) has q
1
,., q
L
poles.
First lets assume that the poles are distinct and not common. Then

( )
( ) ( ) ( ) ( ) ( )
4 4 3 4 4 2 1 4 4 3 4 4 2 1
response force
L
k
n
k k
response natural
N
k
n
k k
L
k
k
k
k
N
k
k
k
k
n u q Q n u P A n y
z q
Q
z P
A
z Y


= =
=

=

+ =

=
1 1
1 1
1 1

Note that natural response zero-input response. It is in fact the no-input response.
Example
( ) ( ) ( ) n x n y n y + = 1
2
1
Find the output when ( ) ( ) n u
n
n x
4
cos 10

=
Solution
( ) ( ) ( ) z X z Y z z Y + =
1
2
1

( )
2
1
2
1
1
1
1
1
=

=

P
z
z H
However, ( )
2 1
1
2
2 1
1
1 10

=
z z
z
z X
4 / *
1 2
4 /
1
j j
e q q e q

= = =
( ) ( ) ( )
4 4 4 4 4 3 4 4 4 4 4 2 1 43 42 1
response force
1 4 j
7 28 j
1 4 j
7 28 j
response natrual
1
z e 1
e 78 6
z e 1
e 78 6
z 2 1 1
3 6
z x z H z Y
o o

= =
/
.
/
.
. .
/
.


( ) ( ) n u n y
n
nr

=
2
1
3 . 6
( ) ( ) n u n n y
o
fr

= 7 . 28
4
cos 5 . 13


28
Testing it with MATLAB:
( )
( )
3 2 1
1
2
1
2
2
1
2
1
2 1
2
1
1 10

|
|
.
|

\
|
+ +
|
.
|

\
|

=
z z z
z
z Y
| | ( )
(
(

|
|
.
|

\
|
+ + = =
2
1
,
2
2
1 ,
2
1
2 , 1 ; 2 / 10 , 10 a b ;
[R, P, K] = residuez (b, a)
(
(
(

=
9074 . 1
2562 . 3 9537 . 5
2562 . 3 9537 . 5
i
i
R ] [
5 . 0
7071 . 0 7071 . 0
7071 . 0 7071 . 0
=
(
(
(

+
= K i
i
poles P
norm (R (1)) = 6.78
angle (R (1)) = -0.5005 rad = - 28.7
o

zplane (b, a) % plots the zero-poles

Example of a Non-Relaxed System
( ) ( ) ( ) ( ) 0 2
2
1
1
2
3
= + n n x n y n y n y
( ) ( ) n u n x
n
|
.
|

\
|
=
4
1
find y(n) if y (-1) = 4 and y (-2) = 10
Solution
Taking Z
+
from both sides:
( ) ( ) ( ) | | ( ) ( ) ( ) | |
1
2 1 1
4
1
1
1
) ( 1 2
2
1
1
2
3

+ + +

= = + + + +
z
z X z Y z y z y z Y z y z Y
( ) ( )
1
1
2 1
2 1
4
1
1
1
2
1
2
3
1

+
+

=
(

+ z
z
z z z Y
Finally: ( )
2 1
2 1
2
1
2
3
1
2
1
4
9
2


+
+
+
=
z z
z z
z Y
or ( )
1
1
1
4
1
1
3
1
1
3
2
2
1
1
1

=
z
z
z
z Y
29
( ) ( ) n u
4
1
3
1
3
2
2
1
n y
response force
n
response natural
n
(
(
(
(

|
.
|

\
|
+ +
|
.
|

\
|
=
3 2 1 43 42 1

Note that natural response is due to system poles and force response is due to the input poles.
Transient response is due to the poles inside the unit circle and steady-state response is due to
poles on the unit circle. In this case,
Transient Response ) (
2
1
4
1
3
1
n u
n n
(
(

|
.
|

\
|
+
|
.
|

\
|

and Steady-State Response ) (
3
2
n u .
Zero-Input (or Initial condition) Response is: ) ( ). ( ) ( z X z H z Y
IC ZI
= and Zero-State Response is:
) ( ). ( ) ( z X z H z Y
ZS
= . In this case,
) (
3
8
2
1
2
4
1
3
1
) ( n u n y
n n
zs
(
(

+
|
.
|

\
|

|
.
|

\
|
=
and ) ( 2
2
1
3 ) ( n u n y
n
zi
(
(

|
.
|

\
|
= .
Note that complete response is either Transient Response+Steady-State Response, or Natural
Response + Force Response, or Zero-Input Response + Zero-State Response. Each response
emphasizes a different aspect of system analysis.

Checking with MatLab:
n = [0: 7]; % just checking the first 8 samples
;

n
4
1
x
|
.
|

\
|
= xic = [1, -2] % terms due to initial conditions (1 2z
-1
)
b = [1, 0];
| |
2
1
2
3
1

= a ;
y
1
= filter (b, a, x, xic);
( ) ( ) ( ) 8 , 1 *
3
2

2
1

4
1
*
3
1
ones n n y + + = ;
30
Since y and y
1
are the same, then our solution is correct!
However, for large order difference equations, it is tedious to determine xic(n) analytically.
MatLab command filtic does find xic as well.
Xic = filtic (b, a, Y, x) where Y and x are initial conditions.
Y = [y(-1), y(-2), , y(-N)]
x = [x(-1), x(-2), , x(-M)]
If x(n) = 0 for n < -1, then x need not to be defined. In our example:
Y = [4, 10];
Xic = 1, -2

Causality and Stability
If h(n) = 0, for n < 0, then the system is causal. Then its ROC is the exterior of a circle. The
stability of a system is quarantined by the condition that the ROC includes the unit circle.
Because the necessary and sufficient condition for a BIBO system is that <

= n
n h ) ( . It
follows that
n
n n
n
n
n
z n h z n h z n h z H

= = ) ( ) ( ) ( ) ( . When evaluated on the unit
circle,

n
n h z H ) ( ) ( . Note that causality and stability are independent of each other. One
doesn't imply the other. However, a causal LTI system is BIBO if and only if all the poles of
H(z) are inside the unit circle.

Pole-Zero Cancellation
Example: y(n) = 2.5 y(n - 1) y(n - 2) +x(n) 5x(n 1) + 6 x(n 2)
( )
2 1
2 1
5 . 2 1
6 5 1


+
+
=
z z
z z
z H
( )
( )( )
( )( ) 2 , 3
2 ,
2
1
2 1
2
1
1
3 1 2 1
2 1
2 1
1 1
1 1
= =
= =


=


z z
P P
z z
z z
z H
It seems that the pole at 2 is cancelled by zero at 2. So, the system is theoretically stable but not
practically.
Do problems: 3,6,7,9,15,22 and 43 of Chapter 3.
31
Lecture 6- Chapter 4
Frequency Analysis of Signals and Systems
Continuous Signals and Discrete-Time Signals


Periodic Aperiodic
Starting with periodic CT signals:
Recall that a linear combination of harmonically related complex exponentials of the form
( )
t kF j
k
k
e C t x
0
2

+
=
= is a periodic signal with fundamental period
o
p
F
T
1
= . In order to find C
k
,
multiply both sides by
Fot 2 j
e
l
and integrate over one period:
dt e C dt e C e dt e t x
l k T
l k
T
t F l k j
k
k
t kF j
k
k
T
t lF j
T
t lF j
P
P P P
43 42 1

+
=
+
=

= =

0
0
) ( 2 2
0
2
0
2
0 0 0 0
) (


( )
p
lFot j
Tp
o
T C e t x =

l
2

( )
Fot j
T
p
e t x
T
C
p
l
l
2
1

= Fourier Series
An important issue is that whether
kFot j
k
k
e C
2
0

+
=
representation is equal to x(t) for every moment
of t. The Dirichlet conditions guarantee that this series is equal to x(t) except at the values of t for
which x(t) is discontinuous. At those values of t, the series converges to the midpoint (average
value) of the discontinuity.
Dirichlet conditions are:
1) x(t) has a finite number of discontinuity in any period.
2) x(t) has a finite number of maxima and minima during each period sufficient but not
3) x(t) is absolutely integrable in any period: necessary
( ) <
Tp
t x
A weaker condition is that signal's energy in one period should be finite: ( ) < dt t x
Tp
2


32
Power Density Spectrum of Periodic Signals
A periodic signal has infinite energy but finite average power.
Parsevals Theorem:

( ) ( ) ( )
( ) ( )
2
K
C T
Tp
KFot 2 j
K
p
KFot 2 j
K Tp
p
Tp
p
2
Tp
p
x
C
dt e t x C
T
1
dt e C t x
T
1
dt t x t x
T
1
dt t x
T
1
P
K p

+

=
= =
= =
4 4 4 3 4 4 4 2 1
* *
*

If x(t) is real then PSD C C C C
k k k k
= =

2
*
2
*
is an even function in frequency and the
phase is an odd function.
Example:







( )
( )

0
0
0
0
2
2
0
2
2 /
2 /
2
sin
sin
2
2
1
0 0
0
kF c
T
A
kF
kF
T
A e e
kT F
A
kF j
e
T
A
dt Ae
T
C
p
p
kF j t kF j
p
kFot j
p
t kF j
p
k
=
=

= =

x(t)
t
-T
p

T
p

-/2 /2
p
o
T
1
F =
p
T
A
F
k
C
A
33
Now if
p
T

decreases ( ),
p
T then C
k
0, which means the signal becomes aperiodic
average power becomes zero.

CT Aperiodic Signals
We can say ( ) ( )


=

t x t x
p
T
p
lim

( ) ( ) ( )
( ) ( ) ( ) dt e t x dt e t x F X
and d e X dF e F x t x
t j t F j
t j Ft j

+

+

+



= =
= =

2
2
2
1

Aperiodic signals are energy signals.

( )
( ) dF F X
dt t x E
x
2
2

+

+

=
=

Energy Density Spectrum: S
xx
(F) = |X(F)|
2

A couple of points:
1) Remember that from only ESD or PSD we cannot reconstruct x(t) because phase information
is lost.
2) C
k
for x
p
(t) is just samples of X(F)
( )
o
p
k
kF X
T
C
1
=
DT Frequency Analysis
First consider a periodic DT signal x(n) = x(n + N)
( )
n
N
k
j
N
o k
k
e C n x

=
2
1

Multiply both sides by
n
N
2 j
e

l

and sum over one period.

34

( )
( )

=
=
=

=

=
else o
N 2 N o K if N
e C e n x
1 N
o n
1 N
o K
n
N
K
2 j
K
n
N
2 j
1 N
o n
, , l
l l




( )
l
l
C N e n x
n
N
j
N
o n
.
2
1
=

( )
( )
( )
2
1
2
1
1
/ 2
/ 2
1
1
1 ,.... ,
1

=
= =
=
= =
N
o n
N
o k
k x
N
o K
N kn j
k
N kn j
N
o n
k
n x
N
C P
e C n x
N o k e n x
N
C

*DTFS is periodic like Periodic DT*
C
k+N
= C
k.
Therefore, the spectrum of a periodic DT, x(n), is also periodic with period N.
( )
( )
( )
k
N
o n
N kn j
N
o n
N n N k j
N k
C e n x
N
e n x
N
C = = =


=

=
+
+
1
/ 2
1
/ 2
1 1


Example
Find DTFS of the following signals:
( )
( ) ( )
15
5
2
sin
3
2
cos
5 3
2
2
1
1
= + =
= =
N n n n x
N
n x
N
n x
43 42 1 43 42 1


For this case, we can directly write it is a sum of exponentials.
( )
( )
( )
(
(
(

+ =
(
(

+ =

43 42 1
3
n 2 2
j
e
3
n 2 3 2
j
3
n 2
j
3
n 2
j
3
n 2
j
1
e e
2
1
e e
2
1
n x

2
1
C
2
1
C
2 1
= = , for x
1
(n)
Interchange the
sums

=
=

=
1 if ,
1
1
1 if ,
1
a
a
a
a N
a
N
N
o n
n

Power
The smallest common denominator
35
( )
( )
(
(

=
(
(

= =


5
n 2 5 4
j
5
n 2
j
5
n 2
j
5
n 2
j
2
e e
j 2
1
e e
j 2
1
n
5
2
n x

sin

j 2
1
C
j 2
1
C
4 1

= = , for x
2
(n) and 0 else where.
C
k
for x(n) is like
2 1
3 5
x
k
x
k
C C

+

=
=
=
else
12
2
1
10 5
2
1
3
2
1
o
k
j
, k
k
j
C
k

Fourier Transform for Aperiodic D.T. Signals

( ) ( ) ( )
( ) ( )

=
=

+
=

d e X n x
e n x X e X
n j
n
n j j
2
2
1

(recall that for C.T. signals it was over
+

and here is over 2 which means that X() is
periodic).
Two Basic Differences Between CTFT and DTFT:
1) X() X(e
j
) is periodic with period 2
( ) ( )
( )
( ) ( )

X e n x e n x k X
n j n k j
n
= = = +

+

+
+
=

2
2
2) Since X() is periodic, (in fact a it is a periodic C.T. signal), then it has a Fourier Series
and in fact x(n) are the coefficients of that Fourier Series.
Before visiting a famous example, lets review the concept of convergence.
If we have a limited observation, we will have the truncation effect, and the famous theory of the
Gibbs. Let ( ) ( )
n j
N
N n
N
e n x X

= if ( ) ( ) 0 lim

4 4 3 4 4 2 1
N
N
X X , then X
N
() converges uniformly
36
to X() as N . This convergence is guaranteed if x(n) is absolutely summable (3
rd
Dirichlet
condition).
( ) <
+

n x this implies ( ) ( ) ( ) < < =

+

+

n x e n x X
n j
, which means X() exists and is
somehow band limited and hence, the uniform convergence.
However, this is a sufficient condition. If x(n) is not absolutely summable but square summable
(finite energy) then X() can exist.
If ( ) < =
+

2
x
n x E , there is not a uniform convergence but there is a mean-square
convergence.
( ) ( ) ( ) 0 lim 0 lim
2
= =

+



error E d X X
N
N
N

Meaning the energy of error goes to zero but not necessarily the error itself.









The example of this particular case is the sinc function.
( ) < < = n
n
n
n x
c
,
sin

This is not absolutely summable. Hence, the ( ) ( )

=
N
N
n j
N
e n x X

doesnt converge to X()
uniformly for all . However, x(n) has a finite energy

c
x
E = . So X
N
() converges to X() in
mean square sense.
- -
c

1
x(n)
X()
c

c
37
( )
n j
N
N
c
N
e
n
n
X

=
sin

Matlab definition: ( )
x
x
x c

sin
sin =
Section 4.2.12
There are two time-domain characteristics that determine the type of signal spectrum and they
are: Periodicity and Continuity
Signals

Continuous Discrete-Time






** Periodicity with period in one domain automatically implies discretion with spacing

1
in
the other domain.
Properties of the Fourier Transform for Discrete-Time Signals
1. Real signals if x(n) is real, the X*() = X(-)
Spectrum magnitude: ( ) ( ) = X X even function
Spectrum phase ( ) ( ) = X X p p odd function.
2. Real and even x(n) Real and Even X()
3. Real and odd x(n) Imaginary and odd X()
4. Imaginary and odd x(n) Real and odd X()
5. Imaginary and even x(n) Imaginary and even X()
6. Linearity a
1
x
1
(n) + bx
2
(n) F aX
1
() +bX
2
()
7. Time-Shifting x(n) F X()
x(n-k) F e
-jk
X()
Aperiodic
&
Continuous
Spectrum
Periodic
Periodic
Aperiodic Aperiodic
Aperiodic
&
Discrete
Spectrum
Periodic
&
Continuous
Spectrum
Periodic
&
Discrete
Spectrum
38

8. Time-Reversal x(n) F X()
x(-n) F X(-) Therefore, FT of an even function is an even function too.
9. Convolution: x
1
(n) * x
2
(n) F X
1
().X
2
()
Proof:
( ) ( ) ( ) k n x k x n x
k
=

+
=
2 1

( ) ( ) ( )
{
( ) ( )
) ( ). (

2 1
) (
2 1
.
2 1
) (

X X
e k n x e k x
e k n x k x X
n
k n j
k
k j
e e
n j
n k
k j k n j
=
=
=


+
=

+
=

+
=
+
=


10. Correlation Theorem: ( ) n r
2 1
x x
F X
1
()X
2
(-)
Proof:
( ) ( ) ( ) n k x k x n r
k
x x
=

+
=
2 1
2 1

FT
( ) ( ) ( ) ( )
( )
3 2 1
K j K n j
2 1 2 1
e e
n j
2
n K
1
n j
n
x x x x
e n K x K x e n r S

+
=
+
=

+
=
= =
( )
( )
( ) [ ]
( )
( )
4 4 4 4 4 3 4 4 4 4 4 2 1 4 4 4 3 4 4 4 2 1

+
=

+
=

=
2 1
x
n
K n j
2
x
K
K j
1
e K n x e K x RHS
Now if x(n) is real, then X
*
() =X(-)
( ) ( ) ( ) ( )
( )
( )
2
x
xx xx
x x x S r

= =
3 2 1
l
*

Energy Spectral Density
11. Frequency Shifting
e
jon
F X( -
o
)


12 Modulation Theorem
Cross-Energy
Density Spectrum
39

( ) ( ) ( ) ( ) [ ]
[ ]
n j n j
o o
e e n
X X n n x
0 0
2
1
) cos(
2
1
cos
0
0

+ =
+ +

13. Parseval Theorem
( ) ( ) ( ) ( )

d X X n x n x
n
*
2 1
*
2 1
2
1

+
=
=
Proof:

( )
( ) ) ( ) (
2
1
) (
) (
2
1
*
2 1
) (
*
2 1
*
2 1
*
2
n x n x d e X n x
d X e n x RHS
n
n x
n j
n
n j
n

=
= =

=
4 4 4 3 4 4 4 2 1

Special case: ( ) ( ) ( ) ( )

d x
2
1
n x n x n x
2
2
2
1 2
= =
( ) ( ) ( )
( )
= = =
+
=


2
d S
2
2
n
xx x
xx
d x
2
1
n x 0 r E
43 42 1

14. Windowing
( ) ( ) n x n x
2 1
F ( ) ( ) [ ]

2 1
*
2
1
X X
( ) ( ) ( ) ( )
n j
2 1
n j
e n x n x e n x x

+

+


= =

( ) ( )
( ) ( )
( ) ( ) ( ) ( ) [ ]

2 1 2
2
1
2
2
1
2
2
1
*
2
1
2
1
2
1
2
1
X X d X X
e e n x d X
e n x d e X RHS
n j n j
n
n j n j
=
=

+
=

+



15. Differentiation in Frequency Domain
n(x)n F j
( )

d
dx

40
Skipping to Section 4.4.8
Correlation Function and Power spectra for Random Input Signals
When the input signal is random, then we have to consider statistical moments of input and
output. So here is a bit of introduction about Stationary Random Process. Starting with the
Definition of Stationary Signals:
If X(t) is a random process with a point Probability Density Function (PDF),
( ) ( ) n x x x x P x P
t t t t
,..., , ,
3 1
2
= for n random variables. ( ) ( )
n I i i
i t x t X
,... 2 , 1
,



















If the joint probability of ( ) ( )
+
=
1 1
,.... ....., , ,
1 1 1
t
n
t
n
x x P x x x P for all t
1
and then the random
process X(t) is stationary in strict sense. In other words, statistical properties of a stationary
random process is time-invariant, meaning that its mean and variance and other moments are
time invariant.

One Observation
Set 1
Set 2
Set 3
t
t
t
t
1
t
1
t
1
t
1
+
t
1
+
t
1
+
x(t,S
1
)
x(t,S
2
)
x(t,S
3
)
x
1
(t
1
)
x
2
(t
1
)
x
3
(t
1
)
x
3
(t+)
x
1
(t
1
+)
x
2
(t+)
41
Statistical (ensemble) Average
( ) ( )
i i i i
t t t t
dx x P x x E

+

=
If we dont have P(x
ti
) but have many observations, then ( ) ( )
1 1 1 1
...
1
2 1
t
N
t t t
x x x
N
x E + + + = , which of
a stationary process it is equal to E(x
ti
) for any t
i
.
Also autocorrelation function:

( ) ( )
[ ]
2 1
2 1 2 1 2 1 2 1
, ,
t t
t t t t t t t t
xx
x x E
dx dx x x P x x x x
=
=

+

If this
xx
depends only on the time difference t
1
t
2
= , then ( ) [ ], ,
1 1

+
=
t t xx
x x E which is the
case for stationary process. If a process has two features:
1) Its
xx
depends only on time difference, , and
2) E(x
t1
) = E(x
t2
) = E(x
ti
)
then the process is said to be stationary in wide sense.
Now if all statistical averages can be obtained by one single realization (or one sample set), then
the process is also Ergodic meaning Ensemble average time average.
( )
n x
x E = and ( )

=
=
1
2
1
N
o n
n x
N

)

x

)
is an estimate of
x
. It will be said it is an unbiased estimate if ( )
x x
E =
)
. Also, it is a
good estimator if
( ) ( ) 0
2 2
=
x x x
E Var
) )
as N
Therefore, time average ensemble average.
Autocorrelation: ( ) ( ) ( ) m n x n x
N
m
N
o n
xx
+ =

=
1
*
1
( ) [ ] ( ) m r m E
xx xx
= the true autocorrelation.
Now back to systems:
x(n) y(n) h(n)
42

( ) { } ( ) ( )
( ) ( ) { } ( ) ( ) 0 H k h k n x E k h
k n x k h E n y E
x x
n
y

= = =
)
`

= =


+

+

+
=

The autocorrelation sequence:

( ) ( ) ( ) { } ( ) ( ) ( ) ( )
( ) ( ) ( ) ( ) { }
( ) ( ) ( ) k m h k h
m n x k n x E h k h
m n x h k n x k h E m n y n y E m
xx
k
k
k
yy
+ =
+ =
)
`

+ = + =

+
=
+
=
l l
l l
l l
l
l
l

.
*
* *

Special Form: when x(n) is a white noise, then ( ) ( ) m m
x xx

2
= and ( ) 0
2
xx x
= . Then
( ) ( ) m m
hh x yy

2
=
( ) ( ) ( )

d H
x hh x yy
2
2
2 2
2
1
0 0

= =
by getting the Fourier transform in general:

2
) ( ) ( real is signal the If
) ( ). ( ). (
) ( ) ( ) ( ) (
. . let
) ( ) ( ) (
) ( ) (

H
H H
e u e l h e k h
e e e e k l m u
e k l m l h k h
e m
xx
xx
u
u j
xx
l
l j k j
k
yy
k j l j u j m j
m j
m k l
xx
m
m j
yy yy
=
=
=
= + =
(

+ =
=

1. The Concept and Representation of Periodic
Sampling of a CT Signal
2. Analysis of Sampling in the Frequency Domain
3. The Sampling Theorem the Nyquist Rate
4. In the Time Domain: Interpolation
5. Undersampling and Aliasing
Signals and Systems
Fall 2003
Lecture #13
21 October 2003
We live in a continuous-time world: most of the signals we
encounter are CT signals, e.g. x(t). How do we convert them into DT
signals x[n]?
SAMPLING
How do we perform sampling?
Sampling, taking snap shots of x(t) every T seconds.
T sampling period
x[n] x(nT), n = ..., -1, 0, 1, 2, ... regularly spaced samples
Applications and Examples
Digital Processing of Signals
Strobe
Images in Newspapers
Sampling Oscilloscope

Why/When Would a Set of Samples Be Adequate?
Observation: Lots of signals have the same samples
By sampling we throw out lots of information
all values of x(t) between sampling points are lost.
Key Question for Sampling:
Under what conditions can we reconstruct the original CT signal
x(t) from its samples?
Impulse Sampling Multiplying x(t) by the sampling function
Analysis of Sampling in the Frequency Domain
Important to
note:
s
1/T
Illustration of sampling in the frequency-domain for a
band-limited (X(j)=0 for ||>
M
) signal
No overlap between shifted spectra
Reconstruction of x(t) from sampled signals
If there is no overlap between
shifted spectra, a LPF can
reproduce x(t) from x
p
(t)
The Sampling Theorem
Suppose x(t) is bandlimited, so that
Then x(t) is uniquely determined by its samples {x(nT)} if
Observations on Sampling
(1) In practice, we obviously
dont sample with impulses
or implement ideal lowpass
filters.
One practical example:
The Zero-Order Hold
Observations (Continued)
(2) Sampling is fundamentally a time-varying operation, since we
multiply x(t) with a time-varying function p(t). However,
is the identity system (which is TI) for bandlimited x(t) satisfying
the sampling theorem (
s
> 2
M
).
(3) What if
s
2
M
? Something different: more later.
Time-Domain Interpretation of Reconstruction of
Sampled Signals Band-Limited Interpolation
The lowpass filter interpolates the samples assuming x(t) contains
no energy at frequencies
c
T
h(t)
Graphic Illustration of Time-Domain Interpolation
Original
CT signal
After sampling
After passing the LPF
Interpolation Methods
Bandlimited Interpolation
Zero-Order Hold
First-Order Hold Linear interpolation
Undersampling and Aliasing
When
s
2
M
Undersampling
Undersampling and Aliasing (continued)
Higher frequencies of x(t) are folded back and take on the
aliases of lower frequencies
Note that at the sample times, x
r
(nT) = x(nT)
X
r
(j) X(j)
Distortion because
of aliasing
Demo: Sampling and reconstruction of cos
o
t
A Simple Example
Picture would be
Modified
M ul t ipl e Choice

1. The system y(t) = x(t) + 2x(t+3) is
(a) causal system (b) non-causal system
(c) partly (a) and ( b) (d) none of these

2. The system
dt
t dy ) (
+3y(t) = x(t) is a
(a) time invariant system (b) time variant system
(c) partly (a) and ( b) (d) none of these

3. The system y(t) = 3x(t) + 4 is
(a) linear system (b) non linear system
(c) partly (a) and ( b) (d) none of these

4. A power signal has infinite energy whereas an energy signal has zero
average power.
(a) true (b) false

5. Find dynamic system of the following
(a) y(n) = 3x(n) (b) y(n) = x(n) + x(n-1)

6. z transform of the signal {0,1,3,2,0,0,0.} is
(a) 1+
z
2
+
2
3
z
(b) 1+
z
3

(c) 1+
z
3
+
2
2
z
(d) 1+
2
2
z


7. The ROC of Question no.6 is
(a) whole of z plane (b) whole of z plane except z =
(c) whole of z plane except z = 0 (d) none of these



8. If x(n) and y(n) are two finite sequences, then x(n) * y(n) is
(a) X(z)/ Y(z) (b) Y(z)/ X( z)
(c) X
2
(z) Y(z) (d) X( z) Y(z)

9. If X(z) =
) 45 . 0 95 . 0 )( 1 (
5 . 0
2
2
z z z
z
, the initial value of x(n) is
(a) 0.5 (b) 1
(c) 0 (d) 0.45

10. z transform of x(nk) is
(a)
k
z

X(z) (b)
n
z

X(z)
(c)
1
z X(z) (d) none of these

11. Find correct representation for the following
(a) x1(n) * x2(n) = x2(n) * x1(n)
(b) x1(n) * [ x2(n) * x3(n)] = [ x1(n)* x2(n)] * x3( n)
(c) x1(n)* [ x2(n)+x3(n)] = x1(n) * x2(n) + x1(n) * x3(n)
(d) all are valid

12. The ROC of the z transform of the discrete time sequence
x(n) =
n
)
3
1
( u(n)
n
)
2
1
( u(-n-1) is
(a)
3
1
<| z| <
2
1
(b) | z| >
2
1

(c) | z| <
3
1
(d) 2<| z| <3

13. The input and output of a continuous time system are respectively denoted
by x(t) and y(t). Which of the following descriptions corresponds to a
causal system?
(a) y(t) = x(t-2) + x(t+4) (b) y(t) = (t-4) x(t+2)
(c) y(t) = (t+4) x(t-1) (d) y(t) = (t+5) x(t+5)

14. The z transform F(z) of the function f(nT) =
nT
a is
(a)
T
a z
z

(b)
T
a z
z

(c)
T
a z
z

(d)
T
a z
z

15. Convolution of x(n+5) with impulse function (n-7) is equal to
(a) x(n-12) (b) x(n+12)
(c) x(n-2) (d) x(n+2)


Q uest ions
PA RT A
1. What is meant by discrete time signals?
2. What are the methods used to represent the discrete time signals?
3. Define Z transform?
4. What is meant by continuous time signals?
5. Define System
6. Sketch the block diagram of DSP system
7. What are the advantages and dis advantages of DSP?
8. Give some application of DSP?
9. Define impulse & unit step signals
10. List the mathematical operations performed on discrete time signals?
11. perform addition of discrete time signals
X
1
(n) = {2,2,1,1} X
2
(n) ={-2,-1,3,2}
12. Classify the system with examples
13. Define signals & describe the classification of signals.

PA RT B
1. Consider the analog Signal X(t)=3 COS (100 t)
i) Determine the minimum sampling rate required to avoid Aliasing
ii) What is the signal obtained if the sampling is 200 Hz
iii) What is the signal obtained after f= 75 Hz

2. Determine whether the system described by the following equations are
liner time invariant
i) Y(n)=nX(n)
ii) Y(n)=aX(n)+b

3. Find the output of the given system by use of convolution
i) x(n)={1,2,1,3,2}
ii) h(n) ={1,1,2,1,3}

4. Find out the convolution by using the Z transform
i) h(n)={1,1,2,3}
ii) X(n)={2,5,6,8}

5. Determine the impulse response of the system described by the difference
equation
Y(k)=Y(n-k)+X(k)


6. Solve the following Difference Equation by Z transform
X(k+2)+3X(k+1) +2X(k)=0 X(0)=0: X(1)=1
7. Define z transform and its properties with proof.
8. Explain the properties of the system?
9. Explain the advantages and disadvantages of DSP over ASP
Find Z transform Of (n)-0.95(n-6)