DSP - SGM
DSP - SGM
DSP - SGM
A PRESENTATION BY
SUGATA MUNSHI
DEPARTMENT OF ELECTRICAL
ENGINEERING
JADAVPUR
of 112
UNIVERSITY
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Discrete-Time Signals
Digital Signal is particular type of discrete-time signal.
A discrete time signal is one which is defined only at
discrete instants of time.
s(t)
samples.
Hence it is also known
t
as :
sampled-data signal.
of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
magnitude quantization
levels
s(t)
5q
4q
3q
2q
q
Digital
signal
ADC
Digital
Signal
Processor
Digital
signal
DAC
Analog
signal
Digital
Memory
BENEFITS
DIGITALLY:
of 112
OF
PROCESSING
SIGNALS
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Reproducibility:
Identical
Can be programmed
achieved,
complex
adaptive
of 112
fitering
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
fs
of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x*p t x t p t
p(t) (carrier)
Pulse
Amplitude
Modulator
x(t)
modulating
signal
x(t)
()
x*p(t)
= Time period of
unit pulse train
= Sampling period
= Pulsewidth =
Sampling duration.
of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
p t u t K u t K
K
x t x t u t K u t K
K
*
p
That is,
for K t K 1
Then,
8
of 112
K=0,1,2,....... ,
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x t x K u t K u t K
*
p
1 es K s
X s x K
e
K
s
*
p
Now,
1 e
1 1 s
...........
2!
s
If is very small compared to , 1 e s
X *p ( s)
Ks
x
(
K
x (t ) x (t )
*
p
x( K ) (t K )
of 112
x( K )
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x*p(t)
x(t)
()
x(t)
x*i(t)
= x*(t)
(very
small)
x*(t)
impulse sampler
x (t )
*
i
x( K ) (t K )
x (t )
*
x( K ) (t K )
Thus, for the sake of mathematical maneuvering a sampleddata signal can be represented by a train of scaled impulses
and the strength of the impulse at any instant is equal to the
value of the sample at that instant. That is, actual sampling
process can be represented mathematically by
sampling or impulse modulation.
10 of 112
impulse
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t)
x*(t)
Impulse
Modulator
t
T(t)
x (t ) x(t ) (t ) x(t ) (t K )
*
- 4
- 3 - 2
11 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Reconstruction
of
Continuous-time
to
x*(t)
12 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
2
x t A sin t A sin 2 ft A sin
T
Sampling rate s
where, n = 0,1,2,..
f
x n A sin 2
n
fs
2 n
A sin
or,
f
T
where, s
f
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x*(t)
Reconstructed analog
sinusoidal signal from
x*(t)
14 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
f
fs f
2
n
x n A sin 2 n 2 2 n
1
A sin 2 n 1
f fs
A sin 2 n
f
s
f
f
fs
2
3
n
x n A sin 4 n 2 4 n
1
A sin 2 n 2
f 2 fs
A sin 2 n
f
s
f
f
fs
3
4
15 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x n A sin 6 n 2 6 n
1
A sin 2 n 3
f 3 fs
A sin 2 n
fs
for
f
f
fs
K 1
K
K 1, 2,3, etc
.x n A sin 2 K n 2 2 K n
f Kf s
A sin 2 n
f
s
16 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Also,
and
i.e,
2 n
x n A sin 2 Mn
n
x n A sin 2 Mn 2
M = 1,2,3, etc.
x n A sin 2 n M
1
x
(
n
ASin
[2
n
(
M
) ]
and
or,
M = 1,2,3, etc.
Mf s f
x n A sin 2
n
fs
Mf s f
x
n
A
sin
2
and
f
s
M = 1,2,3, etc.
represented
by
f f Kf s
17 of 112
x n , has a frequency
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Frequency
f Kf s is the Alias of
frequency f .
Phenomenon is called aliasing.
It is the
effect of undersampling.
Case III :
= 1, i.e., fs = f, = T
x n A sin 2 n A sin
for all values of n.
x(t)
x*(t)
A sin
Reconstructed
signal
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
CASE IV:
1 2, i.e. , f f s 2 f ,
T
T
2
x(t)
x*(t)
t
x n A sin 2 A sin 2 n 2 1 n
A sin 2
n
f fs
A sin 2
n
f
s
f f
A sin 2 s
n
fs
19 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Also
2 n
x n A sin 2 Mn
M = 1,2,3,etc
2 n
x n A sin 2 Mn
M = 2,3,4 etc.
Mf s f
x n A sin 2
n
fs
M=1,2,3,..
and
Mf s f
x n A sin 2
n
fs
M=2,3,4..
20 of 112
fs
.
2
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Frequency
fs f
is
alias of f for
f fs 2 f
Case V:
2, i.e., f s 2 f , T
x n A sin 2 A sin n
2
x(n ) ASin
x n A sin
for even n
&
for odd n .
such that
B sin A sin
B sin A sin .
21 of 112
yield
Hence, if
0,
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t)
ion
Reconstruct
x*(t)
2, i.e. f s 2 f , T
n
x n A sin 2
A sin 2 Mn 2
n
A sin 2 Mf s f
fs
22 of 112
(M = 1,2,3......)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Also,
n
x n A sin 2 Mf s f (M = 1,2,3,..)
fs
x(t)
Reconstruction
x*(t)
Observations :
1
x t A sin 2 ft sampled at f s starting
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
0 f
1
fs .
2
x n A sin 2 fn
(1)
&
y n A sin 2 Mf s f n
A sin 2 nM 2 fn
or
y n A sin 2 fn
Sequences x n and
(2)
y n are identical.
24 of 112
fs
.
2
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Frequency
1
fs
2
folding frequency
or Nyquist
frequency.
Frequency f is alias of Mf s f .
z t As i n 2 Ms f f tare sampled at
Again, if
f s , 0 f s , then
2
z n A sin 2 Mf s f n
A sin 2 nM 2 fn
or
z n
A sin 2 fn
Hence frequencies Mf s f
(3)
fs
0
interval
2.
Frequency
Spectra
of
Discrete-Time
Signals
x(t) is sampled at f s 1 to obtain a D.T. signal x*(t).
25 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t )e jt dt
F.T. of x*(t) is ,
X j F x n t n
n
F x t t n
n
*
t t n , with a period .
n
( )
Fm
(t-n )=
n=-
(t ) e
jms t
dt
26 of 112
e jmst
t 0
(t ) e
Fm e jmst
jms t
for all m,
dt
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Therefore,
( )
where,
(t-n )=
n=-
s 2 f s
e jmst
Hence,
X ( j )
*
X j m
s
m -
X j X
and
X * j X *
27 of 112
X ms f s X ms
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
ms .
f s X ( )
centred at
m 0, 1, 2,.....
f m BW
. . in Hz =
s 2W
i.e.,
W
2
x*(t) by
passing x*(t) through an ideal (brick-wall type) L.P. filter with cutoff
freq.
s
1
& pass band gain
(i.e. ).
fs
2
28 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
If s 2W i.e.
f s 2 f m adjacent replications of
X*()
29 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
ANTI-ALIAS
FILTER
x t is processed (prior to
30 of 112
fs
2 .
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
31 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
THE Z-TRANSFORM
What is Z-Transform ?
x(t) sampled at a rate
fS 1
x * (t )
x(n ) (t n )
(1)
where n = 0, 1, 2,
or, x * (t )
where
x (t n )
(1a)
xn x(n) x[n] x n x t t n
L[ x * (t )] L[ xn (t n )]
n
or,
X * ( s)
(2)
sn
x
e
n
(3)
32 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
s
e
z . Then z is a complex variable.
Let
Substitution of
X * ( s)
z es
in (3) gives
n
x
z
n X ( z)
(4)
Z [ xn ] X ( z )
xn z n
(5)
Z [ xn ] X ( z ) xn z n
n 0
(6)
33 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
The Complex-Variable z
z e s
1
s ln z
z 1 1 z 1 3 1 z 1 5
ln z 2
..............
z
1
3
z
1
5
z
1 z 1
ln( z ) 2
1
1
Hence,
2 1 z 1
s
1 z 1
s j
z es e e j
or,
z e
e cos j sin
u jv
34 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
z e
if 0,
z 1
if 0,
z 1
if 0,
z 1
Z-TRANSFORMS OF SOME
ELEMENTARY
SEQUENCES:
1. Unit Step Sequence
un 1 for n 0
0 for n0
Un
1
-2 -1
01 2 3 4 5 6
Z un un z
n
z n
n 0
1 z 1 z 2 ........
35 of 112
1
z
1 z 1 z 1
; ROC : z 1
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
L u t
1
s
jv
z-plane
s-plane
xn e an u n
Z xn e an z n
n 0
1 e a z 1 e 2 a z n e 3a z 3 .........
1
z
a
;
ROC
:
z
e
1 e a z 1 z e a
36 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Alternatively,
xn e anun ;
where
number of samples.
Then,
1
X ( z ) Z xn
1 e a z 1
;
ROC : z> e-a
If a is real:
37 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
n
1
n
0
-5 -4 -3 -2 -1
1 2 3 4 5
xn 0 for n0
A sin o n for n 0
i.e. xn = A sin o n un
xn
38 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Z [ A Sin 0 n un ]
Az sin o
z e jo
z e jo
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
then,
Az 1 Sin0
X ( z)
; ROC : z 1
-1
2
1-2z Cos0 z
xn f n n
, then
X z Z[ xn ] F z z
40 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
X z Z xn Z f n n
f n n z n
n 0
n 0
n 0
f n z n n z n
Z f n Z n
Proof:
F z z
X(z) = Z[ xn ] ; ROC : r2 z r1 ,
then ,
Z a n xn X a 1 z
ROC : a r2 z a r1
Z a n xn a n xn z n
n
xn a z
X a 1 z
1 n
1
Z un
Example :
2
1 1 z 1
2
Z xnk z K X z
ROC:same as of X(z) except z = 0
41 of 112
(1)
K
Z
x
z
X z
nk
and
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
(2)
*
delaying the time function x t by K time units, i.e. xn by
K samples.
shift.
Multiplication of a Z-transform by zk
*
advancing time function x t by K time units, i.e. by K
samples.
g n is given by
yn xn g n xk g n k
k
for n 0, 1, 2, 3 etc.
42 of 112
(1)
xn
&
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
yn
xn
& g n , then,
Y z X zG z ;
ROC: At least the intersection of those for
X z and G z .
That is, D.T. convolution of 2 sequences results in
multiplication in z-domain.
Proof:
Y z yn z
[ xK g n K ] z n
n K
or,
Y(z) = xK g n K z n
xK z
n K
gnK z
Let m=n-K
Y ( z )
Note:
43 of 112
z m X ( z ).G ( z )
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
yn xn g n xk g nk
for n = 0, 1, 2,
k 0
etc.
E im xn xn
2
N n N
Normalized power:
N
1
xn
N 2 N 1 n N
P im
If E is finite, obviously P = 0
In this case,
xn
is an energy signal.
xn
is a power signal.
Correlation Property
The cross-correlation of finite energy sequences x(n) is y(n) is
given by
44 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
rxy (l )
x ( n) y ( n l ) x ( n l ) y ( n )
(1)
for
l 0, 1, 2, 3 etc.
and,
ryx (l )
y ( n) x ( n l ) y ( n l ) x ( n )
for
(2)
l 0, 1, 2, 3 etc.
Then,
S xy z Z rxy l X z Y z 1
; ROC is at least
1
the intersection of those for X z and Y z
Note:
rxy (l ) ryx (l )
Autocorrelation of a finite-energy sequence x(n) is
rx (l ) rxx (l )
x ( n) x ( n l ) x ( n l ) x ( n )
S x ( z ) Z rx (l ) X ( z ) X ( z 1 )
45 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
If
X ( z ) Z [ xn ]
Z [nxn ] -z
dX(z)
; both transforms have the same ROC
dz
Example :
Z [un ] U ( z )
z
;
z 1
Z[nu n ] -z
dU(z)
d z
z
z
dz
dz z 1 ( z 1) 2
x0 im X z
z
x im
z 1
46 of 112
z 1 X
z
IVT
z 1) X ( z ) .FVT
z lim(
z 1
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
z e e cos j sin
j
jv
S-plane
Z-Plane
Z
Trnasform
For = 0, |z| = 1.
47 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
:
Mapping of Left half of the s-plane into the z-plane:
in direction of j axis.
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
1
1
Each strip extends from j ( N ) s to j(N )s in direction
2
2
s
2
N=1
Complementary strip
Primary strip
Complementary strip
N=0
j3
N=-1
s
2
N = 0 primary strip,
Others are complementary strips.
Mapping of left half of Primary Strip
j
jv
j
C
-
Primary
strip A
D
s
2
49 of 112
S-plane
s
2
Z-plane
u
A
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
counter clockwise
direction.
During
traversal,
varies
in
following
At A, z 10o
Interval [A,B]
z 1
180
At B, z 1180o
Interval B, C
z z 180o ,
1 z 0
0 l
At C , z 0 180o
Interval C , D ,
z z ,
1800 1800 ;
At D, z 0 180o .
Interval D , E :
50 of 112
z 0
manner.
z z 180o ,
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
1 z 0
At E , z 1 180o
Interval E , A :
z 1 , 1800 0o
Sc
Sp
js j 3 s
2
p j s
2
0
Gc
j
s
2
j3
where
s p c j p is point in
L.H. of primary.
In general, any point sc in L.H. of any compl. strip can be
represented by
sc s p jNs
where
51 of 112
primary strip.
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Hence, in z-domain,
zc e
sc
s p jNs
s p
e e
j 2 N
s p
e e jNs
e
s p
zp
Inference:
Correspondence between
z-plane &
s-plane is not
unique.
A point in z-plane corresponds to infinite no. of points in
s-plane, although a point in s-plane corresponds to a
single point in z-plane.
So L.H. of every complementary strip is mapped into unit circle
in z-plane.
Aliasing of complementary strips into primary strip .
52 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
j
N=2
N=1
jv
Z
Trnasform
= 0, + s , + 2s------
Primary
Strip (N=0)
N=-1
N=-2
Z-Plane
S-Plane
Entire L.H. of s-plane is mapped into interior of unit circle in zplane, & entire R.H. of s-plane is mapped into exterior of unit
circle in z-plane.
j axis in s-plane maps into circumference of unit circle in zplane.
X z Z xn xn z n
n
53 of 112
(1)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
where,
z re j and is in radians.
Now,
r z e
X z xn r n e jn
(2)
X z xn r e
n j n
or,
X z xn r
or,
xn r n
n
xn r n
n0
xn
X z x n r n
n 1
n0 r
54 of 112
n j n
xn r e
n
(3)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x n r n , (1 n )
is absolutely
n
summable, i.e. x n r .
n 1
< r1.
n 0
xn
rn
r r2 .
55 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
r2 r r1 ,
an annular region
i.e. r2 z r1 , which is a
Example:
n 0
n 0
az 1 1, i.e. z a ,
If
power
series
1 az .
1
X z
1
; ROC : z a
1 az 1
56 of 112
1
; ROC : z 1
1 z 1
converges
to
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
jv
ROC
Example:
a 1 z
m 1
where, m = n .
2
3
Now, C C C ..........
C
1 C
a 1 z
1
X z
,
1
1
1 a z 1 az
if,
a 1 z 1, i.e.
z a
1
n
a
u
; ROC : z a .
n 1
That is,
1 az 1
ROC is interior of a circle of radius |a| & centred at origin in
z-plane.
57 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Observation:
Z-transforms of
anti-causal
xn a nun bnu n1
n 0
X z a n z n bn z n
az 1 b 1 z
n 0
58 of 112
m 1
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
z b
Case I:
If
not exist.
Example :
b 1 ,a 1
3
2
Case II:
If b a , there is a ring in z-plane with inner radius |a| and
outer radius |b| where both power series converge simultaneously.
Thus,
a b z 1
1
1
X z
; ROC: a z b
1
1
1
2
1 az
1 bz
1 a b z abz
59 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
jv
jv
b
u
a
ROC
b < a
X(z) does not exist
Examples:
1.
[r 2 = 0 ]
sequence.
60 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
[r1 =]
NOTE :
1. A rational Z-transform always has at least one pole
located on any boundary separating its regions of
convergence and divergence.
INVERSE Z-TRANSFORM
If
Z xn X z ,
then,
xn Z 1 X z
z2
X ( z)
; ROC : z 1
( z 1)( z 0.2)
Find Z -1 X ( z )
61 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
z2
1
X z 2
xo Lim X z Lim
1
z
z 1 1.2 z 1 0.2 z 2
z 1 X z Lim 1 1 1.25
x Lim
z 1
z 1
z
1 0.2 z 1 0.8
The long division is illustrated below:
1 1.2 z 1 0.2 z 2
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
But,
X z xo x1 z 1 x2 z 2 x3 z 3 ..............
xo 1
x1 1.2
x2 1.24
x3 1.248
If X z 2
; ROC : |z| < 0.2,
1
z 1.2 z 0.2 1 1.2 z 0.2 z 2
63 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
xn ..........780,155,30,5, 0, 0
X z log 1 az 1 ;
z a
1 x n
log 1 x
n 1
n
Here,
x az 1
1
X z
n 1
an z n
n 1
xn Z 1 X z
or, xn
64 of 112
n 1
n 1
a n for n 1
n
0 for n 0
n 1
a nun 1
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Alternatively
X z log 1 az 1 ;
z a
dX z az 2
dz
1 az 1
xn Z 1 X z
Let,
dz
1 az 1
a
n
Z a a un
1
1 az
n 1
Z nxn Z a a un 1
Z [nxn ] a a
or, xn
n 1
un 1
n 1
a nun 1
Problem:
X z sin z, ROC includes z 1 .
65 of 112
2
K
dX z
z2 d X z
zK d X z
......
.....
dz z 0 2! dz 2 z 0
K ! dz K z 0
z3 z5
z 2 n 1
n
..... 1
n 0
3! 5!
2n 1!
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
X z xn z n
n
xn 1
1
,
2
n
1
!
n 1, 3, 5,.........
x t
u t
X s
1
s
1
e at u t
sa
1
at u t
s ln a
X z
z
z 1
z
e anun
z ea
z
a n un
za
xn
un
table for Z-
X z
Hence we expand
into partial fractions & then
z
66 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Problem:
X z
z
z2 z 1
is
Z-transform of
Determine xn.
Solution:
X z
1
2
z
z z 1
1
3
1
3
z 2 z 1 z j
z
2
2
2
2
j
j
3
z e z e 3
67 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
X z
K1
K2
j
j
z
z e 3 z e 3
1
1
K1 j
e 3 e 3
j 2sin
3
1
1
K 2 j
e 3 e 3 j 2sin
3
X z
j
j
3
3
z
e
z
j 2sin
j 2sin
3
3
1
z
1
z
j
j
j 3 (z e 3 ) j 3 (z e 3 )
xn Z 1 X z
jn
1 jn 3
3
e
e
un
j 3
2
n
sin
3
3
un
Problem:
z3 z2
X ( z) 2
; ROC : z 3
z 4z 3
68 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Solution:
X z
z2 z
z
z 1 z 3
K1 z 1
K 2 z 3
69 of 112
5 z 3
1
z 1 z 3 z 1
5 z 3
6
z 1 z 3 z 3
X z
1
6
1
z
z 1 z 3
1
6
or, X z z
1 z 1 1 3z 1
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
xn n 1 un 6 3 un
n
X z Z xn
xn Z 1 X z
X z z
n 1
dz
2 j C
n 1
n 1
X z z dz Residues of X z z at the poles inside C
2 j C
z zo is
z zo X z z n1 z z
z zo ,
1
d m1
m
z
X ( z ) z n 1
o
m
Res. m 1 ! dz
zz
70 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Proof:
X z xn z n
n
n k 1
dz
X z z dz xn z
k 1
C n
xn z k n 1dz
n
dz 2 j kn
k n 1
Now,
By Cauchy
1 if k n
0 if k n
kn
Hence
X z z
C
dz 2 j xn kn
K 1
2 jxk
71 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
xK
X z z
K 1
dz
2 j C
1
n 1
xn
X z z dz
2 j C
x*(t)
1.
input
Discrete-time
System
xn
input
Discrete-time
System
y*(t)
output
yn
output
Discrete-Time Differentiator
x t
72 of 112
Differentiator
y t
dx t
dt
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
y t
dx t
dt
y (t )
(1)
dy (t )
slope of x(t) at t=t
dt t t
x(t)
x(n)
Discrete-time
Differentiator
y n
y(n)
dx t
dt t n
n=0,1,2,..
73 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t)
Actual slope
x n
Approximate x n x n
slope
x n
n 1
y(n )
dx(t)
dt t n
x(n )- x(n - )
( 2)
y n
or,
1
x n x n
yn ao xn a1 xn1
(3)
(3)
1
1
a
,
and
a
1
where o
74 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
ao
xn
yn
Unit
delay
xn-1
a1
G z
Y z
ao a1Z 1
X z
1 z
(4)
D.T. Integrator
Integral of C.T. fn x(t) over interval 0 to t is:
t
y t x t dt
o
(1)
75 of 112
Integrator
y(t)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t)
x(n)
Discrete-time
Integrator
y(n)
Then,
n
y n x t dt
(3)
or,
y n x t dt x t dt
n
or
y n y n x t dt
n
(4)
n t n .
area shaded rectangle shown below.
Smaller better approx..
x(t)
x n
x n
2 n x n
76 of 112
t'
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Hence,
n
x(t)
dt
x(n )
(5)
y n y n x n
or,
yn yn1 xn
or,
yn ao xn yn1
where,
(6)
(6a)
ao
Equation (6a) gives the 1st order difference equation for D.T.
integrator using rectangular integration.
Algo can be pictorially represented as:
xn
ao
yn
unit delay
yn-1
Y ( z ) ao X ( z ) z -1Y ( z )
(5)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
ao
Y(z)
X(z)
1 z 1
GRI ( z )
1 z 1
z-1
(8)
X(Z)
G z
Y(Z)
z 1
jv
j
S-plane
Z-plane
u
True integrator
Approximate discrete-time
integrator
x t dt
78 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Hence,
(9)
From equation (4) and (9),
y n y n x n x n
2
yn
or,
yn ao xn a1 xn1 yn1
xn
or,
xn 1 yn 1
(10)
where, ao 2 , a1 2
Eq. (10) 1st order difference equation representing algo for
D.T. integrator based on trapezoidal integration.
79 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
xn
ao
yn
a1xn-1
yn-1
Taking Z-transforms :
Y ( z) a0 X ( z ) a1 z 1 X ( z ) z 1Y ( z )
The Z-transfer function is
Y z ao a1 z 1 1 z 1
GTI z
X z
1 b
2 1 z 1
z 1 (11)
2 z 1
z 1
X(Z)
Y(Z)
2 z 1
jv
j
S-plane
Z-plane
80 of 112
u
-1
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
1 z 1
s 2 z 1
2 z 1 2 1 z 1
s
Observations:
S(t)
* If
S(n)
or
Sn
DTLTI
System
ds t
z n
dt t n
zn ao1sn a11sn 1
1
ao1 , a11
* If
81 of 112
d 2s t
z n
dt 2 t n
Z(n)
or
Zn
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
z n
ds t
ds t
dt t n
dt t n
x n x n x n x n 2
zn ao 2 sn a12 sn 1 a22 sn 2
ao 2 a22
, a12
2
* Similarly, if
d 3s t
z n
dt 3 t n
* Hence, if
82 of 112
d i s t
z n
dt i t n
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Continuous-time
LTI System
y(t)
d i y t M d i x t
bi
;
ai
i
i
i 0
i 0
dt
dt
N
M N
(1)
83 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t)
xn
Causal
Discrete-time
LTI System
yn
Signal
M
d i y t
d i x t
ai
i
i 0
dt t n i 0 dt i t n
N
i 0
i 0
i yn 1 i xn 1
(2)
If N 0
Equation (2) can be expressed as
N
1 M
yn
i xni i yn 1
i 1
o i 0
84 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
N
M
i z i Y z i z i X z
i 0
i 0
i z
i
i z
o
i 0
M
Y z i 0
G z
N
X z
i z i
i i
z
i 1 o
N
i 0
Properties
of
Discrete-time
(3)
Convolution
and
xn
G Z
Y Z
yn
gn
g n z 1 G Z
Y Z G Z X Z
yn xn g n xm gn m
(1)
Also,
85 of 112
yn g n xn g m xn m
m
(2)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
xn n
If
yn g n n g m n m
m
Now
nm 1 for n m
0 for n m
and
yn gn
Hence gn is the impulse response of the system.
Following are the properties of convolution:1. Commutative Property:-
xn gn gn xn
(3)
xm g nm g m xn m
i.e. m
0
m0
xn
gn
(4)
yn
gn
Y(Z)
G(Z)
xn
yn
In the Z-domain,
X(Z)
G(Z)
Y(Z)
X(Z)
2. Associative Law:
xn g1n g 2n
86 of 112
x n g1n g 2n
(5)
xn g1n y1n
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
&
y1n g 2 n xn g1n g 2 n yn
87 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
gn g1n g2n
and yn xn g n
From commutative law we also have
g n g 2n g1n
As
g1n
g2n
gLn
yn
xn
g n g1n g 2 n ,......., g Ln
yn
In the Z-domain
X(Z)
Y(Z)
G1(Z)
G2(Z)
G3(Z)
3. Distributive to Property:
88 of 112
GL(Z)
X(Z)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
(6)
gn g1n g2 n
As a Generalization the interconnection of L LTI systems
in parallel with impulse responses g1n, g2n, .., gLn
and excited by the same input xn is equivalent to one
overall system with impulse response
L
g n g jn
j 1
g1n
g2n
xn
g3n
gLn
In the Z-domain
89 of 112
yn
xn
gn = g1n + g 2n
+ ...... + g Ln
yn
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
G1(Z)
G2(Z)
X(Z)
Y(Z)
GL(Z)
causal.
# yn 2 xn 1.5 xn 3
# yn x2 n
# yn xn2
# yn x n y2 x2
90 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
2z2 4z 1
X z
z 2 z 1 z 2 Y z
G z
2 4 z 1 z 2
X z
Y z 0.5 zX z X z 0.5 z 1 X z 0.5 z 2 X z 2 z 1Y z 0.5 z 2Y z
yn 0.5 xn 1 xn 0.5 xn 1 0.5 xn 2 2 yn 1 0.5 yn 2
So system is non-causal.
So, order of numerator polynomial in z Order of denominator
polynomial in z for causality.
xn Bx
for all n
91 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
yn By
for all n
yn g m xn m
m
; n 0, 1, 2 etc.
yn g m xn m g m xn m
xn m Bx
yn Bx g m
m
gm
or, g n
n
92 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
G z Z gn gn z n
n
G z gn z
gn z n
n
z 1)
G z gn
n
g n should be finite.
If the system is BIBO stable n
circle.
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Since the ROC can not contain any pole of G(z), it can be seen that
a causal DTLTI system is BIBO stable if and only if all the poles
of G(z) are inside the unit circle.
Example: A DTLTI system has transfer function
G z
3
4
1
1
1 z 1 1 2 z
4
a stable system.
ii)
a causal system.
iii)
1
4
and z = 2
ROC must include unit circle and poles can not be within
ROC. So ROC is
1
z 2.
4
g n is non-causal
n
n
1
g n 3 un 4 2 un 1
4
94 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
ii)
gn 3 1
un 4 2 un
n
1
4
n
1
g n 3 u n 1 4 2 u n 1
4
Systems
with
Finite-Duration
and
gn 0 for n 0 and n K
95 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
K 1
m0
m0
m K
yn g m xnm g m xn m g m xn m
K 1
or,
yn g m xn m
m0
yn g m xn m
m0
96 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
i 0
i 0
i yni i xn i
(1)
If N 0,
On rearranging the terms, we have
N
1 M
yn
i xni i yni
i 0
o i 0
(2)
If N 0,
This equation directly expresses the output at time n in terms of
previous values of input and output. It can be immediately
seen that in order to calculate yn, we need to know yn-1, yn-2
,, yn-N. An equation of this form is called a recursive
equation, since it specifies a recursive procedure for
determining the output in terms of the input and previous
outputs. The system is hence, called a recursive system.
In the special case, when, in equation (1), N is zero, the
equation reduces to
98 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
i
yn xn i
i 0
o
M
(3)
i
g n n i
i 0
o
M
ni will be 0 for n i
which is obtained as
n
, for 0 n M
g n o
0 ,
otherwise
99 of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Note that the impulse response has finite duration, i.e. it is nonzero only over a finite time interval. Because of this property
the system is an FIR system.
xn
F xn , xn1,....., xnM
yn
Non-recursive System
xn
F yn1,...., ynN,
yn
xn , xn1,....., xnM
delay (s)
Recursive System
100of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
i 0
i 1
yn bi xn i ai yni
(4)
101of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
g n bn ai g n i
i 1
(5)
yn bo xn a1 yn1
(6)
g o bo
g1 a1bo
g 2 a12bo
g3 a13bo
g n a1nbo a1nboun
The response has clearly infinite duration.
102of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
gn
a1>1
a1 = 1
bo
0<a1<1
The sketches of the impulse response show that the nonrecursive coefficient (bo) merely controls the amplitude of the
response at n = 0. The recursive coefficient (a1) determines the
shape of the response. If a1 is negative, the response alternates
sign. If a1 is between 0 and 1, the response exponentially decays
to zero. If a1 is greater than 1, the response grows. If a1 is
unity, the response has a steady value.
103of 112
x t F
1
F
fs
2 f s
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
W 2 f s
*
jt
X e d
jn jt
e d
x n e
2 f s W n
1
where f = sampling period.
s
Inter-changing the order of integration and summation and
considering that
W 2 f m f s
sin W t t
sin W t n sin W t t
t n dt
W t n
W t t t n W t t
2 f s
x t
x n e j t n d
W
x n e jwt n e jwt n
n 2W
j
t
x n
n
sin W t n
W n
(1)
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
sin W t t
x(t ) x n t n
dt
W t t
n
x* t gi t t dt
x* t g i t
sin Wt
sin Wt
g
t
where i
Wt
t
sin
2
t
t
is the impulse
continuous-time
converter
or
ideal
digital-to-analog
converter.
The frequency response function of the ideal-interpolator is
s
s
for
Gi j gi t
2
2
otherwise.
0
Hence the ideal interpolator is nothing but an ideal low pass filter
with pass band of
s
2
105of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
PRACTICAL DAC
In practice a D/A converter employs a zero order hold (ZOH) for
reconstructing the analog signal from its sampled version.
The function of the DAC is to decode the input digital words
(binary numbers) and then convert the digital signal into an analog
signal. Hence a DAC can be represented by a decoder followed by
a ZOH.
106of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x*(t)
xr(t)
Decoder
ZOH
DAC
xr(t)
ZOH
xr(t)
Samples
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
xr t x n u t n u t n
n
n 0, 1, 2, 3,..............
Taking the Laplace Transform of both sides,
e n s e n 1 s
X r s x n
n
s
1 e s
n s
x
n
s n
Now,
x(n )e
n s
X * ( s) L[ x* (t )]
G ho ( s)
1-e-s
108of 112
[A]
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Now,
[B]
1 - e -s
G ho ( s)
s
109of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
Gho ( j )
Sin
[C]
Gho ( j ) Sin
or,
Gho ( j ) m -
m = 0,1,2,..
110of 112
[D]
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
G j
ideal filter Gi j
ZOH Gho j
0.637
3s
2s
s
2
s
2
2s
3s
111of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
2 to remove
ZOH
Post
filter
xo(t)
xo(t)
xr(t)
xr(t)
112of 112
Sugata Munshi
Dept. of Electrical Engineering,
Jadavpur University, Kolkata.
x(t)
113of 112
PreFilter
xb(t)
ADC
x*(t)
Signal
Processing
Algorithm
y*(t)
DAC
yr(t)
Post
Filter
yo(t)