o g and Realization of Digital Filters)

XnTe jonT 1.1.2 Linear Shift Invariant Systems,

(1)
Stability and Causality, Linear Differential
tiplication property, Equation to Difference Equation, Linear
Constant Coefficient Difference Equations
[X(jo)* p(jo)] .. (2) Q24. Write short notes on classification of systems.

Ans: Model Paper-41, Q2

Discrete time systems are classified as,

st-n7) 1. Linear and non-linear systems

2. Time variant and Time invariant systems

3. Causal and non-causal systems

+oo
4. Stable and unstable systems
2d(0-ko,)
k=
5. Static and dynamic systems

corresponding values in equation (2), 6. Memory and memoryless. systems.

Linear and Nonlinear Systems
A systern that satisfies the superposition principle is
X,jo) 0-ko, called a linear system.

(or)
Xjo)*s(0-ko,)
k=-o
A system whose response to the sum of the weighted
inputs is same as the sum of the weighted responses then
the system is said to be a linear system.
XG(o-ko,)) (3
A relaxed system is said to be linear if and only if,

crete-time Fourier transform of xn] Hla )+azm)]-a, H()]+a,H[m))

Where, x (n), x,n) are the arbitrary input sequences and
aa, are the arbitrary constants.
n]e The above condition is illustrated in figure below.

x,(n)

(4) yn) Hfa, z,()+a, x,(n)

-o

uations (1) and (4), we get,

a2

) x(n)

x,(n)
xGa-27k)/T)
k=
yn)- a, H[x,(n)] +a,H[x,(n)]

x(n).
are related as,
d H,(e) Figure: Diagrammatic Explanation of Linearity
H(eT lol<o, 12 In the above representation, if y(n) ="y(»), hen the
L>o, 12 cetem.is known as linear otheriee it ia.nomlineor
2. Time Variant and Time DERABADJ
Invariant Systems Then, by using convolution sum
written as,
formula, y(n) can be .

If the
input-output relationships
of the discrete time
system do not change with time, then the
a time invariant
system is called yn)=2hk)x(n -k
system.
Ifx(n) is the input and y(n) is the output of the system. b-2()xtn -
Then, x(n) -n)
The. above given system is said to be time-invariant iT
2(k)lx(n -k)|
-1 (1)
and only if, If the input is bounded, then
| x(n k) | -
=
M.
x(n -k) Therefore, equation (1) becomes,
n - k)
Here the input and output of the system are shifted byk Ix)=M k=-o
2h)
samples. The response due to shifted input is given by
, k) =
T[a(n k)).
-
If hk)| o, then the output is bounded.
<

k=-
For time-invariant system, yln, k) = y{n - k) for all Therefore, a LTI
system is said to be stable, if its impulse
possible values of k. response is absolutely summable.
f yn, k) * yn for one value of k
-

k) even then the 5. Static and Dynamic Systems

system is said to be time-variant.
Static systems are those systems whose
output depends
Causal and Non-causal Systems: If the output of the only on the present input. They are also known as
system at any time depends only on present and past
memory less system. In these systems,
input values but does not depend on future input values impulse
response h(7) is zero for t <0.
then the system is known as causal system.

The mathematical expréssion for causal

Dynamic systems are system whose output depends on
system is given the present and past inputs.
by,
yn) = Fx(n), x{n - 1), x(n - 2), x(n - 3), ..
Memory and Memory Less Systems
If the output of a system
yn-1), ytn-2), n-3)... at any time depends only
Where,
on
the present values of input, then the system is
known as "memory less system", otherwise it is called
y{n) - Present output memory system".
x(n) - Present input For example,

x(n- 1), x(n-2), x{n-3),... are past inputs A discrete-time system represented by the
relationship
y(n)=[2x(n) x°(n)]P is a memoryless system because
n- 1),y{n-2), y(n-3),.. are past outputs. -

the value
If the output of the system at any of y(n) at any instant of time n, depends only
time depends on on the value of
x(n) at that time.
present, past and future input values then the
known as non-causal system is A discrete-time sy_tem with the
system. description.
Stable and Unstable
Systems: A system is called yn)=x(k) has memory because the output at any
bounded-input bounded-output stable if for all bounded S-o

inputs, the corresponding outputs are bounded. A

is called unstable for system instant depends on all past and
if some bounded inputs, the present inputs.
corresponding outputs are infinite. Similarly, A continuous-time
y)=x{) is a memoryless system with desc on
Consider, z(n) as bounded inputa
system as the output depen ds
only on the present
y{n) as a bounded output input.
A
hn) as the impulse
response of the system.
continuous-time
v) x{t -5) system represented by the lation
=

has rcianstant
Look for the SIA GROUP inGO depends on the inputmemory. The output at any
5 "instants" earlier
 Con onsider a discrete timeognal Processing and Realization of Digital Filters)
system characterized by the following input-output relationsnip
29.

x(n-2)-2x(n-17). Determine whetherthe system

yin) =
causal
time-invariant, linear,
and stable. *Determine is memory less,

|April-18, (R13), a2(b))

Ans (Model Paper-4, Q2(b)
Given that,
The system difference equation is,
. (1)
yn) =x(n-2)-2r(n-17)
Memory or Memory Less System
Ifthe output ofan LTI system at any instant depends only on present values ofinput, then the syst

Jess stem else

systerm
it is called
memory system. is not a
the system meO
ro
Since the
output (n) depends on (n-2yh and (n-7yh value of input x(n),
l e s ss y s t e n m .

Thus, the system is not memory less system.

Time-Invariant
The output response of the system is given as,

n)= x(n-2)-2x(n-17)
The output response of the system for delayed input response is obtained as, .. .(2
n-k)=rn-k-2)-2x(n-k-17)
The delayed output response is obtained as, (3
yn-k)= x(n -k-2)-2x(n-k- 17)
From equations (1) and (2), we have,

n-k)= y(n-k)
. The system is time-invariant

3. Linearity
system for an input x,(n)
is given by,
The response of a
n ) =x,(n-2)-2x,(n-17)
for an input x,(n) is given by,
Similarly, the response of system
y,n)=x,(n-2)-2x,(n-17)

and y,(r) is given by,

The linear combination of system responsey,(n)b(r,(n--2)-2x,(n-17)
atx,(n-2)-2x,(n-17)) +
ay, +by,(n)= + b x , ( n - 2 ) - 2 [ a x , ( n - 1 7 ) +bx,(n
-17)]
=ax,(n-2) by,
i.e., ax, (n) + bx,(n) given
is
combination of input signal
of system for linear
The response ax,(n-2)+bx(n-2)-2[x,(n-17) +bx,m-17)1
y,m)=
(4).and (5), we have,
Comparing equation
)+by,(n) =y,n)
is linea
system
Thus, the given
Causality
response 1s,
The given system x{n-2)-2x(n-
17)
n) =
For, 0 ) = x ( - 2 ) - 2 x ( - 1 7 )

n=0
Ml) =(-1)-2x(-16)
n=1
n 2 2)=x(0)-2x(-15)
values of inputs.
S.
or system
depends only on past
analysis, the output
From the above SIA GRO
is causal.
Thus, the system nn FNGINEERING STUDENTS
 Stability
The necessary and sutticient condition for the system to be stable is,

i.e., 2xtn-2)- 21r( 17)<o

n-

Assuming x(n) = S(n), which implies.

l ) 8-2)-28(n-17
n-o
=

S(m)= 1, n = 0
=|1 -2
=0, elsewhere
= 1 <o

Hence, the system is stable.

Q80. Test
the following systems for linearity, time invariance, causality and stability.
y(n)= sin(2nfr/F)x(n).
Ans:
(Model Paper-l, Q3(a)| May-16, (R13), Q2[a))
The given system is,
yn)-sin( x)
Linearity
For x(n) as input,

For xn) as input,

yn)=sin (2q0
The weight output sum is,

ay,(r)+byn) -a»(r)sin +bz2)sin()

sin la z()+bx))
The output due to weighted sum of inputs is,

,n)=[ax,) +b x,tn)] sin(n

y,n)=ay,(n) + by,(n)
The weighted sum of outputs is
equal to output due to weighted sum of inputs.
Hence, the given system satisfies the princ1ple of
Superposition. Therefore the system is linear.
ime-invariance
The given system is,

tn)= sin

Consider,
n ) = sin{2ryfh/F) x{n) = 7[a(n)]

o k for the SlA GROUP LOGO

2On the TITLE COVER
before you buy
 delayed input is, Ans:
yn, k)= Tx{n-k)] =yi) ln - (1) y(n) nelz()|
sin(27fnlF) x{n - k) The given system 1s,
Mn) = neltn}
The output delayed by 'k' units is,
Causality
yn-k) =y(n)], 1 n-k
Given that,

sin( 2/R x(m-) Mn) n e "to

n =0 O) 0
=

For
M,k)yn, k) *n-k) n=1> y(1) =elt
the n=2 2) = 2el2
Hence, system is
time-variant.
Causality

analysis, the output of the syster

From the above
The given system is,
depends on present inputs only.

r)-sin x Hence, the system is

Linearity
causal.

Given that
For n--2;=2)= sin( -2) n) =n n)y
For n
0; (0)= sin (0) x(0) Let, y,n)=H {x()]= net
y,n)= H [x,(n)]= nem)
For n-2:2)= sin 2) For inputs x,(n) and x,(n) the linear
combination is,
For all values of 'n', the output depends only on present a,x,n) t a,x,(n)
vaues of input and thus the system is causal. combination iz
The response of the system for linear
y,m)= H [a, x,(n) + a,x,(n)]
Stability
The given system is, =n elaan) tat20)|
The linear combination of outputs is given by,
n)= sin x(n)
For the system to be stable, it has to satisfy the following ,y) +a,ym)-a,|nel|+a,net
condition, Equations (1) and (2) are not equal.
Hence, the aystem is non-linear.
Time Invariance
n-oo

ön) Given that,

lfa(n) =
8n) then, h(n)= sin(2rjfnl/F) n ) = nelx(m)

h0) =
sin(0) S(0) =(0) (1)=0 Response for unshifted input,
Forn =0;
==0
Forn 1 ; H1)= sin(2nf1F) 8(1) n)-H[xn)
=0 n e In
sin(4nfF) 8(2)
Forn=2; h{2)= Delayed input response,
2h(m)=0<o y-k)=Hn -k)}]
F(n-k)e - |

is stable. Delayed response,

The given system
for linearity,
C1.hChec
e c k the
following
systems yn-k)=z*H[x{n)]|
invariance
and stability using zr ne lxa)|

causality, time n e n-)1

appropriate test,
Thus,
() y(n)= nelko yn-k) #y(n-k)
The system is time variant.
cos (2n/N).
(i) yin)=a" IGINGFRING STUDENTS -SIAGROUE
 DIGITAL SIGNAL

1.18
h(oo) =
0
Stability n= o

Given that,
Mn) = nela()1
n-2.cos(2xn/)8)= 1
nel6 (0)I stable.
hn) The system is
=

For n=0 h 0 ) = 0
Causality
n=1 h 1 ) = 1
Given that,

yn)= d" cos(27tm/N)

n =o h(o) = oo

And m- 1 h-1) = - 1 n ) = d cos(2rn/) x{n)

For, n=0 X0)=x0)

n-0h{-o) =- 0 n-1)-acos| )
2h- ineo-. n-2 M2)=d'cos 2)
Hence, the system is unstable.
yn)=a cos (2rn/N) From the above ånalysis, the output of the system
The given system is, n) =
d cos(27tn/N) depends on present inputs only.
me Invariance
Hence, the system is causal.
Given that,

n) =d'cos(2nn/)x{n) Linearityy
Response for unshifted input, Given that,

yn)=H[] n) = cocosT
=d cos(2n/)x{r)
Delayed input response, Consider, y,m)- H{z,m)]= d cos r) -()
n-k d-*cos(2x(n -kY)r{n -)
Delayed response, (4)
n-k) =z*H n)] Considering the response of the system for
combination of inputs
a
linear
z*l cos (2nn/) z{n)] Ax,n)+Ax(n) is y,(n).
n-k) =d cos (2tw/) {n-k) y)=HAFM)+47,
n-k) and yn -k) are not equal.
Hence, the system is time
variant.
ity A,cos +d4,cos )
Given that,
The (5)
n) =#
cos(27tn/N) x{n) following condition should be
1.e., satisfied for linearity
hm) =# cos (2nn/M) &n)
For -oh(-o) =0 y,m)-4,y,n)+Ay,n)
From . (6)
equations (3), (4) and (6), we
n=-1 h-1)=0 get,
n=0 h(0) =1
v)Ar cos )+ Aat cos
From equationis (5) au 1 n)0)
 Oduction to
(Introd Digital Signal Processing and Realization of Digital Filters
and
T-1 combination of outputs,
otermine whether each of the following
D e t e
The linear
s defined below
s y s t e m s

is, a y , ) t ay,{n) = a |
Causal

...(2)
(il) Linear
a2 +a e z
k n .3
(H) Dynamic
Equations (1) and (2)
are equal
(iv Time invariant
is linear.
) Stable Hence, the system

(i) Dynamic
(a) y(n)= 2 the past
value of input.
k=n-3
The output y(n)
depends upon
nature.
is dynamic in
(b) yn)=x(-n-2). Therefore, the system
May-17, (R13), Q2(a) Time Invariaint
(dv) because dynamic
The given system
is time variant
The given system is, cannot be time invariant.
systems

(V) Stable
=n-3 to
sufficient condition for the system
The necessary and
Causal
be stable is,
Forn0,0) =2,"
k-3

=e3) +e-2)+ e-) +er0)

Let, x(n) =&tn)
n) hn)
For,n-1, 1 ) 2 "
-

hn)
=e-) +e-)+ e0)+ e ) k=n-

of the system
From the above analysis,
the output

inputs. Hence,
the given system
For n 0, h(n)
=

2
k=-3.

epends on past and present + e0

=e)4+ e ) + er-)
s non-causal.

i)Linear Here,
1, k 0
x)o, k<0
Let,y,)= 2 = H , ( ) ]
h(n) = 0
k=n-3

Similarly,
=H,(n))
k=n-3
is, Hr)2
and x,(n) the
linear combination k=n-3
For inputs x,(n)

an) +a>,0) Forn-1,M-1)=

k4
combination is,
system linear
The response of the = e4) + e)+e-2) +.e=

-Hla,) ar,(n)
+
= 0
Y)
it can be concluded
lae+a,e Therefore, from the above analysis,
k=n-3 that the impulse response
of the system is less than oo, Hence,
.(1) the given system is stable.

k=n-3
k=n:
ENGINEERING
STUDENTS -siA GROoUP
JOURNAL
FOR
ECTRO ALLHN-ONE
 T i m e l n v a r i a n t : The g i
1.20 (iv)
(1) yn)= x(-n -2) n) = x(- n - 2 )
and the statie
The given system is, As the given system is
static
syst
invariant.
be time
Mn) =(-n-2) cannot
time variant.
given system is
Causal: Given that, Hence, the
system is,
Mn) =x(-n-2) Stable: The given
(v)
Forn=0. {0) =(-2) Mn) = x(-n-2)
condition for the system to
Forn1:)=a(-3) necessary
and sufficient
The
Forn 2, 2) =a(-4) be stable is,
F o r n - 1, M-1) =x(-1)
2h)k
Forn-2,-2)=(0)
From the above analysis, it is clear that the output ofthe
Let, x(n)
=
ô(n)
depends o n present, past and future inputs.
system
Then, y(n)=ö(-n -2)
Hence, the system is non-causal.
Linear: Given that, Thus, 0
8(n) =1 for n
=

Mn)=x(-n-2) I = -n-2)| = 0 elsewhere

Let, y,(n)=x,-n-2) Hl,()] N-
n=-o

= 1 <o
n)=x-n-2) =Hr,)]
linear combination is, is stable.
For inputs x,(n) and x,(n), the Hence, the system
listed below,
response
a,)+as) impulse
Q33. For each whether the
corresponding
system
determine
Then,
for linear
combination is,
The response of the system is,
() Causal
)-Hlas,)+a,())
. (1)
a-n-2) +a, x,n-2) (i) Stable
individual outputs is, 2" u(-n)
The linear combination of (a) h(n) =
ayn)+ayn)=a,r,-n-2)] +a,r,(-n-2) 1).
h(n) en u(n
-

2) (2) (b) (R131,02(b))

 From the Time Shing
definition of the step input u(k) =
1, fork20 . Z-transform is
defined as,
property of
h(k) = (10¥ = o
r(n- m) 2 , z " X(Z)
k-o
k * Proof
The Z-transform
of the discrete sequence x(n) is give
Hence, the system is unstable.
1.2 REALIZATION OF as,
DIGITAL FILTERS Ma) -
2m):
"

121 Applications of Z-transforms, n-

ofDifference Equatlons of Digital Solution Then,

Q42. Define Filters
Z-transform and prove all its
Ans: properties. z{x(n -m)} =xtnm).z"
Z-transform: The z-transform ofa
sequence r{n) is defined as, - xn-m)"

r))- 22x():" - xtn-m)

X(z) 2xn-m).z
-m)

Where, represents
z
the
expressed in polar form as z complex
be and can alsovariable
re n, where r =

radius. denotes circle xp) Cp=n-m

Properties of Z-transform
Linearity: The linearity property ofz-transform of
discrete sequences x,(n) and x,(n) is stated as,
two
z X()
Z{ax,n)+ bx,n)} aX,e)+ bX,) =

Proof xnm}=z"X()
Z-transforms of two discrete sequences are defined as (ii) Scaling in the Z-domain (or) Time Scaling
follows, Time scaling property of z-transform states that,
Ifxn X) with ROC =Rthen,

Then, an with ROC=[a|. R

L.H.S Z{ax,(n) + bxn)}
Proof

2(an (r)+brn)a" The x-transform of the

discrete sequence, x(n) is given
n=-

as,
ax(m)z"+br,(n)a*
n=-0

[ar)]=2n)*

- aX,() + b X )

=R.H.S
Where 'a is a constant.
LHS=R.H.S
Note
ztann)+brz()}=al)+bX,()
Poles and zeros of X(z) have their radii changed hy '
MT-1 (Introduction to Digital Signal
Time Reversal:
The time
Processing and
ar Realization of Digital Filters) 1.29

7-transform is defined as,v e rreversal propert

s a l property of Q43. State and prove the following,

x-n)Z X(1/2) ) Convolution property

(1) Differentiation property
proof

The Z-transform of discrete sequence x(n) is (111) Initlal and final value theorem.
given as,
Ans:
X)= xm)."
() Convolution property
Then,
The convolution property of Z-transform
of two signals

x(n) and x,(n) is defined as,

z{x-)}= x-n)"
x,(n)x,m) X,2)X,(2)
2xp) (:p=-n) Proof
The Z-transform of two discrete sequences
are given as,

X,a)=zn)z"
X{z")=X(1/2)
z-r)} = X(1/2)| ns-

The expression for convolution of two signals is given as,

Time Expansion
Time expansion of Z-transform states that, x,m) x_n)- 2#)za(n-k)
If X), with ROC R Then,
Then, x,(n) Z X{E, with ROC= R\k

Where,
if n is an integer multiple of k
z)=}°\kJ
0 Otherwise
Proof
The z-transform of discrete sequence, x({n) is given by, .
E-00

.(1) -22)xzn-
Zxtr)) Xe)-2*
k)rz(n-k).z-in=*)*
On substituting x(n) =x(n) in equation (1),
we
get,

- * *
n=o

(2)

Assume,
z(m)*z2n)}= Xi(2)X2(7)|

i) Differentiation Property
Equation (2) becomes
oo Differentiation property of Z-transform is defined as,
n.xtr) zX
Proof
xoety
The Z-transform of discrete sequence x(n) is given as,
X()
X()=n)="
CIA ROIB A
 DIGITAL SIGNAL PROCES.

1.30 Lt xn)-x(n-1))

Then,
Lt x(rh) =x(co)

x(o)= L(12)X()|

- 2
Or

r)-n)
about
Z-transform and Iistout
detail
Q44. Explain in
- 7 . axin)z"* its applications.

e)--rinztn)} Ans tool for analysis and design

dz Z-transform is a powerful
that cannot be analyzed
signals and systems
X ) =-z(nztr)}
dz
of discrete time
using Laplace transform.

transform of signalf ) is defined as,

The Laplace
(X())= zinz(r)}
Fijo)-Jfedt ..(1)
(ii) Initial and Final Value Theorem
Ifx{n] =0; n <0, be used to get the discrete-time
Laplace transform can
relevant substitutions. That is, by
thenx(0) Lt X() representation by making
' f of the Laplace
changing the continuous-time variable
Proof transform with discrete time variable
eiol
'n' and replacing with
and integral () with summation () we get Z-transform.
If r(n) is a discrete time signal, then its transform is
obtained as,
=xn)" [' xin) is causal] .(2)
n=0
n-0o

x(a) =x(0) +x(1)zl+r(2)z2+.. Equation (2) is the Z-transform of the-discrete time signal
Applying limits on both sides of above equation, we get, x{n).
On substituting z = re" in equation (2), x(z) is

Lt. obtained as,

x (0) Xred)- atnkre)"

. (O)=Lt X(2)|
Final Value Theorem
The Z-transform exists, if the above
Ifx(n) is causal, then Z {nol-(1-z) X{). summation
converges,
o
Since, ar) is causal applying z-transformto zn]-x{n--1]
ie, t | <
Zx(n)-x{n-1)]=X{z) -r{E)
Therefore, for a
signal X{n) its Z-transform X(z) is
-atm)-x(n-Dg defined as,

(1-r)X)= u 2tr)-x(n-) n-o

And x(n) is said to be

inverse Z-transform of
Taking the limit as z >1 on both sides, above equation X{E) if,
implies to,

u-)X) = 2km-xn-DlE" Applications of z-transform

For answer
refer Unit-I, Q17.
ok for the SIA GROUP LOGO on the
TITLE COVER before your buy
 uence f(n) defined below.
3 n0

fin) l3n=0,2,4.

-13. 1,3,5....
Given sequence is, (Model Paper-l, 23{b) ] April-18, (R13),
3 n <0

Sn)={3n0,2,4..

- 1n 1,3,5.
Z-transform offn) =?
Using the expression of z-transform,
r)- )z*
n-a

Substituting the given sequence in equation (1), z{n) can be written as,
zn)=
n-o n=0

Substituting n = 2p for even values (0, 2,4.. ) and n = 2q+1 for odd
values in equation (2), we get,
)-3+E n=1
z

n)-(3-1"+
m 1

1- -
 1-13 1 2z
1-3 1-9 1-

1 1
-+ 1 2z
1- 1
3 922 422

922 422
3-Z 92-1 422-1 2
3
92 2z
3-2

+
3-z
9
Z transform of the sequence,An) is,
Z Z
+
9
1.2.3 Realization of Digital Filters Direct,
anonic Cascade and Parallel Forms
Q58. Explain the different structures for realization
of lIR system and explain how conversion can
be made from direct form-l structure to directt
form-l structure.
Ans: Model Paper-ll, Q3

The realization structures of IIR system are,

1. Direct form-I realization
2 Direct form-II or canonic realization
3 Cascade form realization
4. Parallel form realization.
1. Direct Form-I Realization: The output ofa finite order
LTI system is given by,

r) -4yn
k =1
-
k) +2
k = 0
Bex(n-) . (1)

In the above equation, the output n ) is expressed as a

linear combination of outputs and inputs.

) - EAya-»B.xa-)
k=1 k =0

On taking Z-transform, we get,

Yz)= Y) 2Az+X)B*
k=1 k=0

Yz) = X(z)BA*
k-1 k=0

Brz
k=0
H)= N
1-4
 DIGITAL SIGNAL
1.40 W(z) is
the oulput
of 7I (a).
From figure (2),
In this case, H(z) is expressed as, W()
X(z)
HC)-H,). H,)
. H(a) : (2) X(2) . (3)
W(z) N.

Figure (1) shows the direct fornm-I structure of

Y(z)
{Z) 2k
equation (2).
I1,(z) ()
X() Y()
.4)
ne)-we) 2h" k 9

the direct form-1f

(3) and (4),
By using equations
B,

is drawn and shown

in figure (3).
structure

WZ) Y
X(7)
B

A
Figure (1)
(+
Direct From-II or Canonic Realization: In this
from-II realization,
direct
A Bu
k=0_
H)=
k=1|
Figure (3)
The number of storage elements required by the direct
HE)=H,). H,?) form-1I structure is less compared to the direct form-1.
Hence,
Where, direct form-II is known as Canonic structure.
3. Cascade Form Realization: The transfer function of
the IIR system is of the form,
H)
1- Ha)=Ha()k

H,) -2B,* Ha) H_z)..)

k=0
Consider the transfer function of a
Figure (2) shows the decomposition for direct fom-II. system,
Iebt
Hz)= b,zb bbz'+b)
(1+a,1z +a,2z )(1+4,,1z tam2)
W(z) Y(z) Hz)- H,(2) H,(2)
Where,
H)-atb+baz3
H(Z) (1+a,2'+422)
Figure 12
H,)= *bz+b
i aok for
(1+a,z +4m2)
the Sla GROUP LOGO
2 on the TiTLE COVER before
you buv
 Let, H,)= (E) M) K(E)_ bn0+ b1z+bn2z
X(E) i(E) I+a,1z+ a,22
Where,
í(z)
Xi(z)
1+an12+a,2z 2

w,e)+ We)+a,W,E)- x,2)

(1)

W,e)X,)-a,'We)-a,*"
ni

... (2)
Y,) ,e)+ b' W,e)+ b W,E)
Let, H) - W() Wa) %E) _ bmo+ bniZ+ bm22
Xa(2) W2E) Itam1z+am2Z4
Where,
W2()
X2() 1+amz+am22 -2
(3)
W_) =X_®)-a W)-aW_)
2(2)
Ph otb +b,z2 m2

.(4)
Y,) =bmW,e) +»m W)+ W,)
and equations (3) and (4) are realized by
Equations (1) and (2) are realized by using direct form-lI structure, we get H,(2)
sing direct form-lII structure, we get H,).
is obtained as shown in figure (4).
H ) and H,2) are cascaded, then the cascade form of H() H,()
H
Y (2
W,(2) Y, X() W,{2)

-an2

Figure (4: Cascade Form of H(z)

of an IIR system is given by.
Realization: The transfer function
Parallel Form

HC)-421RE n=l
AM
. He)=A+AT1-AT 1-
H ) = A + H , ) + H , ) + . . + H J )

B,7)AE)+,.t H)
X2)
Yz)=A XE)
+H,2) X() +
ngPNAL
FOR ENGINEERING
STUDENTS SIA GROUP
 PROCESSING
IJNT
SIGNAL
DIGITAL
1.42
Figure (5)) shows the parallel form realization of equation (1),.
A
X(z)

H,

H,)

Y()

Figure (5: Parallel Form Realization Parallel a n d

cascade

and draw various lIR realization structures like

directform-1, Direct form-ll,
DIscuss
forms for the difference equation given by,
ytn)-yn-1)*3ytn-2)*8yn-3)*x{n)+ 3x(n-1)+2x(-2).
Anss
Given that,
The difference equation of IIR system is,
. (1)
r)=y(n-1)+2yn-2)+ytn -3) +x(n) +3r(n -1)+ 2x(n
2)

On applying Z-transform to the above expression,

we get,

e)- Yt)+ Y)+4Yt)+Xt)+3:x)+22*x

ro1-a*|-Me +3z1+2)
.(2)
Y =H)=* 1+3 +2z
- (2) is shown
given IIR filter having a system function given by equation
as
of the
Direct Form-I: The direct form-I realization
in figure (1).
Y(2)
X()

-3/8

3/32

1/64
Figure (1)
 WTitten as.
Y(2)
H)X2) 3+22
Thisisis in the form of
of
X() UE) UY()
Where
U)
X()
Y(E) (3)
and U *1+3z'+2z
..(4)
Then. equation (3) can be
written as,
) -U)
and equation (4) can be
+:'U)- y3UG)x U) .()
written as,
YE) =
U() +3zU(E) +2zU(E)
Equations (5) and (6) can be realized .. (6)
using direct form-ll structure as shown in figure (4)
X(Z) U(z)
+ Y()

3/8
Z

3/32

1/64
Figure (2: Direet Form-H
arallel Form: Equation (1) can be written as,

1+32+22
H(z)=i g -- - z+3z+2)

+32+1 +3z+2
+ 322 64
 DIGITAL SIGNAL PROLI
1ABAD
obtain,
.44 the above
expression,
we

partial fractions to
On applying
3408 804
H) 354

Me) 1 1- 1+
is in the form of,
he above expression
H ) H,)+H)+H,)
form as shown in figure (3).
be implemented
in parallel
It can -3408/35

-1/2

804/35 Y(z)
X()

1/4

377/5

-1/8
Figure (3: Parallel Form

be expressed as,
ade Form: Equation (1)
can

1+3+222 1t3+2
1+ +2-4(+8
H{E) =

2x4X8(1+3+2 64(1 +3z +22)

2
H)
323
t h e sIA GROUP LOGO o n the TITLE CoVER before you buy
 S9nal ProcEssing din
The above cquation can be implemented in cascade form as shown in i g u igure (4).
X(2) > 64 Y(z)

-1/2 (+)--1/8
z

-1/32 2

Figure (4): Cascade Form

hardware requiremens.
Compare a i r e c t form-l and direct form-ll structures with respect to MaylJune-13, (RO9), Q4(a)
060.
Ans:
of direct form-I structure and direct form-II structure is
shown below:
Comparison
Direct Form-II Structure
Direct Form-I Structure
which is also
Direct form-I structure is a structure, which Direct form-II structure is a structure,

used in IIR filters.

is used in IIR filters. and
for both input
2.
This structure uses a separate delay for both 2. This structure uses a single delay
input and output signal sample. output signal sample.
canonical as it requires
3. This is also referred
structure
as
This structure is also known as non-canonical,
3. less number of delays.
as it requires more number of delays.
is efficient
Memory requirements in this
structure
is not 4.
Memory requirements in this
structure
4. when compared to direct form-I
structure.
efficient.
order IIR filter 5. Todesign a (M-1) or (N 1h order IIR filters it
-

5. To design a (M - 1)th or (N- 1)th requires (M+N-1) multipliers, (M +N-2)

adders
2)
itrequires (M +N-1) multipliers, (M+N and max {(M- 1) (N- 1)} delays.
address and (M + N-2) delays. cascade but with first
linear time- 6. This structure is also viewed as
b. This structure is viewed as two recursive and then with non-recursive as
shown in

invariant systems in cascade as shown in figure.

figure
wn)
yn) y(n)
x(n) bo x(n) w(n) bo

b b
Recursivve
Non-recursive
Figure
igure
strucures or
tne following transfer function
realization
O1. Obtain all the possible i í - 2 b c o s 6 z + b'z*).
bcosez-1/ (1+27")
AH2)=1-
Ans
1s,
Given transfer function b cosez
z'+b?z)
)(1-2b cos
B ) T +az
6 z " + b * z ) - bcos
ez-
) 1 - 2 bcos
(1 +az a z ) ( 1 - 2 bcos6z + b z )
(1+ z-2ab cos 8z+ ab?z- b cos ez
1 - 2 bcos +b2z2+
az-)(1-2bcos 0z-1+
bz2
Ha)=
1+
I G STIIDRNTS
 Define DFT and IDFT. State and prove the propes

state and prove any three properties of DFT.

Refer any three properties)

AnS
DETandIDFT

For answer refer Unit-1I, Q1.

Propertiesof DFT

Linearity Property
DFT
If x(n)- XK)and x,(n) X,(H), then
x,n) + bx,(n)} DFT
ax,(k) + bX(

Proof
The N-point DFT ofa sequence x(n) is given as,

n=0

Where,

Consider, xn)= ax,") + bx,n)

Substituting equation (2) in equation (1), we get,

XU)=2,ay(n)+bx,n)]
n=0

2a,n) +br,n)*
n=0 n=0

a2m"+bz)
n=0 n0

ax,(k)+ bXK)
DFT bX,(k
ax )+bX,(m)}
Circular Time Shifting Property
If)DTX), then,
x(n-m) - enbm/N X(k)

Proof
From the definitions of DFT and IDFT,
DFT x(7)]=X() and
x(H).eN, 0SnSn-1
DFT LXK] =x{n)=
 DIGITAL SIGNA TK{N-)
2.6
(N-n)}
=

2 x()e
Replacing n by n - m in equation (3), we get, DFT ( /=0

x{n- m) x)eate- =0

X(k).e ane - r()e 2tk

e2nkmN Xk) From equation (3)]

( - m )DFT
- 2Kkm/N X( x(M[cos (2Tk)-sin(2nk)]e
Hence proved.
3. Circular Frequency Shifting Property
(DI1-0]e
Ifn) DFT , then
- N-1 D1-01e"e 2R
dnlxm/ND X({k m)DN
-

Proof = X{N-k)=(X-R)),
The N-point DFT of a sequence x(n) is given as, Hence proved.
5. Periodicity
DFT ((n)] =
x(n)e2xb/ The N-point DFT of a
discrete sequence x(n) is said to
n0 the below condition,
be periodic ifand only if it satisties
Similarly, the DFT of x(n)e N is given as, Xk+ M) =X(K) for all values of '*.
Proof
DFT (Xn)e =
x(n).e
n=0
onN Rt The N-point DFT sequence of x(n) is given as,
-1
2x(n)e2m(k - m)n/N
N-I
Xk)= E x(n)er-aankN, k=0, 1, .. N-1
n =0 n=0
Then,
2,r(n).e2tn{N+k -m/N N-1 j27(k+M)
N
n0
X(k+ N) x(n).e
n=0

X{N+k-m)
N-1 2Tnk-2 jtN)
- X [(k-m)yl N
x(n).e
n=0
x(m)e 2Tmn
N ( X ( k - m)) y
N-1
x(n).ej2nnk/ N2zNk/N
n=0
Hence proved.

Time Reversal Property N-1 j2 7onk

4.
x(n).e Nj2m
n=0
IFDFT (X(n)]- X{K), then,
N-1 j2nk
DFT{(-) = ((-k),, E x(n)e N (cos 2run-jsin 20)
n=0

Proof
N-1 j2nk
The DFT of (r(-n) is given as,
.
x(n).e N ()-j0)
n=0

N-1
jank -j2tmk
DFT (-),}= 2«-n)ye N x(n)e
n=0 N.1
n0

N-1 -j2tnk
x{n).e N = X(k)
n=0
X(k+ ) =X(k) |
Assuming N-n=l>n=N-1
Hence proved.
tor the SIA GROUP LOGO on the TITi t
 Merketedb

UNIT DISCRETE FOURIER

TRANSFORMS AND FAST
2 FOURIER TRANSFORMS SIA GROUP

PART-A
SHORT QUESTIONS WITH SOLUTIONS
af. Define DFT and IDFT.

Ans
Model Paper-, Q1e
DFT (Discrete Fourier Transform)

Ifxtn) is the finite duration sequence then, the

N-point Discrete Fourier Transform (DFT) of s{n) is defined as,
X)= x(m).e""k=0,1,2,.N-1
n=0

DFT (Inverse Discrete Fourier Transform)

The time domain sequence can be obtained by taking the inverse DFT of the
sequence. The IDFT is expresed as,
-
k=0
n0,1, 2, N-1
02. List the properties of DFT.

Ans:
Model Paper4 a

The properties of DFT are,

Periodicity
2. Linearity
3
Symmetry
4. Linear convolution
Circular convolution
6. Time reversal
7. Circular time shift
Circular frequency shift

Circular correlation
10. Parseval's theorem
Convolution of even and odd sequences

12. Multiplication of two sequences.

PFITRIMAldN-ONE J0URNAL FOR ENGINEERING STUDENTS SIA GROUR
 (n)
h(n) 0 0 0 0 0
-

0 0/0 0
1
4 0 0 0 0
0
0 0 0 0 0
0
yn)
0 4-4 0
0 0 0

000 0 0 0 0
y)4,-4,0,0, 0,0,0,0,0,0
The consuruetion of
actual output of the svstem is done
Thus, the final output is by removing (M- 1) zeroes from
equal to the linear y,n) and
y)
Table below shows a convolution of the system.
pictorial representation of
)-1 2-3 4 -5 6
overlap-add method.
-7 4
P2(n)
y(n)-1 2 +4-4 0 0 0
-3 4 -5 6
-7 840 0 0

n);1,2-3,4,-5,6,-7;8,-4)
(32. Perform linear
convolution the two sequences x(n)
using overlap save of
method and verify the =
{1, 2,3,-1,
-2,-3,4,5, 6} and
result using overlap add method. h{n)= {2,1,-
By taking an (or)
example compute DFT by
Refer onl Overlap Save using overlap save method.
Method May-16, (R13),
Ans
Given that,

Input sequence, x{n) =

Impulse response, h(n) {L,2,3,-1,-2,-3,4,5,6)
{2, 1,-1} =

Finding the linear convolution of

given sequences using overlap save
Overlap Save Method: The method as follows,
lengths of input sequence (L) and
Thus, the length of impulse response (M) are L =9 and
output sequence n) must L +M-1 M=3
respectively.
Then, the input sequence is divided into 9 +3-1= 11
hock along with L new data blocks of data and each block consists of
points. For first block, at initial last (M- 1) data
i.e. positions (M-1) number of zeros points of prev
are
placed.
M 1=3 1=2 Zero

-00 12:3L 3 new

data point
m)2,3 -1-2-3
L=3 new data point
Last M-1=3-1 2
data form
previous block
#TRUM ALL-IN-ONE 1OURNAL FOR
ENGINEERING STIIRENTT
 P R O C E S S I N G
J
Method: In ABADI
this nethod
2.30
SIGNAL

to each bloc
O v e r l a p - a d d

DIGITAL
Using appended ck.
Similarly, 2zerosare
Verification

M-1 -3
-
1 =
x,(n)= {-2, -3, 4, 5,6) s,()(1,2,3,0,0
xn)= {5, 6, 0, 0, 0) x,(n){-1,-2,-3,0, 0
To perform convolution,
5, 6, 0, 0}
Increase the length of h{n) by adding L- 1 = 3 -1 =2 s,n)= {4,
zeros

hn) = {2, 1,-1,0,0} Similarly,

Perform the circular convolution on each block with ) {2, 1, -1,0, 0

impulse response, each block with
circular
convolution on
impulse
Perform

n)=n) h(n) response.

2 0 0 -1, 1 ym)=z (7)Wh)

1 2. 0 -1
)=-1 2 0
0 -1 1 2 0
1 2 0
0 0-1 1 2
-112 0 0
0 -1 1 2 0
0 0 -1 1 2

200-1 1} )x(n) Wh(n)

2 0
V,)= -1 1. 2 0 0
0 -1 1
o0 0-1 -12 12 0-2 2 0
1 2 0 0-1
-1 1 2 0 0
y,m)*3(n) Whm) 0 -1 1 2 0
0 0 -1 1 0 0 -1 1
1 2 0 0 -1-3 -14
1 1 2 0 0 ym) (n) W h(n)
0-11 2 0 17
0 -1 11
yn) =x,(n) Whn) 1 2 0 0 -1|5 14
-1 1 2 0 0 13
2 0 0-1 1|s10 0-1 1 20
12 0 0 -1617 0 0 -1 1 2|
-1 2 o o
0 -1.1 2 0 .'

0 0 -1 1 yn (2,5, 7,1,-33
y,n)= {1,-3, 2, 5, 7} v n)--2,-5,-7,-1,3}-
yr) 13,11,-1,-8,-7 yn)- {8, 14, 13, 1, -6}3
y ) = {-3,-14, 7,17,13)
yAm) = {10, I7, 1,-6,0
2,5,7,1,-3
LL-32,7 add
M-1=2.Discard3,11,-1,-8,-1 -2,-5-7,-i,3
oints add
Discard-3, -14,7, 17,13 8,14, 13, 1,-6
Discard | 10, 17,1,-6,0
r{2, 5, 7,-I,-8,-1,7,17,13,1-6
Discard
Neglect It
Here, yln) the linear
is
convolution of x{n) and hn)
yn)= {2,5, 7,-1,-8,-7, 7, 17, 17 13, 1,-6}
can be observed that the results of
Hence verified. both methods are sau