Systems

5 -46
Signals and Systems

System Modelling :
. 5 3 1 This system has input x(t), outpu t y(t) and
• Consider the system of F1g. . . •
impuJse response h(t).
• - --i•/
System modelling is used to relate the Input x(_t)
j
System h(I)
Output y(t)

input and outpu t of the system with

the help of mathematical equations or Fig. 5.3.1 System modeling
block diagr ams.
• Following are the various meth ods of system modelling
1. lmpulse response
2. lntegrodifferentia l equations
3. Block diagram representations
4. State- space representation.

5.4 Convolution Integral

unit impu lse respo nse of the
The convolution integral relates input outpu t and
continuous time systems.

Applications of convolution
1. It is used for system analysis such as causality,
stability, step respo nse, impu lse

respo nse, invertibility etc.

input and impu lse respo nse is
2. It is used to deter mine outpu t of the system if
given .

3. It relates input outpu t and impu lse response.

ted syste ms using conv olutio n
4. Conv olutio n can be used to resolve interconnec
prope rties.
doma in using fourie r, lapla ce
5. Convolutio n helps to repre sent system in frequency
, stability, filtering etc.
& Z-transforms. This is used to study pole- zero plots

5.4.1 Representation of x(t) in terms of Impulses

of impu lses. We know that
• The signa l x(t) can be represente d as a weigh ted sum
ider the funct ion x(t) and
area of the impu lse function is equal to one. Cons
weigh ted impu lse of Fig. 5.4.1.
Signals and Systems 5 -47 Syatem•
• Fig. 5.4. 1 (n) shows lhe funclion x(I) which is conlinuuus in lime.
There iH a
poinl o_f__amplitud e A nt / = T. In x(I'). ThiH puinl can be rcprcHcnl eJ a8
11 weigh led
impulse as shown in Fig. 5.4.1 (b). Observe Lhnl lhe omplilud c of the
impulRc iA
A. The impulse is plnced nl / = r.
• The function x(I) is conl'inuous. Hence ii wi ll hnvt· infinilc poinls. We
ca n Lake
infinite weighted impulses for nil these pointR. The infinilc weighted
lmpulsc!'.1
means they are so close to each other such lhal they complete ly reprcsenl
x(l).
Since the impulses nrc distinct from each ulhcr we hnve lo add all
of them to
represent x(t). i.e.,

x(t) = f~.r(r) ll(I - r) dr ... (5.4.1)

Area under this point

Is equal to x(t) 6(1 - r), I.e.
x(t) x(I) 6 (I - t) = A

(a)

0 ;T
I
I
I
x(t)6(t- r) I
,A

(b)

Fig. 5.4.1 An impulse function is used to represent point A of x(t)

Here integration is used since ,: is continuous function.

 Signals and Systems 5 - 52 Systems

,,. . Example 5.4.2 : Step - step fimctio 11 co11volutio11.

Tht· impulse re:::po11se and the i11p11t to the system is giue11 ns,
x(t) = u(t+1)
h(t) = u(t - 2)

Deten11i11e the output of tlte system.

Solution : The output y(t) is given as the convolu tion of x (t) and h(t). i.e.,
00

y(t) = x(t) •h(t) = f x (-r) h (t -T) dT ... (5.4.12)

The two functions x(T) and h(r) will be,

x(-r) = u(T+l) =l for T ~ -1

and h(-r) = U(T-2) =l for T~ 2

These two functions are plotted in Fig. 5.4.9 below. In equa tion 5.4.12, we need the
function h (t - r). We know tha t

h (r) = u(-r - 2)

h(r) = 1 for T- 2 ~ 0 or T ~ 2

- +- -T -_,
x('r)=u('r+ 1)
- , ca)
I 1 I
I
~
- ·-
I
- -

I
__,_
I
- i- _j__ l - I
-
t-

"--f -~-~ I -~
--
0 J - -, - - -

,-
-

r- -,
I
h(T) = U(T - 2) I
--1,b, -- - 1-
1 ,
-
I
--,-
I
1
_J_ I
~ -- -
I

r---,---1-
I I I T-
I
I
/0 '
II )
I , - - ,-

I I
-~
h(t-r) I
_J(c) ·- --+--
I
1I
- t- -~~ - - - --1-
I __ lI____ ,

I I T-
I
I
- I 0
i t- 2 , - 1 I
I
·- -
11 1 I
Fig. 5.4.9 Waveforms of x (t ), h (t) and h (t - t)
Signal• and 8yst .m1
6 · 53 Syatams
·1h '" w,ivc ,,rm I';"h,,wn ,n ( r:
J~A .IJ (b ) ilb1,v,,. From . we
'
r-1 ~ ths· ,.,b<,v1; i::9uatHm c;;n . ,
wnte
hll tJ f1,r I - 1 i:: 2
H,.,1, I - t c 2 c;;n i.J li,<; be writt,-n a" / - 2 ,:: t 1,r t-:;; t -
2 Thur,
hU tJ - I f1Jr t ~ I - 2 .. (5.4.13}
·nw, wavd orm i-, ·•hown 1n Fi~. 5A.9 (c ) t.161.,v(·.

1(1) • lJ(t• 1)
(a)

•
[ ·J:
-1
,.
.. t '
A-, p<:r equation 5.4.12 , the
produ ct x( t ) h (l - t') ic, integrated for
1.:v<:ry value: of t. The variation of t
h(l - t) t -,hift., h (I - t). Fig. 5.4.10 c,hows the
(b) pl,,1 of h(l - tJ for arbitrary value: oft.
I
In th.i~ f1gu n: ob<,<.-rvt that h (I - t) = 1
for 1 ~ I - 2 a., per cqua tion 5.4.13 .
I
-1 0 I- 2

l -

Fig. 5.4.10 Plots of x( t) and h(t - -ri The

From the above figure we have
shaded region showa overtap
three ~ara te regions as follows,

From - rn < t,... - 1, x(t) h (I - t ) = 0. Since thc:rc ~ no overlap.

Prom - 1 ~ t ~ l - 2, x{t)h (l - t )-t 0. Since there is overl
ap.
Prom l - 2 < t <<A, x(t)h (l - t) = 0. Since there is no overl
ap.
Based on the above we can write conv olution integration
of equation 5.4.12 as,
, - I 1- 2
y (t) = f (0 / 1) dt+ f .(L < I) dt+ f (1 x 0) dt
- "'· - I 1- 2

1- 2 1- 2
= f dt = It] - I
- 1

= 1- 1
 Signals and Systems 5 - 54 Syste~

Thu~ the rt":-lllt of (·o n,·olut:ion of th,.,

I y(t) = (t-1 )u(t-1 )
two h tnction.., is.
- J J-- ~ - _ J _ - - -~ - - - -
y(t) = f-1 fo r t - ~ ~ -1 i. L' . I ~ I

This htnction can also bt? written ,1s.

y(t) = t -I u(t-1)
,-1
1-- - - This is unit ramp dela~·ed b,· 1. Ifs
plo t is shown in Fig. 5.-!. l l .
0 5

Fig. 5.4.11 Plot of y(t)=u(t+1)• u(t-2)

,,_,. Example 5.4.3 : Exponential - exponential function co11volutio11.

The impulse response of the cirrnit ii given a.., h (I) = e-1 11(t). T/1i..<:: circuit is excited by a11
input of x(t) =e-31 { u(t)-11(t-2)} Detmnine the output of the circuit.

Solution : Here the input and impulse response are :

x(t) = e- 31 {z, (t)-u(f - 2)}

Here u(t)-u(t-2) has the value of 1 from t = 0 to 2. Hence above equation can be
written as,

x(t) = e-3t for OSf S 2

Similarly impulse response is,

lz(t) = e-1 for t?. 0

The output of the circuit can be obtained by convolution of x(t) and Jr(t). From
equation 5.4.5 we can write convolution as,
00

y(t) = x(t)>+h(t)= Jx(r)h(t-r)dr ... (5.4.14)

-oo

Let us write x(t) and h(t) in terms of r. i.e.,

x(r) = e -3, for O5 T 5 2

and
for-r~O ... (5.4.15)
The plots of x(T) and h(-r) are shown in Fig. 5.4.12.
Signals and Systems
5- 55 Systems

(a) : (t)
I -1- i --~
e-3t

-,-
I

ht)
---i_J
-,
(b)
e-t - - -

1-r-
r--rrt- --r--- ,----- !----- ,-
1
Fig. 5.4.12 Plots of x (T) and h (T)

For the convolution of equation 5.4.14 we need h (t-T~ From equation 5.4.15 we
can
obtain h (t--c) by replacing Tby t-T. i.e.
h(t - -c)= e-(t-t) for t-T~O
i.e. h(t - t)=e-( t-t ) fortS't

Case - 1 : 0 st s 2
The plot of h(t-t) is shown in Fig. 5.4.13 (b) for Ost ~ 2 In this figure observe
that
x (t) and h (t-t) overlap as shown by shaded area. For OSt S 2 we can
write the
convolution of equation 5.4.14 as,
0 I 00

y(t) = f Ox h(t-t)d t+f x(T) · li(t-t)dT+ J x{t) x Odt

- 00 0

I I I
-3t
,__..I (a)
i

I
x(r)=e
I
- ·-
I
= f x(t) h (t-t) d-c
I I I 1 I I 0
I I ::.,.
I
I
I
I
I
I
I
I
I 0
m
t~. I 2
_I
7
T
=
t

f0 e-3-r -e - (t - t) dt
I I I
I I I I I I I
I h(t-t) fort< 2 t
,_.. (b) - r - --1- ___J_
I
--'
I = f e-3 t -e-1 -e' dt
I
: 1 I 0
I I k{
--.- 1. / :/~:! I
I
l I
-1 ::)/(:
•:.:.:-!: I I
= e- t f e- 2-r dt
u
·;.·: 1.
' T
0 t I
f
I
! - I 0
I I

Fig. 5.4.13 Plots of x(-c) and h(t -t) for

t s2
 Sys tem,
Slgnala and Sys loms 5 · 56

,. · t I
fl
l--2 ,.-2,11
21 1 ... (5.4. 16)
~(1-,,- ),·~ Fnr 0 S I S 2

Caso - 2 : t > 2
5.4. 13 arc red rawn for I > 2 .
No w cons ider !he c.,sc when I > 2 . Tlw w .,v,•fc,rm ~ 11( Pig.
Fig . 5.'l. 1•1. In !hi, flg un: ob, c r ve Iha ! x( r) and Mt - I )
These w aveform~ ,ire shown in
1
•rl,
ovL p ,
1. , hown by ,
. h,,d i:d arc,1 in fig ure.
overlap from U lo 2. This
r JI
(11) x( r)so

/,,,_,,°l,, > /

I (b)
I
r . 1

0
j
.L__

Fig. 5.4.14 Plots of x (r ) and h (t - r) for t > 2

Hence we can write the convolution equation as,

0 2 ,

y(t) = f Ox h(l - r)dr+ J x (I) · h(l - r)dr+ f Ox h(l - r) dr

0 2

2
= Jx(r)• h (l -r)dr
0
2
- Jr -e-(1-r) dr
=
Je
0

= f2 e - 3r -e -t -er dr
0
2
= e-t f e-2rdr
0

= e-1[-½·e-2r J:
-
Signals and Systems
5-57

for t > 2
Systems

Result

y(t) Thus we obtained convolution as,

t ---·--1-!.
I
0.0664282 0 for t < 2

y(t) = ½(1-e-2t)e-t for 0 S tS2

0 2 I t
½(1-e-4 )e-t for t >2

Fig. 5.4.15 A plot of y(t) This function is plotted in

Fig. 5.4.15.
At t = 2, the value of y(t) = ½(1-e-4 )e-2
= 0.0664282
In the above figure observe that y(t) increases from o to 2. It ha~ the maximum value
at t = 2. Then y(t) decays exponentially.

,,_. Example 5.4.4 : Rectangular pulse to rectangular pulse convolution.

Obtain the convolution of two functions given below.
2 for - 2 $ t $ 2
x(t) = { 0 elsewhere

4 for 0 $ t :52
h (t) = { 0 elsewhere ... (5.4.17)

Solution : The two functions are plotted in Fig. 5.4.16. Observe that x(t) is a pulse of
I
I x(t) : 'I
I
I
I -- amplitude 2 from - 2 to 2. And h (t) is
..
(a) I I 2 I I the pulse of amplitude 4 from O to 2.
I I
! I I
I
I The convolution is given as,
I I I I
I I I
00

I -2
I
I
I II D
'
t-
y(t) = J X (t) h (t-t) dt ... (5.4.18)
I I
I I I
- 00

I I ht)
I
- Here we require h (t-t). Hence we
I
(b) 14
I
can write from equation 5.4.17,

I I 4 for O$ t - t $ 2
I I
i.e. -t$-t$-t+2
I I I
I -
I
h(t-t) =
i.e. t - 2 $ t $ t
I
I
I
t- 0 elsewhere
_j__ -- 0 )

--L-
-i- I
--- 1-- -
-
! Fig. 5.4.17 shows the plots of x(t)
Fig. 5.4.16 Plots of x(t) and h(t) and h(t-t) as per above equation.
of example 5.4.4
 S6gnals and Systems 5 · 58 Sy atern1

Now the nex t <.tep i, to ,1Ctu

- ally cva Iu,-- te convolu tio n intcgrnl.
--
-

r
B J_
--r

I
I
\
- X{'t)

'
2
I
-
I

t
1
-2 0
I - -
l I I I
h(t- 1) -
__ i__ - - r -
I - I
(b) I
I 41 ' - --
-
I
I --

7
I
I I
- - I
I
1 I - 1'
t-2 10
' ( -\ t •- I I
I i
I I I I
Fig. 5.4.17 Plots of X('t)
and h( t- 't)
Case - 1 : To evaluate y(t
) fo rt< - 2
In the above figure observe
tha t t < - 2. Hence there is
h(l -t) . Therefore the pro no ov erlap be tw een x(' t) an
duct x(t ) l1( f-'t ) will be alw ays d
zero.
y(t ) = 0 for t < - 2

Case • 2 : To evaluate y(
t) for -2 $ t < O
Fig. 5.4.18 sho ws the wavef
orms of x(' t) and h(t - 1) for
and h (t - t) starts from t = - - 2 $t < 0. Th e ov erlap of x(t
2. For - 2 $ t < 0, bo th the pu )
area shows the overlap of x(' lses pa rtia lly ov erlap. Th e
t) and h( t- t ). sha de d
,----.- -., ... .-- -.- --
-- - - r - - - ~- ~
-- ----- -2 x(t) _ _ _J
- I

' I

;___,;._
•

,-- -+,-___...,2'-'---""'~
~-
-l:{l1i:lil -
- -

_ _1 _
-o-+
- --+- - 4 - ---

hr -t) for -2 ~ t < 0

1'- --l

-- -- !--
~---1-2-=-..... t
- -l

(b) +--
1

4 --+ --- -+ --- --l

::::\.~·: r.~:
-- -:·-;·~·.:..:.~·-::::· -
~ -- -- -/ } : /
. ).'""- - - - t - -4 ---- .... i.-_
_J

;{\:/);'
r----:-1t-:2::---_--:-2.:.:.:..:.:.:..:J..._-;;o:+ ---l-
-- i. - t
I

Fig. 5.4 .18 Plots of x( t)

and h( t - t) for : $t < O
2
-
signals and Systems

The convolution c.i n be wri tten ""•

5- 59 Sy■temt

00

y{I) = f X(T)/1(/ - t) rh
-2
= f Ox h(t - r)dt + f .r(t) · h (/ - T)dtt J' x(t) x O,/r
- w -2

I
= f X(T)· I, (/ - T)r/T
-2

I
= f 2 X 4 tit
-2

I
=8 f dt
-2

= s rT1,- 2
= 8 (I + 2) for -2 'S I <0

case • 3 : To evaluate y (t) for 0 -s t s 2

Fig. 5.4.19 shows the waveforms of .r(r) and I, (I - T) for OS I S 2 . For this range of t,
both the pulses fully overlap each other. The shaded are.i shows the overlap of the pulses.

I I
,-,- x(tj - -

(a)
I I I 2 I
1-

I -
I I

I T
-2 10
I ? I
I
I I I i
h(l-t) for O~ I ~ 2
I
(b)
I I
I
4 ~ I I '
I I
I

I
'T
I
I I T
t-2 0 I I
I I

: I I I I

Fig. 5.4.19 Plots of x (-i:) and h (t - T) for Ost -s 2

 Signals and Systems 5 • 60 System,

We can write convolution for th.is range as,

y(t) = f x(r) h (I -r) dr

I ~
1-2
f x(T) x Odr+ f x (r) • h (I - r) dr+ f x(r) x Odr
1- 2 I
-~

I
f x(r )- h (t -r)dr
1-2

I I

f 2x4dr= f Bdr
1-2 1- 2

I
B[r] 1-2 =B[t- 1+2]= 16

y(t) 16 for Ost 52

Case - 4 : To evaluate y(t) for 2 s t s 4

54 20 '" 2 st H Fo, tltis ,ang, x(,)

n,, ploo of '(,) and h(H ) "' shown m Fig. d regio n show s the overl ap in figur e.
and h(l-r J partially overlap each other. The shade

I / : I ' x(rJ~/ /
! / _J
(a) - t-- /-1 12 ; / --1

ct tt_ o/ f- ,... ,.,J. =r ir

(b) j ~ - - h ~

---- r-+ -1-i------,

--~-.- .;._:... ~
0 ._ '> I T

7-- 1-1-
Fig, 5.4.20 Plots of x (r) and h (t - r) for 2 s
ts4
this range we can Write the convo lution as,
For

'
 I J
f
)
~-
I/ ( /) I( 'l / (I,/' I J , /,,;) I, (I () ,I I, I f (J / /,(/ 'f/ ,h
I 'J
I
)

- r II , , ,, (/ ,J ,h
I 'J

I 'J

I
I J
') r ti ,I·,
I 11,J,,
' J

J
1
II j ,h . I\ ITl
I J I ')

II ('I I) f11r 2 ;-_ / ". 1

1'11r I • '1, 1111•1·,· wlll 111· 1111 11v1·rl.ip IJ1•lwt·,·1t x('C) ,, 11 d /i(T) ,ind n:~ull o f convolution will
Ill' ,',l'l'tl , l.1•,,

I/(/) (l for I ,,. 1

J ,l'I 111-1 wrllt· 1111 llw v11lt11•H n( 1/(/) rnn1hlncly,

() for I ~ - 2
H(/ ~ 2) for - 2 ". / < 0
!I (I) Ir, fur () ~ / .,.. 2
H(1 /) fur 2 -: 1-; 1
() for I ,,. 1

Thl1-1 ii, lhc rcHult uf convolution.

Example 5.4.5 : Expunc11tial /1111ctio11-Recta11g11lar function convolution.

Impulse rei:1pons1' of r111 L'J'I RYHl,·111 ii; xivm /Jy :

1,(1) = I"-2,
0 ; olherwise

Fi11d llw sys/c111 oulpul dw· lo the inpul :

A ; 0 $/ $ 2
x(I) =
l .
0 ; olherw,se

Also sketch !he output.

Solution : h(t) = e- 21 fort ~O

and x (t) = A for0 $1$ 2

x ('t) = A for O$ 't $ 2