Chapter-2: Impulse Response and Convolution:

 Linear Time Invariant(LTI) System:

 A time invariant system which holds the linearity property in its input-output
relationship is called linear time invariant(LTI) system. A LTI system can be
completely characterized by its impulse response.
 Impulse response of the LTI system is the response of the system when it is exited by
impulse function.
Click to add text

 In the above block diagram, h(t) is the CT impulse response.

 Thus, a LTI system can be represented as shown below:
 where, 𝑥(𝑡) = input or excitation
ℎ(𝑡) = impulse response of the continuous time LTI system
𝑦(𝑡) = output or response of the LTI system

 Continuous time LTI system:

1) Representation of continuous time signals in terms of impulses:
 A continuous time LTI system can be approximated as a linear combination of
staircase as shown in fig. below and represented as 𝑥ො 𝑡 .
 If we define
1
, 0≤𝑡<∆
𝛿∆ 𝑡 = ൝∆
0, 𝑜𝑡ℎ𝑒𝑟𝑒𝑤𝑖𝑠𝑒
since 𝛿∆ 𝑡 have unit amplitude, then

𝑥ො 𝑡 = σ∞
𝑘=−∞ 𝑥(𝑘∆) 𝛿∆ 𝑡 − 𝑘∆ ∆
 As we let ∆→ 0, the approximation 𝑥ො 𝑡
becomes better and better and the limit
equals 𝑥(𝑡). Therefore,
𝑥 𝑡 = lim 𝑥ො 𝑡
∆→0
∞

= lim ෍ 𝑥(𝑘∆) 𝛿∆ 𝑡 − 𝑘∆ ∆
∆→0
𝑘=−∞
 Also, as ∆→ 0 the summation in above
equation approaches an integration. Also,
as ∆→ 0, 𝑥ො 𝑡 → 𝑥(𝑡) ,and ∆→ 𝑑𝜏 and
𝑘∆→ 𝜏.
 Hence,
∞
𝑥 𝑡 = ‫׬‬−∞ 𝑥(𝜏) 𝛿 𝑡 − 𝜏 𝑑𝜏
 Above eq. is also the shifting property of
impulses.
Fig.: Staircase approximation to a CT signal
2) Convolution Integral:
 We know that the arbitrary input signal 𝑥(𝑡) can be expressed in terms of delayed
impulses as
∞
𝑥 𝑡 = ‫׬‬−∞ 𝑥(𝜏) 𝛿 𝑡 − 𝜏 𝑑𝜏 …..i
 If this signal is passed through a LTI system , then the output or response of the LTI
system is expressed as
∞
𝑦 𝑡 = 𝑇 𝑥 𝑡 = 𝑇{‫׬‬−∞ 𝑥(𝜏) 𝛿 𝑡 − 𝜏 𝑑𝜏} …..ii
 Since the system is linear , we have
∞
𝑦 𝑡 = ‫׬‬−∞ 𝑥(𝜏) 𝑇{𝛿 𝑡 − 𝜏 } 𝑑𝜏 …..iii
 Again, since the system is time invariant , we obtain
ℎ 𝑡 = 𝑇{𝛿(𝑡)}
and ℎ 𝑡 − 𝜏 = 𝑇{𝛿(𝑡 − 𝜏)} …..iv
 From equations iii and iv , we have
∞
𝑦 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏 …..v
 Eq. v is the convolution integral and we see that a LTI system is completely
characterized by its impulse response ℎ 𝑡 .
 Eq. v can be also written as
∞
𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏
where ∗ denotes the convolution operation.
 Since the convolution integral is commutative , we can write
∞
𝑦 𝑡 = ℎ 𝑡 ∗ 𝑥 𝑡 = ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
 Convolution integral is also known as superposition integral.
• This computation can be done in four steps:
1. Folding: Fold ℎ(𝜏)about 𝜏 = 0 to obtain ℎ(−𝜏).
2. Shifting: shift ℎ −𝜏 by 𝑡 to the right (left) if 𝑡 is positive (negative) to obtain
ℎ(𝑡 − 𝜏).
3. Multiplication: Multiply 𝑥(𝜏)by ℎ 𝑡 − 𝜏 to obtain the product 𝑣(𝑡) =
𝑥(𝜏)ℎ 𝑡 − 𝜏 .
4. Integration: Integrate all the values of the product sequence 𝑣(𝑡) to obtain the
output at time 𝑡 .
 Note that steps 2 through 4 in the summary must be repeated, for all the possible
time shift −∞ < 𝑡 < ∞.
Example: Let 𝑥 𝑡 be the input to an LTI system with unit impulse response ℎ 𝑡 , where
𝑥 𝑡 = 𝑒 −𝑎𝑡 𝑢 𝑡 , 𝑎 > 0
and ℎ 𝑡 = 𝑢(𝑡)
∞
a) Determine the output 𝑦 𝑡 such that 𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏
∞
b) Determine the output 𝑦 𝑡 such that 𝑦 𝑡 = ℎ 𝑡 ∗ 𝑥 𝑡 = ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
Solution:
a) We know that the convolution integral is given as
∞
𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏
 Then 𝑥 𝜏 = 𝑒 −𝑎𝜏 𝑢 𝜏 , 𝑎 > 0
ℎ 𝜏 = 𝑢(𝜏) and ℎ −𝜏 = 𝑢(−𝜏)
therefore, ℎ 𝑡 − 𝜏 = 𝑢(𝑡 − 𝜏)
1) Interval 1: For 𝑡 < 0
 In this case, there is no nonzero overlapping between 𝑥 𝜏 and ℎ 𝑡 − 𝜏 . Hence the
output 𝑦 𝑡 = 0.
𝑥 𝜏 ℎ 𝑡 − 𝜏 ,𝑡 < 0

2) Interval 2: For 𝑡 ≥ 0
𝑒 −𝑎𝜏 , 0 < 𝜏 < 𝑡
 In this case, 𝑥 𝜏 ℎ 𝑡−𝜏 =ቊ
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
This is shown in fig. below:
 𝑡 −𝑎𝜏 𝑒 −𝑎𝜏 𝑡 1
 Therefore, 𝑦 𝑡 = ‫׬‬0 𝑒 𝑑𝜏 = =− 𝑒 −𝑎𝑡 − 𝑒 0
−𝑎 0 𝑎
1
Or 𝑦 𝑡 == 1 − 𝑒 −𝑎𝑡 , 𝑡 ≥ 0
𝑎
1
This can also be written as 𝑦 𝑡 = 1 − 𝑒 −𝑎𝑡 𝑢(𝑡)
𝑎
 The output 𝑦 𝑡 is shown in fig. below:

b) Assignment(try yourself)
 Properties of LTI System:
 We know that the convolution integral is given by
∞
𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏
Then the properties of LTI systems are given below:
1) Commutative
2) Distributive
3) Associative
4) Causality
5) Stability
6) Invertibility
7) Unit step response
8) Memory
• Note that commutative, distributive and associative properties are also the
properties of convolution.
1) Commutative:
 A basic property of convolution(either CT or DT) is that it is commutative operation.
That is, 𝑦 𝑡 =𝑥 𝑡 ∗ℎ 𝑡 =ℎ 𝑡 ∗𝑥 𝑡
Proof:
∞
 We know, 𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏
Let, 𝑡 − 𝜏 = 𝑠, then 𝜏 = 𝑡 − 𝑠
As 𝜏 → −∞, 𝑠 → ∞ and 𝜏 → ∞, 𝑠 → −∞
∞
 Therefore, 𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = ‫׬‬−∞ 𝑥(𝑡 − 𝑠)ℎ 𝑠 𝑑𝑠
∞
= ‫׬‬−∞ ℎ 𝑠 𝑥(𝑡 − 𝑠) 𝑑𝑠
𝑦 𝑡 =𝑥 𝑡 ∗ℎ 𝑡 =ℎ 𝑡 ∗𝑥 𝑡
2) Distributive:
 Another basic property of convolution is distributive. Specifically, convolution
distributive over addition. Then we have,
𝑥 𝑡 ∗ ℎ1 𝑡 + ℎ2 𝑡 = 𝑥 𝑡 ∗ ℎ1 𝑡 + 𝑥 𝑡 ∗ ℎ2 (𝑡)
 Fig.: Interpretation of distributive property of convolution for parallel interconnection of LTI system

3) Associative:
 Another important and useful property of convolution is associative. That is,
𝑥 𝑡 ∗ ℎ1 𝑡 ∗ ℎ2 𝑡 = {𝑥 𝑡 ∗ ℎ1 𝑡 } ∗ ℎ2 (𝑡)
4) Causality:
 A causal system is one whose output depends only on the present and past inputs.
In case of LTI systems, causality is measured in terms of the impulse response of the
LTI system.
 We know the convolution integral is
∞
𝑦 𝑡 = ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
0 ∞
𝑦 𝑡 = ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏 + ‫׬‬0 ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
(𝐼) (𝐼𝐼)
 The (𝐼) terms include future inputs and the (𝐼𝐼) includes present and past inputs. If
we decompose first term, we get,
𝑦 𝑡 = … + ℎ −2 𝑥 𝑡 + 2 + ℎ −1 𝑥(𝑡 + 1) ; thus the inputs
are future inputs.
 The LTI system is causal if the first term become zero. For this the impulse response
must satisfy the condition
ℎ 𝑡 = 0, for 𝑡 < 0.
This is both necessary and sufficient condition for causal LTI system.
5) Stability: (Bounded Input Bounded Output) (finite)
 A system is BIBO stable if every bounded input produces bounded output. In LTI
system stability is also measured in terms of its impulse response.
 Consider an input signal 𝑥(𝑡) is bounded in magnitude, there exists a constant 𝑀𝑥
such that
𝑥(𝑡) ≤ 𝑀𝑥 < ∞
 Suppose this bounded input is applied to a LTI system, the output is
∞
𝑦 𝑡 = ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
The magnitude of the output is
∞
𝑦 𝑡 = ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
∞
 But we know, 𝑦 𝑡 ≤ ‫׬‬−∞ ℎ 𝜏 𝑥 𝑡 − 𝜏 𝑑𝜏
Also we know, 𝑥(𝑡 − 𝜏) ≤ 𝑀𝑥 , therefore we can write
∞
𝑦 𝑡 ≤ 𝑀𝑥 ‫׬‬−∞ ℎ 𝜏 𝑑𝜏 …..i
 From eq. i , it is clear that the output is bounded if the impulse response of the LTI
system is absolutely integrable. That is , if
 ∞
ℎ 𝜏 𝑑𝜏 < ∞
‫׬‬−∞
This is both necessary and sufficient condition for stability of a LTI system.
6) Invertibility:

𝛿 𝑡 = ℎ(𝑡) ∗ ℎ1 (𝑡)

Fig.: An inverse system for continuous-time LTI system

 Consider ℎ 𝑡 be the impulse response of LTI system. Then the LTI system is
invertible only when an inverse system exists that when cascaded with the original
system, produces an O/P equal to the I/P to the first system.
 In fig., ℎ1 (𝑡) is the inverse system and
ℎ(𝑡) ∗ ℎ1 (𝑡) = 𝛿 𝑡
Example: Determine the inverse system of the delay system given below:
𝑦 𝑡 = 𝑥(𝑡 − 𝑡0 ) …..i
Solution: (not needed)
 Given, 𝑦 𝑡 = 𝑥(𝑡 − 𝑡0 )
𝑦 𝑡 = 𝑥 𝑡 − 𝑡0 = 𝑥(𝑡 − 𝑡0 ) ∗ 𝛿(𝑡 − 𝑡0 )
{𝑥 𝑡 ∗ 𝛿 𝑡 = 𝑥(𝑡) }
 But from convolution integral , we have
𝑦 𝑡 =𝑥 𝑡 ∗𝛿 𝑡
therefore, ℎ 𝑡 = 𝛿(𝑡 − 𝑡0 )
and 𝑥(𝑡 − 𝑡0 ) = 𝑥(𝑡) ∗ 𝛿(𝑡 − 𝑡0 )
 To find inverse system, let 𝑡 − 𝑡0 = 𝑠 in eq. I, then
𝑦 𝑠 + 𝑡0 = 𝑥(𝑠)
𝑥 𝑠 = 𝑦 𝑠 + 𝑡0
𝑥 𝑠 = 𝑦 𝑠 ∗ 𝛿(𝑠 + 𝑡0 )
 In general, 𝑥 𝑡 = 𝑦 𝑡 ∗ 𝛿(𝑡 + 𝑡0 )
 Compare with 𝑥 𝑡 = 𝑦 𝑡 ∗ ℎ1 (𝑡) we obtain the inverse system is
ℎ1 𝑡 = 𝛿(𝑡 + 𝑡0 )
7) Unit Step Response:
 The unit step response is the response of LTI system when the input is 𝑥 𝑡 = 𝑢(𝑡).
Therefore, the unit step response of LTI system is given by
s(t) = 𝑢 𝑡 ∗ ℎ 𝑡
∞ 𝑡
 That is, s(t) = ‫׬‬−∞ ℎ 𝜏 𝑢 𝑡 − 𝜏 𝑑𝜏 = ‫׬‬−∞ ℎ 𝜏 𝑑𝜏
which is the running integral of its impulse response. Then we have
𝑑{𝑠(𝑡)}
ℎ 𝑡 = = 𝑠′(𝑡)
𝑑𝑡
8) Memory:
 The response LTI system is
∞
𝑦 𝑡 = ‫׬‬−∞ 𝑥(𝜏)ℎ 𝑡 − 𝜏 𝑑𝜏 = 𝑥 𝑡 ∗ ℎ 𝑡 …..i
 A continuous-time LTI system is memoryless if
ℎ 𝑡 = 0 for 𝑡 ≠ 0 …..ii
and such a memoryless system has the form
𝑦 𝑡 = 𝑘 𝑥(𝑡) …..iii
for some constant 𝑘 and has the impulse response
ℎ 𝑡 = 𝑘 𝛿(𝑡) …..iv
 For 𝑘 = 1 in eq. iv, the system is then called identity system. And we can write the
convolution integral formula as
∞
𝑦 𝑡 = ‫׬‬−∞ 𝑥(𝜏)𝛿 𝑡 − 𝜏 𝑑𝜏 = 𝑥 𝑡 ∗ 𝛿 𝑡
{𝑥 𝑡 ∗ 𝛿 𝑡 = 𝑥(𝑡)}
 If the impulse response ℎ 𝑡 ≠ 0 for 𝑡 ≠ 0, the system is called dynamic system or
LTI system with memory.

Frequency Response of LTI System:

 If 𝑥 𝑡 and ℎ 𝑡 is the unit impulse response of LTI system then the response or
output 𝑦 𝑡 of the LTI system is expressed in terms of convolution integral as
∞
𝑦 𝑡 = ‫׬‬−∞ 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 …..
=𝑥 𝑡 ∗ℎ 𝑡
 To analyze LTI system, it is convenient to utilize the frequency domain because
differential equation and convolution operation in the time domain become
algebraic operation in frequency domain.
 Applying convolution property of CTFT in above equation, we get
𝑌 𝑗𝜔 = 𝑋 𝑗𝜔 𝐻( 𝑗𝜔)
𝑌( 𝑗𝜔)
and 𝐻 𝑗𝜔 = …..i
𝑋 𝑗𝜔
where 𝐻 𝑗𝜔 is the frequency response of LTI system. It is also called transfer
function or system function.
• Magnitude and Phase Representation of Frequency Response:
 The frequency response 𝐻 𝑗𝜔 can be written in polar form as
𝐻 𝑗𝜔 = 𝐻 𝑗𝜔 𝑒 𝑗 𝜃( 𝑗𝜔) …..i
where, 𝐻 𝑗𝜔 = amplitude (or magnitude) response
and 𝜃 𝑗𝜔 = phase response
 Note that phase response does not affect the amplitude of the individual frequency
components but only provides information concerning the relative phases of
exponentials that make up ℎ(𝑡).
 𝐻 𝑗𝜔 exhibits conjugate symmetry. That is,
𝐻 𝑗𝜔 = 𝐻 −𝑗𝜔 , symmetric about origin.
and 𝜃 −𝑗𝜔 = −𝜃( 𝑗𝜔) , antisymmetric about origin.
 Discrete-Time LTI Systems:
1. Representation or Resolution of Discrete-Time Signals in terms of Impulses:
 A discrete-time (DT) signal 𝑥[𝑛] can be resolved into the sum of weighted impulse
sequences. The concept is illustrated below:
Fig: Decomposition of a DT signal into a weighted sum of shifted impulses
 From above fig. , we can write
𝑥 −2 , 𝑛 = −2
𝑥 −2 𝛿 𝑛 + 2 = ቊ
0 , 𝑛 ≠ −2
𝑥 −2 , 𝑛 = −1
𝑥 −1 𝛿 𝑛 + 1 = ቊ
0 , 𝑛 ≠ −1
𝑥 0, 𝑛=0
𝑥 0𝛿 𝑛 =ቊ
0 , 𝑛≠0
𝑥 1, 𝑛=1
𝑥 1 𝛿 𝑛−1 =ቊ
0 , 𝑛≠1
 In general, we can write
𝑥 𝑛 = ⋯ + 𝑥 −2 𝛿 𝑛 + 2 + 𝑥 −1 𝛿 𝑛 + 1 + 𝑥 0 𝛿 𝑛 + 𝑥 1 𝛿 𝑛 − 1 + ⋯

 Therefore, 𝑥 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘] …..1

Equation (1) corresponds to a representation of an arbitrary sequence 𝑥 𝑛 as a
linear combination of shifted impulses 𝛿[𝑛 − 𝑘] where the weights in this linear
combination are 𝑥 𝑘 .
2. Response of LTI Systems to Arbitrary Inputs: The Convolution Sum
 Any arbitrary input signal 𝑥[𝑛] can be expressed i terms of weighted impulses as
 𝑥 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘] …..1
 Now, if the LTI system is excited by x[n], then the output 𝑦[𝑛] can be expressed as
𝑦 𝑛 = 𝑇{𝑥 𝑛 }
𝑦 𝑛 = 𝑇{σ∞ 𝑘=−∞ 𝑥 𝑘 𝛿[𝑛 − 𝑘]} …..2
 Since, the system is linear, we can write
𝑦 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 𝑇{𝛿 𝑛 − 𝑘 } …..3
 Again, since the system is time-invariant , we have
ℎ 𝑛 = 𝑇{𝛿 𝑛 }
then ℎ 𝑛 − 𝑘 = 𝑇{𝛿 𝑛 − 𝑘 } …..4
 From equations (3) and (4), we can write
𝑦 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘 …..5
Equation (5) gives the response 𝑦 𝑛 of the LTI system as a function of input signal
𝑥[𝑛] and the unit impulse response ℎ[𝑛], called convolution sum.
 Equation (5) can be written as 𝑦 𝑛 =𝑥 𝑛 ∗ℎ 𝑛 …..6
 Suppose we wish to compute output of the system at time 𝑛 = 𝑛0 , then we have
𝑦 𝑛0 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛0 − 𝑘 …..7
This computation can be done in four steps:
1. Folding: Fold ℎ 𝑘 about k=0 to obtain ℎ −𝑘 .
2. Shifting: shift ℎ −𝑘 by 𝑛0 to the right (left) if 𝑛0 is positive (negative) to obtain
ℎ 𝑛0 − 𝑘 .
3. Multiplication: Multiply 𝑥 𝑘 by ℎ 𝑛0 − 𝑘 to obtain the product 𝑣𝑛0 𝑘 =
𝑥 𝑘 ℎ 𝑛0 − 𝑘 .
4. Summation: Sum all the values of the product sequence 𝑣𝑛0 𝑘 to obtain the
output at time 𝑛 = 𝑛0 .
 Note that steps 2 through 4 in the summary must be repeated, for all the possible
time shift −∞ < 𝑛 < ∞.

Example: The impulse response of a LTI system is

ℎ 𝑛 = {1, 𝟐, 1, −1}
Determine the response of the system to the input
 𝑥 𝑛 = 𝟏, 2, 3, 1
where bold number represents value at origin (𝑛 = 0)
Solution:
1. Graphical Method:
 Given, ℎ 𝑛 = {1, 𝟐, 1, −1} , impulse response, and
𝑥 𝑛 = 𝟏, 2, 3, 1 , input signal
 The output 𝑦 𝑛 is given by the convolution sum as
𝑦 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘
= ⋯ + 𝑥 −1 ℎ 𝑛 + 1 + 𝑥 0 ℎ 𝑛 + 𝑥 1 ℎ 𝑛 − 1 + 𝑥 2 ℎ 𝑛 − 2 +
𝑥 3 ℎ 𝑛−3 +𝑥 4 ℎ 𝑛−4 +⋯
=0+𝑥 0 ℎ 𝑛 +𝑥 1 ℎ 𝑛−1 +𝑥 2 ℎ 𝑛−2 +𝑥 3 ℎ 𝑛−3 +0
 Therefore, 𝑦 𝑛 = ℎ 𝑛 + 2ℎ 𝑛 − 1 + 3ℎ 𝑛 − 2 + ℎ 𝑛 − 3 …..i
 Now, we draw the individual signal of ℎ 𝑛 , 2ℎ 𝑛 − 1 , 3ℎ 𝑛 − 2 , ℎ 𝑛 − 3 , and add
them to obtain output 𝑦 𝑛 as illustrated in the following figures:
 Therefore, the output is
𝑦 𝑛 = {1, 𝟒, 8, 8, 3, −2, −1}
2. Mathematical Method:
 The output 𝑦 𝑛 is given by the convolution sum as
𝑦 𝑛 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘
 Now, 𝑦 0 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ −𝑘
= ⋯ + 𝑥 −1 ℎ 1 + 𝑥 0 ℎ 0 + 𝑥 1 ℎ −1 + 𝑥 2 ℎ −2 + ⋯
=0+1×2+2×1+0=4
𝑦 1 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 1 − 𝑘
= ⋯ + 𝑥 0 ℎ 1 + 𝑥 1 ℎ 0 + 𝑥 2 ℎ −1 + ⋯
=1×1+2×2+3×1=8
𝑦 2 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 2 − 𝑘
= ⋯ + 𝑥 0 ℎ 2 + 𝑥 1 ℎ 1 + 𝑥 2 ℎ 0 + 𝑥 3 ℎ −1 + ⋯
= 1 × −1 + 2 × 2 + 3 × 2 + 1 × 1 = 8
 𝑦 3 = σ∞
 Similarly , 𝑘=−∞ 𝑥 𝑘 ℎ 3 − 𝑘
= ⋯+ 𝑥 1 ℎ 2 + 𝑥 2 ℎ 1 + 𝑥 3 ℎ 0 + ⋯
= 2 × (−1) + 3 × 1 + 1 × 2 = 3
𝑦 4 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 4 − 𝑘
= ⋯+ 𝑥 2 ℎ 2 + 𝑥 3 ℎ 1 + ⋯
= 3 × −1 + 1 × 1 = −2
𝑦 5 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 5 − 𝑘
= ⋯+ 𝑥 3 ℎ 2 + ⋯
= 1 × (−1) = −1
𝑦 6 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 6 − 𝑘
= ⋯+ 𝑥 3 ℎ 3 + ⋯
=1×0=0
and 𝑦 𝑛 = 0 for 𝑛 ≥ 6
 Again, 𝑦 −1 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ −1 − 𝑘
= ⋯ + 𝑥 0 ℎ −1 + ⋯
=1×1=1
 Similarly, 𝑦 −2 = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ −2 − 𝑘
= ⋯ + 𝑥 0 ℎ −2 + ⋯
=1×0=0
and 𝑦 𝑛 = 0 for 𝑛 ≤ −2
 Therefore, the output is, 𝑦 𝑛 = {1, 𝟒, 8, 8 , 3, −2, −1}

3. Overlap and Add Method:

ℎ𝑛 1 2 1 Click-1
to add text
𝑥𝑛
1 1 2 1 -1
𝑦 𝑛 = {1, 𝟒, 8, 8, 3, −2, −1}
2 2 4 2 -2

3 3 6 3 -3

1 1 2 1 -1
o Note:
• Determination of length of 𝒚[𝒏]:
 Let, length of 𝑥[𝑛] = 𝐿1 and length of ℎ[𝑛] = 𝐿2
then length of 𝑦 𝑛 , 𝐿 = 𝐿1 + 𝐿2 − 1
 For above example, 𝐿 = 4 + 4 − 1 = 7
• Range of convoluted signal 𝒚[𝒏]:
 It is given by
sum of lower limits ≤ 𝑛 ≤ sum of upper limits
0 + −1 ≤ 𝑛 ≤ 3 + 2
−1 ≤ 𝑛 ≤ 5
 Therefore, output signal ranges from −1 to 5.

Example: Find the output of the signal 𝑦[𝑛] of a relaxed LTI system with impulse
response ℎ 𝑛 = 𝑎𝑛 𝑢 𝑛 , 𝑎 < 1, when the input is a unit step sequence.
Solution:
1, 𝑛 ≥ 0
 Given, 𝑥 𝑛 = 𝑢[𝑛], where, 𝑢 𝑛 = ቊ
0, 𝑛 < 0
and ℎ 𝑛 = 𝑎𝑛 𝑢 𝑛 , 𝑎 < 1
 The output is given by the convolution sum as
𝑦 𝑛 = σ∞ 𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘 (convolution is commutative)
𝑦 𝑛 = σ∞ 𝑘=−∞ ℎ 𝑘 𝑢 𝑛 − 𝑘
• Interval 1:
 For 𝑛 < 0,
 Properties of LTI System:
 We know that the convolution integral is given by
𝑦 𝑛 = σ∞
𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘 = 𝑥 𝑛 ∗ ℎ 𝑛
Then the properties of LTI systems are given below:
1) Commutative
2) Distributive
3) Associative
4) Causality
5) Stability
6) Invertibility
7) Unit step response
8) Memory
• Note that commutative, distributive and associative properties are also the
properties of convolution.
1) Commutative:
 A basic property of convolution(either CT or DT) is that it is commutative operation.
That is, 𝑦[𝑛] = 𝑥[𝑛] ∗ ℎ[𝑛] = ℎ[𝑛] ∗ 𝑥[𝑛]
Proof:
 We know, 𝑦[𝑛] = 𝑥[𝑛] ∗ ℎ[𝑛] = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘
Let, 𝑛 − 𝑘 = 𝑠, then 𝑘 = 𝑛 − 𝑠
As 𝑘 → −∞, 𝑠 → ∞ and 𝑘 → ∞, 𝑠 → −∞
 Therefore, 𝑦[𝑛] = 𝑥[𝑛] ∗ ℎ[𝑛] = σ∞ 𝑠=−∞ 𝑥 𝑛 − 𝑠 ℎ 𝑠
= σ∞𝑠=−∞ ℎ 𝑠 𝑥 𝑛 − 𝑠
𝑦[𝑛] = 𝑥[𝑛] ∗ ℎ[𝑛] = ℎ[𝑛] ∗ 𝑥[𝑛]
2) Distributive:
 Another basic property of convolution is distributive. Specifically, convolution
distributive over addition. Then we have,
𝑥 𝑛 ∗ ℎ1 𝑛 + ℎ2 𝑛 = 𝑥[𝑛] ∗ ℎ1 [𝑛] + 𝑥[𝑛] ∗ ℎ2 [𝑛]
3) Associative:
 Another important and useful property of convolution is associative. That is,
𝑥[𝑛] ∗ ℎ1 [𝑛] ∗ ℎ2 [𝑛] = {𝑥[𝑛] ∗ ℎ1 [𝑛]} ∗ ℎ2 [𝑛]
4) Causality:
 A causal system is one whose output depends only on the present and past inputs.
In case of LTI systems, causality is measured in terms of the impulse response of the
LTI system.
 We know the convolution integral is
𝑦[𝑛] = σ∞𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘
𝑦[𝑛] = σ−1 ∞
𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘 + σ𝑘=0 ℎ 𝑘 𝑥 𝑛 − 𝑘
(𝐼) (𝐼𝐼)
 The (𝐼) terms include future inputs and the (𝐼𝐼) includes present and past inputs. If
we decompose first term, we get,
𝑦[𝑛] = … + ℎ[−2]𝑥[𝑛 + 2] + ℎ[−1]𝑥[𝑛 + 1] ; thus the inputs
are future inputs.
 The LTI system is causal if the first term become zero. For this the impulse response
must satisfy the condition
ℎ[𝑛] = 0, for 𝑛 < 0.
This is both necessary and sufficient condition for causal LTI system.
5) Stability:
 A system is BIBO stable if every bounded input produces bounded output. In LTI
system stability is also measured in terms of its impulse response.
 Consider an input signal 𝑥[𝑛] is bounded in magnitude, there exists a constant 𝑀𝑥
such that
𝑥[𝑛] ≤ 𝑀𝑥 < ∞
 Suppose this bounded input is applied to a LTI system, the output is
𝑦[𝑛] = σ∞𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘
The magnitude of the output is
𝑦[𝑛] = σ∞ 𝑘=−∞ ℎ 𝑘 𝑥 𝑛 − 𝑘
 But we know, 𝑦[𝑛] ≤ σ∞ 𝑘=−∞ ℎ[𝑘] 𝑥[𝑛 − 𝑘]
Also we know, 𝑥[𝑛 − 𝑘] ≤ 𝑀𝑥 , therefore we can write
𝑦[𝑛] ≤ 𝑀𝑥 σ∞ 𝑘=−∞ ℎ[𝑘] …..i
 From eq. i , it is clear that the output is bounded if the impulse response of the LTI
system is absolutely integrable. That is , if
 σ∞𝑘=−∞ ℎ[𝑘] < ∞
This is both necessary and sufficient condition for stability of a LTI system.
6) Invertibility:

𝛿[𝑛] = ℎ[𝑛] ∗ ℎ1 [𝑛]

Fig.: An inverse system for continuous-time LTI system

 Consider ℎ[𝑛] be the impulse response of LTI system. Then the LTI system is
invertible only when an inverse system exists that when cascaded with the original
system, produces an O/P equal to the I/P to the first system.
 In fig., ℎ1 [𝑛] is the inverse system and
ℎ[𝑛] ∗ ℎ1 [𝑛] = 𝛿[𝑛]
7) Unit Step Response:
 The unit step response is the response of LTI system when the input is 𝑥[𝑛] = 𝑢[𝑛].
Therefore, the unit step response of LTI system is given by
s n =𝑢 𝑛 ∗ℎ 𝑛 =ℎ 𝑛 ∗𝑢 𝑛
 That is, s 𝑛 = σ∞ 𝑛
𝑘=−∞ ℎ 𝑘 𝑢 𝑛 − 𝑘 = σ𝑘=−∞ ℎ 𝑘 …..i
s 𝑛 = σ𝑛−1
𝑘=−∞ ℎ 𝑘 + ℎ[𝑛] …..ii
 From (i), also we have, s 𝑛 − 1 = σ𝑛−1
𝑘=−∞ ℎ 𝑘 …..iii
 From (ii) and (iii), we have s 𝑛 = s 𝑛 − 1 +ℎ[𝑛]
 Therefore, ℎ 𝑛 =s 𝑛 − s 𝑛−1 …..iv
8) Memory:
 The response LTI system is
𝑦[𝑛] = σ∞𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘 = 𝑥[𝑛] ∗ ℎ[𝑛] …..i
 A discrete-time LTI system is memoryless if
ℎ[𝑛] = 0 for 𝑛 ≠ 0 …..ii
and such a memoryless system has the form
𝑦[𝑛] = 𝑘 𝑥[𝑛] …..iii
for some constant 𝑘 and has the impulse response
ℎ[𝑛] = 𝑘 𝛿[𝑛] …..iv
 For 𝑘 = 1 in eq. iv, the system is then called identity system. And we can write the
convolution integral formula as
𝑦[𝑛] = σ∞ 𝑘=−∞ 𝑥 𝑘 𝛿 𝑛 − 𝑘 = 𝑥[𝑛] ∗ 𝛿[𝑛]
{𝑥 𝑛 ∗ 𝛿 [𝑛] = 𝑥[𝑛]}
 If the impulse response ℎ[𝑛] ≠ 0 for 𝑛 ≠ 0, the system is called dynamic system or
LTI system with memory.

Frequency Response of LTI System:

 If 𝑥 𝑛 and ℎ[𝑛] is the unit impulse response of LTI system then the response or
output 𝑦 𝑛 of the LTI system is expressed in terms of convolution integral as
𝑦[𝑛] = σ∞ 𝑘=−∞ 𝑥 𝑘 ℎ 𝑛 − 𝑘 …..i
= 𝑥[𝑛] ∗ ℎ[𝑛]
 To analyze LTI system, it is convenient to utilize the frequency domain because
differential equation and convolution operation in the time domain become
algebraic operation in frequency domain.
 Applying convolution property of DTFT in above equation, we get
𝑌 𝑒 𝑗𝜔 = 𝑋 𝑒 𝑗𝜔 𝐻( 𝑒 𝑗𝜔 )
𝑗𝜔 𝑌( 𝑒 𝑗𝜔 )
and 𝐻 𝑒 = …..i
𝑋 𝑒 𝑗𝜔
where 𝐻 𝑒 𝑗𝜔 is the frequency response of LTI system. It is also called transfer
function or system function.
• Magnitude and Phase Representation of Frequency Response:
 The frequency response 𝐻 𝑒 𝑗𝜔 can be written in polar form as
𝐻 𝑒 𝑗𝜔 = 𝐻 𝑒 𝑗𝜔
𝑒 𝑗 𝜃( 𝑒 𝑗𝜔 ) …..i
where, 𝐻 𝑒 𝑗𝜔 = amplitude (or magnitude) response
and 𝜃 𝑒 𝑗𝜔 = phase response
 Note that phase response does not affect the amplitude of the individual frequency
components but only provides information concerning the relative phases of
exponentials that make up ℎ[𝑛].
 𝐻 𝑒 𝑗𝜔 exhibits conjugate symmetry. That is,
𝐻 𝑒 𝑗𝜔 = 𝐻 − 𝑒 𝑗𝜔 , symmetric about origin.