Chapter 2

Linear Time-Invariant Systems

(LTI Systems)
 Linear Time-Invariant Systems
• A system is said to be Linear Time-Invariant (LTI) if it possesses the basic system properties of linearity and
time-invariance.
• The input-output relationship for LTI systems is described in terms of a convolution operation.
DT Signal Decomposition in terms of shifted unit impulses 𝑥𝑛 𝑦𝑛
LTI System
𝑥𝑛
𝑥2
𝑥2 𝑥 −1 𝑥1
𝑥 −1 𝑥 1 𝑥 −2 𝑥0
𝑥 −2 𝑥0 ∙∙∙ + ∙∙∙
+ + +
−2 −1 0 1 2
𝑥 −2 𝛿[𝑛 + 2] 𝑥 −1 𝛿[𝑛 + 1] 𝑥 0 𝛿[𝑛] 𝑥 1 𝛿[𝑛 − 1] 𝑥 2 𝛿[𝑛 − 2]
−2 −1 0 1 2 𝑡
∞ ෍𝛽 𝛿 𝑛 − 𝑘 ෍𝛽 ℎ 𝑛 − 𝑘

𝑥 𝑛 = ෍ 𝑥[𝑘]𝛿[𝑛 − 𝑘] 𝛿𝑛 ℎ𝑛 𝛿 𝑛−𝑘 ℎ 𝑛−𝑘

LTI System LTI System LTI System
𝑘=−∞
Sum of scaled impulses Impulse response Time invariance Linearity
∞
∞
𝑥𝑛 𝑦 𝑛 = ෍ 𝑥 𝑘 ℎ[𝑛 − 𝑘] 𝑦 𝑛 = 𝑥[𝑛] ∗ ℎ[𝑛] = ෍ 𝑥 𝑘 ℎ[𝑛 − 𝑘]
LTI System 𝑘=−∞
𝑘=−∞
Convolution sum
 Convolution 1
Example:
Impulse response of
an LTI system

Convolution

input

Find output

𝑦 𝑛 = 𝑥[𝑛] ∗ ℎ[𝑛] = ෍ 𝑥 𝑘 ℎ[𝑛 − 𝑘]

𝑘=−∞
There are only two non-zero values for the input

𝑦 𝑛 = 𝑥 0 ℎ 𝑛 − 0 + 𝑥 1 ℎ 𝑛 − 1 = 0.5ℎ 𝑛 + 2 ℎ[𝑛 − 1]
 Convolution 2
Example:

3 𝑛=0 3 𝑛=0
𝑥𝑛 = 2 𝑛=1
ℎ𝑛 = 1 𝑛=1
1 𝑛=2 2 𝑛=2
0 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒 0 𝑒𝑙𝑠𝑒𝑤ℎ𝑒𝑟𝑒
Length =3 Length =3

Solution:

Convolution Length = 𝑁1 + 𝑁2 − 1 = 3 + 3 − 1 = 5
 Convolution 3
Example: Find the output of an LTI system having a unit impulse response ℎ 𝑛 = 𝑢[𝑛], for the input 𝑥 𝑛 =∝𝑛 𝑢 𝑛

∝ 𝑘 0≤𝑘≤𝑛 𝑘 >𝑛 →ℎ 𝑛−𝑘 =0

𝑥 𝑘 ℎ[𝑛 − 𝑘] = ቊ 𝑘<0 →𝑥 𝑘 =0
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

Thus, for 𝑛 ≥ 0,

𝑛
𝑛≥0 1−∝𝑛+1
𝑦𝑛 =෍ ∝𝑘 =
1 −∝
𝑘=0
𝑛<0
Thus, for all 𝑛,
𝑛
1−∝𝑛+1 𝑠 = ෍ ∝𝑘 = 1 +∝ +∝2 + ⋯ +∝𝑛
𝑦𝑛 = 𝑢[𝑛]
1 −∝ 𝑘=0
∝ 𝑠 = ∝ +∝2 + ⋯ +∝𝑛+1 𝑠 −∝ 𝑠 = 1−∝𝑛+1
 The Representation of Continuous-Time Signals in Terms of Impulses
𝑥(𝑡) is approximated in
𝑥(𝑡)
terms of pulses or staircase Defining:

1ൗ 0≤𝑡≤Δ
𝛿Δ 𝑡 = ൝ Δ
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

𝑡 1 0≤𝑡≤Δ
Δ𝛿Δ 𝑡 = ቊ
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
𝑥ො 𝑡 : the approximation of 𝑥(𝑡)
At 𝑡 = 0 Δ𝛿Δ 𝑡 has a unit amplitude
𝑥(0) 0 ≤ 𝑡 ≤ Δ
𝑥ො 0 = 𝑥(0)Δ𝛿Δ 𝑡 = ቊ
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
The complete pulse/staircase approximation of
𝑥(𝑡)is the sum
At 𝑡 = Δ
𝑥(Δ) Δ ≤ 𝑡 ≤ 2Δ
𝑥ො Δ = 𝑥(Δ)Δ𝛿Δ 𝑡 − Δ = ቊ 𝑥ො 𝑡 =∙∙∙ +𝑥ො −Δ + 𝑥ො 0 + 𝑥ො Δ +∙∙∙
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
∞

In general 𝑥ො 𝑡 = ෍ 𝑥(𝑘Δ)𝛿Δ 𝑡 − 𝑘Δ Δ
At 𝑡 = 𝑘Δ
𝑘=−∞
𝑥(𝑘Δ) 𝑘Δ ≤ 𝑡 ≤ (𝑘 + 1)Δ
𝑥ො 𝑘Δ = 𝑥(𝑘Δ)Δ𝛿Δ 𝑡 − 𝑘Δ = ቊ
0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
 The Representation of CT Signals in Terms of Impulses2
If we let Δ approach 0 𝑥ො 𝑡 becomes closer and closer and equals 𝑥(𝑡) in the limit of 0

𝑥 𝑡 = lim 𝑥ො 𝑡 = lim ෍ 𝑥(𝑘Δ)𝛿Δ 𝑡 − 𝑘Δ Δ

Δ→0 Δ→0
𝑘=−∞

In the limiting case the sum approaches integral (area):

Δ → 0; 𝛿Δ 𝑡 → 𝛿(𝑡)
∞
∞ ∞
෍ ∙∙∙ Δ → න ∙∙∙ 𝑑𝜏 Consequently: 𝑥 𝑡 =න 𝑥 𝜏 𝛿 𝑡 − 𝜏 𝑑𝜏
𝜏=−∞ 𝜏=−∞
𝑘=−∞
 The Convolution Integral
• The impulse response ℎ(𝑡) of a continuous-time LTI system 𝑆 𝛿(𝑡) ℎ(𝑡)
CT LTI System
ℎ 𝑡 = 𝑆{𝛿 𝑡 } Sum (Integral) of weighted shifted impulses
∞ linearity ∞
𝑦 𝑡 =𝑆 𝑥 𝑡 =𝑆 න 𝑥 𝜏 𝛿 𝑡 − 𝜏 𝑑𝜏 = න 𝑥 𝜏 𝑆{𝛿 𝑡 − 𝜏 } 𝑑𝜏
For the input 𝑥(𝑡): 𝜏=−∞ 𝜏=−∞
weight impulse
• Since the system is time-invariant: ∞
Time-invariance 𝑦 𝑡 =න 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 = 𝑥(𝑡) ∗ ℎ(𝑡)
𝑆 𝛿 𝑡−𝜏 = ℎ(𝑡 − 𝜏)
𝜏=−∞

Example 1
Let, the input x(t) to an LTI system with unit impulse
response ℎ 𝑡 be given as 𝑥 𝑡 = 𝑒 −𝑎𝑡 𝑢(𝑡) for 𝑎 > 0
and ℎ 𝑡 = 𝑢(𝑡). Find the output 𝑦 𝑡 of the system.
∞
𝑦 𝑡 = 𝑥(𝑡) ∗ ℎ(𝑡) = න 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 𝑥 𝑡 ≠ 0 𝑓𝑜𝑟 𝑡 ≥ 0
∞ 𝜏=−∞ ℎ 𝑡 = 𝑢(𝑡).
= න 𝑒 −𝑎𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 𝑓𝑜𝑟 𝑡 > 0 ℎ 𝑡−𝜏 =ቊ
1 0<𝜏<𝑡
0 𝜏>𝑡
0 Thus, for all t, we can write
𝑡 𝑡
1 −𝑎𝜏 1 1
= න 𝑒 −𝑎𝜏 𝑑𝜏 = 𝑒 ቤ = 1 − 𝑒 −𝑎𝑡 𝑦 𝑡 = 1 − 𝑒 −𝑎𝑡
0 −𝑎 0
𝑎 𝑎
 The Convolution Integral2
Example 2 𝑥 𝜏 = 𝑒 2𝑡 𝑢(−𝜏)
𝑥 𝑡 = 𝑒 2𝑡 𝑢(−𝑡)
Find 𝑦 𝑡 = 𝑥(𝑡) ∗ ℎ(𝑡) , where
ℎ 𝑡 = 𝑢(𝑡 − 3)

∞
The system response is 𝑦 𝑡 = ‫=𝜏׬‬−∞ 𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
these two signals have regions of nonzero overlap 𝜏

For 𝒕 − 𝟑 ≤ 𝟎: nonzero overlap for −∞ ≤ 𝜏 ≤ 𝑡 − 3 ℎ 𝑡−𝜏

𝑡−3
1
𝑦 𝑡 =න 𝑒 2𝜏 𝑑𝜏 = 𝑒 2(𝑡−3) For 𝑡 ≤ 3
−∞ 2

For 𝒕 − 𝟑 ≥ 𝟎: nonzero overlap for −∞ ≤ 𝜏 ≤ 0 𝜏

0 𝑦 𝑡
1
𝑦 𝑡 = න 𝑒 2𝜏 𝑑𝜏 = For 𝑡 ≥ 3
−∞ 2

1 2(𝑡−3)
𝑒 For 𝑡 ≤ 3
𝑦 𝑡 =൞ 2
1ൗ For 𝑡 ≥ 3 𝑡
2
 𝑦 𝑡 =න
∞
𝑥 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏
The Convolution Integral3
𝜏=−∞

Convolution

Shift ℎ 𝑡 − 𝜏

Aggregate the overlapping

𝒕<𝟎 No overlapping
𝑦 𝑡 =0
𝑡
𝑦 𝑡 = න 2 ∙ 1 𝑑𝜏 0<𝜏<𝑡
0 𝟎<𝒕<𝟐
𝟎<𝒕<𝟐 𝑡
𝑦 𝑡 = 2 𝜏ቚ
0

𝑦 𝑡 = 2𝑡
 The Convolution Integral4

𝟐<𝒕<𝟒

𝟒<𝒕<𝟔

No-overlap 𝑦 𝑡 = 0
𝒕>𝟔
𝟎 𝟏 𝟐 𝟑 𝟒 𝟓 𝟔 𝟕
𝒕−𝟒 𝒕
 Properties of LTI Systems
• The characteristics of an LTI system are completely determined by its impulse response. This property holds in
general only for LTI systems only.

• The unit impulse response of a nonlinear system does not completely characterize the behavior of the system.

Consider a discrete-time system with unit impulse response:

If the system is LTI, we get the system output (by convolution):

There is only one such LTI system for the given h[n].

However, there are many nonlinear systems with the same response, h[n]. Two different Non-Linear systems
with same impulse response

2 1, 𝑛 = 0,1
ℎ 𝑛 = 𝛿 𝑛 +𝛿 𝑛−1 ℎ 𝑛 =ቊ
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
1, 𝑛 = 0,1
ℎ 𝑛 = 𝑀𝑎𝑥 𝛿 𝑛 , 𝛿 𝑛 − 1 ℎ 𝑛 =ቊ
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
 1 Commutative Property

Proof: (discrete domain)

Put 𝑟 = 𝑛 − 𝑘 ⇒ 𝑘 = 𝑛 − 𝑟

Similarly, we can prove it for continuous domain.

 2 Distributive Property 𝑥 𝑛 ∗ ℎ1 𝑛

Convolution is distributive over addition,

in discrete time
𝑥 𝑛 ∗ ℎ2 𝑛

𝑥 𝑛 ∗ ℎ1 𝑛 + 𝑥 𝑛 ∗ ℎ2 𝑛
in continuous time

Example:
and

x[n] in nonzero for entire n, so direct convolution is difficult. Therefore, we will use commutative property.

1𝑓𝑜𝑟 𝑘 ≥ 0 1𝑓𝑜𝑟 𝑘 ≤ 𝑛

𝑙 = −𝑘 𝑙 =𝑚−𝑛
𝑛 ∞ 𝑙 ∞ 𝑚−𝑛 ∞ 𝑚
𝑘
1 1 𝑛
1
෍ 2 =෍ =෍ =2 ෍ = 2𝑛+1
𝑘=−∞ 𝑙=−𝑛 2 𝑚=0 2 𝑚=0 2

1𝑓𝑜𝑟 𝑘 ≤0 1𝑓𝑜𝑟 𝑘 ≤ 𝑛
 3 Associative Property

in continuous time

In discrete time

Proof:
 4 LTI Systems With and Without Memory
A system is memory-less if its output at any time depends only on the value of the input at that same time.

System output: 𝑦[𝑛] depends on only 𝑥[𝑛] only if k = 𝑛, so for ℎ 𝑛 = 0 for 𝑛 ≠ 0

impulse response 𝑥 𝑛 = 𝛿[𝑛]

A discrete-time LTI system can be memory-less if only: 𝑦 𝑛 = 𝑥 𝑛 ℎ 0 = 𝐾𝛿[𝑛]

Thus, the impulse response have the form: ℎ 𝑛 = 𝐾𝛿 𝑛 , with 𝐾 = ℎ 0 is a constant

the convolution sum reduces to

If k = 1, then the system is called identity system.

Similarly for continuous LTI systems.

a continuous-time LTI system is memory-less if
 5 Invertibility of LTI Systems
A system is invertible only if an inverse system exists

The system with impulse response ℎ1 (𝑡) is inverse of

the system with impulse response ℎ 𝑡 if:

Example:
𝑡0 > 0 delay
Consider the LTI system consisting of a pure time shift 𝑦 𝑡 = 𝑥(𝑡 − 𝑡0 ) 𝑡0 < 0 advance
The impulse response for the system (for 𝑥 𝑡 = 𝛿(𝑡)): ℎ 𝑡 = 𝛿(𝑡 − 𝑡0 ) impulse response 𝑥(𝑡) = 𝛿(t)

the system’s output (the convolution): 𝑦 𝑡 = 𝑥 𝑡 ∗ ℎ 𝑡 = 𝑥 𝑡 ∗ 𝛿 𝑡 − 𝑡0 = 𝑥(𝑡 − 𝑡0 )

To recover the input from the output (invert the system), all that is required is to shift the output back.

The inverse system impulse response: ℎ1 𝑡 = 𝛿(𝑡 + 𝑡0 )

then ℎ 𝑡 ∗ ℎ1 𝑡 = 𝛿 𝑡 − 𝑡0 ∗ 𝛿 𝑡 + 𝑡0 = 𝛿(𝑡) identity system (𝑦 𝑡 = 𝑥 𝑡 ∗ 𝛿 𝑡 = 𝑥(𝑡))

 Invertibility of LTI Systems: Example 2
Consider an LTI system with impulse response: ℎ 𝑛 = 𝑢[𝑛] 𝑢 𝑛 − 𝑘 = 0 𝑓𝑜𝑟 𝑘 > 𝑛
summer or accumulator

∞ 𝑛
Response of this system (convolution sum): 𝑦 𝑛 =෍ 𝑥 𝑘 𝑢[𝑛 − 𝑘] 𝑦 𝑛 =෍ 𝑥𝑘
𝑘=−∞ 𝑘=−∞
first difference equation
𝑛−1
Inverse system
𝑦 𝑛 =𝑥 𝑛 +෍ 𝑥𝑘 𝑦 𝑛 = 𝑥 𝑛 + 𝑦[𝑛 − 1] 𝑥 𝑛 = 𝑦 𝑛 − 𝑦[𝑛 − 1] 𝑦 𝑛 = 𝑥 𝑛 − 𝑥[𝑛 − 1]
𝑘=−∞

Impulse response (𝑥 𝑛 = 𝛿[𝑛]): ℎ1 𝑛 = 𝛿 𝑛 − 𝛿[𝑛 − 1]

Verification: ℎ 𝑛 ∗ ℎ1 𝑛 = 𝛿 𝑛

ℎ 𝑛 ∗ ℎ1 𝑛 = 𝑢[𝑛] ∗ 𝛿 𝑛 − 𝛿[𝑛 − 1]
= 𝑢[𝑛] ∗ 𝛿 𝑛 − 𝑢[𝑛] ∗ 𝛿[𝑛 − 1]
ℎ 𝑛 ∗ ℎ1 𝑛 = 𝛿 𝑛
= 𝑢[𝑛] − 𝑢[𝑛 − 1]
=𝛿 𝑛 the impulse responses are inverses of each other
 6 Causality of LTI Systems
• The output of a causal system depends only on the present and past values of the input to the system.
∞

𝑦 𝑛 = ෍ 𝑥 𝑘 ℎ[𝑛 − 𝑘] 𝑦 𝑛 must not depend on 𝑥 𝑘 for 𝑘 > 𝑛 to be causal

𝑘=−∞

Therefore, for a discrete-time LTI system to be causal: 𝑥 𝑘 ℎ 𝑛 − 𝑘 = 0 for 𝑘 > 𝑛 ℎ 𝑛 − 𝑘 = 0 for 𝑘 > 𝑛
𝑓𝑜𝑟 𝑘 > 𝑛 → 𝑛 − 𝑘 < 0 ℎ 𝑛 =0 for 𝑛 < 0
Causality for LTI system is equivalent to the condition of initial rest (output must be 0 before applying the input)
𝑓𝑜𝑟 𝑘 > 𝑛 ℎ 𝑛 − 𝑘 = 0 𝑓𝑜𝑟 𝑘 < 0 ℎ 𝑘 = 0

Both the accumulator (ℎ 𝑛 = 𝑢[𝑛]) and its

inverse (ℎ 𝑛 = 𝛿 𝑛 − 𝛿 𝑛 − 1 ) are causal.

Inverse system of the accumulator

• Similarly, for a continuous-time LTI system to be causal:
 7 Stability of LTI Systems
• A system is stable if every bounded input produces a bounded output (BIBO).

Consider, an input x[n] to an LTI system that is bounded in magnitude:

Suppose that we apply this to the LTI system with impulse response h[n].
We take 𝑥 𝑛 = 𝐵

Example:
Therefore, if
An LTI system with pure time shift is stable.

The system is stable if the impulse response ℎ 𝑛 is

absolutely summable.
• Similar case in continuous-time LTI system. An accumulator (DT domain) system is unstable.

the system is stable if the impulse response is

absolutely integrable. Similarly, an integrator (CT domain) system is unstable.
 8 Unit Step Response of An LTI System
• the unit step response, 𝑠[𝑛] or 𝑠(𝑡), the output corresponding to the input 𝑥 𝑛 = 𝑢[𝑛] or 𝑥 𝑡 = 𝑢 𝑡 .
• it is worthwhile relating the unit step response to the impulse response
commutative property Response to the input h[n] of a
LTI system with unit impulse
response u[n].
impulse response of an accumulator
𝑢[𝑛] is the unit impulse response of the accumulator. 𝑛
1 𝑛≥0
ℎ𝑛 =෍ 𝛿[𝑘] = ቄ = 𝑢[𝑛]
𝑘=−∞ 0 𝑛<0
 LTI Systems Described by Differential Equation
(Linear Constant-Coefficient Differential Equation)
𝑁 𝑀
A general Nth-order linear constant-coefficient differential equation 𝑑 𝑘 𝑦(𝑡) 𝑑 𝑘 𝑥(𝑡)
that relates the input 𝑥(𝑡) to the output 𝑦(𝑡) is given by ෍ 𝑎𝑘 = ෍ 𝑏𝑘
𝑑𝑡𝑘 𝑑𝑡𝑘
𝑘=0 𝑘=0

Example1:
consider a first-order differential equation

where the input to the system is:

𝑦𝑝 (𝑡) the particular solution

The complete solution is
𝑦ℎ (𝑡) The homogeneous solution,
 Finding the particular solution 𝑦𝑝 (𝑡) (signal of the
 Finding the homogeneous solution(hypothesize a solution)
same form as the input)
forced response 𝑦𝑝 𝑡 = 𝑌𝑒 3𝑡 𝑦ℎ 𝑡 = 𝐴𝑒 𝑠𝑡 Determine 𝑠 and A
Determine 𝑌
From differential equation:
From differential equation:
𝐾 𝑠𝐴𝑒 𝑠𝑡 + 2𝐴𝑒 𝑠𝑡 = 0 ⇒ 𝐴 𝑠 + 2 𝑒 𝑠𝑡 = 0 ⇒ 𝑠 = −2
3𝑌𝑒 3𝑡 + 2𝑌𝑒 3𝑡 = 𝐾𝑒 3𝑡 ⇒ 3𝑌 + 2𝑌 = 𝐾 ⇒ 𝑌 = 𝑦ℎ 𝑡 = 𝐴𝑒 −2𝑡
5
𝐾 Complete
𝑦𝑝 𝑡 = 5 𝑒 3𝑡 , 𝐾 𝑟𝑒𝑎𝑙 𝑎𝑛𝑑 𝑡 > 0
solution:
 Example_contd
• To find A suppose that the auxiliary condition is 𝑦 0 = 0, i.e. , at t = 0, 𝑦 𝑡 = 0
Using this condition into the complete solution, we get:
𝐾 3𝑡
𝑦 𝑡 = 𝑒 + 𝐴𝑒 −2𝑡 With 𝑦 0 =0
5

Example2:
1
Find 𝑦 𝑛 of the system with the difference equation 𝑦 𝑛 − 𝑦 𝑛 − 1 = 𝑥[𝑛]
2
1
We have the output 𝑦 𝑛 = 𝑥 𝑛 + 𝑦 𝑛−1
2
Consider the input 𝑥 𝑛 = 𝑘 𝛿[𝑛] and initial condition 𝑦 −1 = 0 (rest)
1
𝑦 0 = 𝑥 0 + 𝑦 −1 = 𝑘
2
1 1 Setting 𝑘 = 1 we obtain the impulse response for the system
𝑦 1 =𝑥 1 + 𝑦 0 = 𝑘 𝑛
2 2 2 1
1 1 ℎ𝑛 = 𝑢[𝑛]
𝑦 2 =𝑥 2 + 𝑦 1 = 𝑘 2
2 2
⋮ 𝑛 impulse response with infinite duration
1 1
𝑦 𝑛 =𝑥 𝑛 + 𝑦 𝑛−1 = 𝑘 infinite impulse response ( IIR) systems.
2 2
 Block Diagram Representations of Systems
systems described by linear constant -coefficient difference and differential equations can be represented in terms
of block diagram interconnections of elementary operations (adder, scaler, unit delay).

adder scaler
unit delay
Example:
Consider the causal system described Consider the causal continuous−time system
by the first-order difference equation described by a first−order differential equation
𝑦 𝑛 + 𝑎 𝑦 𝑛 − 1 = 𝑏 𝑥[𝑛] 𝑑𝑦 𝑡
+ 𝑎𝑦 𝑡 = 𝑏 𝑥(𝑡)
𝑑𝑡