Chapter 2
Linear Time-Invariant Systems
Prepared by Dr Nidal Kamel
Introduction
Linearity and time invariance, play a fundamental role in
signal and systems analysis for two reasons:
Many physical processes can be modeled as linear time-invariant
(LTI) systems.
LTI systems can be analyzed intensively providing both insight into
their properties and a set of powerful tools form the core of signal
and system analysis.
Prepared by Dr Nidal Kamel
Introduction
LTI systems are amenable to analysis because it possess the
superposition property.
This fact will allow us to develop complete characterization
of any LTI system in term of its response to unit impulse.
Such representation is referred to as convolution sum in
DT systems and convolution integral in CT systems.
Prepared by Dr Nidal Kamel
DT Systems: The Convolution Sum
Representation of DT Signals in term of impulses
x[1],
x[1] [n 1]
0,
x[0],
x[0] [n]
0,
x[1],
x[1] [n 1]
0,
n -1
n -1
n0
n0
n 1
n 1
x[n] ... x[3] [n 3] x[2] [n 2] x[1] [n 1]
x[0] [n] x[1] [n 1] x[2] [n 2] x[3] [n 3] ...
xk n - k
k -
Prepared by Dr Nidal Kamel
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
x[n]
x k n-k
k -
... x[3] [n 3] x[2] [n 2] x[1] [n 1]
x[0] [n] x[1] [n 1] x[2] [n 2] x[3] [n 3] ...
If h[n] is the output of the system to [n] and hk [n] is the output
to [n-k], then the output of linear system to x[n] is:
y[n] ... x[3]h3[n] x[ 2]h2[n] x[ 1]h1[n]
x[0]h[n] x[1]h1[n] x[2]h2[n] x[3]h3[n] ..
x k h n
k -
Prepared by Dr Nidal Kamel
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
x[n] x[1] [n 1] x[0] [n] x[1] [n 1]
y[n] x[1]h1[n] x[0]h[n] x[1]h1[n]
Prepared by Dr Nidal Kamel
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
The input-output relationship of a linear systems is given as:
x[n]
xk n - k
y[n]
xk h n
k -
k -
If the linear system is also time-invariant, then,
hk [n] h[n k ]
Thus the output of LTI system is given
y[n]
x[k ]h[n k ] x n h n
Prepared by Dr Nidal Kamel
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
Consider an LTI system with h[n] and x[n]
as shown in the figure. Find the system
output, y[n].
y[n]
x[k ]h[n k ]
x[0]h[n 0] x[1]h[n 1]
0.5h[n] 2h[n 1]
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
Prepared by Dr Nidal Kamel
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
y[n]
x[k ]h[n k ]
10
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
1, 0 n 4
x[n]
0, otherwise
n ,
h[n]
0,
0n6
otherwise
Prepared by Dr Nidal Kamel
11
DT Systems: The Convolution Sum
Unit Impulse Response and Convolution Sum of LTI
k 0
k 0
y[n] h[ n k ] nk
k 0
k 0
y[n] h[ n k ] nk
1, 0 k 4
x[k ]
0, otherwise
k ,
h[k ]
0,
y[n]
k n 6
k n 6
h[n k ]
0k 6
otherwise
Prepared by Dr Nidal Kamel
12
nk
CT Systems: The Convolution Integral
Representation of CT Signals in Terms of Impulses
If we define
1
,
(t )
0,
0t
otherwise
Since (t) has unit amplitude, we have
x (t )
x(k ) t k
Prepared by Dr Nidal Kamel
13
CT Systems: The Convolution Integral
Convolution Integral Representation of LTI systems
The approximate representation of
x(t), is given as
x (t )
x(k)
t k
Consequently, the response
linear system is given as
y (t )
of
xk hk (t )
14
CT Systems: The Convolution Integral
Convolution Integral Representation of LTI systems
If the system is linear -time invariant (LTI), then
y (t )
x k h (t k )
As 0 the summation becomes integral
y (t ) x( )h(t )d
x (t ) h (t )
Prepared by Dr Nidal Kamel
15
CT Systems: The Convolution Integral
Convolution Integral Representation of LTI systems
Let x(t) be the input to LTI system with unit impulse response
h(t), where
x(t ) e atu(t ),
a0
h (t ) u(t )
y(t ) x( )h(t )d
t
0 e a d
1
1 e at u(t )
a
16
CT Systems: The Convolution Integral
Convolution Integral Representation of LTI systems
Consider the convolution of the
following two signals:
1,
x (t )
0,
t ,
h (t )
0,
0t T
otherwise
0 t 2T
otherwise
17
CT Systems: The Convolution Integral
Convolution Integral Representation of LTI systems
There are three intervals:
y (t ) x( )h(t )d
0,
1
t2,
2
1
y (t ) Tt T 2 ,
2
3 2
1 2
Tt
T ,
2
2
0,
t0
0tT
T t 2T
2T t 3T
3T t
18
Properties of Linear-Time Invariant System
The LTI systems are represented in terms of their unit impulse
responses.
y[n]
x[k ]h[n k ] x[n] h[n]
y (t ) x h t d x t h t
Prepared by Dr Nidal Kamel
19
Properties of Linear-Time Invariant System
Commutative Property
Convolution is a commutative operation.
x n h n h n x n
h k x n k
x t h t h t x t h x t d
Prepared by Dr Nidal Kamel
20
Properties of Linear-Time Invariant System
Distributive Property
Convolution is distributive over addition:
x n h1 n h2 n x n h1 n x n h2 n
x t h1 t h2 t x t h1 t x t h2 t
Parallel interconnected LTI systems
y(t ) y1 (t ) y2 (t )
x(t ) h1 (t ) x(t ) h2 (t )
x(t ) h1 (t ) h2 (t )
Prepared by Dr Nidal Kamel
21
Properties of Linear-Time Invariant System
Distributive Property
Find the convolution of the following two sequences:
n
1
x[n] u n 2n u n
2
h[n] u[n]
We may use the distributive property
y n x1[n] x2 [n] h[n]
x1[n] h[n] x2 [n] h[n]
= y1 n y2 n
Prepared by Dr Nidal Kamel
22
Properties of Linear-Time Invariant System
Distributive Property
n
k 0
k 0
y1[n] x[k ]
y2 [ n ]
x[k ]
k
0
x[k ]
1 k
2
k
2
for n 0
k
0
k
2
= 2 for n 0
y[n] y1[n] y2[n]
23
Properties of Linear-Time Invariant System
Associative Property
Convolution is associative:
x n h1 n h2 n x n h1 n h2 n
x t h1 t h2 t x t h1 t h2 t
Prepared by Dr Nidal Kamel
24
Properties of Linear-Time Invariant System
Systems with and without Memory
System is memoryless if
y[n] bx[n]
This is true if
y t bx t
h[n] b [n]
h t b t
The convolution sum reduces to
y[n] x[n] b [n] bx[n]
y t x t b t bx t
Prepared by Dr Nidal Kamel
25
Properties of Linear-Time Invariant System
Invertibility of LTI Systems
The h(t) system is invertible if an
inverse system h1(t) exists.
The overall impulse response is
h(t ) h1 (t ) (t )
The same applies to DT system
h[n] h1[n] [n]
Prepared by Dr Nidal Kamel
26
Properties of Linear-Time Invariant System
Invertibility of LTI Systems
Consider an LTI with impulse response h[n] = u[n], find the
output of the system.
y[ n ]
x[k ]u[n k ]
x[k ]
This system is invertible and its inverse has impulse response
h1[n] [n] [n 1]
Prepared by Dr Nidal Kamel
27
Properties of Linear-Time Invariant System
Invertibility of LTI Systems
The impulse response of inverse system is
h1[n] [n] [n 1]
We can verify this result by direct calculation:
h[n] h1[n] u[n] [n] [n 1
u[ n]* [ n] - u[ n]* [ n -1]
u[ n] - u[ n -1]
[ n]
Prepared by Dr Nidal Kamel
28
Properties of Linear-Time Invariant System
Causality of LTI Systems
LTI system is causal if
h(t ) 0
for t < 0
h[n] 0
for n 0.
For causal LTI system, the convolution become
y[ n ]
x[k ]h[n k ]
k
t
y ( t ) x ( ) h ( t )
Prepared by Dr Nidal Kamel
29
Properties of Linear-Time Invariant System
Causality of LTI Systems
Both the accumulator and its inverse are causal.
h[n] u[n]
h[n] [n] [n 1]
Causality for linear system is equivalent to the condition of
initial rest.
Initial rest: if the input to causal system is zero up to some point
in time, then the output must also be zero up to the same time.
Initial rest = no input no output
Prepared by Dr Nidal Kamel
30
Properties of Linear-Time Invariant System
Stability of LTI Systems
A system is stable if BIBO
Consider input to LTI that is bounded in magnitude:
x[n] B for all n.
The magnitude of the output is:
h k x n k
y[n]
y[n]
h k x n k
B h[k ]
Prepared by Dr Nidal Kamel
for all n
31
Properties of Linear-Time Invariant System
Stability of LTI Systems
Thus, DT-LTI system is stable if
h[k ]
CT-LTI, the system is stable if
h( ) d
Prepared by Dr Nidal Kamel
32
Properties of Linear-Time Invariant System
Stability of LTI Systems
Find whether pure time shift system is stable or not.
h[n]
n n 1
0
h( ) d ( t0 d 1
thus the system is stable.
Find whether accumulator system, h[n] = u[n] is stable or not.
u[n] u[n]
n0
u ( ) d d
0
thus system is unstable. Prepared by Dr Nidal Kamel
33
Properties of Linear-Time Invariant System
Unit step Response of an LTI
step response is obtained when u[n] is applied at the input
of the system
s[n ] u[n ] h[n ] h[n ] u[n ]
h[k ]u[n k ] h[k ]
Thus h[n] can be recovered from s[n] using the relation
[n] u[n] u[n 1] h[n] s[n] s[n 1]
Prepared by Dr Nidal Kamel
34
Properties of Linear-Time Invariant System
Unit step Response of an LTI
For the CT system, the unit step response is given as:
s (t ) u(t ) h(t ) h(t ) u(t )
h( )u(t )d h( )d
In analogy to the DT part,
(t )
du(t )
ds(t )
h (t )
dt
dt
Prepared by Dr Nidal Kamel
35
Causal LTI Systems Described by Differential
and Difference Equations
Important class of CT systems has input-output relationships in
form of linear constant-coefficient differential equations.
A general Nth-order equation is given by
d k y(t ) M
d k x(t )
ak
bk
k
k
dt
dt
k 0
k 0
N
In case of N = 0, we have
1
y (t )
a0
d k x(t )
bk
k
dt
k 0
M
Prepared by Dr Nidal Kamel
36
Causal LTI Systems Described by Differential
and Difference Equations
auxiliary conditions are required for determination of inputoutput relationship.
Different auxiliary conditions result in different input-output
relationships.
In practical systems we use auxiliary condition of initial rest.
dy(t0 )
d N 1 y(t0 )
y(t0 )
...
0
N 1
dt
dt
Under condition of initial rest, the system is causal and LTI.
Prepared by Dr Nidal Kamel
37
Causal LTI Systems Described by Differential
and Difference Equations
DT systems have input-output relationships in form of linear
constant-coefficients difference equations:
N
a
k 0
y[n k ] bk x[n k ]
k 0
N
1 M
y[n] bk x[n k ] ak y[n k ]
a0 k 0
k 1
This form is called the recursive form.
Prepared by Dr Nidal Kamel
38
Causal LTI Systems Described by Differential
and Difference Equations
When N = 0, we have the following form for input-output:
bk
y[n]
k 0 a0
M
x[n k ]
y[n] h[k ]x[n k ]
k 0
This often called nonrecursive equation.
By direct computation, this system has a finite impulse
response (FIR) of the form:
bn
,
h[n] a0
0,
0nM
otherwise
Prepared by Dr Nidal Kamel
39
Causal LTI Systems Described by Differential
and Difference Equations
Consider the difference equation with x[n] = b[n]. Find y[n].
Suppose we impose condition of initial rest .
1
y[n] y[n 1] x[n]
2
1
y[n] x[n] y[n 1]
2
Condition of initial rest implies x[n] = 0 for n < 0, and y[n] =
0 for n < 0.
Prepared by Dr Nidal Kamel
40
Causal LTI Systems Described by Differential
and Difference Equations
Starting with this initial condition, we may solve for
successive values of y[n] for n 0 as follows:
1
y[1] b,
2
1
1
y[1] x[1] y[0] b,
2
2
y[0] x[0]
1
1
y[2] x[2] y[1] b
2
2
n
1
1
y[n] x[n] y[n 1] b
2
2
System impulse response is
1
h[n] u[n]
2
41
Summary
In this chapter we developed important representations
for LTI systems.
In discrete time we derived the representation of signals
as weighted sums of shifted impulse.
Next, we used this representation to derive the
convolution sum representation for LTI discrete systems.
In continuous time we derived representation of signals as
weighted integral of shifted unit impulse.
Later, this representation is used to derive the convolution
integral for LTI continuous systems.
Prepared by Dr Nidal Kamel
42
Summary
Moreover, the convolution sum and integral provided us
with a means of analyzing the properties of LTI systems,
including causality and stability.
Important class of continuous time systems described by
linear constant-coefficients differential equations, is
discussed.
Discrete time systems represented by linear constantcoefficients difference equations, are also discussed.
Prepared by Dr Nidal Kamel
43