Analysis of DT Linear Invariant Systems Using Convolution

Discrete-Time Signals & Systems
Analysis of DT Linear Invariant Systems

5
In this section we will analyze the Linear Time-Invariant
(LTI) systems.
System Analysis Techniques:
Two methods are presented in for analyzing the response of a
system to a given input:

Direct Solution of the Input-Output Equation (or
difference equation).
Signal Decomposition (Convolution).
5
System Input output equation:
The general input output equation for any system:
where F[ ] denotes some functions of the quantities.
We can rewrite the general input output equation as
( )
( )
( )
( ) ( )
1
N M
k k
k k L
y n a n y n k b n x n k
= =
= +

( )
( )
( )
( ) ( )
1 0
N M
k k
k k
y n a n y n k b n x n k
= =
= +

where a
k
and b
k

are constant
parameters
If the system is casual L = 0, so the equation:
This system is linear , time-variant system. This
system called Adaptive Linear System.
Note that both a and b
vary with time
5
( ) ( )
( )
1 0
N M
k k
k k
y n a y n k b x n k
= =
= +

If a and b are constant over time, then the previous equation
further simplifies into the general equation for causal, Linear,
Time-Invariant (LTI) Systems:
Linear Time-Invariant (LTI) Systems:
Note: a
k
and b
k
are independent of time (n) or simply
constant for all time n.
Input, Output Equation is called Difference Equation.
The order for the LTI system is N.
5
System Interconnections:
C
a
s
c
a
d
e

P
a
r
a
l
l
e
l

y(n) = T
2
{y
1
(n)} = T
2
{T
1
{x(n)}}
y
1
(n) = T
1
{x(n)}
T
c
= T
2
T
1
T
1
T
2
Specifically, for LTI Systems:
T
2
T
1
= T
1
T
2
y
1
(n) = T
1
{x(n)} y
2
(n) = T
2
{x(n)}
y
3
(n) = y
1
(n)+y
2
(n) = T
1
{x(n)} + T
2
{x(n)}
y(n) = (T
2
+ T
1
)x(n)

T
p
= (T
2
+ T
1
)

5
Non-linear Difference Equation:
Difference Equations can also describe non linear systems:
( ) ( )
y n A x n B = +
( )
( )
y n x n =
( ) ( ) ( )
2
2 3 y n x n x n =
( )
( ) ( )
( )
3 log y n x n x n = +
5
Signal Decomposition:
Decompose input into weighted basis functions:
Generally we choose the basis
functions and compute C
k
.
( ) ( )
k
k
k
x n c x n =
Decomposition to Impulses:
Decompose input sequence
to sum of unit samples.
( ) { }
3,5,2,1 x n =
( ) { } ( )
1
3,0,0,0 3 x n n o = =
( ) { } ( )
2
0,5,0,0 5 1 x n n o = =
( ) { } ( )
3
0,0,2,0 2 2 x n n o = =
( ) { } ( )
4
0,0,0,1 3 x n n o = =
( ) ( ) ( ) ( ) ( )
1 2 3 4
x n x n x n x n x n = + + +
( ) ( ) ( ) ( ) ( )
3 5 1 2 2 3 x n n n n n o o o o = + + +
Consider
5
Decomposition to Impulses:
( )
( )
( )
3
0 k
x n x k n k o
=
=
( )
( )
( )
k
x n x k n k o
=
=
( ) ( ) ( )
,
k k
c x k x n n k o = =
We can write the following
equation for the previous
expression:
In general form
where
5
Convolution Summation:
( )
( )
( )
k
x n x k n k o
=
=
( )
( )
( )
k
k
y n x k h n
=
=
Consider the following system:

Recall input can be decomposed
as follows:
Therefore:
If T system is Linear:
Impulse Response:
Useless, infinite set of responses:
(dependent on both n and k)
T
y(n)
x(n)
)} ( { ) , ( k n T k n h = o
)} ( { ) ( k n T k n h = o
) ( ) , ( k n h k n h =
5
Convolution Summation:
Impulse Response
Note if system is time invariant
Therefore
and
For LTI systems, the convolution sum is as follows:
( ) ( ) ( ) ( ) ( ) ( ) * ( ) ( ) * ( )
k k
y n h k x n k x k h n k h n x n x n h n

= =
= = = =

5
Convolution Summation Calculation:
Graphical method:
To calculate y(n
0
) using the convolution summation, do
the following steps:
Folding: Fold h(k) about k = 0 to obtain h(-k).
Shifting: Shift h(-k) by n
0
to the right (left) if n
0
is
positive (negative), to obtain h(n
0
k)
Multiplication: Multiply x(k) by h(n
0
k) to obtain the
product sequence x(k)h(n
0
k)
Summation: Sum all the values of the product
sequence x(k)h(n
0
k) to obtain the value of the output
at time n = n
0

5
Graphical Convolution Example
Consider the following input and impulse response
{ }
( ) 1,2,1, 1 h n =
|
{ }
( ) 1,2,3,1 x n =
|
3
0 1 2 3 4
x(k)
k
-1 -2 -3
-4
2
0 1 2 3 4
h(k)
k
-1 -2 -3
-4
Input Sequence Impulse Response
5
Graphical Convolution Example Cont.
2
0 1 2
3 4
h(-k)
k
-1 -2 -3
-4
2
0 1 2
3 4
h(1-k)
k
-1
-2
-3
-4
Folded Impulse Shifted Impulse
3
0 1 2 3 4
x(k)
k
-1 -2 -3 -4
2
0 1 2
3 4
h(1-k)
k
-1
-2
-3 -4
5
Graphical Convolution Example Cont.
4
0 1 2
3 4
x(k)h(1-k)
k
-1
-2
-3 -4
Multiply input by folded shifted h(n):
Product Sequence x(k)h(n-k):
( )
( )
( )
3
0
1 1 1 4 3 8
k
y x k h k
=
= = + + =
Sum of Product Sequence:

5
Mathematical Convolution Example
Consider the following input and impulse response
x(n) = [5 ,2 , 4 , -1] h(n) = [5 , 2 , 4 , -1]
y[0] y[1] y[2] y[3] y[4] y[5] y[6]
25 20 44 6 12 -8 1
h(k)
x(k)
5 2 4 -1
5 25 10 20 -5
2 10 4 8 -2
4 20 8 16 -4
-1 -5 -2 -4 1
y(n) = [25, 20 , 44 , 6 , 12 , -8 , 1]
Note: The convolution of two FINITE-LENGTH sequences is that if
x(n) is of length L
1
and h(n) is of length L
2
, y(n) = x(n)*h(n) will be
of length : L = L
1
+ L
2
-1
5
Convolution Properties
( )
( )
( ) ( ) ( )
k
y n x k h n k x k h n
=
= = -
( ) ( ) ( ) ( )
x k h n h n x k - = -
( )
( )
( ) ( )
k k
x k h n k h k x n k

= =
=

Convolution Symbol:
Convolution is Commutative:
Convolution is Distributive:
( )
( ) ( )
( ) ( ) ( ) ( )
1 2
1 2
h n x n x n h n x n h n x n - + = - + - (

Convolution is Associative:
( ) ( )
( ) ( )
( ) ( )
1 2 2 3
3 1
x n x n x n x n x n x n - - = - - ( (

5
Useful Geometric Summation Formulas
5
Causal Linear Time-Invariant Systems
An LTI system is Causal IFF its impulse
response is 0 for negative values of n.
( )
0, 0 h n for n = <
Convolution sum for Causal, LTI system
reduces to the following:
Also, for Causal LTI:
If the input is Causal, LTI system is Causal
then the Convolution further simplifies to:
( )
( )
( )
0 k
y n h k x n k
=
=
( )
( )
( )
0
n
k
y n x k h n k
=
=
( )
( )
( )
n
k
y n x k h n k
=
=

5
Causal Linear Time-Invariant Systems
Another way of examining Causal LTI Difference Equation
If Future Inputs are not
needed then Causal LTI
system reduces as follows:
( )
( )
( )
k
y n x n k h k
=
=
( )
( )
( )
( )
( )
1
0
future
inputs
k k
y n x n k h k x n k h k

= =
= +

For causality, the reference to future
inputs must be equal to 0, or simply
there must not be future inputs
existent in the systems difference
equation.
If k is negative then x(n-k)
corresponds to future input
references
Therefore:
Then for the causal system the original
equation can be rewritten:
( )
( )
( )
0 k
y n x n k h k
=
=
( )
( )
1
0
future
inputs
k
x n k h k
=
=

5
Non-Causal Systems
Non Causal systems will produce an output BEFORE the input
is applied.
Consider the following non-causal equation:
If the input is applied at n=0, then at n= -1, we
already have an output (before the input is applied):
The output of course is dependant on a future input (which
does not make much sense in real time implementations).
( )
( ) 2 1 y n x n = +
( )
( 1) 2 0 y x =
5
Stability of Linear Time-Invariant Systems
BIBO Stable Defined as:
For all bounded inputs x(n), the outputs
y(n) are bounded if the system is BIBO
stable:
and
Where __ and __ are finite numbers
Reminder Triangular or Schwartz
Inequality:
y
M
x
M
( )
x
x n M s <
( )
y
y n M s <
1 2 1 2
x x x x + = +
1 2 1 2
x x x x + s +
5
Consider the following system:
Note the output is defined as follows:

Assume the input is stable:

The above can then be rewritten as follows:

Therefore:
( ) ( ) ( ) ( ) ( ) ( ) ( )
k
k k
y n h k x n k h k x n k h k x n k

=
= =
= s =

( )
x
x n k M s
( ) ( ) ( )
x
k k
h k x n k h k M

= =
s

( ) ( )
x
k
y n M h k
=
s

T
y(n)=T{x(n)}
x(n)
5
Now, if the system T is stable then the following
holds true:

Multiplying both sides and further simplifying:

Defining new finite integer:

Therefore if the system T is stable and the input is
bounded, then the following relation must be true:
( ) ( )
x y
k
y n M h k M
=
s s
( )
1 1
x y
k
x x
M h k M
M M
=
s
1
h y
x
M M
M

( )
h
k
h k M
=
s <

5
Therefore a stable systems impulse response is
absolutely summable:

That summation may go to infinity (the system will be
unstable) if any of the following occur:
The Summation is Infinite and does not converge
The Summation is finite but contains at least one
term which is infinite
If a system is FIR of length N then h(n) is absolutely
summable (stable):
assuming for all n
( )
h
k
h k M
=
s <
( )
0
N
k
h k
=
<
( )
h n =
5
System with Finite Duration & Infinite
Duration Impulse Response:
If impulse response, h(n), is of finite length, then the system
is categorized as finite impulse response FIR
If impulse response, h(n), is of infinite length, then the system
is categorized as infinite impulse response IIR
Causal FIR System
Recursive systems use previous outputs to
compute present output:
( )
( ) ( ) ( ) ( )
{ }
1 ,..., , 1 ,..., 1 ,... y n H y n x n x n x n = +
( )
( ) ( ) ( )
{ }
, 1 ,..., 1 ,... y n H x n x n x n = +
5
Non-recursive systems do NOT use
previous outputs to compute present
output:
5
Difference Equations:
LTI System is characterized by a linear Constant-Coefficient difference
equation (LCCDE):
Order of difference equation is N
Example:
n=0
Order: N=2
Mathematical Initial Conditions:
( )
( ) ( )
1
M
N
k k
k L
k
y n a y n k b x n k
=
=
= +
( ) ( ) ( ) ( )
5 1
1 2
6 6
y n y n y n x n = +
( ) ( ) ( ) ( )
5 1
0 1 2 0
6 6
y y y x = +
( )
1 y
( )
2 y
and
5
For a given difference equation, there exists as many
initial conditions as the order of the difference
equation (N initial conditions for a difference equation
of order N).
Output sequence computed recursively:
n = 1
n = 2
n = 3
( ) ( ) ( ) ( )
5 1
1 0 1 1
6 6
y y y x = +
( ) ( ) ( ) ( )
5 1
2 1 0 2
6 6
y y y x = +
( ) ( ) ( ) ( )
5 1
3 2 1 3
6 6
y y y x = +
etc.
5
Total Solution:
The procedure for computing the solution of LCCDE is:
y
h
(n) is the homogenous solution
y
p
(n) is the particular solution
Linear Constant-Coefficient difference equation (LCCDE):
( )
( ) ( )
1
M
N
k k
k L
k
y n a y n k b x n k
=
=
= +

5
The Homogenous Solution:
Now set input x(n) = 0 ( )
0
0
N
k
k
a y n k
=
=
Assume exponential solution:

( )
n
y n =
where is constant to
be determined.
Make substitutions into D.E.:
0
0
N
n k
k
k
a

=
=
( )
1 2
1 2 1
... 0
N N N
n N
N N
a a a a

+ + + + + = or
5
From the previous:
The polynomial in parenthesis is called characteristic polynomial
Has N roots:
In practice, typically coefficients (a
1
,a
2
,.a
N
) are real.
( )
1 2
1 2 1
... 0
N N N
n N
N N
a a a a

+ + + + + =
1 2
, ,...,
N

The Homogenous Solution:
( )
1 1 2 2 3 3
...
n n n n
h N N
y n C C C C = + + +
The general form of the homogenous solution is:
5
Recursive Difference Equations Stability:
Assuming N-distinct roots, the homogeneous solution is written as:
If
For as then all the
roots must satisfy the following relation:
( )
1 1 2 2 3 3
...
n n n n
h N N
y n C C C C = + + +
( )
0
h
y n
system is
n
stable

)
( )
0
h
y n
n
1, 1,2,...
i
for all i N < =
5
Recursive Difference Equations Stability Example:
Consider:
Assume an exponential solution:
Where __ is a constant to be determined
Plug into difference equation:

Note above must = 0 (input = 0)
Factor:
( ) ( ) ( ) ( )
5 1
1 2
6 6
y n y n y n x n = +
( )
n
y n =
2 1
1 5
0
6 6
n n n

+ =
2 1
2
5 1
0
6 6
n

(
+ =
(

5
Characteristic Equation:
Eliminate Fractions:
Quadratic Equation:
Solution:

Our Example:
2
5 1
0
6 6
+ =
2
6 5 1 0 + =
2
0 a b c + + =
2
1,2
4
2
b b a c
a
1,2
5 25 24 5 1 1 1
,
12 12 2 3

= = =
5
Two Solutions:
The Solution is Linear Combination of
two solutions:

Note C1 and C2 are arbitrary constants
computed from initial conditions
( ) ( )
1 2
1 1
,
2 3
n n
y n y n
| | | |
= =
| |
\ . \ .
( ) ( ) ( )
1 1 2 2
y n C y n C y n = +
( )
1 2
1 1
2 3
n n
y n C C
| | | |
= +
| |
\ . \ .
5
homogeneous solution
If

If

For
System is stable
( )
h
y n =
( )
h
y n
system is
n
unstable

)
( )
0
h
y n
system is
n
stable

)
( )
1 2
1 1
2 3
n n
y n C C
| | | |
= +
| |
\ . \ .
5
The Particular Solution:
The particular solution y
p
(n) is required to satisfy
the DE for the specific input signal x(n), n>0:
For example if x(n) = a
n
u(n) the particular solution will
be in the form:
5
The Particular Solution:
The particular solution to a DE for different several inputs:
5
Example:
Determine the total solution for n 0 of a DT system characterized
by the following difference equation:
For x(n) = u(n) assuming the initial conditions of
y(-1) = 1 and y(-2) = 0.
Solution:
For x(n) = u(n)
Particular solution:
Substitute this
solution into the DE
5
Example: (cont.)
Homogenous solution:
( )
n
y n =
0 25 . 0
) 2 (
=
n n

0 ) 25 . 0 (
2 ) 2 (
=

n
set
0 ) 5 . 0 )( 5 . 0 ( = +
n
n y ) 5 . 0 ( ) (
1
=
n
n y ) 5 . 0 ( ) (
2
=
) ( ) ( ) (
2 2 1 1
n y A n y A n y
h
+ =
n n
h
A A n y ) 5 . 0 ( ) 5 . 0 ( ) (
2 1
+ =
then
substitute
The homogenous solution
5 . 0
1
=
5 . 0
2
=
5
Example: (cont.)
Total Solution:
The total solution is:
at n = 0 and n = 1
The solution is:
5
Zero-Input & Zero-State Response:
An alternate approach to determining the total
solution of DE.
) ( ) ( ) ( n y n y n y
zs zi
+ =
y
zi
(n) : zero-input response
y
zs
(n) : zero-state response
y
zi
(n) is obtained by solving DE by setting the input x(n) = 0.
y
zs
(n) is obtained by solving DE by applying the specified input
with all initial conditions set to zero (0).
5
Determining Impulse Response:
Given a LTI system in the form of a
difference equation,

For LTI Systems:

For input:
( ) n o ( ) h n
LTI
zero state
( )
( ) ( )
1
M
N
k k
k L
k
y n a y n k b x n k
=
=
= +
( ) ( )
x n n o =
5
Case 1 Non Recursive Systems
For unit impulse input, output:
For non recursive system:
Expanding above:

Impulse Response:

Impulse Response:
Length = M+L+1 (FIR)
( )
( )
M
k
k L
h n b n k o
=
=
( ) ( )
y n h n =
( ) ( ) ( ) ( ) ( ) ( )
1 0 1
... 1 0 1 ...
L M
h n b n L b n b b n b n M o o o o o

= + + + + + + + +
( ) ( ) ( ) ( ) ( )
0 1 1
;... 0 ; 1 ; 1 ;..
M L
h M b h b h b h b h L b

= = = = =
( ) { }
1 0 1
,... , , ,...,
L M
h n b b b b b

=
Impulse response is same as coefficients in difference equation
( )
,
k
h k b for L k M = s s
5
Case 2 Recursive Systems
Procedure for finding Impulse Response given recursive
difference equation:

Find Homogeneous Solution:
( )
1 1 2 2
...
n n n
h N N
y n C C C = + + +
( ) ( )
x n n o =
1 2
, ,...,
N
C C C
let
and compute
5
implementation of DT Systems
Structure for the Realization of LTI Systems:
Here, the LCCDE structure for the realization of systems is
described, additional structures for these system will introduce in
later chapters.
Consider the first-
order system:
Direct Form I Structure
5
The previous system can be viewed as 2 LTI systems in cascade,
the first is a non-recursive system described by the equation:
Whereas the second is a recursive system described by the
equation:
5
If we interchange the order of the recursive and non-recursive
systems, we obtain an alternative structure for the realization of
the system described previously:
5
Direct Form II Structure
Minimizing the 2 delay in the structure form I to 1 delay in
structure form II.
5
In General
| | | |

= =
=
M
0 k
k
N
0 k
k
k n x b
k n y a
| | | | | |

= =
=
M
0 k
k
N
1 k
k
k n x b k n y a n y
D
i
r
e
c
t

F
o
r
m

I

S
t
r
u
c
t
u
r
e

or
5
We can change the
order of the cascade
systems.
| | | | | |
| | | |
=
=
=
+ =
M
0 k
k
N
1 k
k
k n w b n y
n x k n w a n w
5
D
i
r
e
c
t

F
o
r
m

I
I

S
t
r
u
c
t
u
r
e

Analysis of DT Linear Invariant Systems Using Convolution

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Analysis of DT Linear Invariant Systems Using Convolution

Загружено:

Авторское право:

Доступные форматы

Discrete-Time Signals & Systems