7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 1/32çiloğlu 1
LTI SYSTEMS
CONVOLUTION AND IMPULSE RESPONSE
The Lengths of Input and Output Sequences
Two interpretations of its computation
LINEAR BUT TIME-VARYING SYTEMS
FINITE/INFINITE IMPULSE RESPONSE SYSTEMS (FIR, IIR)
PROPERTIES OF LTI SYSTEMS
Convolution is commutative
Convolution is associative
Cascading LTI Systems
Parallel LTI Systems
BIBO Stability
Causal LTI Systems
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 2/32çiloğlu 2
LTI System
T{.}
y n x nT
x n y n
IMPULSE RESPONSE AND CONVOLUTION
Linearity 1 2 1 2a x n b x n a x n b x n T T T
Time-invariance 0 0 x n y n x n n y n n T T .
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 3/32çiloğlu 3
An input signal can be written as
[] ⋯ + [1][ + 1] + [0][] + [1][ 1] + ⋯
[][ ]=−
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 4/32çiloğlu 4
Using the linearity and time-invariance of the system the output of the LTI system
can be written as
[] []ℎ[ ]=−
ℎ[][ ]
=−
where ℎ[] is the “impulse response” of the LTI system.
This operation is called convolution of
[], and
ℎ[].
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 5/32çiloğlu 5
Derivation [] []
{ [][ ]
=− }
[][ ]=−
[]ℎ[ ]
=−
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 6/32çiloğlu 6
Convolution is shown by a ′ ∗ ′,
[] [] ∗ ℎ[]
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 7/32çiloğlu 7
It is easy to show that
[]ℎ[ ]=− ℎ[][ ]=−
i.e., [] ∗ ℎ[] ℎ[] ∗ []
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 8/32çiloğlu 8
Note that
[] []ℎ[ ]
=−
ℎ[][ ]=− +
[] ∗ ℎ[]
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 9/32çiloğlu 9
Ex:
[] ℎ[]⏟ [] + ℎ[ 1]⏟ [] or equivalently
[] [] + [ 1] + [ 2]
[]
……
ℎ[]
… …
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 10/32çiloğlu 10
Ex: The output can be considered as the superposition of responses to individual
samples of the input.
The response to [2]
The response to [0]
The response to
[3]
The complete response
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 11/32çiloğlu 11
To compute one output sample at a time:
For example, to get [3]
1) multiply the two sequences [] and ℎ[3 ]
2) Add the sample values of this product
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 12/32çiloğlu 12
Ex:
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 13/32çiloğlu 13
THE LENGTHS OF INPUT AND OUTPUT SEQUENCES
Suppose that the input signal and the impulse response have finite durations:
Input
[] : length
Imp. Resp. ℎ[] : length
Then the length of the output signal is + 1.
Output [] : length + 1
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 14/32çiloğlu 14
Ex:
Input
[] : length is 6 samples
Imp. Resp. ℎ[] : length is 3 samples
Output [] : length is 8 samples
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 15/32çiloğlu 15
CONVOLUTION IN MATLAB
>> help conv
conv Convolution and polynomial multiplication.
C = conv(A, B) convolves vectors A and B.
The resulting vector is lengthMAX([LENGTH(A)+LENGTH(B)-1,LENGTH(A),LENGTH(B)]).
If A and B are vectors of polynomial coefficients,
convolving them is equivalent to multiplying the two
polynomials.
C = conv(A, B, SHAPE) returns a subsection of the
convolution with
size
specified by SHAPE:'full' - (default) returns the full convolution,
'same' - returns the central part of the convolution
that is the same size as A.
'valid' - returns only those parts of the convolution
that are computed without the zero-padded edges.
LENGTH(C)is MAX(LENGTH(A)-MAX(0,LENGTH(B)-1),0).
Class support for inputs A,B:
float: double, single
See also deconv, conv2, convn, filter and,
in the signal Processing Toolbox, xcorr, convmtx.
Overloaded methods:
cvx/conv
gf/conv
gpuArray/conv
Reference page in Help browser
doc conv
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 16/32çiloğlu 16
Exercise: Compare “conv” and “filter” commands
Exercise for enthusiastic readers: Study “deconv” command.
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 17/32çiloğlu 17
Play and experiment with code1_LN3.m
clear all close all
h = [2 -3 1]; x = [1 0 0 0 -2];
% h = [2 -3 1 -1 4]; % x = rand(1,5); % uniformly distributed numbers from
[0,1] % x = 2*(rand(1,5)-0.5); % uniformly distributed numbers from
[-1,1]%
% x = rand(1,randi(7,1)) % x = rand(1,randi(7,1)) % uniformly distributed numbers from[0,1], signal length is also random y = conv(x,h)
nh = 0:length(h)-1; nx = 0:length(x)-1; ny = 0:length(y)-1;
mini = min([h x y]);
maxi = max([h x y]);
figure subplot(3,1,1); stem(nh,h,'linewidth',2); v = axis; v = [v(1)-1 length(y) mini maxi]; axis(v) subplot(3,1,2); stem(nx,x,'r','linewidth',2) v = axis;
v = [v(1)-1 length(y) mini maxi]; axis(v) subplot(3,1,3); stem(ny,y,'k','linewidth',2) v = axis; v = [v(1)-1 length(y) mini maxi]; axis(v)
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 18/32çiloğlu 18
and also with code2_LN3.m
clear all close all
h = [2 -3 1 ] x = [1 0 0 0 -2] y = conv(h,x) yy = filter(h,1,x) yyy = filter(x,1,h)
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 19/32çiloğlu 19
LINEAR BUT TIME-VARYING SYTEMS
If the system is linear but time-varying, the derivation on the first page yields
[] []ℎ[]=−
where ℎ[] is the response of the system to an impulse at time k , i.e, [ ].
(time-varying impulse response)
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 20/32çiloğlu 20
LTV systems are used, for example, to model mobile communication channels.
In mobile communication channel impulse response changes due to the motion of
the transmitter and receiver, and also due to the change or moving objects in the
environment.
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 21/32çiloğlu 21
FINITE/INFINITE IMPULSE RESPONSE SYSTEMS
If ℎ[] has a finite number of samples, i.e.,
ℎ[] 0 , < , > , <
then the system is said to be a Finite Impulse Response (FIR) system, otherwise an
Infinite Impulse Response (IIR) system.
FIR
IIR
n
h n
n
h n
1 N
2 N
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 22/32çiloğlu 22
Ex:
FIR or IIR ?
[] 0.3 [] 2 [ 4]
[] 0.7 [ 1] []
[] 0.4 [] 2 [ 4] + 5
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 23/32çiloğlu 23
PROPERTIES OF LTI SYSTEMS
CONVOLUTION IS COMMUTATIVE
[] ∗ ℎ[] ℎ[] ∗ []
[] ∗ ℎ[] []ℎ[ ]=−
Let
[ ]ℎ[]=−
ℎ[] ∗ []
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 24/32çiloğlu 24
CONVOLUTION IS ASSOCIATIVE
([] ∗ []) ∗ [] [] ∗ ([] ∗ [])
Prove!
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 25/32çiloğlu 25
CASCADING LTI SYSTEMS
[] ∗ ℎ[]
[] ([] ∗ ℎ[]) ∗ ℎ[]
[] ∗ (ℎ[] ∗ ℎ[])
[] ∗ (ℎ[] ∗ ℎ[])
([] ∗ ℎ[]) ∗ ℎ[]
If the systems are LTI, the order of cascade can be changed!
h1 h2 x n
h2 h1
x n
y n
n y
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 26/32çiloğlu 26
Ex: Consider the following two systems described by
[] [] + [ 1] + [ 2]
and
[] [ ]=
[] + [ 1] + [ 2] + ⋯
They are LTI (check!), so their order is arbitrary in their cascade connection.
Show that their impulse responses are
ℎ[] [] + [ 3]
[] + [ 1] + [ 2]
ℎ[] []
What is the impulse response of their cascade?
Can you write a recursive description for the second system?
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 27/32çiloğlu 27
Ex: Let two systems be described by
[] [ 2]
and
[] []
Their impulse responses
ℎ[] [ 2] and
ℎ[] [] ‼!
However, one of them is nonlinear so their order cannot be changed in their
cascade.
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 28/32çiloğlu 28
PARALLEL LTI SYSTEMS
is equivalent to
h1[n]+h2[n] x n y n
h2[n]
h1[n]
x n y n+
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 29/32çiloğlu 29
BIBO STABILITY
An LTI system is BIBO stable iff
|ℎ[]|=−
where is a finite constant.
i.e., impulse response is absolutely summable.
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 30/32çiloğlu 30
Proof:
Sufficiency:
|[]| ℎ[][ ]
=− |ℎ[]| [ ]⏟ ≤
=−
|ℎ[]|
=−
so if ∑ |ℎ[]|=− then |[]|
Necessity: (by contradiction)
Assume that the impulse response is not absolutely summable, i.e. n
h n
|ℎ[]|=− → ∞
Also let
[] ℎ[]|ℎ[]| ℎ[] ≠ 0
0 ℎ[] 0
so that [] is bounded.
Now consider
[0] ℎ[][]=− ℎ[] ℎ[]|ℎ[]|
=− |ℎ[]|
=− → ∞
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 31/32çiloğlu 31
FIR (LTI) SYSTEMS ARE ALWAYS STABLE
Why?
7/25/2019 LN3 LTI Systems v1 2015
http://slidepdf.com/reader/full/ln3-lti-systems-v1-2015 32/32
CAUSAL LTI SYSTEMS
An LTI system is causal iff
ℎ[] 0 < 0
Top Related