Convolution is such an effective tool that can be utilized to determine a linear time-invariant (LTI) system’s output from an input and the impulse response knowledge.
Given two discrete time signals x[n] and h[n], the convolution is defined by
$x\left[ n \right]*h\left[ n \right]=y\left[ n \right]=\sum\limits_^<\infty >>x\left[ i \right]h\left[ n-i \right]~~~~~~~~~~~~~~~~~~~~~~~\left( 1 \right)$
The summation on the right side is called the convolution sum.
It should be noted that the convolution sum exists when x[n] and h[n] are both zero for all integers n
Since the summation in (2) is over a finite range of integers (i=0 to i=n), the convolution sum exists. Hence any two signals that are zero for all integers n
To compute the convolution (1) or (2)
Suppose that x[n]=a n u[n] and h[n]= a n u[n] Where u[n] is a discrete-time unit-step function and a and b are fixed non zero real numbers.
Step 1:
Change discrete time signal index n to i in both signals:
$h\left[ i \right]=^>u\left[ i \right]$
Step 2:
Flip h[i] to get h[-i]:
$h\left[ -i \right]=^>u\left[ -i \right]$
Step 3:
Shift by n to get h[n-i]:
$h\left[ n-i \right]=^>u\left[ n-i \right]$
Step 4:
Find y[n] by summing the product x[i]h[n-i] over a finite range of i:
$y\left[ n \right]=\sum\limits_^<\infty >>x\left[ i \right]h\left[ n-i \right]~~~~~~~~~~~\left( 3 \right)$
Thus the summation on i in (3) may be taken from i=0 to i=n and the convolution operation is given by,
$y\left[ n \right]=\sum\limits_^>x\left[ i \right]h\left[ n-i \right]$
If a≠b, then standard math relation gives
The convolution operation satisfies a number of useful properties which are given below:
Commutative Property
If x[n] is a signal and h[n] is an impulse response, then
Associative Property
If x[n] is a signal and h1[n] and h2[n] are impulse responses, then
Distributive Property
If x[n] is a signal and h1[n] and h2[n] are impulse responses, then