1
$\begingroup$

Having $h[n]=u_0[n]+0.8u_0[n-1]+1.6u_0[n-2]$ and $x[n]=(u_1[n]-u_1[n-3])$

The goal is to determine if the system $y[n]=h[n]*x[n]$ is linear.

I know that I would have to test it for Homogenity and Additivity in order to determine if the system is linear or not, but the actual proccess testing it is getting me confused.

My professor did it the following way:

$$y_1[n]=h[n]*x_1[n]$$

$$y_2[n]=h[n]*x_2[n]$$

$$h[n]*(\alpha x_1[n]+\beta x_2[n])$$

$$y[n]=h[n]*\alpha x_1[n]+h[n]*\beta x_2[n]$$

Wasn't this insufficient to prove linearity? Wouldn't we have to test it with the actual values of $x[n]$ and $h[n]$?

$\endgroup$
  • $\begingroup$ Your second set of four equations does not prove anything. It is just the set of constraints that the system needs to satisfy for you to prove the system satisfies superposition (is linear). $\endgroup$ – Peter K. Jan 30 at 16:11
  • $\begingroup$ Convolution itself is a linear operation. And zero input results in zero output here. So there you have it - additivity and homogenity. $\endgroup$ – jithin Jan 30 at 16:28
  • $\begingroup$ @PeterK.That's what I suspected. Then in order to determine if the system is linear I would have to make the substitutions? $\endgroup$ – 1stTimeStackOverflow Jan 30 at 16:28
  • $\begingroup$ Yes, the substitutions would need to be made. Though I'm not sure what $u_0$ nor $u_1$ is. And you'd need an $x_1$ and an $x_2$ not just an $x$. $\endgroup$ – Peter K. Jan 30 at 16:36
1
$\begingroup$

To show that a system

$$y[n] = T\{x[n]\} $$

is linear, you have to show that it's additive and homogeneous which is codified in

$$T\{ a ~x_1[n] + b ~ x_2[n] \} = a ~T\{x_1[n]\} + b ~T\{x_2[n] \} $$ for all complex $a,b$ and all $x_1,x_2$.

The property set above is satisfied for a system given by : $$y[n] = T\{x[n]\} = \sum_{k=-\infty}^{\infty} h[k] x[n-k] = h[n] \star x[n] $$ where $\star$ is the convolution operator and $h[n]$ is some sequence to be defined (incidentally it will be the impulse response of the given LTI system).

Of course you didn't have to prove it, as it's already known that if the input-output relation of a system is given by convolution sum; i.e. $y[n] = h[n] \star x[n]$, then it's by definition an LTI (linear and time-invariant) system. Indeed input output relation of any LTI system is given by the convolution sum...

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.