0
$\begingroup$

I learned about the inverse system. INV

Suppose I have an impulse response $h(t)$ which is a box function. ($A = 1, T = 2$) BOX

If I take the Fourier transform and referring the transform pairs, $$H(j\omega) = 2\operatorname{sinc}(\omega), \\ H(j\omega)H_1(j\omega) = 1, \\ H_1(j\omega) = \frac{1}{H(j\omega)} = \frac{1}{2\operatorname{sinc}(\omega)}$$ where $\operatorname{sinc}(u) \triangleq \frac{\sin(u)}{u}$ , right?

Then, what is the reverse of this Fourier Transform $\frac{1}{2\operatorname{sinc}(\omega)}$?

Or instead of using the Fourier Transform, should I use a different approach to get the inverse system of the box function?

$\endgroup$
  • 3
    $\begingroup$ Note that not all systems are invertible. For your example, consider a cosine with period $T$ as an input signal $x(t)$. Then your box function will yield zero, i.e., $y(t)=0$. Therefore, no matter how you design $h_1(t)$, you can never recover $x(t)$. More generally, almost any system will filter out some frequencies (consider e.g. lowpass filters), which are lost and cannot be recovered afterwards. $\endgroup$ – Florian May 28 at 8:35
  • 3
    $\begingroup$ A general inverse does not exist. $H(j \omega)$ is zero for some values of $\omega$ and inverting this would result in "divide by zero". You can still create "approximate" inverse, but that's more complicated and depends a lot on the specific application. $\endgroup$ – Hilmar May 28 at 12:41
1
$\begingroup$

As pointed out in the comments, the inverse system is not realizable. However, it is possible to write down an expression for the impulse response of the inverse system. We want a system that results in a Dirac impulse when convolved with the given rectangular pulse $h(t)$. This can be done as follows. We delay the rectangular impulse by $T/2$, so it starts at $t=0$. Then we add an infinite number of delayed versions of it such that the result becomes a step function $u(t)$. Each delay is an integer multiple of $T$:

$$u(t)=\sum_{n=0}^{\infty}h(t-T/2-nT)\tag{1}$$

The system that achieves this has an impulse response given by

$$f(t)=\sum_{n=0}^{\infty}\delta(t-T/2-nT)\tag{2}$$

Now we have a step function, from which we can generate a Dirac impulse by taking the derivative. So the inverse system is given by the concatenation of the system with impulse response $f(t)$ given by $(2)$ and an ideal differentiator. The total impulse response of the inverse system is then

$$g(t)=(f\star\delta')(t)=\sum_{n=0}^{\infty}\delta'(t-T/2-nT)\tag{3}$$

where $\delta'(t)$ is the distributional derivative of the Dirac impulse.

Clearly, the system with impulse response $g(t)$ given by $(3)$ is not realizable, but as a theoretical concept it is interesting, just like other ideal systems (e.g., ideal frequency selective filters, differentiators, Hilbert transformers, etc.).

$\endgroup$

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.