I have used the energy-type signal autocorrelation function:


I have rewritten the equation as: $$\begin{align} \int_{-\infty}^{\infty}\big[u(t)-u(t-1)\big]\big[u(t+\tau )-u(t+\tau-1)\big]dt \\ \end{align}$$

How do I simplify this equation?

  • 1
    $\begingroup$ This is the deterministic auto-correlation computation of a time-domain signal x(t). Can you interpret the first integral as a convolution of $x(t)$ with $h(t)$ given by $$ y(\tau) = \int_{-\infty}^{\infty} x(t)h(\tau-t)dt $$? What's $h(t)$ in this case? $\endgroup$ – Fat32 Feb 13 at 16:40
  • 3
    $\begingroup$ Whenever the difference of two $u(t)$ is involved it's a good idea to draw it. Than it becomes obvious what the actual integration boundaries need to be. $\endgroup$ – Hilmar Feb 13 at 18:37
  • $\begingroup$ @Hilmar I have tried drawing the graph, I am still unsure of the integration limits. am I right to say that the limits will change as tau goes from negative to positive? $\endgroup$ – Dugong98 Feb 14 at 6:56

Don't make this more complicated than it really is. $x(t)$ is non-zero in the interval $t\in[0,1]$, and $x(t+\tau)$ is non-zero in the interval $t\in[-\tau,1-\tau]$. The integrand is non-zero only if the two functions overlap. There is no overlap for $1-\tau<0$ and $-\tau>1$, i.e., for $|\tau|>1$. So for $|\tau|>1$ the autocorrelation is zero.

For $-1<\tau<1$, we have, according to above considerations, the following integral:



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.