Are there any signals whose autocorrelation $R(\tau)$ has the following form? Assuming $\tau_c > 0$ and $R_0 > 0$ a constant,

$$R(\tau) = \begin{cases}R_0, \text{ for $|\tau| < \tau_c$} \\ 0, \text{ otherwise}\end{cases}$$


It is not possible for a (WSS) random process $\{X(t)\}$to have an autocorrelation function of the form $$R_X(\tau) = E[X(t)X(t+\tau)] = \begin{cases}\sigma^2, & |\tau|\leq T,\\0, & |\tau| > T. \end{cases}$$

Assuming a zero-mean process for simplicity, note that $R_X(0) = E[X^2(t)] = \sigma^2$ where $\sigma^2$ is the common variance of the random variables. Suppose that the highlighted statement above is true and that $T = 2$ and thus $R_X(1) = E[X(t)X(t+1)] = \operatorname{cov}(X(t), X(t+1))$ also has the same value $\sigma^2$. Then, $X(t)$ and $X(t+1)$ are perfectly correlated random variables and thus are equal in the mean-square sense -- their difference has variance $0$: \begin{align}\operatorname{var}(X(t) - X(t+1)) &= \operatorname{var}(X(t))+\operatorname{var}(X(t+1)) - 2\operatorname{cov}(X(t), X(t+1))\\ &= \sigma^2+\sigma^2-2\sigma^2\\ &=0.\end{align} and thus $X(t) = X(t+1)$ in the mean-square sense. But then, since $t$ is arbitrary, we have that \begin{align}X(t+1) &= X((t+1) + 1) = X(t+2)\\ X(t+2) &= X((t+2) + 1) = X(t+3). \end{align} So, $X(t) = X(t+1) = X(t+2) = X(t+3)$ so that $R_X(3) = E[X(t)X(t+3)]=\sigma^2$ also, contrary to the hypothesis that $R_X(\tau) = 0$ when $|\tau| > 2$.

In summary, $R_X(\tau)$ cannot have a fixed nonzero value in a finite-length interval including the origin and value $0$ outside this finite-length interval.

  • $\begingroup$ +1. Much better than my misguided attempt. $\endgroup$
    – Jason R
    Feb 27 '19 at 4:47
  • $\begingroup$ So, this form of autocorrelation is not acceptable? $\endgroup$ Feb 27 '19 at 22:08
  • $\begingroup$ @ChonglinZhu You are correct. The brick wall autocorrelation you are asking about is not a valid autocorrelation function and there cannot be any signal whose autocorrelation looks like a brick wall. $\endgroup$ Mar 3 '19 at 23:54

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.