I have a homework problem that I am just not sure where to start with. I have to take the Discrete Time Fourier Transform of a signal represented by:

$$x[n] = n^2 a^n u[n]$$

given that $|a| < 1$, $\Omega_0 < \pi$, and u[n] being the unit step function.

There is a hint saying that "Calculus and derivatives will help!" but that actually confuses me more than it helps. However, regardless, I'm just not sure how to even get started evaluating that. I know how to take the DTFT of a signal, but just can't figure out how to get it into a usable form. I also am allowed to use the following conversion:

$$ x[n] = a^nu[n] \iff X(\Omega) = \frac{e^{j\Omega}}{e^{j\Omega} - a}$$

I am not looking for the answer, but rather just some pointers as to how to start manipulating the original signal.

  • 1
    $\begingroup$ I suppose $u$ is the unit step function? You should add its definition to the post. The hint probably means you should first solve the problem with $y[n] = a^n u[n]$ and then use the fact that multiplication by $n$ is related to taking derivative. $\endgroup$ – user147263 Jan 21 '15 at 5:46
  • $\begingroup$ Good point, I updated it with that edit. And I didn't think about it that way, I was trying to do the manipulation first, but nothing was coming out right. But you make a good point that makes more sense now that I think about it more, I'll try that, thanks. $\endgroup$ – JMoser Jan 21 '15 at 5:49
  • $\begingroup$ @Fundamental, your solution suggestion seemed to be the most helpful. I completely forgot about the multiplication by n in the time domain being a derivative in the Fourier domain. I would say this helped me the most and would have liked to have it marked as the answer. But I know that the other answer posted would work as well $\endgroup$ – JMoser Jan 21 '15 at 6:13
  • $\begingroup$ Sorry, don't have the time to write it out. Feel free to post your own answer. $\endgroup$ – user147263 Jan 21 '15 at 6:22
  • $\begingroup$ Oh alright, I wasn't meaning to post the actual answer to the question. Just the method to try to go about solving it. But I can post that. $\endgroup$ – JMoser Jan 21 '15 at 6:25

You haven't given a definition for the discrete Fourier transform but I presume it is something like $$X(\omega) = \sum_{n=-\infty}^{\infty}x(n) e^{-i\omega n}$$ which in this case would be $$X(\omega) = \sum_{n=0}^{\infty}n^2 a^n e^{-i\omega n} = \sum_{n=0}^{\infty}n^2 z^n$$ where $z = ae^{-i\omega}$. Now suppose we set $$f(z) = \sum_{n=0}^{\infty} z^n$$ This series converges provided that $|z| < 1$. Fortunately, we are given that $|z| = |a| < 1$. We can differentiate power series term by term in the interior of their disc of convergence, so $$f'(z) = \sum_{n=1}^{\infty} nz^{n-1}$$ and $$f''(z) = \sum_{n=2}^{\infty}n(n-1)z^{n-2} = z^{-2}\sum_{n=2}^{\infty}n(n-1)z^{n}$$ Try rearranging this to see if you can get an expression for $\sum_{n=2}^{\infty}n^2 z^n$ in terms of $f'(z)$ and $f''(z)$. Then you can manually add the $n=0$ and $n=1$ terms.

You will also want to use the fact that $$f(z) = \frac{1}{1-z}$$ for $|z| < 1$.

  • $\begingroup$ Yes sorry, I didn't provide that definition, but you have provided the correct one. Alright I'll give that a shot as well. Thanks $\endgroup$ – JMoser Jan 21 '15 at 5:54
  • $\begingroup$ I can see where and how this would work out, but I think that the answer provided by Fundamental in the comments seems more straight forward. $\endgroup$ – JMoser Jan 21 '15 at 6:14

To include this as the answer that most helped me, since it was just as a comment to my original post. The most straightforward solution seemed to be by using the property of the Discrete Time Fourier Transform for when you are multiplying by n in the time domain, that corresponds to a derivation in the Fourier/frequency domain.

Therefore, the solution would just be to take the Fourier transform from the provided conversion, and then taking the second derivative of that. This is the solution that I got after doing so:

$$X(\Omega) = \frac{a^3e^{j\Omega} - ae^{j3\Omega}}{(e^{j\Omega} - a)^4}$$


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.