Suppose $x[n]=[x_0,x_1, \ldots,x_{N-1}]$ and $y[n]=[x_0,0,x_1,0,\ldots,0,x_{N-1}]$. What is the relationship between the Fourier Transforms of both the signals?

When I try for some specific examples in MATLAB, I see that $Y[k]$ has an image of $X[k]$ but how do we show this mathematically?


Let's be more general and consider the insertion of $L-1$ zeroes between each sample. Mathematically, this can be written as $$ y[m]= \begin{cases} x[n] & \text{if } m = nL\\ 0 & \text{otherwise} \end{cases}. $$ The DTFT of $y[n]$ is then given by $$Y(e^{j\Omega}) = \sum_{m=-\infty}^{\infty} y[m]e^{-jm\Omega}.$$ Because $y[m]$ is non-zero only when $m$ is a multiple of $L$, that is $m = nL$, this becomes $$Y(e^{j\Omega}) = \sum_{n=-\infty}^{\infty} x[n]e^{-jnL\Omega} = X(e^{jL\Omega}).$$ From this, you can see that the DTFT of $y$ is the DTFT of $x$ with a period $L$ times smaller (i.e. $2\pi/L$ instead of $2\pi$). Intuitively, this makes sense: you know that "expanding" a signal in the time-domain is equivalent to "squeezing" its spectrum in the frequency domain (see the time scaling property of the Fourier transform).

| improve this answer | |
  • $\begingroup$ Thank you Anapar. Between, instead of DFT in the answer, shouldn't it be DTFT? $\endgroup$ – user7080 Jun 1 '17 at 6:17
  • $\begingroup$ Indeed, that's fixed now. $\endgroup$ – anpar Jun 1 '17 at 6:22

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.