0
$\begingroup$

In terms of analog signals, we can represent digital signal as :

$$ x[n] \triangleq x_{a}(nT) = \int_{-\infty}^{\infty}X_{a}(f) \, e^{j2\pi f nT} \ \mathrm{d}f $$

While if we focused on the integral on the right side and according to the Digital Signal Processing by John Proakis, chapter 6.1, we can rewrite it into :

$$ \int_{-\infty}^{\infty}X_{a}(f) \, e^{j2\pi f nT} \ \mathrm{d}f = \sum_{k=-\infty}^{\infty} \int_{(k-1/2)F_{s}}^{(k+1/2)F_{s}}X_{a}(f) \, e^{j2\pi nf/F_{s}} \ \mathrm{d}f $$

where $ F_{s} \triangleq \frac1T $. My question is how the second equation comes up ? what does the interval of $(k-1/2)F_{s}$ to $(k+1/2)F_{s}$ means ?

Furthermore, it is stated in the book that "observing the $X_{a}(f)$ in the interval of $(k-1/2)F_{s}$ to $(k+1/2)F_{s}$ is identical to $X_{a}(f-kF_{s})$ in the interval of $-F_{s}/2$ to $F_{s}/2$". Is there any explanation or derivation how both things are identical ?

Thank you so much, hope that my question is clear enough

$\endgroup$
1
$\begingroup$

Because sampled signals Spectrum will have copies of original spectrum at multiples of $f=F_s$. Even if aliasing happened due to choice of $F_s$, the spectrum of sampled signal $x(n.T_s)$ will be periodic in frequency always with period $F_s$. And you can take any one such $k^{th}$ copy or period and integrate from $\frac{k-1}{2}F_s$ to $\frac{k+1}{2}F_s$. Now if you vary $k$ from $-\infty$ to $\infty$, then it is equivalent to integration on LHS.

Key: Sampling at $F_s$ will create a spectrum which is periodic in frequency with period $F_s$.

$\endgroup$
4
  • $\begingroup$ Thanks sir, I think now I understand, just to clarify, instead of using integral for continuous summing in the interval of $ -\infty $ until $ \infty $, I can just sum a continuously in the interval of $ -\frac{1}{2} F_{s} $ until $ \frac{1}{2} F_{s} $. And because the spectrum outside that interval is just a copy, than I can just repeatly sum it for $ -\infty $ until $ \infty $ $\endgroup$ – Anthony Lauly May 6 '20 at 7:54
  • $\begingroup$ @AnthonyLauly No, you have to integrate the spectrum from $-F_s/2$ to $F_s/2$ because it is continuous in frequency. Also, $k$ indexing is necesaary because the integration has to be summed for $k=-\infty$ to $k=\infty$. $\endgroup$ – DSP Rookie May 6 '20 at 9:43
  • $\begingroup$ yup that is what I mean, I think I chose the wrong sentences to delivered. Btw thank you, I get the point. $\endgroup$ – Anthony Lauly May 6 '20 at 11:35
  • $\begingroup$ @AnthonyLauly If DSP Rookie answered your question, best practice would be to check it off as correct (or if you are looking for better / alternate answers, leave as is). Thanks! $\endgroup$ – Dan Boschen May 6 '20 at 23:16

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.