I need to calculate the output power spectral density of the following digital filter

enter image description here

My calculations are as follows:





Did I get this equation right? would somebody help me please because this diagram is kind of confusing for me.


1 Answer 1


A digital filter doesn't have power spectral density, a signal has. I guess what you want is the transfer function of this filter, i.e., $H(z)=Y(z)/X(z)$.

Let's add two intermediate signal $u(n)$ and $v(n)$ in the diagram.

enter image description here

We have:

$$ y(n) = v(n-1) + du(n) \tag{1} $$ $$ u(n)=v(n-1)+x(n)\tag{2} $$ $$ v(n) = x(n) -du(n) \tag{3} $$

Take Z-transform and we get $$ Y(z) = V(z)z^{-1} + dU(z) \tag{4} $$ $$ U(z) = V(z)z^{-1}+X(z)\tag{5} $$ $$ V(z)=X(z)-dU(z)\tag{6} $$

Furthermore, we can derive that

$$ V(z)=X(z)-d\Big[V(z)z^{-1} + X(z)\Big] $$

$$ \frac{V(z)}{X(z)} = \frac{1-d}{1+dz^{-1}} $$

$$ Y(z)=V(z)z^{-1} + d\Big[V(z)z^{-1}+X(z) \Big] =(1+d)V(z)z^{-1} + dX(z) $$

Thus, $$ H(z) = \frac{Y(z)}{X(z)} = (1+d) \frac{V(z)}{X(z)} z^{-1} + d = (1+d) \frac{(1-d)z^{-1}}{1+dz^{-1}} +d $$

  • $\begingroup$ very impressive. Thank you very much. This for some reason was very confusing. I see how to solve something like that from now on. Basically to introduce new signals. $\endgroup$
    – JordenSH
    Apr 23, 2021 at 8:21
  • $\begingroup$ So if the signal, has an input power spectral density say $\sigma _x^2$ how would someone find the output power spectral density? I would appreciate it if you provide this update to the answer. $\endgroup$
    – JordenSH
    Apr 23, 2021 at 8:25
  • $\begingroup$ Power spectral density of input signal is $P_{xx} = X(e^{j\omega}) X^*(e^{j\omega})$. Since you already have the transfer function $H(z)$ and also the frequency response $H(e^{j\omega})$, just take $Y(e^{j\omega}) = H(e^{j\omega}) X(e^{j\omega})$ into account and you will see the answer. $\endgroup$
    – ZR Han
    Apr 23, 2021 at 8:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.