so, here is a very simple model. here

The CCDE is given, but I was trying to derive it on my own and now I am stuck.

first of all, the reverse y[n] when goes through the delayed system, it turns to y[n-M]. then when it passes through the leaky integrator system, $y'[n-M] = (1 - \lambda) * y[n-M] + \lambda * y'[n-M-1]$

combining two, $y[n] = x[n] + \alpha * y'[n-M]$ $ = x[n] + \alpha * (1 - \lambda) * y[n - M] + \alpha * \lambda * y'[n-M-1]$

Now, I am stuck. Can anyone help me with it? Did I understand something wrong?


It's probably more straightforward to first compute the total transfer function in terms of $H(z)$, then plug in the actual expression for $H(z)$, and from that write down the difference equation:

$$Y(z)=X(z)+\alpha z^{-M}H(z)Y(z)\tag{1}$$

From $(1)$ we can derive the total transfer function:

$$\frac{Y(z)}{X(z)}=\frac{1}{1-\alpha z^{-M}H(z)}\tag{2}$$

Plugging $H(z)=(1-\lambda)/(1-\lambda z^{-1})$ into $(2)$ gives

$$\begin{align}\frac{Y(z)}{X(z)}&=\frac{1}{1-\alpha z^{-M}\frac{1-\lambda}{1-\lambda z^{-1}}}\\&=\frac{1-\lambda z^{-1}}{1-\lambda z^{-1}-\alpha(1-\lambda)z^{-M}}\tag{3}\end{align}$$

Eq. $(3)$ is equivalent to

$$Y(z)=\big(1-\lambda z^{-1}\big)X(z)+\big(\lambda z^{-1}+\alpha(1-\lambda)z^{-M}\big)Y(z)\tag{4}$$

from which you can directly see the corresponding difference equation.

  • $\begingroup$ Thanks a lot. Now I understand how to solve similar kind of problems. At least I think so. $\endgroup$ – Rima Feb 24 at 9:37

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.