Solution to Q3 of Week 10 Quiz Pool


a. Note that systems g1 and g2 are Length-2 FIR filters which are of the form –

Qp10q3fir.jpg

$ \begin{align} g1[n] &= a_1g1'[n] + b_1g1'[n-1] \\ g2[n] &= a_1g2'[n] + b_1g2'[n-1] \\ \end{align} $

or, looking at it in terms of impulse responses,

$ \begin{align} g1[n] &= a_1\delta[n] + b_1\delta[n-1] \\ g2[n] &= a_2\delta[n] + b_2\delta[n-1]\\ \\ \text{Given:} \\ g1[0] &= a_1 = (1/2) \\ \\ \text{To Find:} \\ g1[1] &= b_1 \\ g2[0] &= a_2 \\ g2[1] &= b_2 \\ \end{align} $

$ \begin{align} g1[n] = (1/2)\delta[n] + b_1\delta[n-1] \\ g2[n] = a_2\delta[n] + b_2\delta[n-1] \\ \end{align} $

$ \begin{align} G1(\omega) &= \frac{1}{2} + b_1z^{-1} \\ G2(\omega) &= a_2 + b_2z^{-1} \\ \end{align} $

It is also given to us that,
y[n] = x[n-1],
so feeding in $ \delta[n] $ as input (x[n]) would result in $ \delta[n-1] $.
Thus we require:

$ \begin{align} h1[n] * g1[n] + h2[n] * g2[n] &= \delta[n-1] \end{align} $

Taking Z transform,
$ \begin{align} H1(z)G1(z) + H2(z)G2(z) &= z^{-1} \\ \end{align} $
$ \begin{align} \frac{1}{2}(1+z^{-1})(\frac{1}{2} + b_1z^{-1}) + \frac{1}{2} (1- z^{-1}) (a_2 + b_2z^{-1}) &= z^{-1} \end{align} $
$ \begin{align} (1+z^{-1})(\frac{1}{2} + b_1z^{-1}) + (1- z^{-1}) (a_2 + b_2z^{-1}) &= 2z^{-1} \end{align} $
$ \begin{align} \frac{1}{2} + b_1z^{-1} + \frac{1}{2}z^{-1} + b_1z^{-2} + a_2 + b_2z^{-1} - a_2z^{-1} - b_2z^{-2} = 2z^{-1} \end{align} $
$ \begin{align} (\frac{1}{2} + a_2) + (\frac{1}{2} + b_1 + b_2 - a_2)z^{-1} + (b_1 - b_2)z^{-2} &= 2z^{-1} \end{align} $

Solve equation by equating coefficients of $ z^0, z^{-1}, z^{-2} $,

$ \frac{1}{2} + a_2 = 0, a_2 = -\frac{1}{2} $
$ \begin{align}b_1 - b_2 = 0, b_1 = b_2\end{align} $
$ \begin{align}\frac{1}{2} + b_1 + b_2 - a_2 = 2\end{align} $
$ \begin{align}\frac{1}{2} + 2b_1 + \frac{1}{2} = 2\end{align} $
$ \begin{align}2b_1 = 1, b_1 = \frac{1}{2}\end{align} $
$ \begin{align}b_2 = \frac{1}{2}\end{align} $


Therefore our two systems are -
$ \begin{align}g1[n] = \frac{1}{2}g1'[n] + \frac{1}{2}g1'[n-1]\end{align} $
$ \begin{align}g2[n] = -\frac{1}{2}g2'[n] + \frac{1}{2}g2'[n-1]\end{align} $


b.

h[n] = h1[n] * g1[n] + h2[n] * g2[n] = $ \delta[n-1] $
$ H(\omega) = H1(\omega)G1(\omega) + H2(\omega)G2(\omega) $

Taking the Fourier transform,
$ \begin{align} H(e^{j\omega}) &= (\frac{1}{2} + \frac{1}{2}e^{-j\omega}) (\frac{1}{2} + \frac{1}{2}e^{-j\omega}) + (\frac{1}{2} - \frac{1}{2}e^{-j\omega}) (-\frac{1}{2} + \frac{1}{2}e^{-j\omega}) \\ &= (\frac{1}{2} + \frac{1}{2}e^{-j\omega})^2 - (\frac{1}{2} - \frac{1}{2}e^{-j\omega})^2 \\ &= e^{-2j\omega /2}(\frac{e^{j\omega /2} + e^{-j\omega /2}}{2})^2 - e^{-2j\omega /2}j^2(\frac{e^{j\omega /2} - e^{-j\omega /2}}{2j})^2 \\ &= e^{-j\omega}cos^2(\omega /2) + e^{-j\omega}sin^2(\omega /2) \\ &= e^{-j\omega} (cos^2(\omega /2) + sin^2(\omega /2)) \\ &= e^{-j\omega} \end{align} $

Notice that by inverting H($ \omega $), we obtain h[n] = $ \delta[n-1] $ proving our answer in part a.

Taking the magnitude of the H($ \omega $),
$ |H(e^{jw})| $ = 1 (Plot is line with constant value 1)
Phase{$ H(e^{jw} $)} = slope = -1 (Plot is line with slope -1 with value 0 at $ \omega $ = 0)



Back to Lab Week 10 Quiz Pool

Back to ECE 438 Fall 2010 Lab Wiki Page

Back to ECE 438 Fall 2010

Alumni Liaison

Basic linear algebra uncovers and clarifies very important geometry and algebra.

Dr. Paul Garrett