Академический Документы
Профессиональный Документы
Культура Документы
Useful Information
(Not all of this information is necessarily required today!)
Parseval’s Theorem
∞ π DFT Exponential Factor
1 jω 2
x[n] X (e
2
E = = ) dω
2π WN = e − j (2π N )
n = −∞ −π
Z-transform Properties
Z-transform of a sequence x[n] Z 1
∞ x[ − n ] ↔ X (1 / z ), ROC =
Rx
X ( z) = x[ n ] z − n Z
n = −∞
z on x[ n] ↔ X ( z / z o ), ROC = z o R x
Question Marks
1 10
2 14
3 12
4 18
5 14
Total 68
(b) Is your choice for T in part (a) unique? If so, explain why. If not, specify
another choice of T consistent with the information given. [5 pts]
Q2.- (i) Consider the finite-length sequence x1[n] shown below. The 5-point Discrete Fourier
Transform (DFT) of x1[n] is denoted by X1[k]. Plot the sequence y1[n] whose DFT is
0 1 2 3 4 n
–3
(ii) Another sequence x2[n] is shown below. The value of x2[n] at time n = 0 is not known
and is denoted by a. The sequence y2[n], also shown below, is claimed to be the 5-point
circular convolution of x1[n] and x2[n] This is, .
Verify if this claim is correct. If it is, find the value of a. If it is not, explain why. [8 pts]
7
x2[n] y2[n]
a
−1 0 1 2 3 4 n 0 1 2 3 4 n
−2
−3
−4
Q3.- Suppose you have a signal x[n] with 1012 nonzero samples whose discrete-time Fourier
transform you wish to estimate by computing the DFT. You find that it takes your
computer 1 second to compute the 1012-point DFT of x[n]. You then add twelve zero-
valued samples at the end of the sequence x[n] to form a 1024-point sequence x1[n]. Now
the computation of X1[k] takes your computer just 5 milliseconds. Reflecting, you realize
that by using x1[n], you are able to compute more samples of in a much shorter
time by adding some zeros to the end of x[n] and pretending that the sequence is longer.
How do you explain this apparent paradox?
Justify the difference in computational time numerically. [12 pts]
Q4.- For the system represented by the signal flow graph shown in the figure below.
(a) Write the set of 3 difference equations for the node variables w[n], w[n] and y[n]. [5 pts]
w[n] = ____________________________
v[n] = ____________________________
y[n] = ____________________________
(b) Using the z-transform of the difference equations above, find the system transfer
function . [5 pts]
(c) Find the poles and zeros of H(z) and write H(z) in factorized form.
Draw the flow graph of an equivalent system that is the cascade of two first-
order systems. Use direct form II for each of the first-order systems. [5 pts]
Q5.- (i) We wish to design an FIR lowpass filter satisfying the specifications
0.99 < H (e jω ) < 1.01, 0 ≤ ω ≤ 0.375π ,
− 0.005 < H (e jω ) < 0.005, 0.425π ≤ ω ≤ π ,
by applying a window w[n] to the impulse response hd[n] for the ideal discrete-time
lowpass filter with cutoff ωc = 0.4π.
(a) Which of the filters listed in the table below can be used to meet these
specifications? [4 pts]
(b) For each window that you claim will satisfy these specifications, give the
minimum length required for the filter. [4 pts]
(ii) We are interested in implementing a continuous-time LTI lowpass filter H(j) using
the system shown in the figure below, where the discrete-time system has frequency
response Hd (e j).
The sampling time T = 1 ms and the input signal xc(t) is appropriately bandlimited with
Xc(j) = 0 for .
Let the specifications on
0.95 ,
.
Determine the corresponding specifications on the discrete-time frequency response
Hd (e j). [6 pts]