Computer Engineering 4TL4: Digital Signal Processing

Day Class Instructor: Dr. I. C. BRUCE

Duration of Examination: 3 Hours
McMaster University Final Examination December, 2003

This examination paper includes eleven (11) pages and eight (8) questions. You are
responsible for ensuring that your copy of the paper is complete. Bring any discrepancy
to the attention of your invigilator.

Special Instructions: Use of Casio fx-991 calculator only is allowed.

All major questions (numbered 1 to 8) are worth equal marks.
Some equations and tables that may assist you are
provided on pages 4–11.

1. The step response y [ n ] of a causal, stable LTI system is:

y [ n ] = ( 12 ) u [ n ] ,
n

where u [ n ] is the unit step function.

a. Find the z-domain transfer function H ( z ) of this system.

b. Find the impulse response h [ n ] of this system.

c. Find the linear constant coefficient difference (LCCD) equation that describes this system.

2. Consider the causal stable LTI system with the transfer function:
1
H ( z) = .
(1 + 1
2 z −1
)(1 + 1
4 z −1 )

Draw the block diagram for this system implemented in parallel form with 1st-order subsystems.

3. A causal LTI system has the transfer function:

H ( z) =
(1 + 0.2 z )(1 − 9 z ) .−1 −2

1 + 0.81z −2
a. Find transfer functions for a minimum-phase system H1 ( z ) and an all-pass system H ap ( z )
such that:
H ( z ) = H1 ( z ) H ap ( z ) .

b. Sketch the pole-zero plots of H ( z ) , H1 ( z ) and H ap ( z ) , clearly show their ROCs by shading
them on the pole-zero plots.

Continued on page 2
 Page 2 of 11

4. The spectrogram of a speech signal is shown below.

a. Is this a narrowband spectrogram or a wideband spectrogram? Explain your reasoning.
b. This signal is a single word, either “shop” or “posh”. From the spectrogram, which of these
two words does it appear to be? Explain your reasoning.

Frequency (kHz) 8

0
0 0.05 0.1 0.15 0.2 0.25 0.3
Time (s)

5. Two alternative methods of digital IIR filter design make use of the:
i. impulse invariance transformation and
ii. bilinear transformation.

Explain:
a. which of these two methods is applicable to a wider range of filter frequency response types
(i.e., lowpass, highpass, bandpass, bandstop, multiband) and why, and
b. which of these two methods is simpler and why.

6. Four values of a stationary random process x[n] have been observed, giving the discrete-time
sequence v[n] = {0.38, 0.66, 1.17, 0.86}.

a. Use the autocorrelation-based method of spectral estimation to estimate the PSD of x[n], using
a triangular correlation window wc[m] = {0.5, 1, 0.5}.

b. Does x[n] appear to be lowpass, bandpass or highpass?

Continued on page 3
 Page 3 of 11

7. Let hlp [ n ] denote the impulse response of an ideal lowpass filter with unity passband gain and
cutoff frequency ω c = π 4 . The figure below shows an ideal LTI frequency-selective filter h [ n ]
that incorporates hlp [ n ] as a subsystem. Sketch H ( e jω ) , the magnitude frequency response of
h [ n ] , indicating explicitly the band-edge frequencies in terms of ω c and specify whether the
system is a lowpass, highpass, bandpass, bandstop or multiband filter.

(−1)n (−1)n

× hlp[n] ×
x[n] y[n]

h[n]

8. Consider the finite sequence x [ n ] = {1, 0.75, 0.09, −0.06, −1, −0.85} . The DFT of this
sequence X [ k ] has been calculated for 0 ≤ k ≤ 5 and lossy compression has been applied to the
sequence of DFT coefficients X [ k ] by ignoring (i.e., setting to zero) the coefficients for k = 2, 3
and 4, to obtain the sequence of compressed DFT coefficients:
X [ k ] = {−0.07, 1.465 − j 2.3296, 0, 0, 0, 1.465 + j 2.3296} .

a. Compute the reconstructed sequence x [ n ] by taking the inverse DFT of X [ k ] .

b. Calculate the root mean squared (RMS) error between x [ n ] and x [ n ] . Express this RMS error
as a percentage of the RMS of x [ n ] .

THE END

Continued on page 4
 Page 4 of 11

Equations, Tables and Other Information

Unit step function:

Unit impulse function:

Trigonometric identities:

Linear convolution:

N-point circular convolution:

Autoregressive Moving Average (ARMA) difference equation:

Continued on page 5
 Page 5 of 11

Discrete-Time Fourier Transform (DTFT):

z –Transform:

Discrete Fourier Transform (DFT):

Continued on page 6
 Page 6 of 11

Phase delay:

Group delay:

Generalized linear phase system:

Impulse invariance transformation:

Bilinear transformation:

Continued on page 7
 Page 7 of 11

Short-Time Fourier Transform (STFT):

Discrete Short-Time Fourier Transform (Discrete STFT):

Spectrogram:

Periodogram:

Continued on page 8
 Page 8 of 11

Welch Periodogram:

Autocorrelation-based spectral estimator:

Continued on page 9
 Page 9 of 11

Continued on page 10
 Page 10 of 11

Continued on page 11
 Page 11 of 11

DFT properties: