END-OF-YEAR EXAMINATIONS 2002

Unit: ELEC321 Communication Systems (D2)

Day and Time: Monday, 25 November 2002, 1:50 p.m.

Time Allowed: Three hours plus 10 minutes reading time.

Total Number of Questions: SIX (6)

Instructions: Answer any FIVE (5) questions only.

Total marks for this paper: 100.

The questions are of equal value.
Use one or more examination booklets.
Non-programmable electronic calculators may be used.
________________________________________________________________________
 ELEC321 (D2) PAGE 2

1. Fourier Analysis

(a) (8 marks)

Find and sketch the autocorrelation function R11(τ) for

x1(t) = e-at u(t), a > 0

where u(t) is a unit step function.

(b) (8 marks)

Consider a filter with transfer function H ( f) = 1 and an input

1+ j 2πf
x(t) = e-2t u(t), where u(t) is a unit step function.

(i) Find the energy spectrum of the output.

(ii) Show that the normalised total energy output is one-third of the normalised
input energy.

(c) (4 marks)

The transfer function of a particular n-th order low-pass filter is

|H(f)| = 1
1+( f / fc )2n

where fc is the 3dB cutoff frequency.

(i) Show that, as n → ∞ , |H(f)| approaches the characteristic of the ideal

low-pass filter.
(ii) Find n so that |H(f)|2 is constant to within 1 dB over the frequency range
|f| < 0.8 fc.
________________________________________________________________________
 ELEC321 (D2) PAGE 3

2. Analogue Modulation

(a) (10 marks)

Sketch block diagrams that show how you could produce the following signals; give a
mathematical derivation of the operation of each system.

(i) A SSBSC signal using the phasing method;

(ii) A VSB signal.

(b) (10 marks)

Vestigial sideband (VSB) modulation is a compromise between DSB and SSB modulation.
It relaxes the sharp-cutoff requirement of a SSB signal by retaining a trace of the other
sideband in the transmitted signal. Let the Fourier transform of a VSB-modulated signal
sc(t) be

Sc(f) = 1 [M(f-fc) + M(f+fc)]H(f)

2

where M(f) is the Fourier transform of an input message signal m(t), fc is the carrier
frequency, and H(f) is the transfer function of a bandpass filter. Show that, for
distortionless coherent demodulation of a VSB signal, H(f) must satisfy the following
condition:

H(f-fc) + H(f+fc) = K, |f|<B

where K is a constant and B is the bandwidth (in Hz) of the message signal.
________________________________________________________________________
 ELEC321 (D2) PAGE 4

3. Nyquist Sampling Rate, PCM, and Delta Modulation

(a) (2 marks)

State the sampling theorem.

(b) (11 marks)

Find the Nyquist sampling rate for each of the following signals:

(i) x1(t) = 5 cos 1000πt cos 4000πt;

sin 200π t
(ii) x2(t) = 200π t ;
2
 sin 200π t 
(iii) x3(t) =  200π t  .

(c) (4 marks)

A compact-disk (CD) recording system samples each of two stereo signals at 44100
samples per second. Each sample is then uniformly quantised into 16 bits.

(i) Determine the output peak signal-to-rms quantisation noise power ratio for a
full-scale sinusoid.
(ii) Determine the number of bits recorded on a CD if the CD can record an
hour’s worth of music.

(d) (3 marks)

Consider a sinusoidal input test signal x(t) = A sin 2π f m t, where A > 0. If the
sinusoidal signal is applied to a delta modulator with step size k', show that slope overload
will occur if

k' fs < 2π Afm

where fs is the sampling frequency.

________________________________________________________________________
 ELEC321 (D2) PAGE 5

4. Line Coding and Waveform Shaping

(a) (2 marks)

A binary sequence may be encoded to various signal formats (called line codes) for
transmission. Describe the difference between a return-to-zero (RZ) and a non-return-to-
zero (NRZ) waveform format.

(b) (11 marks)

State the encoding rules for the following types of signals:

(i) Bipolar RZ signal;

(ii) Manchester-coded signal;
(iii) Miller-coded signal.

Figure 4.1 shows a Manchester-coded and a Miller-coded signal waveform, where T is

the bit duration and A > 0. Determine the corresponding input binary sequences.

A A
0 t 0 t
-A -A
T (a) Manchester coding T (b) Miller coding
Figure 4.1 (a) Manchester coding and (b) Miller coding.

(c) (7 marks)

In digital communication, eye patterns (diagrams) are widely used as a qualitative/visual

performance indicator of a system. Figure 4.2 shows an eye pattern for binary signals.

A
Decision
0
threshold
-A
Best sampling instant
Figure 4.2 Eye pattern for binary signals.

Sketch the diagram in your answer and locate in it the eye opening, the noise margin of the
system, the amount of amplitude distortion (ISI), and the amount of timing jitter.

Draw an eye pattern for 3-level signals.

________________________________________________________________________
 ELEC321 (D2) PAGE 6

5. Matched-Filter Detection

(a) (12 marks)

A known signal s(t) together with additive white Gaussian noise with two-sided noise
n
0
power spectral density 2 watt/Hz is applied to the input of a matched filter. If the filter
transfer function is H(f) and its output is sampled at time t0 , use Parseval’s Theorem
and Schwarz’s inequality to show that the maximum peak signal-to-noise power ratio at the
output of the matched filter is

|so (t0 )|2

= n2E
N 0

where E is the energy of the input signal s(t), |s o (t0 )| is the amplitude of the output
signal at time t0, and N is the output noise power.

(b) (8 marks)

Find the output of the matched filter and find an expression for the maximum output signal-
to-noise power ratio if the input s(t) is a rectangular pulse of amplitude A and duration
T.
________________________________________________________________________
 ELEC321 (D2) PAGE 7

6. Source and Error-Control Coding

(a) (2 marks)

Write down the Huffman encoding procedure for a discrete memoryless source with
symbols xi and corresponding probabilities pi, i = 1, 2, ..., m.

(b) (8 marks)

Consider a discrete memoryless source with symbols x 1 , x 2 , x 3 , x 4 , and x 5 , and

corresponding symbol probabilities p 1 = 0.4, p 2 = 0.2, p 3 = 0.2, p 4 = 0.1, and p 5 =
0.1. Construct a Huffman code for the source, and calculate the efficiency of the code.

(c) (10 marks)

Consider the binary block code C composed of the following four code words.

C = {(00100), (10010), (01001), (11111)}

(i) What is the number of information bits, k?

(ii) What is the number of parity-check bits, c ?
(iii)What is the minimum distance of this code?
(iv) What is the maximum weight for which the detection of all error patterns is
guaranteed?
(v) What is the maximum weight for which the correction of all error patterns is
guaranteed?
(vi) Is this code linear? Prove your answer.
________________________________________________________________________