Telecommunications Engineering
Dr. David Tay
Room BG434
x 2529
d.tay@latrobe.edu.au
DT
Formatting and baseband modulation
• Formatting is to ensure that the message is compatible
with digital processing.
• Transforms message into sequence of bits - 0 or 1.
• Topics: character coding, sampling, quantization, PCM
(Pulse Code Modulation)
Baseband Systems
• Simplified system:
DT
• Baseband signals - spectrum from DC to some relatively
small finite frequency.
• Baseband channels - copper wires or coaxial cable.
• Pulse modulation - converts bit streams (0 or 1) into
waveforms suitable for transmission through channel.
DT
Textual Data - character coding
ASCII - transforms text into bits:
EBCDIC - not so popular (see book)
DT
Messages, characters and symbols
Text message - sequence of alphanumeric characters.
Steps in digital transmission:
1. Encode each character into a sequence of bits - bit stream.
2. Group k bits to form a symbol. There are M = 2k
possible symbols.
3. Transmit each symbol with a different waveform - there
needs to be M unique waveforms, i.e. a M -ary system.
Example: different k (M ) can be used to transmit the same
message.
DT
DT
Formatting analog information
Convert analog waveform into digital format.
A bandlimited signal having no spectral components above
fm Hz can be determined uniquely by values sampled at uni-
form intervals of
1
Ts ≤ seconds
2fm
Sampled signal values are;
x(nTs ) for n = 0, ±1, ±2, . . .
Nyquist criterion
1
fs = ≥ 2fm (Nyquist rate)
Ts
Ideal impulse sampling: define periodic impulse train
∞
xδ (t) = δ(t − nTs )
n=−∞
Ideal sampled (Pulse Amplitude Modulated (PAM)) signal
∞
xS (t) = x(t)xδ (t) = x(nTs )δ(t − nTs )
n=−∞
DT
It can be shown that the spectrum of xS (t) is
∞
1
XS (f ) = X(f − nfs )
Ts n=−∞
spectrum repeats itself periodically in frequency every fs Hz.
DT
Example: fs = 2fm
DT
Example: fs > 2fm and fs < 2fm
Natural sampling: instead of ideal impulse function, use a
train of rectangular pulse of non-zero width and finite ampli-
tude for sampling. Still get periodic replication of spectrum
but with scaling of the aliases - see book for more details.
DT
Aliasing
• Overlapped regions of replicated spectrum due to under-
sampling
DT
• Can be avoided by:
1. Increasing sampling rate
DT
2. Prefiltering by anti-aliasing filter before sampling
• In practice filters do not have sharp cut-off.
• Increase sampling rate by around 10% to 20% to account
DT
for the transition bandwidth of filter - Engineer’s Nyquist
criterion
fs ≥ 2.2fm
Example: CD audio. fm = 20kHz gives fs = 44.1 ksam-
ples/sec.
Oversampling: cost effective solution that utilizes the ef-
ficiency of digital signal processing. Essentially use lousy
anti-aliasing filter but can get away with it but using much
higher sampling rate - see book for details.
Interface with digital system: quantization
• After sampling, need quantization to finally convert an
analog signal to digital format.
• Quantization is basically some form of rounding: limit
the allowable amplitude values to be from a finite set:
DT
• If step size q is sufficiently small, the quantization error
e can be assumed to be uniformly distributed between
−q/2 to q/2: p(e) = 1/q for −q/2 < e < q/2.
• Quantization noise variance which is the average noise
DT
power given by:
q/2 q/2
2 2 q221
σ = e p(e)de = e de =
−q/2 −q/2 q 12
• If the analog signal voltage levels is from −Vp to Vp the
voltage range is Vpp = 2Vp . If the number of quantization
levels is L, then
Vpp 2Vp
L= =
q q
• Note that the formula in the book S/N = 3L2 only ap-
plies if one wants peak signal power over average noise
power.
• Non-uniform quantization: use different step sizes for dif-
ferent amplitude levels. Can improve SNR - see book for
details.
Pulse Code Modulation (PCM)
• Represent each discrete quantized ampltitude level with
a unique l bit codeword.
• PCM sequence: the bit stream from sequence codewords
from the quantized levels
DT
• PCM waveform: the analog baseband waveform that is
used to represent the bits in the bit stream.
DT
Baseband Transmission
• PCM transforms analog information waveforms into bits
which are abstract quantities.
• To transmit these bits over a physical communication
channel, need electrical waveforms to physically repre-
sent these bits.
• PCM waveform types (line codes): represent the same
bit sequence but different waveform shapes - see book
for details.
DT
DT
M-ary Pulse Modulation (Multilevel Signalling)
• Instead of binary pulses, use pulses which can have M (=
2k ) number of amplitude (voltage) levels.
• Group k bits together to form a k-bit symbol which is
represented by one pulse.
• If data rate is R bit/sec, symbol rate is
R
RS = symbols/sec
k
• Advantage - reduced bandwidth compared to binary PCM.
Bandwidth proportional to RS .
• Disadvantage - more sensitive to noise and require a
greater signal energy for similar performance to binary
PCM
DT
Compare k = 3 with k = 1.
DT
Example I: Want to transmit 800 characters/sec, where
each character is represented by its 7-bit ASCII codeword,
followed by an eight bit for error detection, per character. A
multilevel PAM waveform with M = 16 levels is used. De-
termine
• Effective transmitted bit rate.
• Symbol rate.
Solution: Bit rate (Rb ) = (character rate) × (bits per char-
acter) = 800 × (7 + 1) = 6400 bits/sec.
Bits per symbol k = log2 M = 4. Symbol rate
Rb 6400
RS = = = 1600 symbols/s
k 4
Example II: Want to transmit an analog information wave-
form with fm = 3 kHz with the following criteria: the quan-
tization distortion must be
e ≤ 1%Vpp
Determine the symbol rate if
• One bit is transmitted at a time.
• Transmit k = 4 bits at a time.
DT
Solution: First determine the require bit-rate to achieve the
criteria.
q Vpp
emax = =
2 2L
Vpp
≤ 0.01Vpp
2L
L ≥ 50
Number of bits for L levels is
l = log2 L ≥ log2 50 = 5.6
Choose l = 6 bits per quantized analog sample.
Using Nyquist criterion, fs = 2fm = 6000 samples/sec. Bit
rate
R = fs l = 36 kbits/sec
• With scheme 1, M = 21 = 2 distinct symbols required.
RS = R = 36, 000 symbols/sec.
• With scheme 2, M = 24 = 16 distinct symbols required.
RS = R/4 = 36, 000/4 = 9000 symbols/sec.
Example III: The maximum frequency of an analog wave-
form is fm = 4000 Hz. A 16-level PAM transmission system
is to be used. The quantization distortion must not exceed
1% of the peak-to-peak analog signal.
DT
• Determine the minimum number of bits per sample that
should be used in the PAM transmission system.
• Determine the required sampling rate and the resulting
bit-rate.
• Determine the symbol rate.
Solution: By definition
Vpp
L=
q
Maximum quantization distortion
q
emax = ≤ 0.01Vpp
2
Therefore
Vpp 1
L≥ = = 50 levels
0.02Vpp 0.02
Number of bits for L levels is
l = log2 L ≥ log2 50 = 5.6
Choose l = 6 bits per quantized analog sample.
Using Nyquist criterion, fs = 2fm = 8000 samples/sec. Bit
rate
Rb = fs l = 48 kbits/sec
DT
M = 16 → k = 4 bits per symbol
Symbol rate
Rb
Rs = = 12, 000 symbols/sec
k
Example IV: In the compact (CD) digital audio system, an
analog signal is digitized so that the ratio of the peak-signal
power to the peak-quantization noise power average power as
is at least 96dB. The sampling rate is 44.1 kilosamples/s.
1. Determine the minimum number of uniform quantization
levels needed for the analog signal.
2. Determine the corresponding number of bits per sample
(assuming a fix codeword length for all levels).
3. Calculate the data rate in bits/s.
Solution:
1. Assume voltage range is from −Vp to Vp , number of quan-
tization levels is L and quantization step size is q. Now by
definition q = 2Vp /L. The peak signal power Speak = Vp2
DT
and the peak quantization power is Npeak = (q/2)2 .
Therefore
Speak Vp2 96/10
Vp2 9.6
≡ = 10 =⇒ = 10
Npeak (q/2)2 Vp2 /L2
√
Minimum number of levels L = 109.6 = 63, 096 levels.
2. Number of bits l = log2 L = log2 63, 096 = 16 bits/sample.
3. Data rate R = 16 bits/sample × 44.1 ksamples/s =
705,600 bits/s (705.6 kbits/s).
Exercise to try yourself from texbook: problems 2.1
(page 101), 2.2 (page 101), 2.3 (page 101), 2.8 (page
102). Solution will be provided at a later date on LMS.
DT
