EEE351, Fall 2022

Principles of Communication
Systems
(Digital Communication Systems)

Dr. Saleem Akhtar

Department of Electrical & Computer Engineering
CUI, Lahore
 Principles of Communication Systems

Resource Persons
• Dr. Saleem Akhtar (Theory)
• Email: [email protected]

• Mr. M. Usman Iqbal (Lab)

• Email: [email protected]

[2]
 Pulse Code Modulation (PCM)

Learning Objectives

– Basic Elements of a PCM System

– Quantization

– Quantization Distortion / Noise and Signal to Quantization Noise Ratio

– The Encoder

– Transmission Bandwidth and its tradeoff with Signal to Quantization

Noise Ratio

[3]
 Pulse Code Modulation
⚫ Basic Elements of a PCM System

[4]
 Pulse Code Modulation
⚫ Basic Elements of a PCM System

Figure 6.13 PCM system diagram.

[5]
 Pulse Code Modulation
 Most useful and widely used of all the pulse modulations.
 PCM is a method of converting an analog signal into a digital signal (A/D
conversion).
 An analog signal’s amplitude can take on any value (infinite) over a
continuous range while digital signal amplitude can take on only a finite
number of values.
 An analog signal can be converted into a digital signal by means of three
steps:
– sampling
– quantizing, that is, rounding off its value to one of the closest
permissible numbers (or quantized levels) as shown in Fig. 6.14.
– Binary coding, that is conversion of quantized samples to 0s and 1s.

[6]
 Pulse Code Modulation
⚫ Quantization

Figure 6.14 Quantization of a sampled analog signal.

[7]
 Pulse Code Modulation
⚫ Quantization
⚫ The sampled amplitudes of the analog signal m(t) lie in the range
−𝑚𝑝 , 𝑚𝑝 , which is partitioned into 𝐿 subintervals, each of magnitude
2𝑚𝑝
∆𝑣 = .
𝐿
⚫ Next, each sample amplitude is approximated by the midpoint value of
the subinterval in which the sample falls (see Fig. 6.14 for L = 16).
⚫ Each sample is now approximated to one of the 𝐿 numbers.
⚫ Thus, the signal is digitized with quantized samples taking on anyone
of the 𝐿 values.
⚫ Such a signal is known as an L-ary digital signal.

[8]
 Pulse Code Modulation
⚫ Encoding
⚫ From practical viewpoint, a binary digital signal (a signal that can take on only
two values) is very desirable because of its simplicity, economy, and ease of
engineering.
⚫ We can convert an L-ary signal into a binary signal by using binary encoding.
⚫ Figure 1.5 shows such a coding for the case of L = 16. This code, formed by
binary representation of the 16 decimal digits from 0 to 15, is known as the
natural binary code (NBC).
⚫ Each of the 16 levels to be transmitted is assigned one binary code of four digits.
The analog signal 𝑚(𝑡) is now converted to a (binary) digital signal.
⚫ A binary digit is called a bit for convenience.
⚫ Thus, each sample in this example is encoded by four bits.
⚫ To transmit this binary data, we can assign a negative pulse to a binary 0 and a
positive pulse to a binary 1 so that each sample is now transmitted by a group of
four binary pulses (pulse code). The resulting signal is a binary signal.

[9]
 Pulse Code Modulation
⚫ Figure 1.5 Example of
PCM encoding
– 𝐿 = 24 = 16
– 𝑛 = log 2 𝐿

[10]
 Pulse Code Modulation
⚫ Example

[11]
 Pulse Code Modulation
⚫ Example: The audio signal of bandwidth about 15 kHz,
– But subjective tests show that signal articulation (intelligibility) is not affected if all the
components above 3400 Hz are suppressed (Recall also the role of Antialiasing Filter).

– Sampling Rate = 8000 samples per second (8 kHz) ⇒ 25% oversamling for practical filters

– Quantization Levels = 256

– Each Quantized Level is encoded into = 8 bits

– Thus, a telephone signal requires 8 × 8000 = 64000 binary pulses per second.

⚫ The music stored on compact disc (CD) is also an application of PCM. This
is a high-fidelity situation requiring the audio signal bandwidth to be 15 kHz.
Although the Nyquist sampling rate is only 30 kHz, the actual sampling rate
of 44.1 kHz is used for the reason mentioned earlier. The signal is quantized
into a rather large number of levels (𝐿 = 65, 536) to reduce the quantizing
error. The binary-coded samples are now recorded on the CD.

[12]
Pulse Code Modulation

13
 Pulse Code Modulation
⚫ Quantization Distortion/Error/Noise
– For quantization, we limit the amplitude of the message signal 𝑚(𝑡) to the range
−𝑚𝑝 , 𝑚𝑝 . Note that 𝑚𝑝 is not necessarily the peak amplitude of 𝑚(𝑡). The
amplitudes of 𝑚(𝑡) beyond ±𝑚𝑝 are chopped off. Thus, 𝑚𝑝 is not a parameter of
the signal 𝑚(𝑡), but is a constant of the quantizer.

– The amplitude range −𝑚𝑝 , 𝑚𝑝 is divided into 𝐿 uniformly spaced intervals,

2𝑚𝑝
each of width ∆𝑣 = . A sample value is approximated by the midpoint of the
𝐿
interval in which it lies . The quantized samples are coded and transmitted as
binary pulses. At the receiver some pulses will be detected incorrectly.

– Hence, there are two sources of error in this scheme: quantization error and
pulse detection error. In almost all practical schemes, the pulse detection error
is quite small compared to the quantization error and can be ignored. In the
present analysis, therefore, we shall assume that the error in the received signal is
caused exclusively by quantization.

[14]
 Pulse Code Modulation
⚫ Quantization Distortion/Error/Noise
– If 𝑚 𝑘𝑇𝑠 is the is the kth sample of the signal 𝑚(𝑡), and if 𝑚ෝ 𝑘𝑇𝑠 is the
corresponding quantized sample
– 𝑞 𝑘𝑇𝑠 is the quantization error in the kth sample
𝑞 𝑘𝑇𝑠 = 𝑚 𝑘𝑇𝑠 − 𝑚 ෝ 𝑘𝑇𝑠
– distortion component 𝑞(𝑡) in the reconstructed signal is thus,
𝑞 𝑡 =𝑚 𝑡 −𝑚 ෝ 𝑡
– The signal 𝑞(𝑡) is the undesired signal, and, hence, acts as noise, known as
quantization noise.

[15]
 Pulse Code Modulation
⚫ Quantization Distortion/Error/Noise

⚫ Calculation of the power or the mean square value of 𝒒(𝒕).

2𝑚𝑝
– The quantum levels are separated by ∆𝑣 = .
𝐿

– Since a sample value is approximated by the midpoint of the subinterval (of

∆𝑣
height ∆𝑣) in which the sample falls, the maximum quantization error is ± 2 .
∆𝑣 ∆𝑣
– Thus, the quantization error lies in the range − , ,
2 2
∆𝒗 𝒎𝒑
– quantization error = = (basic Design Equation for uniform
𝟐 𝑳
quantizer)
∆𝑣 ∆𝑣
– Assuming that the error is equally likely to lie anywhere in the − , and the
2 2
quantization error be denoted by the random variable 𝑄 of sample value 𝑞.

[16]
 Pulse Code Modulation
⚫ Quantization Distortion / Error / Noise
– the mean square quantizing error or power of quantization noise 𝑁𝑞 = 𝑞෪2 = 𝜎𝑞2
is given by

1 Δ𝑣 Δ𝑣
, − <𝑞 ≤
𝑓𝑄 (𝑞) = ൞Δ𝑣 2 2
0, otherwise

Δ𝑣 Δ𝑣
2 1 2
𝜎𝑞2 = න 𝑞 2 𝑓𝑄 (𝑞)𝑑𝑞 = න 𝑞 2 𝑑𝑞
−
Δ𝑣 Δ𝑣 −Δ𝑣
2 2

Δ𝑣 2
=
12

𝑚𝑝2 2𝑚𝑝
= 2 as ∆𝑣 =
3𝐿 𝐿

[17]
 Pulse Code Modulation
⚫ Quantization Distortion / Error / Noise
– Assuming that the pulse detection error at the receiver is negligible, the
reconstructed signal 𝑚ෝ 𝑡 at the receiver output is
𝑚ෝ 𝑡 =𝑚 𝑡 +𝑞 𝑡
– The desired signal at the output is 𝑚(𝑡), and the (quantization) noise is 𝑞(𝑡). In
case 𝑚 𝑡 is a WSS random signal, the power 𝑃 of the message signal 𝑚(𝑡)
∞ ∞
𝑆𝑜 = 𝑃 = 𝑚෫
2 𝑡 =𝑅 𝜏
𝑀 ȁ𝜏=0 = න 𝑆𝑀 𝑓 𝑑𝑓 = න 𝑚2 𝑓𝑀 𝑚 𝑑𝑚
−∞ −∞
𝑚𝑝2
𝑁𝑜 = 𝑁𝑞 =
3𝐿2
𝑆𝑜 𝑃 𝑃
= 3𝐿2 𝑚2 = 3𝐿2 𝑚2 (another Design Equation)
𝑁𝑞 𝑝 𝑝
– In this equation 𝑚𝑝 is the peak amplitude value that a quantizer accept, and is
therefore a constant of the quantizer.
𝑆
– This means 𝑜 , the SQNR, is a linear function of the message signal power 𝑃.
𝑁𝑞

[18]
 Pulse Code Modulation
⚫ The Encoder (Gray Coding)
– The multiplexed PAM output is applied at the input of the encoder, which
quantizes and encode each sample into a group of 𝑛 binary digits.

– The encoding involves answering successive questions, beginning with whether

or not the sample is in the upper or lower half of the allowed range.

– The first code digit 1 or 0 is generated, depending on whether the sample is in the
upper or the lower half of the range.

– In the second step, another digit 1 or 0 is generated, depending on whether the

sample is in the upper or the lower half of the subinterval in which it has been
located.

– This process continues until the last binary digit in the code is generated.

[19]
 Pulse Code Modulation
⚫ The Encoder (Gray Coding)

[20]
 Pulse Code Modulation
⚫ Maximum Information Rate:

– A knowledge of the maximum rate of information that can be transmitted

over a channel of bandwidth 𝐵 Hz is of fundamental importance in digital
communication.

– A maximum of 𝟐𝑩 independent pieces of information per second can be

transmitted, error free, over a noiseless channel of bandwidth 𝑩 Hz.

– In other words, we can transmit error free at most two pieces of

information per second per hertz bandwidth.

– This is one of the fundamental relationships in communication theory.

[21]
 Pulse Code Modulation
⚫ The Transmission Bandwidth
– For a binary PCM, we assign a distinct group of 𝑛 binary digits (bits) to each of
the 𝐿 quantization levels.
– Because a sequence of 𝑛 binary digits can be arranged in 2𝑛 distinct
patterns
𝐿 = 2𝑛 or 𝑛 = log 2 𝐿
– Each quantized sample is, thus, encoded into 𝑛 bits. Because a signal 𝑚(𝑡) band-
limited to 𝐵 Hz requires a minimum of 2𝐵 samples per second, we require a total
of 2𝑛𝐵 bits per second (bps), that is, 2𝑛𝐵 pieces of information per second.
– Because a unit bandwidth (l Hz) can transmit a maximum of two pieces
of information per second , we require a minimum channel of bandwidth
𝐵𝑇 Hz, given by 𝐵𝑇 = 𝑛𝐵 𝐻𝑧.
– This is the theoretical minimum transmission bandwidth required to
transmit the PCM signal. We will see later on that for practical reasons
we may use a transmission bandwidth higher than this.

[22]
 Pulse Code Modulation
⚫ Quantization Distortion/Error/Noise (Revisit …)
– For a binary PCM, we assign a distinct group of 𝑛 binary digits (bits) to each of
the 𝐿 quantization levels.
– We have SQNR
𝑆𝑜 𝑃
= 3𝐿2 2
𝑁𝑞 𝑚𝑝
𝑚𝑝2
𝑁𝑞 = 2
3𝐿
– Since
𝐿 = 2𝑛 or 𝑛 = log 2 𝐿
𝑆
– Increasing 𝐿 (or 𝑛) decreases 𝑁𝑞 or improves 𝑁𝑜
𝑞

– This equation can be written in the form

𝑆𝑜 𝑃 𝑃 𝑃
= 3 𝑚2 𝐿2 = 3 𝑚2 22𝑛 = 3 𝑚2 22𝑛
𝑁𝑞 𝑝 𝑝 𝑝

[23]
 Pulse Code Modulation
𝑺
⚫ Tradeoff between SNR ( 𝒐 ) and bandwidth required 𝑩𝑻
𝑵𝒒

𝑆𝑜 𝑃
= 3 2 22𝑛
𝑁𝑜 𝑚𝑝

𝑃
⚫ Let 𝑐 = 3 𝑚2
𝑝

𝐵
𝑆𝑜 2𝑛 2 𝐵𝑇
=𝑐 2 =𝑐 2 since 𝐵𝑇 = 𝑛𝐵
𝑁𝑜

𝑆𝑜
in dBs = 10 log10 𝑐 + 2𝑛 log10 2 = 𝛼 + 6𝑛 dB
𝑁𝑜

𝛼 = 10 log10 𝑐

[24]
 Pulse Code Modulation
𝑺
⚫ Tradeoff between SNR ( 𝒐 ) and bandwidth required 𝑩𝑻
𝑵𝒒
𝑆𝑜
= 10 log10 𝑐 + 2𝑛 log10 2 = 𝛼 + 6𝑛 dB
𝑁𝑜
– SQNR increases exponentially with increasing 𝒏 or 𝑳 (bandwidth) => 6dB
per bit

– This shows that increasing 𝑛 by 1 (increasing one bit in the code word)
quadruples the output SNR (6-dB increase).

– Thus, if we increase 𝑛 from 8 to 9, the SNR quadruples, but the transmission

bandwidth increases only from 32 to 36 kHz (an increase of only 12.5%).

– This shows that in PCM, SNR can be controlled by transmission bandwidth.

– Frequency and phase modulation also do this. But it requires a doubling of the
bandwidth to quadruple the SNR. In this respect, PCM is strikingly superior to
FM or PM.
25
 Pulse Code Modulation
⚫ Example : Temperature measurements covering the range −40℃ to 40℃ with
accuracy ±0.5℃ or better, are taken as 1 second intervals. They are converted to
binary format for PCM transmission. What is the bit rate required?

⚫ Solution: quantization error

∆𝑣 𝑚𝑝
=
2 𝐿
40
0.5 =
𝐿
40
𝐿 ≥ 0.5 = 80 also 𝐿 = 2𝑛

⚫ Therefore, 𝐿 = 128, 𝑛 = 7 (bit per sample)

⚫ Bit Rate = sampling rate × bit per sample = 1 × 7 = 7 𝑏𝑝𝑠

[26]
 Pulse Code Modulation
⚫ Example 6.2:

[27]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
Motivation
– Speech signals have the
characteristic that small-
amplitude samples occur more
frequently than large-amplitude
ones
– Human auditory system
exhibits a logarithmic
sensitivity
▪ More sensitive at small-
amplitude range (e.g., 0
might sound different from
0.1)
▪ Less sensitive at large-
amplitude range (e.g., 0.7 histogram of typical
might not sound different speech signals
much from 0.8)
[28]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:

⚫ In speech signals, very low speech volumes predominates. Only 15% of the time,
the voltage exceeds the RMS value

⚫ These low level signals are under represented with uniform quantization. Same
∆𝑣 2
noise power but low signal power
12

[29]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
⚫ Recall that SQNR, is an indication of the quality of the received signal. Ideally,
we would like to have a constant SQNR (the same quality) for all values of the
message signal power 𝑃.
⚫ Unfortunately, the SQNR is directly proportional to the signal power, which
varies from talker to talker by as much as 40 dB (a power ratio of 104 ).
⚫ The signal power can also vary because of the different lengths of the connecting
circuits. This means the SQNR in can vary widely, depending on the talker and
the length of the circuit.
⚫ Even for the same talker, the quality of the received signal will deteriorate
markedly when the person speaks softly.
⚫ Statistically, it is found that smaller amplitudes predominate in speech and larger
amplitudes are much less frequent. This means the SNR will be low most of the
time.
⚫ The answer is non uniform quantization

[30]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
– The root of this difficulty lies in the fact that the quantizing steps are of
2𝑚𝑝
uniform value where ∆𝑣 = .
𝐿

∆𝑣 2
– The quantization noise 𝑁𝑞 = is directly proportional to the square of
12
the step size.
– The problem can be solved by using smaller steps for smaller amplitudes
(nonuniform quantizing) as shown in Figure 6.15a.

[31]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:

[32]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
– The same result is obtained by first compressing signal samples and then
using a uniform quantization (Figure 6.15b).

Non-uniform Quantization = Compression + Uniform quantization

[33]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
– The horizontal axis is the normalized input signal (i.e., the input signal
amplitude 𝑚 divided by the signal peak value 𝑚𝑝 ).
– The vertical axis is the output signal 𝑦.
– The compressor maps input signal increments ∆𝑚 into larger increments
∆𝑦 for small input signals, and vice versa for large input signals.
– Hence, a given interval ∆𝑚 contains a larger number of steps (or smaller
step size) when 𝑚 is small.
– The quantization noise is smaller for smaller input signal power.
– An approximately logarithmic compression characteristic yields a
quantization noise nearly proportional to the signal power 𝑃, thus making
the SNR practically independent of the input signal power over a large
dynamic range

[34]
 Pulse Code Modulation
⚫ Non-Uniform Quantization: µ-Law
– Among several choices, two compression laws have been accepted as desirable
standards by the CCITT: the µ-law used in North America and Japan, and the A-
law used in Europe and the rest of the world and international routes.

– The μ-law algorithm (μ-law) is a companding algorithm, primarily used in the

digital telecommunication systems of North America and Japan.

– Its purpose is to reduce the dynamic range of an audio signal.

– In the analog domain, this can increase the SNR achieved during
transmission, and in the digital domain, it can reduce the quantization error
(hence increasing signal to quantization noise ratio).

– μ-law (for positive amplitudes only) is given by:

1 𝜇𝑚 𝑚
𝑦= ln 1 + 0≤ ≤1
ln 1 + 𝜇 𝑚𝑝 𝑚𝑝

[35]
 Pulse Code Modulation
⚫ Non-Uniform Quantization: A-Law

– A-law algorithm used in the rest of worlds.

– A-law algorithm provides a slightly larger dynamic range than the μ-law at
the cost of worse proportional distortion for small signals.

– By convention, A-law is used for an international connection if at least one

country uses it.

– A-law (for positive amplitudes only) is given by:

𝐴 𝑚 𝑚 1
0≤ ≤
1 + ln 𝐴 𝑚𝑝 𝑚𝑝 𝐴
𝑦=
1 𝐴𝑚 1 𝑚
1 + ln ≤ ≤ 1
1 + ln 𝐴 𝑚𝑝 𝐴 𝑚𝑝

[36]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
⚫ The compression parameter μ (or A) determines the degree of
compression.
⚫ To obtain a nearly constant SQNR over an input-signal-power dynamic
range of 40 dB, μ should be greater than 100.
⚫ Early North American channel banks and other digital terminals used a
value of 𝜇 = 100, which yielded the best results for 7-bit (128-level)
encoding.
⚫ An optimum value of 𝜇 = 255 has been used for all North American 8-
bit (256-level) digital terminals, and the earlier value of μ is now almost
extinct.
⚫ For the A-law, a value of 𝐴 = 87.6 gives comparable results and has
been standardized by the ITU-T.
[37]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:

Figure 6.16 (a) μ-Law characteristic. (b) A-Law characteristic.

[38]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:
⚫ The compressed samples must be restored to their original values at the
receiver by using an expander with a characteristic complementary to
that of the compressor.
⚫ The compressor and the expander together are called the compandor.

Figure 6.17 Utilization of compressor and expander for nonuniform quantization.

[39]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:

⚫ We know that 𝐿2 = 22𝑛 , and the output SNR can be expressed as

𝑆𝑜 2𝑛
=𝑐 2
𝑁𝑜
⚫ where
3𝑚෫2 𝑡
2 [uncomressed case]
𝑚𝑝
𝑐=
3
[comressed case]
ln 1+𝜇 2

⚫ Therefore,
𝐵
𝑆𝑜 2 𝑇
=𝑐 2 𝐵
𝑁𝑜

[40]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:

Figure 6.18 Ratio of signal to quantization noise in PCM with and without compression.

[41]
 Pulse Code Modulation
⚫ Example 6.3:

[42]
 Pulse Code Modulation
⚫ Non-Uniform Quantization:

⚫ In the standard audio file format used by Sun, Unix and Java, the audio in "au" files can be
pulse-code-modulated or compressed with the ITU-T G.711 standard through either the μ-
law or the A-law.

⚫ The μ-law compressor (μ = 255) converts 14-bit signed linear PCM samples to
logarithmic 8-bit samples, leading to storage saving.

⚫ The A-law compressor (A = 87.6) converts 13-bit signed linear PCM samples to
logarithmic 8-bit samples.

⚫ In both cases, sampling at the rate of 8000 Hz, a G.77 encoder thus creates from audio
signals bit streams at 64 kbit/s.

⚫ Since the A-law and the μ-law are mutually compatible, audio recoded into "au" files can
be decoded in either format.

⚫ Microsoft WAV audio format also has compression options that use μ-law and A-law.

[43]
 Pulse Code Modulation
⚫ PCM is not a very efficient system because it generates so many bits and
requires so much bandwidth to transmit.
⚫ Many different ideas have been proposed to improve the encoding
efficiency of A/D conversion.
⚫ Some of these techniques are
⚫ Differential Pulse Code Modulation (DPCM)
⚫ Adaptive Differential Pulse Code Modulation (ADPCM)
⚫ Delta Modulation (DM)
⚫ Sigma-Delta Modulation
⚫ Adaptive Delta Modulation (ADM)

[44]
 Pulse Code Modulation
⚫ Differential Pulse Code Modulation (DPCM)

⚫ The transmitter determines the estimate 𝑚

ෝ 𝑞 𝑘 of quantized sample
value 𝑚𝑞 𝑘 from past sample values 𝑚𝑞 𝑘 − 1 , 𝑚𝑞 𝑘 − 2 , ⋯

⚫ The difference between 𝑑 𝑘 = 𝑚 𝑘 − 𝑚 ෝ 𝑞 𝑘 is quantized and 𝑑𝑞 𝑘

transmitted via PCM. Therefore, 𝑑𝑞 𝑘 actually represents the difference
𝑑 𝑞 𝑘 = 𝑚𝑞 𝑘 − 𝑚
ෝ𝑞 𝑘

⚫ The receiver after receiving 𝑑𝑞 𝑘 , also estimate the 𝑚

ෝ 𝑞 𝑘 and also
find the quantized sample value as
𝑚𝑞 𝑘 = 𝑚
ෝ 𝑞 𝑘 + 𝑑𝑞 𝑘

⚫ The received samples 𝑚𝑞 𝑘 are decoded and passed through a low-pass

filter for D/A conversion.

[45]
 Pulse Code Modulation
⚫ Differential Pulse Code Modulation (DPCM)

Figure 6.28 DPCM system: (a) transmitter; (b) receiver.

[46]
 Pulse Code Modulation
⚫ Differential Pulse Code Modulation (DPCM)
⚫ Thus, the peak amplitude 𝑚𝑝 of the transmitted values is reduced considerably.
2𝑚𝑝
⚫ Because the quantization interval Δ𝑣 = , for a given L (or n), this reduces the
𝐿
quantization interval Δ𝑣, thus reducing the quantization noise, which is given by
Δ𝑣 2
.
12

⚫ This means that for a given n (or transmission bandwidth), we can increase the
SNR, or for a given SNR, we can reduce n (or transmission bandwidth).

⚫ For the same SNR, the bit rate for DPCM could be lower than that for PCM by 3
to 4 bits per sample. Thus, telephone systems using DPCM can often operate at
32 kbit/s or even 24 kbit/s.

[47]
 Pulse Code Modulation
⚫ Adaptive Differential Pulse Code Modulation (ADPCM)
⚫ Adaptive DPCM (ADPCM) can further improve the efficiency of DPCM
encoding by incorporating an adaptive quantizer at the encoder.

⚫ For a fixed number of quantization levels, when a fixed quantization step Δ𝑣 is

applied, either the quantization error is too large because Δ𝑣 is too big or the
quantizer cannot cover the necessary signal range when Δ𝑣 is too small.

⚫ Therefore, a better approach is that quantization step Δ𝑣 to be adaptive so that

Δ𝑣 is large or small depending on whether the prediction error for quantizing is
large or small.

⚫ Compared with DPCM, ADPCM can further compress the number of bits
needed for a signal waveform.

⚫ For example, it is very common in practice for an 8-bit PCM sequence to be

encoded into a 4-bit ADPCM sequence at the same sampling rate. This easily
represents a 2:1 bandwidth or storage reduction with virtually no loss.

[48]
 Pulse Code Modulation
⚫ Adaptive Differential Pulse Code Modulation (ADPCM)
⚫ ADPCM encoder has many practical applications.
⚫ The ITU-T standard G.726 specifies an ADPCM speech coder and decoder (called codec)
for speech signal samples at 8 kHz.
⚫ For different quality levels, G.726 specifies four different ADPCM rates at 16, 24, 32, and
40 kbit/s. They correspond to four different bit sizes for each speech sample at 2 bits, 3
bits, 4 bits, and 5 bits, respectively, or equivalently, quantization levels of 4, 8, 16, and 32,
respectively. The most common ADPCM speech encoders use 32 kbit/s.
⚫ In addition to ITU-T G.726, other variations of ADPCM include the OKI ADPCM codec,
the Microsoft ADPCM codec supported by WAVE players, and the Interactive Multimedia
Association (IMA) ADPCM, also known as the DVI ADPCM.
⚫ The 32 kbit/s ADPCM speech codec is also widely used in the DECT (Digital Enhanced
Cordless Telecommunications) system and Personal Handy-phone System (PHS).
⚫ low-power, short-range mobile radio services with limited on no roaming

[49]
 Pulse Code Modulation
⚫ Delta Modulation (DM)
⚫ Sample correlation used in DPCM is further exploited in delta modulation (DM) by
oversampling (typically four times the Nyquist rate) the baseband signal.
⚫ This increases the correlation between adjacent samples, which results in a small
prediction error that can be encoded using only one bit (L = 2).
⚫ Thus, DM is basically a 1-bit DPCM, that is, a DPCM that uses only two levels (L = 2) for
quantization of 𝑑 𝑘 = 𝑚 𝑘 − 𝑚 ෝ𝑞 𝑘 .

⚫ In comparison to PCM (and DPCM), it is a very simple and inexpensive method of A/D
conversion.
⚫ This strategy allows us to use fewer bits per sample for encoding a baseband signal.
⚫ In DM, we use a first-order predictor, which, is just a time delay of 𝑇𝑠 , (the sampling
interval). Thus, the DM transmitter (modulator) and receiver (demodulator) are identical to
those of the DPCM , with a time delay for the predictor.
𝑚𝑞 𝑘 = 𝑚𝑞 𝑘 − 1 + 𝑑𝑞 𝑘
𝑚𝑞 𝑘 − 1 = 𝑚𝑞 𝑘 − 2 + 𝑑𝑞 𝑘 − 1 𝑚𝑞 𝑘 = 𝑚𝑞 𝑘 − 2 + 𝑑𝑞 𝑘 + 𝑑𝑞 𝑘 − 1
𝑚𝑞 𝑘 = σ𝑘𝑚=0 𝑑𝑞 𝑚

[50]
 Pulse Code Modulation
⚫ Delta Modulation (DM)

Figure 6.31 Delta modulation: (a) and (b) delta demodulators;

Figure 6.30 Delta modulation is a special case of DPCM.

[51]
 Pulse Code Modulation
⚫ Delta Modulation (DM)

Figure 6.31 (c) message signal versus integrator output signal

(d) delta-modulated pulse trains; (e) modulation errors.

[52]
 Pulse Code Modulation
⚫ Delta Modulation (DM)
⚫ Issues with DM
⚫ Slope overload
⚫ Granular noise (similar to quantization noise)
⚫ The slope overload noise can be reduced by increasing 𝐸 (the step size). This
unfortunately increases the granular noise.
⚫ There is an optimum value of 𝐸, which yields the best compromise giving the
minimum overall noise. This optimum value of 𝐸 depends on the sampling frequency
𝑓𝑠 and the nature of the signal.
⚫ The maximum voice signal amplitude 𝐴𝑚𝑎𝑥 that can be used without causing slope
overload in DM is related with the step size 𝐸 and the sampling frequency 𝑓𝑠 .
𝐸𝑓𝑠
𝐴𝑚𝑎𝑥 𝑣𝑜𝑖𝑐𝑒 ≅ 𝜔𝑟 ≅ 2𝜋 × 800
𝜔𝑟

⚫ The performance of DM can be enhanced through

⚫ Sigma-Delta Modulation (Σ − ΔΜ)
⚫ Adaptive Delta Modulation (ADM)

[53]
 Pulse Code Modulation
⚫ Delta Modulation (DM)
⚫ Sigma-Delta Modulation (Σ − ΔΜ)

⚫ Since signal transmission inevitably is subjected to channel noise, such noise will be
integrated and will accumulate at the receiver output, which is a highly undesirable
phenomenon that is a major drawback of DM.
⚫ To overcome this critical drawback of DM, minor modification can be made in the
conventional DM and the resultant structure in known as Sigma-Delta Modulation
(Σ − ΔΜ).
⚫ The channel noise no longer accumulates at the demodulator.
⚫ The important low-frequency content of the message 𝑚 𝑡 is preemphasized by
the integrator. This helps many practical signals (such as speech) whose low-
frequency components are more important.
⚫ The integrator effectively smooths the signal for encoding. Hence, overloading
becomes less likely.
⚫ The low-pass nature of the integrator increases the correlation between
successive samples, leading to smaller encoding error.
⚫ The demodulator is simplified.

[54]
 Pulse Code Modulation
⚫ Delta Modulation (DM)
⚫ Sigma-Delta Modulation (Σ − ΔΜ)

Figure 6.33 (a) Conventional delta modulator. (b)Σ-Δ modulator. (c) Simpler Σ-Δ modulator.

[55]
 Pulse Code Modulation
⚫ Delta Modulation (DM)
⚫ Adaptive Delta Modulation (ADM)

⚫ The DM discussed so far suffers from one serious disadvantage. The dynamic range
of amplitudes is too small because of the threshold and overload effects discussed
earlier. To address this problem, some type of signal compression is necessary.
⚫ In DM, a suitable method appears to be the adaptation of the step value 𝐸 according
to the level of the input signal derivative.
⚫ For example, when the signal 𝑚 𝑡 is falling rapidly, slope overload occurs. If we
can increase the step size during this period, the overload could be avoided. On the
other hand, if the slope of 𝑚 𝑡 is small, a reduction of step size will reduce the
threshold level as well as the granular noise.
⚫ The slope overload causes 𝑑𝑞 𝑘 to have several pulses of the same polarity in
succession. This calls for increased step size.
⚫ Similarly, pulses in 𝑑𝑞 𝑘 alternating continuously in polarity indicates small-
amplitude variations, requiring a reduction in step size.
⚫ In ADM, we detect such pulse patterns and automatically adjust the step size. This
results in a much larger dynamic range for DM.

[56]
 Pulse Code Modulation
⚫ PCM, DPCM, ADPCM, DM, and Σ − ΔΜ are all examples of what are known as
waveform source encoders
⚫ Waveform encoding schemes are designed to reproduce the waveform output of the
source at the destination with as little distortion as possible.
⚫ Basically, waveform encoders do not take into consideration how the signals for
digitization are generated.
⚫ In these techniques, no attention is paid to the mechanism that produces the
waveform; all attempts are directed at reproducing the source output at the destination
with high fidelity. The structure of the source plays no role in the design of waveform
coders and only properties of the waveform affect the design
⚫ Thus, waveform coders are robust and can be used with a variety of sources as long as
the waveforms produced by the sources have certain similarities.
⚫ Hence, the amount of compression achievable by waveform encoders is highly
limited by the degree of correlation between successive signal samples.
⚫ For a low-pass source signal with finite bandwidth B Hz, even if we apply the
minimum Nyquist sampling rate 2B Hz and 1-bit encoding, the bit rate cannot be
lower than 2B bit/s.

[57]