Chapter-3

Waveform Coding Techniques

PCM [Pulse Code Modulation]
PCM is an important method of analog to-digital conversion. In this modulation
the analog signal is converted into an electrical waveform of two or more levels. A
simple two level waveform is shown in fig 3.1.
Fig:3.1 A simple binary PCM waveform
The PCM system block diagram is shown in fig 3.2. The essential operations in the
transmitter of a PCM system are Sampling, Quantizing and Coding. The Quantizing and
encoding operations are usually performed by the same circuit, normally referred to as
analog to digital converter.
The essential operations in the receiver are regeneration, decoding and
demodulation of the quantized samples. Regenerative repeaters are used to reconstruct
the transmitted sequence of coded pulses in order to combat the accumulated effects of
signal distortion and noise.
PCM Transmitter:
Basic Blocks:
1. Anti aliasing Filter
2. Sampler
3. Quantizer
4. Encoder
An anti-aliasing filter is basically a filter used to ensure that the input signal to sampler is
free from the unwanted frequency components.
For most of the applications these are low-pass filters. It removes the frequency
components of the signal which are above the cutoff frequency of the filter. The cutoff
frequency of the filter is chosen such it is very close to the highest frequency component
of the signal.
Sampler unit samples the input signal and these samples are then fed to the Quantizer
which outputs the quantized values for each of the samples. The quantizer output is fed
to an encoder which generates the binary code for every sample. The quantizer and
encoder together is called as analog to digital converter.
Continuous time
message signal PCM Wave
(a) TRANSMITTER
Distorted
PCM wave
(b) Transmission Path
Input
(c) RECEIVER
Fig: 3.2 - PCM System : Basic Block Diagram
LPF Sampler Quantizer Encoder
Regeneration
Circuit
Decoder Reconstruction
Filter
Destination
User
Regenerative
Repeater
Regenerative
Repeater
REGENERATIVE REPEATER
REGENERATION: The feature of the PCM systems lies in the ability to control the
effects of distortion and noise produced by transmitting a PCM wave through a channel.
This is accomplished by reconstructing the PCM wave by means of regenerative
repeaters.
Three basic functions: Equalization
Timing and
Decision Making
Distorted Regenerated
PCM PCM wave
Wave
Fig: 3.3 - Block diagram of a regenerative repeater.
The equalizer shapes the received pulses so as to compensate for the effects of
amplitude and phase distortions produced by the transmission characteristics of the
channel.
The timing circuit provides a periodic pulse train, derived from the received
pulses, for sampling the equalized pulses at the instants of time where the signal to noise
ratio is maximum.
The decision device is enabled at the sampling times determined by the timing
circuit. It makes its decision based on whether the amplitude of the quantized pulse plus
noise exceeds a predetermined voltage level.
Amplifier -
Equalizer
Decision
Making
Device
Timing
Circuit
Quantization Process:
The process of transforming Sampled amplitude values of a message signal into a
discrete amplitude value is referred to as Quantization.
The quantization Process has a two-fold effect:
1. the peak-to-peak range of the input sample values is subdivided into a finite set
of decision levels or decision thresholds that are aligned with the risers of the
staircase, and
2. the output is assigned a discrete value selected from a finite set of representation
levels that are aligned with the treads of the staircase..
A quantizer is memory less in that the quantizer output is determined only by the value of
a corresponding input sample, independently of earlier analog samples applied to the
input.
Fig:3.4 Typical Quantization process.
Types of Quantizers:
1. Uniform Quantizer
2. Non- Uniform Quantizer
Ts 0 2Ts 3Ts
Time
Analog Signal
Discrete Samples
( Quantized )
In Uniform type, the quantization levels are uniformly spaced, whereas in non-
uniform type the spacing between the levels will be unequal and mostly the relation is
logarithmic.
Types of Uniform Quantizers: ( based on I/P - O/P Characteristics)
1. Mid-Rise type Quantizer
2. Mid-Tread type Quantizer
In the stair case like graph, the origin lies the middle of the tread portion in Mid Tread
type where as the origin lies in the middle of the rise portion in the Mid-Rise type.
Mid tread type: Quantization levels odd number.
Mid Rise type: Quantization levels even number.
Fig:3.5 Input-Output Characteristics of a Mid-Rise type Quantizer
Input
Output
2 3 4
/2
3/2
5/2
7/2
Fig:3.6 Input-Output Characteristics of a Mid-Tread type Quantizer
Quantization Noise and Signal-to-Noise:
The Quantization process introduces an error defined as the difference between the input
signal, x(t) and the output signal, yt). This error is called the Quantization Noise.
q(t) = x(t) y(t)
Quantization noise is produced in the transmitter end of a PCM system by
rounding off sample values of an analog base-band signal to the nearest permissible
representation levels of the quantizer. As such quantization noise differs from channel
noise in that it is signal dependent.
Let be the step size of a quantizer and L be the total number of quantization levels.
Quantization levels are 0, ., 2 ., 3 . . . . . . .
The Quantization error, Q is a random variable and will have its sample values bounded
by [-(/2) < q < (/2)]. If is small, the quantization error can be assumed to a
uniformly distributed random variable.
Consider a memory less quantizer that is both uniform and symmetric.
L = Number of quantization levels
X = Quantizer input
Y = Quantizer output

2
/2 3/2
Input
Output
The output y is given by
Y=Q(x) ------------- (3.1)
which is a staircase function that befits the type of mid tread or mid riser quantizer of
interest.
Suppose that the input x lies inside the interval
I
k
= {x
k
< x x
k+1
} k = 1,2,---------L ------- ( 3.2)
where x
k
and x
k+1
are decision thresholds of the interval I
k
as shown in figure 3.7.
Fig:3.7 Decision thresholds of the equalizer
Correspondingly, the quantizer output y takes on a discrete value
Y = y
k
if x lies in the interval I
k
Let q = quantization error with values in the range
2 2
A
s s
A
q then
Y
k
= x+q if n lies in the interval I
k
Assuming that the quantizer input n is the sample value of a random variable X of
zero mean with variance
2
x .
The quantization noise uniformly distributed through out the signal band, its interfering
effect on a signal is similar to that of thermal noise.
Expression for Quantization Noise and SNR in PCM:-
Let Q = Random Variable denotes the Quantization error
q = Sampled value of Q
Assuming that the random variable Q is uniformly distributed over the possible range
(-/2 to /2) , as
f
Q
(q) = 1/ - /2 q /2 ------- (3.3)
0 otherwise
I
k
I
k-1
X
k-1
X
k
X
k+1
y
k-1
y
k
where f
Q
(q) = probability density function of the Quantization error. If the signal
does not overload the Quantizer, then the mean of Quantization error is zero and
its variance
Q
2
.
f
Q
(q)
1/
- /2 0 /2 q
Fig:3.8 PDF for Quantization error.
Therefore
} {
2 2
Q E
Q
=
dq q f q
q Q
) (
2
2
}

=
---- ( 3.4)
12
1
2 2
2
2
2
A
=
A
=
}
A
A

dq q
Q

--- (3.5)
Thus the variance of the Quantization noise produced by a Uniform Quantizer,
grows as the square of the step size. Equation (3.5) gives an expression for Quantization
noise in PCM system.
Let
2
X
= Variance of the base band signal x(t) at the input of Quantizer.
When the base band signal is reconstructed at the receiver output, we obtain
original signal plus Quantization noise. Therefore output signal to Quantization noise
ration (SNR) is given by
12 /
) (
2
2
2
2
A
= = =
X
Q
X
O
Power Noise
Power Signal
SNR

---------- (3.6)
Smaller the step size , larger will be the SNR.
Signal to Quantization Noise Ratio:- [ Mid Tread Type ]
Let x = Quantizer input, sampled value of random variable X with mean X, variance
2
X
. The Quantizer is assumed to be uniform, symmetric and mid tread type.
x
max
= absolute value of the overload level of the Quantizer.
= Step size
L = No. of Quantization level given by
1
2
max
+
A
=
x
L ----- (3.7)
Let n = No. of bits used to represent each level.
In general 2
n
= L, but in the mid tread Quantizer, since the number of representation
levels is odd,
L = 2
n
1 --------- (Mid tread only) ---- (3.8)
From the equations 3.7 and 3.8,
1
2
1 2
max
+
A
=
x
n
Or
1 2
1
max

= A
n
x
---- (3.9)
The ratio
x
x

max
is called the loading factor. To avoid significant overload distortion, the
amplitude of the Quantizer input x extend from
x
4 to
x
4 , which corresponds to
loading factor of 4. Thus with
x
x 4
max
= we can write equation (3.9) as
1 2
4
1

= A
n
x

----------(3.10)
2 1
2
2
] 1 2 [
4
3
12 /
) ( =
A
=
n X
O
SNR

-------------(3.11)
For larger value of n (typically n>6), we may approximate the result as
) 2 (
16
3
] 1 2 [
4
3
) (
2 2 1 n n
O
SNR ~ =

--------------- (3.12)
Hence expressing SNR in db
10 log
10
(SNR)
O
= 6n - 7.2 --------------- (3.13)
This formula states that each bit in codeword of a PCM system contributes 6db to the
signal to noise ratio.
For loading factor of 4, the problem of overload i.e. the problem that the sampled
value of signal falls outside the total amplitude range of Quantizer, 8
x
is less than 10
-4
.
The equation 3.11 gives a good description of the noise performance of a PCM
system provided that the following conditions are satisfied.
1. The Quantization error is uniformly distributed
2. The system operates with an average signal power above the error threshold so
that the effect of channel noise is made negligible and performance is there by
limited essentially by Quantization noise alone.
3. The Quantization is fine enough (say n>6) to prevent signal correlated patterns in
the Quantization error waveform
4. The Quantizer is aligned with input for a loading factor of 4
Note: 1. Error uniformly distributed
2. Average signal power
3. n > 6
4. Loading factor = 4
From (3.13): 10 log
10
(SNR)
O
= 6n 7.2
In a PCM system, Bandwidth B = nW or [n=B/W]
substituting the value of n we get
10 log
10
(SNR)
O
= 6(B/W) 7.2 --------(3.14)
Signal to Quantization Noise Ratio:- [ Mid Rise Type ]
Let x = Quantizer input, sampled value of random variable X with mean X variance
2
X
.
The Quantizer is assumed to be uniform, symmetric and mid rise type.
Let x
max
= absolute value of the overload level of the Quantizer.
A
=
max
2x
L ------------------(3.15)
Since the number of representation levels is even,
L = 2
n
------- (Mid rise only) ---- (3.16)
From (3.15) and (3.16)
n
x
2
max
= A
-------------- (3.17)
12 /
) (
2
2
A
=
X
O
SNR

-------------(3.18)
where
2
X
represents the variance or the signal power.
Consider a special case of Sinusoidal signals:
Let the signal power be Ps, then Ps = 0.5 x
2
max
.
n
O
L
Ps Ps
SNR
2 2
2 2
2 5 . 1 5 . 1
12
12 /
) ( = =
A
=
A
=
-----(3.19)
In decibels, ( SNR )
0
= 1.76 + 6.02 n -----------(3.20)
Improvement of SNR can be achieved by increasing the number of bits, n. Thus
for n number of bits / sample the SNR is given by the above equation 3.19. For every
increase of one bit / sample the step size reduces by half. Thus for (n+1) bits the SNR is
given by
(SNR)
(n+1) bit
= (SNR)
(n) bit
+ 6dB
Therefore addition of each bit increases the SNR by 6dB
Problem-1: An analog signal is sampled at the Nyquist rate fs = 20K and quantized
into L=1024 levels. Find Bit-rate and the time duration Tb of one bit of the binary
encoded signal.
Solution: Assume Mid-rise type, n = log
2
L = 10
Bit-rate = Rb = nfs = 200K bits/sec
Bit duration Tb = 1/ Rb = 5sec.
Problem-2: A PCM system uses a uniform quantizer followed by a 7-bit binary encoder.
The bit rate of the system is 56Mega bits/sec. Find the output signal-to-quantization
noise ratio when a sinusoidal wave of 1MHz frequency is applied to the input.
Solution:
Given n = 7 and bit rate Rb = 56 Mega bits per second.
Sampling frequency = Rb/n = 8MHz
Message bandwidth = 4MHz.
For Mid-rise type
(SNR)
0
= 43.9 dB
CLASSIFICATION OF QUANTIZATION NOISE:
The Quantizing noise at the output of the PCM decoder can be categorized into
four types depending on the operating conditions:
Overload noise, Random noise, Granular Noise and Hunting noise
OVER LOAD NOISE:- The level of the analog waveform at the input of the PCM
encoder needs to be set so that its peak value does not exceed the design peak of Vmax
volts. If the peak input does exceed Vmax, then the recovered analog waveform at the
output of the PCM system will have flat top near the peak values. This produces
overload noise.
GRANULAR NOISE:- If the input level is reduced to a relatively small value w.r.t to the
design level (quantization level), the error values are not same from sample to sample and
the noise has a harsh sound resembling gravel being poured into a barrel. This is granular
noise.
This noise can be randomized (noise power decreased) by increasing the number
of quantization levels i.e.. increasing the PCM bit rate.
HUNTING NOISE:- This occurs when the input analog waveform is nearly constant.
For these conditions, the sample values at the Quantizer output can oscillate between two
adjacent quantization levels, causing an undesired sinusoidal type tone of frequency
(0.5fs) at the output of the PCM system
This noise can be reduced by designing the quantizer so that there is no vertical
step at constant value of the inputs.
ROBUST QUANTIZATION
Features of an uniform Quantizer
Variance is valid only if the input signal does not overload Quantizer
SNR Decreases with a decrease in the input power level.
A Quantizer whose SNR remains essentially constant for a wide range of input power
levels. A quantizer that satisfies this requirement is said to be robust. The provision for
such robust performance necessitates the use of a non-uniform quantizer. In a non-
uniform quantizer the step size varies. For smaller amplitude ranges the step size is small
and larger amplitude ranges the step size is large.
In Non Uniform Quantizer the step size varies. The use of a non uniform
quantizer is equivalent to passing the baseband signal through a compressor and then
applying the compressed signal to a uniform quantizer. The resultant signal is then
transmitted.
Fig: 3.9 MODEL OF NON UNIFORM QUANTIZER
At the receiver, a device with a characteristic complementary to the compressor
called Expander is used to restore the signal samples to their correct relative level.
The Compressor and expander take together constitute a Compander.
Compander = Compressor + Expander
Advantages of Non Uniform Quantization :
1. Higher average signal to quantization noise power ratio than the uniform quantizer
when the signal pdf is non uniform which is the case in many practical situation.
2. RMS value of the quantizer noise power of a non uniform quantizer is
substantially proportional to the sampled value and hence the effect of the
quantizer noise is reduced.
COMPRESSOR
UNIFORM
QUANTIZER EXPANDER
Expression for quantization error in non-uniform quantizer:
The Transfer Characteristics of the compressor and expander are denoted by C(x)
and C
-1
(x) respectively, which are related by,
C(x). C
-1
(x) = 1 ------ ( 3.21 )
The Compressor Characteristics for large L and x inside the interval I
k
:
1 ,..... 1 , 0
2 ) (
max
=
A
= L k for
L
x
dx
x dc
k
---------- ( 3.22 )
where
k
= Width in the interval I
k.
Let f
X
(x) be the PDF of X .
Consider the two assumptions:
f
X
(x) is Symmetric
f
X
(x) is approximately constant in each interval. ie.. f
X
(x) = f
X
(y
k
)
k = x
k+1
- x
k
for k = 0, 1, L-1. ------(3.23)
Let p
k
= Probability of variable X lies in the interval I
k,
then
p
k
= P (x
k
< X < x
k+1
) = f
X
(x) k = f
X
(y
k
) k ------ (3.24)
with the constraint _

=
=
1
0
1
L
k
k
p
Let the random variable Q denote the quantization error, then
Q = yk X for xk < X < xk+1
Variance of Q is

Q
2
= E ( Q
2
) = E [( X y
k
)
2
] ---- (3.25)
}
+

=
max
max
) ( ) (
2
2
x
x
X k Q
dx x f y x
---- ( 3.26)
Dividing the region of integration into L intervals and using (3.24)
}
_
+

A
=

=
1
2
1
0
2
) (
k
k
x
x
k
k
k
L
k
Q
dx y x
p

----- (3.27)
Using y
k
= 0.5 ( x
k
+ x
k+1
) in 3.27 and carrying out the integration w.r.t x, we
obtain that
_

=
A =
1
0
2
2
12
1
L
k
k k Q
p
------- (3.28)
Compression Laws.
Two Commonly used logarithmic compression laws are called - law and A law.
-law:
In this companding, the compressor characteristics is defined by equation 3.29.
The normalized form of compressor characteristics is shown in the figure 3.10. The -
law is used for PCM telephone systems in the USA, Canada and Japan. A practical
value for is 255.
----( 3.29)
Fig: 3.10 Compression characteristics of -law
1 0
) 1 ln(
) / 1 ln( ) (
max
max
max
s s
+
+
=
x
x x x
x
x c

A-law:
In A-law companding the compressor characteristics is defined by equation 3.30. The
normalized form of A-law compressor characteristics is shown in the figure 3.11. The
A-law is used for PCM telephone systems in Europe. A practical value for A is 100.
------------- ( 3.30)
Fig. 3.11: A-law compression Characteristics.
Advantages of Non Uniform Quantizer
Reduced Quantization noise
High average SNR
A x
x
A
x x A
1
0
ln 1
/
max
max
s s
+
=
max
) (
x
x c
1
1
l
n
1
) / ln
(
1
ma
x
ma
x
s s
+
+
x
x
A A
x x A
Differential Pulse Code Modulation (DPCM)
For the signals which does not change rapidly from one sample to next sample, the PCM
scheme is not preferred. When such highly correlated samples are encoded the resulting
encoded signal contains redundant information. By removing this redundancy before
encoding an efficient coded signal can be obtained. One of such scheme is the DPCM
technique. By knowing the past behavior of a signal up to a certain point in time, it is
possible to make some inference about the future values.
The transmitter and receiver of the DPCM scheme is shown in the fig3.12 and fig 3.13
respectively.
Transmitter: Let x(t) be the signal to be sampled and x(nTs) be its samples. In this
scheme the input to the quantizer is a signal
e(nTs) = x(nTs) - x^(nTs) ----- (3.31)
where x^(nTs) is the prediction for unquantized sample x(nTs). This predicted value is
produced by using a predictor whose input, consists of a quantized versions of the input
signal x(nTs). The signal e(nTs) is called the prediction error.
By encoding the quantizer output, in this method, we obtain a modified version of the
PCM called differential pulse code modulation (DPCM).
Quantizer output, v(nTs) = Q[e(nTs)]
= e(nTs) + q(nTs) ---- (3.32)
where q(nTs) is the quantization error.
Predictor input is the sum of quantizer output and predictor output,
u(nTs) = x^(nTs) + v(nTs) ---- (3.33)
Using 3.32 in 3.33, u(nTs) = x^(nTs) + e(nTs) + q(nTs) ----(3.34)
u(nTs) = x(nTs) + q(nTs) ----(3.35)
The receiver consists of a decoder to reconstruct the quantized error signal. The quantized
version of the original input is reconstructed from the decoder output using the same
predictor as used in the transmitter. In the absence of noise the encoded signal at the
receiver input is identical to the encoded signal at the transmitter output. Correspondingly
the receive output is equal to u(nTs), which differs from the input x(nts) only by the
quantizing error q(nTs).
Fig:3.12 - Block diagram of DPCM Transmitter
Fig:3.13 - Block diagram of DPCM Receiver.
Output
Decoder

Predictor
Input
b(nTs)
v(nTs) u(nTs)
x^(nTs
)
x(nT
s
)
Sampled Input
x(nT
s
)
e(nT
s
)
v(nT
s
)
Quantizer
Predictor

+
u(nT
s
)
Output
^
Prediction Gain ( Gp):
The output signal-to-quantization noise ratio of a signal coder is defined as
2
2
0
) (
Q
X
SNR

=
--------------------( 3.36)
where
x
2
is the variance of the signal x(nTs) and
Q
2
is the variance of the
quantization error q(nTs). Then
P P
Q
E
E
X
SNR G SNR ) ( ) (
2
2
2
2
0
=

\
|

\
|
=

------(3.37)
where
E
2
is the variance of the prediction error e(nTs) and (SNR)
P
is the prediction
error-to-quantization noise ratio, defined by
2
2
) (
Q
E
P
SNR

=
--------------(3.38)
The Prediction gain Gp is defined as
2
2
E
X
P
G

=
--------(3.39)
The prediction gain is maximized by minimizing the variance of the prediction error.
Hence the main objective of the predictor design is to minimize the variance of the
prediction error.
The prediction gain is defined by
) 1 (
1
2
1

=
P
G
---- (3.40)
and ) 1 (
2
1
2 2
=
X E
----(3.41)
where
1
Autocorrelation function of the message signal
PROBLEM:
Consider a DPCM system whose transmitter uses a first-order predictor optimized
in the minimum mean-square sense. Calculate the prediction gain of the system
for the following values of correlation coefficient for the message signal:
825 . 0
) 0 (
) 1 (
) (
1
= =
x
x
R
R
i 950 . 0
) 0 (
) 1 (
) (
1
= =
x
x
R
R
ii
Solution:
Using (3.40)
(i) For 1= 0.825, Gp = 3.13 In dB , Gp = 5dB
(ii) For 2 = 0.95, Gp = 10.26 In dB, Gp = 10.1dB
Delta Modulation (DM)
Delta Modulation is a special case of DPCM. In DPCM scheme if the base band signal
is sampled at a rate much higher than the Nyquist rate purposely to increase the
correlation between adjacent samples of the signal, so as to permit the use of a simple
quantizing strategy for constructing the encoded signal, Delta modulation (DM) is
precisely such as scheme. Delta Modulation is the one-bit (or two-level) versions of
DPCM.
DM provides a staircase approximation to the over sampled version of an input base band
signal. The difference between the input and the approximation is quantized into only two
levels, namely, corresponding to positive and negative differences, respectively, Thus,
if the approximation falls below the signal at any sampling epoch, it is increased by .
Provided that the signal does not change too rapidly from sample to sample, we find that
the stair case approximation remains within of the input signal. The symbol denotes
the absolute value of the two representation levels of the one-bit quantizer used in the
DM. These two levels are indicated in the transfer characteristic of Fig 3.14. The step
size Aof the quantizer is related to by
A = 2 ----- (3.42)
Fig-3.14: Input-Output characteristics of the delta modulator.
Let the input signal be x(t) and the staircase approximation to it is u(t). Then, the basic
principle of delta modulation may be formalized in the following set of relations:
Output
Input
+
-
0
) ( ) ( ) (
)] ( sgn[ ) (
) ( ) ( ) (
) ( ) ( ) (
^
s s s s
s s
s s s
nT b T nT u nT u
and nT e nT b
Ts nTs u nTs x nTs e
nT x nT x nT e
+ =
=
=
=

----- (3.43)
where T
s
is the sampling period; e(nT
s
) is a prediction error representing the difference
between the present sample value x(nT
s
) of the input signal and the latest approximation
to it, namely ) ( ) (
^
s s s
T nT u nT x = .The binary quantity, ) (
s
nT b is the one-bit word
transmitted by the DM system.
The transmitter of DM system is shown in the figure3.15. It consists of a summer, a two-
level quantizer, and an accumulator. Then, from the equations of (3.43) we obtain the
output as,
_ _
= =
= =
n
i
n
i
iTs b iTs e nTs u
1 1
) ( )] ( sgn[ ) (
----- (3.44)
At each sampling instant, the accumulator increments the approximation to the input
signal by , depending on the binary output of the modulator.
Fig 3.15 - Block diagram for Transmitter of a DM system
x(nT
s
)
Sampled Input
x(nT
s
)
e(nT
s
)
b(nT
s
)
One - Bit
Quantizer
Delay
Ts

+
u(nT
s
)
Output
^
In the receiver, shown in fig.3.16, the stair case approximation u(t) is reconstructed by
passing the incoming sequence of positive and negative pulses through an accumulator in
a manner similar to that used in the transmitter. The out-of band quantization noise in
the high frequency staircase waveform u(t) is rejected by passing it through a low-pass
filter with a band-width equal to the original signal bandwidth.
Delta modulation offers two unique features:
1. No need for Word Framing because of one-bit code word.
2. Simple design for both Transmitter and Receiver
Fig 3.16 - Block diagram for Receiver of a DM system
QUANTIZATION NOISE
Delta modulation systems are subject to two types of quantization error:
(1) slope overload distortion, and (2) granular noise.
If we consider the maximum slope of the original input waveform x(t), it is clear that in
order for the sequence of samples{u(nT
s
)} to increase as fast as the input sequence of
samples {x(nT
s
)} in a region of maximum slope of x(t), we require that the condition in
equation 3.45 be satisfied.
dt
t dx
T
s
) (
max >

------- ( 3.45 )
Otherwise, we find that the step size A = 2 is too small for the stair case
approximation u(t) to follow a steep segment of the input waveform x(t), with the result
that u(t) falls behind x(t). This condition is called slope-overload, and the resulting
quantization error is called slope-overload distortion(noise). Since the maximum slope of
the staircase approximation u(t) is fixed by the step size A, increases and decreases in
u(t) tend to occur along straight lines. For this reason, a delta modulator using a fixed step
size is often referred ton as linear delta modulation (LDM).
Low pass
Filter

Delay
Ts
Input
b(nTs)
u(nTs)
u(nTs-Ts)
The granular noise occurs when the step size A is too large relative to the local
slope characteristics of the input wave form x(t), thereby causing the staircase
approximation u(t) to hunt around a relatively flat segment of the input waveform; The
granular noise is analogous to quantization noise in a PCM system.
The e choice of the optimum step size that minimizes the mean-square value of
the quantizing error in a linear delta modulator will be the result of a compromise
between slope overload distortion and granular noise.
Output SNR for Sinusoidal Modulation.
Consider the sinusoidal signal, x(t) = A cos(2fot)
The maximum slope of the signal x(t) is given by
A f
dt
t dx
0
2
) (
max =
----- (3.46)
The use of Eq.5.81 constrains the choice of step size A = 2, so as to avoid slope-
overload. In particular, it imposes the following condition on the value of :
A f
dt
t dx
T
s
0
2
) (
max

= >
----- (3. 47)
Hence for no slope overload error the condition is given by equations 3.48 and 3.49.
Ts f
A
0
2

s
------ (3.48)
s
AT f
0
2 >
------ (3.49)
Hence, the maximum permissible value of the output signal power equals
2 2
0
2
2 2
max
8 2
s
T f
A
P

= =
---- (3.50)
When there is no slope-overload, the maximum quantization error . Assuming that the
quantizing error is uniformly distributed (which is a reasonable approximation for small
). Considering the probability density function of the quantization error,( defined in
equation 3.51 ),
otherwise
q for q f
Q
0
2
1
) (

+ s s =
----- (3.51)
The variance of the quantization error is Q
2
.
2
2
2
3 2
1

= =
}
+

dq q
Q
----- (3.52)
The receiver contains (at its output end) a low-pass filter whose bandwidth is set equal to
the message bandwidth (i.e., highest possible frequency component of the message
signal), denoted as W such that f
0
W. Assuming that the average power of the
quantization error is uniformly distributed over a frequency interval extending from -1/T
s
to 1/T
s
, we get the result:
Average output noise power

\
|
=

\
|
=
3 3
2
2

s
s
c
o
WT
f
f
N
----- ( 3.53)
Correspondingly, the maximum value of the output signal-to-noise ratio equals
3 2
0
2
max
8
3
) (
s o
O
T Wf N
P
SNR

= =
----- (3.54)
Equation 3.54 shows that, under the assumption of no slope-overload distortion, the
maximum output signal-to-noise ratio of a delta modulator is proportional to the sampling
rate cubed. This indicates a 9db improvement with doubling of the sampling rate.
Problems
1. Determine the output SNR in a DM system for a 1KHz sinusoid sampled
at 32KHz without slope overload and followed by a 4KHz post
reconstruction filter.
Solution:
Given W=4KHz, f0 = 1KHz , fs = 32KHz
Using equation (3.54) we get
(SNR)
0
= 311.3 or 24.9dB
Delta Modulation:
Problems
2. Consider a Speech Signal with maximum frequency of 3.4KHz and
maximum amplitude of 1volt. This speech signal is applied to a delta modulator
whose bit rate is set at 60kbit/sec. Explain the choice of an appropriate step size for
the modulator.
Solution: Bandwidth of the signal = 3.4 KHz.
Maximum amplitude = 1 volt
Bit Rate = 60Kbits/sec
Sampling rate = 60K Samples/sec.
STEP SIZE = 0.356 Volts
3. Consider a Speech Signal with maximum frequency of 3.4KHz and
maximum amplitude of 1volt. This speech signal is applied to a delta modulator
whose bit rate is set at 20kbit/sec. Explain the choice of an appropriate step size for
the modulator.
Solution: Bandwidth of the signal = 3.4 KHz.
Maximum amplitude = 1 volt
Bit Rate = 20Kbits/sec
Sampling rate = 20K Samples/sec.
STEP SIZE = 1.068 Volts
4. Consider a Delta modulator system designed to operate at 4 times the Nyquist
rate for a signal with a 4KHz bandwidth. The step size of the quantizer is 400mV.
a) Find the maximum amplitude of a 1KHz input sinusoid for which the delta
modulator does not show slope overload.
b) Find post-filtered output SNR
Solution: Bandwidth of the signal = f0 =1 KHz.
Nyquist Rate = 8K samples/sec
Sampling Rate = 32K samples/sec.
Step Size = 400 mV
a) For 1KHz sinusoid, Amax = 2.037 volts.
b) Assuming LPF bandwidth = W= 4KHz
SNR = 311.2586 = 24.93 dB
Adaptive Delta Modulation:
The performance of a delta modulator can be improved significantly by making the step
size of the modulator assume a time-varying form. In particular, during a steep segment
of the input signal the step size is increased. Conversely, when the input signal is varying
slowly, the step size is reduced. In this way, the size is adapted to the level of the input
signal. The resulting method is called adaptive delta modulation (ADM).
There are several types of ADM, depending on the type of scheme used for adjusting the
step size. In this ADM, a discrete set of values is provided for the step size. Fig.3.17
shows the block diagram of the transmitter and receiver of an ADM System.
In practical implementations of the system, the step size
) (
s
nT A or ) ( 2
s
nT
is constrained to lie between minimum and maximum values.
The upper limit,
max
, controls the amount of slope-overload distortion. The lower limit,
min
, controls the amount of idle channel noise. Inside these limits, the adaptation rule
for ) (
s
nT is expressed in the general form
(nTs) = g(nTs). (nTs Ts) ------ (3.55)
where the time-varying multiplier ) (
s
nT g depends on the present binary output ) (
s
nT b
of the delta modulator and the M previous values ) ( . . . . . . . ), (
s s s s
MT nT b T nT b .
This adaptation algorithm is called a constant factor ADM with one-bit memory,
where the term one bit memory refers to the explicit utilization of the single pervious
bit ) (
s s
T nT b because equation (3.55) can be written as,
g(nTs) = K if b(nTs) = b(nTs Ts)
g(nTs) = K
-1
if b(nTs) = b(nTs Ts) ------ (3.56)
This algorithm of equation (3.56), with K=1.5 has been found to be well matched to
typically speech and image inputs alike, for a wide range of bit rates.
Figure: 3.17a) Block Diagram of ADM Transmitter.
Figure: 3.17 b): Block Diagram of ADM Receiver.
Coding Speech at Low Bit Rates:
The use of PCM at the standard rate of 64 kb/s demands a high channel
bandwidth for its transmission. But channel bandwidth is at a premium, in which case
there is a definite need for speech coding at low bit rates, while maintaining acceptable
fidelity or quality of reproduction. The fundamental limits on bit rate suggested by speech
perception and information theory show that high quality speech coding is possible at
rates considerably less that 64 kb/s (the rate may actually be as low as 2 kb/s).
For coding speech at low bit rates, a waveform coder of prescribed configuration
is optimized by exploiting both statistical characterization of speech waveforms and
properties of hearing. The design philosophy has two aims in mind:
1. To remove redundancies from the speech signal as far as possible.
2. To assign the available bits to code the non-redundant parts of the speech signal in
a perceptually efficient manner.
To reduce the bit rate from 64 kb/s (used in standard PCM) to 32, 16, 8 and 4
kb/s, the algorithms for redundancy removal and bit assignment become increasingly
more sophisticated.
There are two schemes for coding speech:
1. Adaptive Differential Pulse code Modulation (ADPCM) --- 32 kb/s
2. Adaptive Sub-band Coding.--- 16 kb/s
1. Adaptive Differential Pulse Code Modulation
A digital coding scheme that uses both adaptive quantization and adaptive
prediction is called adaptive differential pulse code modulation (ADPCM).
The term adaptive means being responsive to changing level and spectrum of the input
speech signal. The variation of performance with speakers and speech material, together
with variations in signal level inherent in the speech communication process, make the
combined use of adaptive quantization and adaptive prediction necessary to achieve best
performance.
The term adaptive quantization refers to a quantizer that operates with a time-varying
step size ) (
s
nT A , where T
s
is the sampling period. The step size ) (
s
nT A is varied so as
to match the variance x
2
of the input signal ) (
s
nT x . In particular, we write
(nTs) = . ^
x
(nTs) ----- (3.57)
where Constant
^
x
(nTs) estimate of the
x
(nTs)
Thus the problem of adaptive quantization, according to (3.57) is one of estimating
) (
s
x nT continuously.
The computation of the estimate ) (
^
s
x nT in done by one of two ways:
1. Unquantized samples of the input signal are used to derive forward estimates of
) (
s
x nT - adaptive quantization with forward estimation (AQF)
2. Samples of the quantizer output are used to derive backward estimates of
) (
s
x nT - adaptive quantization with backward estimation (AQB)
The use of adaptive prediction in ADPCM is required because speech signals are
inherently nonstationary, a phenomenon that manifests itself in the fact that
autocorrection function and power spectral density of speech signals are time-varying
functions of their respective variables. This implies that the design of predictors for such
inputs should likewise be time-varying, that is, adaptive. As with adaptive quantization,
there are two schemes for performing adaptive prediction:
1. Adaptive prediction with forward estimation (APF), in which unquantized
samples of the input signal are used to derive estimates of the predictor
coefficients.
2. Adaptive prediction with backward estimation (APB), in which samples of the
quantizer output and the prediction error are used to derive estimates of the
prediction error are used to derive estimates of the predictor coefficients.
(2) Adaptive Sub-band Coding:
PCM and ADPCM are both time-domain coders in that the speech signal is
processed in the time-domain as a single full band signal. Adaptive sub-band coding
is a frequency domain coder, in which the speech signal is divided into a number of
sub-bands and each one is encoded separately. The coder is capable of digitizing
speech at a rate of 16 kb/s with a quality comparable to that of 64 kb/s PCM. To
accomplish this performance, it exploits the quasi-periodic nature of voiced speech
and a characteristic of the hearing mechanism known as noise masking.
Periodicity of voiced speech manifests itself in the fact that people speak with a
characteristic pitch frequency. This periodicity permits pitch prediction, and
therefore a further reduction in the level of the prediction error that requires
quantization, compared to differential pulse code modulation without pitch
prediction. The number of bits per sample that needs to be transmitted is thereby
greatly reduced, without a serious degradation in speech quality.
In adaptive sub band coding (ASBC), noise shaping is accomplished by adaptive
bit assignment. In particular, the number of bits used to encode each sub-band is
varied dynamically and shared with other sub-bands, such that the encoding accuracy
is always placed where it is needed in the frequency domain characterization of the
signal. Indeed, sub-bands with little or no energy may not be encoded at all.
Applications
1. Hierarchy of Digital Multiplexers
2. Light wave Transmission Link
(1) Digital Multiplexers:
Digital Multiplexers are used to combine digitized voice and video signals as
well as digital data into one data stream.
The digitized voice signals, digitized facsimile and television signals and
computer outputs are of different rates but using multiplexers it combined into a single
data stream.
Fig. 3.18: Conceptual diagram of Multiplexing and Demultiplexing.
Two Major groups of Digital Multiplexers:
1. To combine relatively Low-Speed Digital signals used for voice-grade channels.
Modems are required for the implementation of this scheme.
2. Operates at higher bit rates for communication carriers.
Basic Problems associated with Multiplexers:
1. Synchronization.
2. Multiplexed signal should include Framing.
3. Multiplexer Should be capable handling Small variations
Multiplex
er
High-Speed
Transmissio
n
line
DeMux
1
:
:
:
:
2
1
N
2
N
Digital Hierarchy based on T1 carrier:
This was developed by Bell system. The T1 carrier is designed to operate at 1.544 mega
bits per second, the T2 at 6.312 megabits per second, the T3 at 44.736 megabits per
second, and the T4 at 274.176 mega bits per second. This system is made up of various
combinations of lower order T-carrier subsystems. This system is designed to
accommodate the transmission of voice signals, Picture phone service and television
signals by using PCM and digital signals from data terminal equipment. The structure is
shown in the figure 3.19.
Fig. 3.19: Digital hierarchy of a 24 channel system.
The T1 carrier system has been adopted in USA, Canada and Japan. It is designed to
accommodate 24 voice signals. The voice signals are filtered with low pass filter having
cutoff of 3400 Hz. The filtered signals are sampled at 8KHz. The -law Companding
technique is used with the constant = 255.
With the sampling rate of 8KHz, each frame of the multiplexed signal occupies a period
of 125sec. It consists of 24 8-bit words plus a single bit that is added at the end of the
frame for the purpose of synchronization. Hence each frame consists of a total 193 bits.
Each frame is of duration 125sec, correspondingly, the bit rate is 1.544 mega bits per
second.
Another type of practical system, that is used in Europe is 32 channel system which is
shown in the figure 3.20.
Fig 3.20: 32 channel TDM system
32 channel TDM Hierarchy:
In the first level 2.048 megabits/sec is obtained by multiplexing 32 voice channels.
4 frames of 32 channels = 128 PCM channels,
Data rate = 4 x 2.048 Mbit/s = 8.192 Mbit/s,
But due to the synchronization bits the data rate increases to 8.448Mbit/sec.
4 x 128 = 512 channels
Data rate = 4 x8.192 Mbit/s (+ signalling bits) = 34.368 Mbit/s
1
2
3
4
1
2
3
4
13
14
15
16
61
62
63
64
1
2
3
4
2.048 Mbit/s
8.448 Mbit/s
34.368 Mbit/s
139.264 Mbit/s
x16 x4
(32 channels x 64
= 2048 channels)
(2) Light Wave Transmission
Optical fiber wave guides are very useful as transmission medium. They have a
very low transmission losses and high bandwidths which is essential for high-speed
communications. Other advantages include small size, light weight and immunity to
electromagnetic interference.
The basic optical fiber link is shown in the figure 3.21. The binary data fed into the
transmitter input, which emits the pulses of optical power., with each pulse being on or
off in accordance with the input data. The choice of the light source determines the
optical signal power available for transmission.
Fig: 3.21- Optical fiber link.
The on-off light pulses produced by the transmitter are launched into the optical fiber
wave guide. During the course of the propagation the light pulse suffers loss or
attenuation that increases exponentially with the distance.
At the receiver the original input data are regenerated by performing three basic
operations which are :
1. Detection the light pulses are converted back into pulses of electrical current.
2. Pulse Shaping and Timing - This involves amplification, filtering and
equalization of the electrical pulses, as well as the extraction of timing
information.
3. Decision Making: Depending the pulse received it should be decided that the
received pulse is on or off.
--END--