Differential Pulse Code

Modulation (DPCM)
DPCM
◦ Voice or video signal is sampled at a rate slightly
higher than Nyquist rate done in PCM.
 Found to be highly correlation between adjacent samples
 The adjacent samples of the signal carry the same information
with little difference.
◦ Consider the difference value of the two samples.
◦ Over all bit rate will decrease and number of bits
required to transmit one sample will also be reduced.
◦ Based on the principle of prediction.
 Value of present sample is predicted from the past.
 Prediction may not be exact but it is very close to
the actual value

8/20/2020
4.2 ECE305 Digital Communication
 PCM – Digitization of a Voice (or) Video
signal-
◦ signal is sampled at a rate slightly higher than
the Nyquist rate.
 Resulting sampled signal exhibit high
correlation between adjacent samples.
 High Correlation – signal does not change
rapidly form one sample to next sample.

8/20/2020 ECE305 Digital Communication 3

 Highly correlated samples are encoded –
PCM-resulting encoded signal contains
redundant information.
 By removing this redundancy before
encoding – more efficient coded signal.

8/20/2020 ECE305 Digital Communication 4

 Use the Linear Prediction Techniques.
 If we know the past behavior of a signal –
it is possible to make some inference
about its future values.

8/20/2020 ECE305 Digital Communication 5

DPCM Transmitter

Sampled
input
x(nTs) e(nTs) v(nTs) b(nTs)
∑ QUANTIZER ENCODER
+ DPCM
_ O/P
+
+
∑
x’(nTs)

PREDICTOR
u(nTs)

8/20/2020 ECE305 Digital Communication 6

 e(nTs) = x(nTs)-x’(nTs) ----------------- (1)
 v(nTs) =Q[e(nTs)]
= e(nTs)+q(nTs) -------------------- (2)
q(nTs) – Quantization Error.
u(nTs) = x’(nTs)+v(nTs) ------------------- (3)
Using (2) & (3)
u(nTs) = x’(nTs)+e(nTs)+q(nTs) ---------(4)
From (1)
u(nTs) = x(nTs)+q(nTs)

8/20/2020 ECE305 Digital Communication 7

DPCM Receiver

OUTPUT
INPUT +
DECODER ∑

PREDICTOR

8/20/2020 ECE305 Digital Communication 8

Signal to Quantization Noise Ratio
 (SNR)o = σ2X / σ2Q
◦ σ2X = Variance of the original input signal x(nTs)
◦ σ2Q= Variance of the Quantization error.
◦ σ2E = Variance of the predictor error.
 Rewrite the Above Equation
◦ (SNR)o = (σ2X / σ2E ) (σ2E / σ2Q )
= Gp (SNR)p
◦ Gp – Predictor gain.
◦ (SNR)p – Predictor error to quantization noise ratio.

8/20/2020 ECE305 Digital Communication 9

Mid tread and Mid riser

8/20/2020 ECE305 Digital Communication 10

Problems
 Using midriser type with step size of 1.
Find the DPCM quantized output for the
following samples ( 0, 0.3, 1.5, 0.7, 1,2.3)

8/20/2020 ECE305 Digital Communication 11

8/20/2020 ECE305 Digital Communication 12
DPCM Encoder

8/20/2020 ECE305 Digital Communication 13

8/20/2020 ECE305 Digital Communication 14
Problems-DPCM Quantization using
midriser type

8/20/2020 ECE305 Digital Communication 15

8/20/2020 ECE305 Digital Communication 16
Why Delta Modulation?
 If we oversample ( typically more than 4 times
of Nyquist rate) a signal, the correlation
between the adjacent samples increases and a
result, the predictor error decreases.
 If the predictor error is small enough, it can be
encoded by only one bit.
 DM provides stair case approximation.
 Delta Modulation (DM) is basically one-bit
version of DPCM.
Applications
 Delta Modulation was used by Satellite Business
Systems or SBS for its voice ports to provide long
distance phone service to large domestic corporations
with a significant inter-corporation communications
need (such as IBM).
 24kbit/s Delta Modulator achieved full voice quality with
no discernable degradation as compared to a high quality
phone line or the standard 64kbit/s µ-law Companded PCM.
 With the increasing availability of FPGAs and game-
related ASICs,
 sample rates are easily controlled so as to avoid
slope overload and granularity issues.

8/20/2020 ECE305 Digital Communication 18

Delta Modulation
 Delta Modulation transmit only one bit
per sample.
 The present sample value is compared
with the previous sample value.
 The difference between x(t) and staircase
approximated signal confined to two
levels i.e +δ and - δ.
Delta Modulation
 Signaling rate and channel bandwidth are
large in PCM, to overcome this problem
DM is used.
 ONE BIT Quantizer or TWO level
Quantizer(+ ∂ or - ∂).
 ∆=2 ∂

 Present sample value is compared with the previous

sample value and the indication whether the amplitude
is increased or decreased is sent.
8/20/2020 ECE305 Digital Communication 20
Delta Modulation

8/20/2020 ECE305 Digital Communication 21

 Difference between the input signal and
staircase approximated signal confined to
two levels.
 +Ve ∂ --signal is increased by one step---- ‘1’
 -Ve ∂ --signal is reduced by one step-------‘0’

 e(n Ts)= x(n Ts)- x’(n Ts)

 x(n Ts)- Present sample
 x’(n Ts)-Last sample

8/20/2020 ECE305 Digital Communication 22

DM Transmitter

Sampled
input
+ e(nTs) ONE BIT- b(nTs)
x(nTs) ∑ QUANTIZER
_ OUT PUT
+
+ ∑
x’(nTs)

DELAY Ts
u(nTs)
Accumulator
e(nTs) = x(nTs)-x’(nTs)
= x(nTs)-u(nTs-Ts)

u(nTs) = u(nTs-Ts)+b(nTs)

b(nTs) = δ sgn[e(nTs)]
 +ve δ if x(n Ts)> x’(n Ts)-----------Symbol ‘1’ transmitted
 -ve δ if x(n Ts) < x’(n Ts)-------------Symbol ‘0’ transmitted
DM Receiver

INPUT OUTPUT
+ LOW-PASS
∑
FILTER
+

DELAY Ts

Accumulator
DM -Features
1. One bit code word for the output, which
eliminates the need for word framing.

2. Simplicity of design for both the transmitter

and receiver.

8/20/2020 ECE305 Digital Communication 26

DM- Types of quantization error
Slope overload Noise
 Granular Noise

8/20/2020 ECE305 Digital Communication 27

 Condition to avoid slope over load
◦

8/20/2020 ECE305 Digital Communication 28

DM-Signal to noise ratio Derivation
 Consider the effect of quantization noise under no slope overload.
When there is no slope overload, the maximum quantization error
is + ∂ or - ∂ (uniformly distributed ).The receiver contains a LPF
whose bandwidth is set equal to the message bandwidth( i.e
highest possible frequency component of the message signal), is
denoted by W and Ts.
 Assume the use of sinusoidal modulation

◦ Assume Avg. signal and Noise Power of the quantization noise is

uniformly distributed over a frequency interval extending from -
1/Ts to 1/Ts.

 Solutions
◦ Max .SNR of a DM is proportional to the sampling rate of cubed.
Therefore 9 dB Improvement is achieved.
8/20/2020 ECE305 Digital Communication 29
8/20/2020 ECE305 Digital Communication 30
Delta Modulation
 Adv
◦ Transmit only one bit per sample.
◦ Signaling rate and channel BW is small.
 Disadv
◦ Granular Noise
 Step size is too large when compared to small variations in the input
signal.
◦ Slope overload distortion
 Error between original signal and staircase approximated
signal . To reduce this error , step size should be increased
when slope of input signal is high.

8/20/2020 ECE305 Digital Communication 31

8/20/2020 ECE305 Digital Communication 32
Non Uniform Quantizer

8/20/2020 ECE305 Digital Communication 33

8/20/2020 ECE305 Digital Communication 34
8/20/2020 ECE305 Digital Communication 35
Non Uniform Quantizer
• In telephonic communication , it is preferable to use a variable separation
between the levels.
• Loud talk
• Weak Talk
• Able to reduce the quantization noise power for low amplitude signals,
the average SNR improved .
• Step size can be made smaller for smaller signals and larger for larger
signals . Its difficult to implement.
• Practical approach
– Predistort the signals by compression characteristics and put in to
uniform quantizer.
– COMPANDING ( Compressed and Quantized signal )
– Block diagram (i/p, Compressor , Uniform quantizer and Expander
,o/p)

• Two Compression Laws

• A-law and μ law .
Non Uniform Quantizer
• Able to reduce the quantization noise power for low amplitude
signals, the average SNR improved .
• Step size can be made smaller for smaller signals and larger for larger
signals . Its difficult to implement.
• Practical approach
– Predistort the signals by compression characteristics and put in to
uniform quantizer.
– COMPANDING ( Compressed and Quantized signal )
– Block diagram (i/p, Compressor , Uniform quantizer and
Expander)
• Two Compression Laws
• A-law- Europe and  law- US .
Non Uniform Quantizer
 For voice comm.channels,50% of the time, the voltage
characterizing detected speech energy is less than one fourth
of the rms value.

 Large amplitude values are relatively rare; 15% of the time

does the voltage exceed the rms value.
◦ Uniform system would be wasteful for speech signals.
◦ With uniform Quantisation, SNR is worse for low level signal than for
high level signals.
◦ Non Uniform Provide fine quantisation of the weak signals .

4.38
4.39
Compression Techniques
A - law compressor
w1 (t )  1, A0
 A w1 (t ) 1
 0  w1 (t ) 
 1 + ln A A
w2 (t ) = 
1 + ln ( A w1 (t ) ) 1
  w1 (t )  1
 1 + ln A A

 - law compressor
(very popular internatio nally)

w1 (t )  1
ln (1 +  w1 (t ) )
w2 (t ) =
ln (1 +  )
In the U.S.,  = 255 is used.
 X=Amplitude of the input signal at a
particular instant of time (v)
 y= compressed output (v)
 µ = unitless parameter used to define yhe
amount of compression
 Xmax = maximum amplitude of the
uncompressed analog input signal (v)
 Ln= natural logarithm

8/20/2020 ECE305 Digital Communication 41

8/20/2020 ECE305 Digital Communication 42
Non Uniform Quantizer

4.43
Adaptive Delta Modulation
 To overcome the quantization errors due to slope overload and
granular noise, step size is made adaptive to variations in the input
signal.
 DM can be improved by making the step size of the modulator by
time varying form.
◦ Step size is adapted to the level of input signal.
◦ Signal approximation -Increased or reduced.
◦ Block diagram(Tx ,RX)-LOGIC for STEP SIZE control
 For e.g. If Quantizer output is high(1) , step size may be increased. If
Quantizer output is Low (0),step size may be reduced
 Step size is constrained to lie between minimum and maximum
values.
 ∂min < = ∂(nTs) <= ∂max
 ∂max- Controls the amount of slope overload distortion.
 ∂min – channel noise

8/20/2020 ECE305 Digital Communication 44

ADM
 In receiver ,first part generates the step
size from each incoming bit.(same as Tx)
 Previous input and present input decides
the step size.
 Accumulator forms staircase waveform.
 LPF-to reconstruct the smooth signal

8/20/2020 ECE305 Digital Communication 45

8/20/2020 ECE305 Digital Communication 46
8/20/2020 ECE305 Digital Communication 47
8/20/2020 ECE305 Digital Communication 48
Line codes-Baseband Shaping
Format
 In telecommunication, a line code (also
called digital baseband modulation) is
a code chosen for use within a
communications system for baseband
transmission purposes.
 Line coding is often used for digital data
transport.
 Performance of line code against two
important parameters
◦ Symbol pulse and Ts
Line codes
 Electrical representation of binary data
stream (single bit)
a) Unipolar NRZ
b) Polar NRZ signalling
c) Unipolar RZ signalling
d) Bipolar NRZ or AMI
e) Manchester code.

 Dibits- Natural Code and Gray Code

Dibits
Level Natural Code Gray Code

-3 00 00

-1 01 01

1 10 11

3 11 10

• Gray Code- Adjacent bits are adjusted by only one bit

8/20/2020 ECE305 Digital Communication 53

Natural and Gray Code

8/20/2020 ECE305 Digital Communication 54

Differential Encoding

◦ Differential Encoding – Is used to provide

polarity reversal protection
◦ A Differential Coding system consists of a
modulo 2 adder operation

◦ For e.g. i/p is 10010011, o/p is 110110111

8/20/2020 ECE305 Digital Communication 55

Formatting alphanumeric text bypasses the sampling and
quantization operation and directly encodes the text.
References:
[1]. Simon Haykin, “Digital Communication”,
John Wiley, 2009
[2]. Simon Haykin, “Communication
system”, John Wiley.

8/20/2020 ECE305 Digital Communication 57