Chapter Two Digital Communication

Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

Chapter Two
Layout: 10 Hrs.
1. Introduction.
2. Pulse Code Modulation (PCM).
3. Differential Pulse Code Modulation (DPCM).
4. Delta modulation.
5. Adaptive delta modulation.
6. Sigma Delta Modulation (SDM).
7. Linear Predictive Coder (LPC).
8. MATLAB programs.

1
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

Lecture Five
Differential Pulse Code Modulation

Objective of Lecture:
Understand the way by which we convert the analog signal to binary bits.

Behavioral goals:
The student will be able to use new analog to bit form conversion and understand
how it possible to reduce bit rate (and improving bandwidth efficiency) by keeping
the quality of signal after reconstruction.

This lecture answer important questions which are:

What is DPCM?
Why DPCM is important?
How is DPCM done?
Where can you exploit DPCM?
What are the problems in the DPCM?

2
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

2.5. Differential Pulse Code Modulation (DPCM)

For yet another form of digital pulse modulation, we recognize that when a voice or video
signal is sampled at a rate slightly higher than the Nyquist rate, the resulting sampled signal
is found to show a high degree of correlation between adjacent samples. The meaning of
this high correlation is that, in an average sense, the signal does not change rapidly from
one sample to the next.

When these highly correlated samples are encoded as in a standard PCM system, the
resulting encoded signal contains redundant information. Redundancy means that
codewords are not absolutely essential to the transmission of information which are
generated as a result of the encoding process. By removing this redundancy before
encoding, we obtain a more efficient bits rate or bandwidth transmission, compared to
PCM (i.e.𝐵𝑃𝐶𝑀 ).

Figure 2.15 DPCM Transmitter and Receiver.

3
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

DPCM is based on linear predication. In particular, if we know the past behavior of a signal
up to a certain point in time, it is possible to make some inference about its future values;
such a process is commonly called prediction. In fact, a good prediction lead to better
performance. The linear predictor function is given as:
𝐾

̂ (𝑛𝑇𝑠 ) = ∑ 𝑎𝑖 𝑚(𝑛𝑇𝑠 − 𝑖𝑇𝑠 )

𝑚
𝑖=1

(1)

̂ (𝑛𝑇𝑠 ) is the predicted value 𝑛𝑡ℎ sample, 𝑎𝑖 is predictor coefficients, 𝑚(𝑛𝑇𝑠 −

In which, 𝑚
𝑖𝑇𝑠 ) is 𝑛𝑡ℎ past samples, and 𝐾 is predictor order. For more details about linear predictor
see appendix.

The DPCM transmitter is describe as follow:

1. First the signal passed through pre-sampled filter.

2. The analog signal is sampled at Nyquist rate or over sampled at sampling rate 𝑓𝑠 , in
which the signal converted from continues-time signal 𝑚(𝑡) to discrete-time
signal 𝑚(𝑛𝑇𝑠 ).
3. The sampled signal amplitude 𝑚(𝑛𝑇𝑠 ) move to comparator and predictor (predictor
is linear predictor can be implemented using tapped-delay line or FIR filter),
̂ (𝑛𝑇𝑠 ) based on amplitude of past
predictor estimate the current sampled value 𝑚
sampled values 𝑚(𝑛𝑇𝑠 ).
̂ (𝑛𝑇𝑠 ) move to comparator to produce error
4. The predicted sampled signal 𝑚
term 𝑒(𝑛𝑇𝑠 ) = 𝑚(𝑛𝑇𝑠 ) − 𝑚
̂ (𝑛𝑇𝑠 ), where 𝑒(𝑛𝑇𝑠 ) ≪ 𝑚(𝑛𝑇𝑠 ) .
5. The error term moved to quantizer which is round-off the value either up or down,
and
6. Then each quantization level encoded with 𝑁 𝑏𝑖𝑡𝑠, which is given as 𝑁 = log 2 (𝑀).
7. The 𝑀 × 𝑁 bits converted from parallel to serial to enter the pulse modulation which
converted the bit to electrical pulse.

4
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

8. Electrical pulse is transmitted over physical channel (we already know channel
consideration is physical).

At the receiver, the DPCM decoder uses same method as the encoder to predict the value
of the present sample. After the error term is received and demodulated, it is added to
prediction ̂ (𝑛𝑇𝑠 ) − 𝑚(𝑛𝑇𝑠 ) = 𝑒(𝑛𝑇𝑠 ) → 𝑚
(i.e.,𝑚 ̂ (𝑛𝑇𝑠 ) + 𝑒(𝑛𝑇𝑠 ) = 𝑚(𝑛𝑇𝑠 )),
producing again the actual amplitude of present sample.

If prediction algorithm is chosen properly, the error term will have lower dynamic range
than original signal (i.e., 𝑒(𝑛𝑇𝑠 ) ≪ 𝑚(𝑛𝑇𝑠 )) and thus need fewer quantizing bits to achieve
the much less quantization error which make DPCM require lower rate than PCM. DPCM
require less quantization level because it depend on error term that is represented with less
dynamic range. Error term depend on the predictor performance, if the signal well
predicted, the error term reduces and dynamic range become less which result less
quantization level (i.e. 𝑀).

2.5.1. SNR of Differential Pulse Code Modulation

SNR is important metric in communication system which measure the systems

performance, let examine DPCM SNR. In fact, of the PCM in dB is given as:
(𝑆𝑁𝑅)𝑞 = 1.8 + 6.02 𝑁
(2)
The SNR of DPCM is given by:
(𝑆𝑁𝑅)𝑞 = 1.8 + 6.02 𝑁𝐷𝑃𝐶𝑀
(3)
The number of Bits of DPCM is given as:
𝑒(𝑛𝑇𝑠 )
𝑁𝐷𝑃𝐶𝑀 = log 2
𝑞𝑒 𝑃𝐶𝑀
(4)
Finally, The SNR of DPCM is given by:

5
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

𝑒(𝑛𝑇𝑠 )
(𝑆𝑁𝑅)𝑞 = 1.8 + 6.02 log2 (5)
𝑞𝑒
𝑃𝐶𝑀

2.5.2. Transmission Bandwidth (Bit Rate) of DPCM

A signal 𝑚(𝑡) bandlimited to 𝑊 Hz, in the sequel, sampling rate 𝑓𝑠 required
is 2𝑊 𝑠𝑎𝑚𝑝𝑙𝑒/𝑠𝑒𝑐, if each quantized samples encoded to 𝑁 𝑏𝑖𝑡𝑠, then total channel
bandwidth required is given as
𝐵𝐷𝑃𝐶𝑀 = 2 𝑁𝐷𝑃𝐶𝑀 × 𝑊 = 2 log 2 (𝑀) × 𝑊 = 𝑏𝑖𝑡𝑠/𝑠𝑒𝑐 (6)
Yields,
𝐵𝐷𝑃𝐶𝑀 = 𝑁𝐷𝑃𝐶𝑀 𝑓𝑠 = 𝑏𝑖𝑡𝑠/𝑠𝑒𝑐
(7)
From (7), minimum bandwidth required for PCM is proportional to the message signal
bandwidth and number of bit per quantization level or step-size. 𝑁𝐷𝑃𝐶𝑀 < 𝑁𝑃𝐶𝑀 due to
dynamic range of DPCM signal is less than dynamic range of PCM signal (i.e., 𝑒(𝑛𝑇𝑠 ) ≪
𝑚(𝑛𝑇𝑠 )). Where, 𝑁𝐷𝑃𝐶𝑀 in term of quantization error of PCM is given as (founded by
Dr.Ahmed Alkhayyat):
𝑒(𝑛𝑇𝑠 )
𝑁𝑃𝐷𝐶𝑀 = log 2
𝑞𝑒 𝑃𝐶𝑀
(8)
Exercise 2.10: suppose we want to convert the analog signal shown in Fig. 2.16 into digital
format. If the sampling rate 8000 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 ⁄𝑠𝑒𝑐 and using uniform quantization
at 8 𝑏𝑖𝑡𝑠/𝑠𝑎𝑚𝑝𝑙𝑒. Find number of levels, Nyquist interval, number of samples,
quantization error PCM, linear predictor function, and DPCM transmitter parameters then
show how DPCM is more efficient than PCM in bit rate term? 𝑒(𝑛𝑇𝑠 ) = 0.5 𝑣.

6
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

Figure 2.16: original signal before DPCM

Solution:

Hence number of quantization level is

𝑀 = 2𝑁 = 28 = 256 𝑙𝑒𝑣𝑒𝑙𝑠

The Nyquist interval is

1 1
𝑇𝑠 = = = 0.125 𝑚𝑠𝑒𝑐
𝑓𝑠 8000

Number of samples using 𝑇𝑠 = 0.125 𝑚𝑠 is

𝑡𝑖𝑚𝑒 𝑒𝑛𝑑 𝑜𝑓 𝑠𝑖𝑔𝑛𝑎𝑙 10 𝑠𝑎𝑚𝑝𝑙𝑒𝑠

Number of Samples = = = 80
𝑠𝑎𝑚𝑝𝑙𝑖𝑛𝑔 𝑖𝑛𝑡𝑒𝑟𝑣𝑎𝑙 0.125 𝑠𝑖𝑔𝑛𝑎𝑙

The maximum quantization error of PCM is given as

𝑑𝑦𝑛𝑎𝑚𝑖𝑐 𝑟𝑎𝑛𝑔𝑒 𝑜𝑓 𝑠𝑖𝑔𝑛𝑎𝑙 4

𝑞𝑒 = ± = ± = ±0.0078125 𝑣𝑜𝑙𝑡𝑠
2𝑁+1 29

7
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

Suppose that linear predictor with order 𝐾 = 3, hence the predictor function is:

̂ (𝑛𝑇𝑠 ) = ∑ 𝑎𝑖 𝑚(𝑛 − 𝑖)
𝑚
𝑖=1

The linear predictor function after evaluating coefficients, 𝑎1 , 𝑎2 𝑎𝑛𝑑 𝑎3 , is written as:
̂ (𝑛𝑇𝑠 ) = 0.75 𝑚(𝑛 − 1) + 0.20 𝑚(𝑛 − 2) + 0.05 𝑚(𝑛 − 3)
𝑚
The amplitude of each sample after sampling process is given as:
𝑚(𝑛𝑇𝑠) = {0.23, 0.38,0.56,0.73,0.9,1.05,1.2,1.35, … . }
𝑚(𝑛𝑇𝑠) = {0𝑇𝑠 , 1𝑇𝑠 , 2𝑇𝑠 , 3𝑇𝑠 , 4𝑇𝑠 , 5𝑇𝑠 , 6𝑇𝑠 , 7𝑇𝑠 , … . }
𝑛 = {0,1,2,3,4,5,6,7, … . }

̂ (𝑛 )
DPCM transmitter is shown in the Table below which consist, 𝑛𝑇𝑠 , predicted value 𝑚
, actual value 𝑚(𝑛𝑇𝑠), error term 𝑒 (𝑛𝑇𝑠 ). Because the dynamic range of DPCM signal
depend on the error term, hence it is smallest that original signal. Where maximum
dynamic range of DPCM is ± 0.5, by which we can estimate the bit rate of DPCM as:

8
 Chapter Two Digital Communication
Waveform Encoding - DPCM BY: Dr.AHMED ALKHAYYAT

𝑑𝑦𝑛𝑎𝑚𝑖𝑐 𝑟𝑎𝑛𝑔𝑒 𝑜𝑓 𝑠𝑖𝑔𝑛𝑎𝑙 0.5

𝑞𝑒 = ± = ± = ±0.0078125 𝑣𝑜𝑙𝑡𝑠
2𝑁+1 2𝑁+1

2𝑁+1 = 64 → log 2 2𝑁+1 = log 2 64 → 𝑁 + 1 = 6 → 𝑁 = 5 𝑏𝑖𝑡𝑠 𝑝𝑒𝑟 𝑙𝑒𝑣𝑒𝑙

Hence number of bits reduced per level from 8 bits to 5 bits and gives same quantization
error. The DPCM bit rate efficiency is given as:

𝐵𝑃𝐶𝑀 = 𝑁 𝑓𝑠 = 8000 × 8 = 64 𝑘𝑝𝑏𝑠

𝐵𝐷𝑃𝐶𝑀 = 𝑁 𝑓𝑠 = 8000 × 5 = 40 𝑘𝑝𝑏𝑠
40
𝑏𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ 𝑒𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = = 62.5 %
64

Figure2.17: Original (actual), predicated and error term signal