Chapter Two Digital Communication

Waveform Encoding - DM 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 - DM BY: Dr.AHMED ALKHAYYAT

Lecture Five
Delta Modulation

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

Behavioral goals:
This lecture answer important questions which are:
What is DM?
Why DM is important?
How is DM done?
Where can you exploit DM?

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

2.6. Delta Modulation (DM)

In fact, the voice information quality is not primary importance, hence it is possible to
reduce the data rate in such the voice could be recognizable at receiver side. In order to
reduce bit rate, we need to design code modulation with less quantization levels or less
number of bits for each level, in such case, we may use another digital pulse modulation
technique known as delta modulation.

2.6.1. Transmitter and Receiver of DM

In delta modulation (DM), an incoming message signal is oversampled (i.e., at a rate much
higher than the Nyquist rate) to purposely increase the correlation between adjacent
samples of the signal. The increased correlation is done so as to allow the use of a simple
quantizing (round-off) strategy for constructing the encoded signal.
In its basic form, DM provides a staircase approximation to the oversampled version of
the message signal. Unlike PCM, the difference between the input signal and its
approximation is quantized into only two levels (only zero and one digit, that is mean
number of bits 𝑁 is 2 bits)—namely,±∆, corresponding to positive and negative differences.
Where, −∆ indicate 1 and +∆ indicate 0. In other word, 1 signify a positive error term
and 0 signify a negative error term.
Thus, if the predicted samples falls below the input signal at any sampling epoch, it is
increased by +∆, the error term is positive, hence we signify it with 1. If, on the other hand,
the approximation lies above the signal, it is diminished (reduced) by +∆, the error term is
negative, hence we signify it with 0. The DM block diagram is given in Fig. 2.16
̂ (𝑛𝑇𝑠 − 1) + ∆
𝑚 ̂ (𝑛𝑇𝑠 − 1) ≤ 𝑚(𝑛𝑇𝑠 − 1)
𝑖𝑓 𝑚
̂ (𝑛𝑇𝑠 ) = {
𝑚
̂ (𝑛𝑇𝑠 − 1) − ∆
𝑚 ̂ (𝑛𝑇𝑠 − 1) > 𝑚(𝑛𝑇𝑠 − 1)
𝑖𝑓 𝑚
(1)
The principle of DM is resemble to (mimics ‫ )مشابه‬guessing game, suppose we ask you to
guess a number between 1 and 10, after you guess, we tell that your guess was too low.
Then, for your next guess you will increase your last guess by some increment. Because

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

we deal with error term, so if error is large, we reduce it by +∆, and if the error is negative,
we increase by +∆.

Figure 2.16 DM system: (a) Transmitter and (b) receiver.

Let denote 𝑚(𝑡 ) is original signal, then 𝑚(𝑛𝑇𝑠 ) is sampled signal. The basic principle of
delta modulation may then be formalized in the following set of three discrete-time
relations:
𝑒(𝑛𝑇𝑠 ) = 𝑚(𝑛𝑇𝑠 ) − 𝑚𝑞 (𝑛𝑇𝑠 − 𝑇𝑠 )
(2)
𝑒𝑞 (𝑛𝑇𝑠 ) = ∆ 𝑠𝑔𝑛 (𝑒(𝑛𝑇𝑠 ))
(3)
Where 𝑇𝑠 is the sampling period; 𝑒(𝑛𝑇𝑠 ) is an error signal representing the difference
between the present sample 𝑚(𝑛𝑇𝑠 ) value of the input signal and output of the accumulator
𝑚𝑞 (𝑛𝑇𝑠 − 𝑇𝑠 ), and 𝑒𝑞 (𝑛𝑇𝑠) is the quantized version of 𝑒(𝑛𝑇𝑠 ) and sgn[.] is the signum

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

function, assuming the value +1 or -1. The quantizer output is finally encoded to produce
the desired DM data.

2.6.2. Transmission Bandwidth (Bit Rate) of DM

A signal 𝑚(𝑡) bandlimited to 𝑊 (𝑓𝑚 ) 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
𝐵𝐷𝑀 = 2 𝑁 × 𝑊 𝑏𝑖𝑡𝑠/𝑠𝑒𝑐 (4)

Delta modulation is signal bit allocation, i.e., 𝑁𝐷𝑀 = 2 𝑏𝑖𝑡, therefore minimum bandwidth
required is given as

𝐵𝐷𝑀 = 2𝑓𝑠 = 𝑏𝑖𝑡𝑠/𝑠𝑒𝑐

(5)

2.6.3. SNR of Delta Modulation

The signal to quantization noise ratio is written as

𝐸 {𝑚(𝑡 )2 }
𝑆𝑁𝑅 =
𝐸 {𝑒𝑞 (𝑡 )2 }
(6)
In which, 𝐸(. ) is average notation. Making the assumption that the quantization noise in
DM is uniformly distributed over ±∆, the mean-square quantization error power is
+∆
1 2 1 𝑒 3 ∆2
𝐸 {𝑒𝑞 (𝑡)2 } = ∫ 𝑒 𝑑𝑒 = =
2∆ 2∆ 3 3
−∆

(7)
We assume that this power is spread evenly over all frequencies up to the sampling
frequency 2𝑓𝑠 , then we re-write (30) as

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

∆2 1 ∆2
)2
𝐸 {𝑒𝑞 (𝑡 } = × =
3 2𝑓𝑠 6𝑓𝑠
(8)
However, there is still the lowpass filter in the DM receiver — if the cutoff frequency is
set to the maximum frequency 𝑓𝑚 , finally we write normalized noise power as:
+𝑓𝑚
∆2 ∆2 𝑓𝑚
𝐸 {𝑒𝑞 (𝑡)2 } = ∫ 𝑑𝑓 =
6𝑓𝑠 3𝑓𝑠
−𝑓𝑚

(9)
Suppose 𝑚(𝑡 ) = 𝐴 sin(2𝜋 𝑓𝑚 ), then average power signal is given as:
𝐸 {𝑚(𝑡 )2 } = 0.5 𝐴2
(10)
To avoid overload slop, the maximum amplitude is given as
∆ 𝑓𝑠
𝐴=
2𝜋𝑓𝑚
Finally, SNR of DM is given as:
𝐴2 3𝑓𝑠 ∆2 𝑓𝑠2 3 𝑓𝑠 3 𝑓𝑠3
𝑆𝑁𝑅𝐷𝑀 = × = × = × 3
2 ∆2 𝑓𝑚 8 𝜋 2 𝑓𝑚 ∆2 𝑓𝑚 8 𝜋2 𝑓𝑚
(11)
It is clear that, 𝑆𝑁𝑅𝐷𝑀 of DM is directly proportional to 𝑓𝑠3 , we can conclude that the
𝑆𝑁𝑅𝐷𝑀 improve rapidly by increasing the sampling rate which make DM performance
better and easy to detect.
In fact, Delta modulation systems are subject to two types of quantization error:

1. Slope Overload Distortions2.

2. Granular Noise

2.6.3.1. Overload Slop and Granular Problem in DM

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

The quantized signal 𝑚𝑞 (𝑡) needs to closely follow the original signal of 𝑚(𝑡) in order to
the recovered quantized signal resembles 𝑚(𝑡), see Fig 2.17. Taking a careful look at the
signal shows situations where quantized signal is unable to follow the original one as the
slope of original signal is higher than that of quantized signal.
Therefore, overload slop problem can be define as divergence between original signal slop
and the quantized signal because the quantized process (quantizer steps) cannot follow the
original signal therefore the received quantized signal will be difficult to recover.

Figure 2.17. Slope-overload distortion and granular noise, in delta modulation.

To avoid slop-overload, the condition is
𝑑
𝑚(𝑡 ) ≤ ∆ 𝑓𝑠
𝑑𝑡
(12)
If the original signal is 𝑚(𝑡 ) = 𝐴 sin(2 𝜋 𝑓𝑚 𝑡), then
𝑑
𝑚(𝑡 ) = 𝐴 (2 𝜋 𝑓𝑚 ) sin(2 𝜋 𝑓𝑚 )
𝑑𝑡
𝑑
| 𝑚 (𝑡 )| = 𝐴 2𝜋𝑓𝑚
𝑑𝑡 𝑚𝑎𝑥

Apply the condition of (35), we obtain

𝐴 2𝜋𝑓𝑚 ≤ ∆ 𝑓𝑠
∆ 𝑓𝑠
𝐴 ≤
2𝜋𝑓𝑚

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

This is to say that, the maximum amplitude of the input signal should be less than or equal
∆ 𝑓𝑠
to to avoid slop over load.
2𝜋𝑓𝑚

Also it possible to conclude that optimum step size for sine wave signal is given as
𝐴 2𝜋𝑓𝑚
∆𝑜𝑝𝑡𝑖𝑚𝑢𝑚 =
𝑓𝑠
Granular noise, In contrast to slope-overload distortion, granular noise occurs when the
step size is too large relative to the local slope characteristic of the original message signal
𝑚(𝑡 ).

Exercise 2.11: delta modulation system, the message signal is 𝑚(𝑡 ) = 0.1 sin(2𝜋 10−3 𝑡),
and 𝑓𝑠 = 20 𝐾𝐻𝑧, does slop overload occur at step size ∆ = 60 𝑚𝑉 or ∆ = 4 𝑚𝑉?

Solution:

𝑑
| 𝑚 (𝑡 )| = 0.1 × 2𝜋 × 10−3 = 200𝜋
𝑑𝑡 𝑚𝑎𝑥

200𝜋 ≤ ∆𝑓𝑠

200𝜋 200 𝜋
∆≥ → → ∆ = 31.4 𝑚𝑉
𝑓𝑠 20 000

The optimum step size is ∆ > 31.4 𝑚𝑉 to avoid slop overload error. Therefore, slop
overload does not occur at step size 60 mV, while slop overload occur at step size 4 mV.

Exercise 2.12: Find sine wave signal amplitude for minimum slop over load error in DM
system. If the step size is 1V with sampling interval (repetition signal) 𝑇𝑠 = 1ms and
message signal frequency is 100Hz.

Solution:

The maximum signal amplitude is given by

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

∆ 𝑓𝑠 1 × 1000 1000
𝐴 ≤ ≤ ≤ ≤ 1.59 𝑉
2𝜋𝑓𝑚 2 × 𝜋 × 𝑓𝑚 2𝜋100
𝐴 = 1.59 𝑉 𝑖𝑠 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑎𝑚𝑝𝑙𝑖𝑡𝑢𝑑𝑒
Exercise 2.13: signal to be transmitted using DM, where message signal is given as
𝑚(𝑡 ) = 10 cos 1000𝜋𝑡 + 5 cos 1500𝜋𝑡
1. Determine an appropriate 𝑓𝑠 and step size for the DM.
2. Determine SNR for DM system.

Solution:

1. The appropriate 𝑓𝑠 is given as

𝑓𝑠 = 2 𝑓𝑚

Where, 𝑓𝑚 is obtain from the message signal as

1500
2𝑓𝑚 𝜋𝑡 = 1500𝜋𝑡 → 𝑓𝑚 = = 750 𝐻𝑧
2

Then simply we can find sampling frequency (Nyquist rate) as

𝑓𝑠 = 2 × 750 = 1.5 𝑘𝐻𝑧

Nyquist rate for DM system should be several time larger than Nyquist rate condition, i.e.,
10 times of normal value, hence 𝑓𝑠 is given as
𝑓𝑠(𝐷𝑀) = 10 𝑓𝑠 = 10 × 1.5 = 15 𝑘𝐻𝑧
The appropriate step-size is obtain as
𝑑
| 𝑚 (𝑡 )| ≤ ∆ 𝑓𝑠
𝑑𝑡 𝑚𝑎𝑥
𝑑
𝑚(𝑡) is obtain as
𝑑𝑡

𝑑
| 𝑚 (𝑡 )| = 10 × 1000𝜋 + 5 × 1500𝜋 = 17500𝜋
𝑑𝑡 𝑚𝑎𝑥

9
Chapter Two Digital Communication
Waveform Encoding - DM BY: Dr.AHMED ALKHAYYAT

Then, step size is obtain as

17500 𝜋
17500 𝜋 ≤ ∆ 15000 → ∆ = = 3.67 𝑉
15000
2. SNR of DM is obtain as
3
3 𝑓𝑠(𝐷𝑀) 3 150003
𝑆𝑁𝑅𝐷𝑀 = × = × = 304.27
8 𝜋2 𝑓𝑚3 8 𝜋2 7503
𝑆𝑁𝑅𝐷𝑀 = 10 log10 304.27 = 24.8 𝑑𝐵

Question: is it possible to improve signal to noise ratio without increasing transmitted

power to detect the received signal easily?

The answer is YES, SNR increase through increasing sampling rate which increase the
correlation between samples that improve signal reconstruction at receiver. Therefore, let
make sampling frequency 100 times more of message frequency, as result we write 𝑓𝑠 as
𝑠𝑎𝑚𝑝𝑙𝑒𝑠
𝑓𝑠 = 20 𝑓𝑚 = 100 × 750 𝐻𝑧 = 75
𝑠𝑒𝑐
Therefore, 𝑆𝑁𝑅𝐷𝑀 is written as
3
3 𝑓𝑠(𝐷𝑀) 3 750003
𝑆𝑁𝑅𝐷𝑀 = 10 log10 × = 10 log10 × = 45.8 𝑑𝐵
8 𝜋2 𝑓𝑚3 8 𝜋2 7503
Question: is it possible to increase sampling rate to infinity?
The answer NO, because increasing sampling rate required sophisticated devices which
increase the device complexity and cost.

Exercise 2.14: if 25 𝑑𝐵 SNR of DM systems, determine the quantized bit rate for analog
signal with bandwidth (𝑓𝑚 ) 3400 kHz?

Solution:

We need SNR in linear form instead of logarithmic form

11
Chapter Two Digital Communication
Waveform Encoding - DM BY: Dr.AHMED ALKHAYYAT

𝑆𝑁𝑅 25
𝑆𝑁𝑅𝐷𝑀 = 10 10 = 1010 = 316.22

Then we apply SNR of DM as

3 3
3 𝑓𝑠(𝐷𝑀) 𝑓𝑠(𝐷𝑀) 316.22 × (3400)3 3
𝑆𝑁𝑅𝐷𝑀 = × → 316.22 = 0.04 × → = 𝑓𝑠(𝐷𝑀)
8 𝜋2 𝑓𝑚3 3400 0.04

3
𝑓𝑠(𝐷𝑀) = 310 × 1014 → 𝑓𝑠(𝐷𝑀) = 67,729
Due to that DM is single bit allocation, hence the bit rate transmission of DM is given as

𝐵𝐷𝑀 = 𝑓𝑠(𝐷𝑀) = 67 𝑘𝑏𝑝𝑠

Exercise 2.15: The input signal is 𝑚(𝑡 ) = 0.01 𝑡. The modulator operates at sampling
frequency of 20𝐻𝑧 and has step size 2 𝑚𝑉. Sketch the delta modulator output?

Solution:

Nyquist interval is obtain as

1 1
𝑇𝑠 = = = 0.05
𝑓𝑠 20

We need to find number of samples, which is obtain as:

𝑡 1
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑎𝑚𝑝𝑙𝑒𝑠 = = = 20 𝑠𝑎𝑚𝑝𝑙𝑒𝑠/𝑠𝑒𝑐
𝑇𝑠 0.05

11
Chapter Two Digital Communication
Waveform Encoding - DM BY: Dr.AHMED ALKHAYYAT

Let describe how delta modulation work is:

1. At 𝑇𝑠 = 0, always 𝑚(𝑡 ) > 0 , allocate 1 digit, next +∆ (added to next step)
2. At 𝑇𝑠 = 0.05, the 𝑚(𝑡 ) < +∆, 𝑚(𝑡) is low, allocate 0 digit, next −∆
3. At 𝑇𝑠 = 0.1, the 𝑚(𝑡 ) > 0, 𝑚(𝑡) is high, allocate 1 digit, next +∆
4. At 𝑇𝑠 = 0.15, the 𝑚(𝑡 ) < +∆, 𝑚(𝑡) is low, allocate 0 digit, next −∆
5. At 𝑇𝑠 = 0.2, the 𝑚(𝑡 ) > 0, 𝑚(𝑡) is high, allocate 1 digit, next +∆
6. At 𝑇𝑠 = 0.25, the 𝑚(𝑡 ) > ∆, 𝑚(𝑡) is high, allocate 1 digit, next +∆
7. At 𝑇𝑠 = 0.3, the 𝑚(𝑡 ) < 2∆, 𝑚(𝑡) is low, allocate 0 digit, next −∆
8. At 𝑇𝑠 = 0.35, the 𝑚(𝑡 ) > ∆, 𝑚(𝑡) is high, allocate 1 digit, next +∆
9. At 𝑇𝑠 = 0.4, the 𝑚(𝑡 ) < 2∆, 𝑚(𝑡) is low, allocate 0 digit, next −∆
10. At 𝑇𝑠 = 0.45, the 𝑚(𝑡 ) > ∆, 𝑚(𝑡) is high, allocate 1 digit, next +∆
11. At 𝑇𝑠 = 0.5, the 𝑚(𝑡 ) < 2∆, 𝑚(𝑡) is low, allocate 0 digit, next +∆
12. At 𝑇𝑠 = 0.55, the 𝑚(𝑡 ) > 3∆, 𝑚(𝑡) is high, allocate 1 digit, next −∆
13. At 𝑇𝑠 = 0.6, the 𝑚(𝑡 ) < 2∆, 𝑚(𝑡) is low, allocate 0 digit, next +∆
14. At 𝑇𝑠 = 0.65, the 𝑚(𝑡 ) < 3∆, 𝑚(𝑡) is low, allocate 0 digit, next +∆
15. At 𝑇𝑠 = 0.7, the 𝑚(𝑡 ) > 4∆, 𝑚(𝑡) is high, allocate 1 digit, next −∆

12
Chapter Two Digital Communication
Waveform Encoding - DM BY: Dr.AHMED ALKHAYYAT

Exercise 2.16: use DM to covert signal given bellow into binary form, then find sampling
frequency, message frequency and 𝑆𝑁𝑅𝐷𝑀 . The optimum step size ∆ = 6.25 𝑚𝑉.

Solution:

From the figure bellow, frequency message is

1 1
𝑓𝑚 = = = 2000𝐻𝑧
𝑡 0.5 × 10−3

The sampling frequency of DM is given as

𝑓𝑠 = 5 × 𝑓𝑚 = 5 × 2000 = 10,000 𝐻𝑧

The SNR of DM is given as

3
3 𝑓𝑠(𝐷𝑀) 10,000^3
𝑆𝑁𝑅𝐷𝑀 = × → 0.04 × =5
8 𝜋2 𝑓𝑚3 2000^3

13
Chapter Two Digital Communication
Waveform Encoding - DM BY: Dr.AHMED ALKHAYYAT

Figure 2.18. DM Process, (a) message signal, (b) quantized signal, (c) binary
representation

14