(CH 6) Pulse and Digital Modulation

Chapter 6

Pulse and Digital Modulation

Sampling Theorem:

A signal band limited to B Hz by regularly-spaced samples, provided sampling rate

is at least 2B sample per second.

F𝑠 ≥ 2𝐵 F𝑠 ≥ 2𝑓𝑚𝑎𝑥

Or …(6-1)

F𝑠 = 2𝑓𝑚𝑎𝑥 Minimum sampling rate or Nyquist sampling rate

…(6-2)

G(f)
g(t)
f Max

t f
-B 0 B
Base band
frequency
gs(t) Gs(f)

1
Ts =
fs
t f
-B 0 B
Ts -2Fs -Fs Fs 2Fs

∞ ∞
1
𝑔𝑠 (𝑡) = 𝑔(𝑡) ∑ 𝛿(𝑡 − 𝑛𝑇𝑠 ) = ∑ 𝑔(𝑡)𝑒 𝑗𝑛𝜔𝑠 𝑡
𝑇𝑠
𝑛=−∞ 𝑛=−∞

∞
1
𝐺𝑠 (𝜔) = ∑ 𝐺(𝜔 − 𝑛𝜔𝑠 )
𝑇𝑠
𝑛=−∞

137
(CH 6) Pulse and Digital Modulation

Effect of sampling on a signal spectrum:

B=3 kHz H(f)

Speech signal

f (k Hz)
-3 0 3
* Fs=8 kHz
Hs(f)
8000 sample / second

f (k Hz)
-11 -8 -5 -3 0 3 5 8 11 13 16 19
* Fs=6 kHz=2B Hs(f)
(Nyqiust rate)

f (k Hz)
-9 -6 -3 0 3 6 9 12 15

* Fs=5 kHz Hs(f) (aliasing error)

(overlap)

f (k Hz)
-10 -5 -3 -2 0 2 3 5 7 8 10

Ex 6-1: Determine the Nyquist rate of the sampling for the signal:

𝑔(𝑡) = 10 cos 100𝜋𝑡 + 15 cos 150𝜋𝑡 + 5𝑐𝑜𝑠300𝜋𝑡

Solution:

300𝜋
𝑓𝑚𝑎𝑥 = = 150 𝐻𝑧
2𝜋

𝑁𝑦𝑞𝑢𝑖𝑠𝑡 𝑟𝑎𝑡𝑒 = 2𝑓𝑚𝑎𝑥 = 2 × 150 = 300 𝐻𝑧

H.W:

Determine the Nyquist rate of sampling required for

a) 𝑔(𝑡) = 10 cos 100𝜋𝑡 cos 200𝜋𝑡

b) 𝑔(𝑡) = 𝑒 −2|𝑡| (approximate the BW where|𝐺(𝜔)| drops to value less than 0.1)

138
(CH 6) Pulse and Digital Modulation

The sampler behaves exactly as a multiplier. It multiplies f(t) by a gating function x(t)

which is a train of impulses with frequency fs then,

x(t)
𝑓𝑠 (𝑡) = 𝑓(𝑡). 𝑥(𝑡)

t
Ts
Reconstruction of f(t) from fs(t) is

done using LPF with cutoff fmax

x(t)

fs(t) LPF f(t)

R fs(t)

(Sampler) (Reconstruction cct)

Note:
F(f)
If f(t) is a baseband signal over a frequency

range fL to fu such that B=fu- fL then sampling

f
FL FU
theorem of such signals states that f(t) can be

completely recovered from fs(t) if:

2𝑓𝑢 𝑓𝑢
𝑓𝑠 = , where k is the largest integer not exceeding
𝑘 𝐵

Ex 6-2:
2∗(25) F(f)
For the signal shown besides, 𝐹𝑠 = where 𝑘 =
𝑘
25
𝑖𝑛𝑡 ( )=2
10
f
2×(25) 15 25
𝑓𝑠 = or 𝑓𝑠 = 25 kHz (kHz)
2

Reconstruction in such case has done using BPF

139
(CH 6) Pulse and Digital Modulation

Pulse Modulation Techniques:

If an analog signal is sampled, the sampled values may be used to modify certain

parameter of a periodic train pulses (amplitude, width or position). Accordingly, we have

Pulse Amplitude Modulation (PAM), Pulse Width Modulation (PWM) or Pulse Position

Modulation (PPM).

f(t)

fs (t)

t
TS 2TS 3TS 4TS

PAM

PWM

PPM

140
(CH 6) Pulse and Digital Modulation

1- PAM
fs(t)
f(t) Sampler Q(ω) Φ PAM(t)
Generation:

f(t) F(ω)

t ω
ωm

fs (t)
Fs(ω)

t ω
2π
Ts
q(t) Q(ω)
τ

t ω
2π
τ
fs(t) q(t)

t ω
2π
T

Pulse Amplitude modulation the same as the output of the sampled at rate 𝑓𝑠 (𝑓𝑠 ≥

2𝑓𝑚𝑎𝑥 𝑓𝑜𝑟 𝑏𝑎𝑠𝑒𝑏𝑎𝑛𝑑 𝑠𝑖𝑔𝑛𝑎𝑙𝑠).

∅𝑃𝐴𝑀 (𝑛𝑇𝑠 ) = ∑∞
𝑛=−∞ 𝑓(𝑛𝑇𝑠 ) 𝑞(𝑡 − 𝑛𝑇𝑠 ) … (6-3)

PAM is usually used as TDM-PAM (TDM=Time Division Multiplexing) to

transmit more than one message at the same channel.

Detection:

Φ PAM(t) LPF Q-1(ω) f(t)

141
(CH 6) Pulse and Digital Modulation

2- PWM and PPM:

Generation:

Sample a + c
d e

M
f(t) Pulse
& Hold + Comp PPM
b Vth
generator
Ramp.
generator PWM

Clock
f(t)
T

τ
t
a
r

C R o/p t
b

𝑟𝑐 ≪ 𝜏
t
𝑅𝐶 ≫ 𝑇 Sample & Hold circuit
c
Detection:
Vth
One method of detection is to
t
convert PWM or PPM signals to d

PAM ones using ramp generator PWM

t

starts at kTS, stops at tk, restarts at e

PPM
(k+1)TS and so forth.
t

142
(CH 6) Pulse and Digital Modulation

PWM

PPM

PAM
kTS tk (k+1)TS

PWM and PPM, are rarely used now in communication systems.

Time Division Multiplexing (TDM):

A mode of transmission in which simultaneous transmission of several

baseband signals on time-sharing basis is possible.

f1(t) TDM-PAM
f2(t)

t
TS TX

1
TS: Sampling time for each signal (𝑇𝑠 ≤ 𝑓𝑜𝑟 𝑛𝑦𝑞𝑢𝑖𝑠𝑡 𝑠𝑎𝑚𝑝𝑙𝑖𝑛𝑔)
2𝑓𝑚𝑎𝑥

Tx: Clock frequency for PAM/TDM system.

𝑇
𝑇𝑥 = 𝑠 ; where N: number of messages (channels)
𝑁

143
(CH 6) Pulse and Digital Modulation By Dr. Hikmat Al-Shamary & Dr. Tariq M. Salman

Timing Circuit
Pulse Generator

f1(t)
f2(t)
f3(t) To Line
Sampler LPF
or
transmitter

fN(t)
Tx

Commutator

Timing Circuit

LPF f1(t)
LPF f2(t)

Received LPF f3(t)

signal

LPF fN(t)

Rx
Synchronus
commutator

If N identical messages have the same 𝑓𝑚𝑎𝑥 , then 𝑓𝑠 ≥ 2𝑁𝑓𝑚𝑎𝑥 at the channel the

sample rate is at least (2Nfmax).

The minimum required bandwidth of the channel is half the

sampling frequency or
BW𝑚𝑖𝑛 ≥ 𝑁𝑓𝑚𝑎𝑥
Hz for TDM-PAM ….(6-4)

144
(CH 6) Pulse and Digital Modulation

Ex 6-3:

Twelve speech signals are TDM-PAM transmitted, find minimum sample rate at the

channel and minimum required BW.

Solution:

𝑓𝑚𝑎𝑥 =3.5 kHz (for speech)

𝑓𝑠 ≥ 2𝑁𝑓𝑚𝑎𝑥 or 𝑓𝑠 ≥ 2 × (12) × (3.5 𝑘𝐻𝑧)

𝑓𝑠𝑚𝑖𝑛 =84 kHz

1
𝐵𝑊𝑚𝑖𝑛 = 𝑓𝑠𝑚𝑖𝑛 = 42 𝑘𝐻𝑧
2

Ex 6-4:

Determine the minimum transmission BW in a TDM system transmitting 20

different messages, each message signal have BW of 5 kHz; compare the result if FDM is

used with AM & SSB techniques.

Solution:

• TDM

𝑓𝑚𝑖𝑛 = 2𝑁𝑓𝑚𝑎𝑥 = 2 × (20) × (5 𝑘𝐻𝑧)=200 kHz

𝐵𝑊𝑚𝑖𝑛 = 100 𝑘𝐻𝑧

• FDM

AM: 𝐵𝑊𝑚𝑖𝑛 = 2(5 ∗ 20)𝑘𝐻𝑧 = 200 𝑘𝐻𝑧

SSB: 𝐵𝑊𝑚𝑖𝑛 = 5 ∗ 20 𝑘𝐻𝑧 = 100 𝑘𝐻𝑧

∴ TDM/PAM is more efficient in terms of BW than FDM/AM

145
(CH 6) Pulse and Digital Modulation

Digital Communication System:

Digital communication has several advantages over analog communication:

1- Digital communication has high immunity to channel noise and channel

distortion.

2- Regenerative repeaters along the transmission path can detect and retransmit a

new, clean signal.

3- Digital hardware implementation is flex able (it may use microprocessors, digital

switching and LSI-ICs)

4- Digital signals can be added to yield low error and high fidelity as well as privacy.

5- It is easier to multiplex digital signals.

6- Exchange of SNR and BW can be done more effectively.

Digital Transmission of Analogue Signals:

1- Pulse Code Modulation (PCM):

This is widely used in digital transmissions. Its block diagram is as shown below:

Channel DAC
f(t) LPF Sampler Quantizer Encoder Trans. Decoder LPF

fs ADC

ADC: Analogue to digital converter.

DAC: Digital to Analogue converter.

146
(CH 6) Pulse and Digital Modulation

The output of the sampler 𝑓𝑠 (𝑘𝑇𝑠 )(𝐴𝐷𝐶). Assuming that f(t) has ±𝑓𝑝 peak voltage

level, (ADC full scale), the quantizer will divide the +𝑓𝑝 to −𝑓𝑝 range into L equally

spaced intervals of size ∆𝑉 (step size) then:

2𝑓𝑝
∆V =
𝐿
volt …. (6-5)

Quantizing Noise:

Since the quantization process introduces some fluctuations about the true value,

these fluctuations can be regarded as noise. As the number of quantization levels L

increases, the quantization noise decreases.

𝑓𝑝2
N𝑞 = 2 Volt2 …. (6-6)
3𝐿

147
(CH 6) Pulse and Digital Modulation

Encoding:

ADC will then encode the quantized values according to a certain binary code. The

uniform PCM with equal step size mostly uses the signed binary code of n bits.

n bits
fS(kTS) +
0101
+
0100
Magnitude +
0011
bits
+
Sign 0010
bit { +- 1
0
+
0001
+
0000
- t
1001
- V V
1010
-
1011
-
1100

For n=4, then the ±𝑓𝑝 values will be encoded as shown above, this is called transfer

characteristic of the PCM encoder. The relation between number of quantizing levels and number

of bits of encoder is:

𝐿 = 2𝑛 Or 𝑁 = 𝑙𝑜𝑔2 𝐿 … … (6-7)

Note:

If n for a given value of L is not integer number,

Then n is computed using 𝑛 = 𝑖𝑛𝑡(𝑙𝑜𝑔2 𝐿) + 1,

and L is corrected using 𝐿 = 2𝑛

148
(CH 6) Pulse and Digital Modulation

The output SNR:

S𝑜 3𝐿2 ̅̅̅̅̅̅̅
𝑓 2 (𝑡)
=
N𝑞 𝑓𝑝2 Volt2 …. (6-8)

Note:

S𝑜 S𝑜
𝑁𝑜 = 𝑁𝑞 ; =
N𝑜 N𝑞

𝐴 2
• For tone modulation: ̅̅̅̅̅̅̅
𝑓 2 (𝑡) = ; 𝑓𝑝 = 𝐴
2

S𝑜 3𝐿2
=
N𝑞 2
F ... (6-9)

S𝑜
( ) = 1.76 + 20 𝑙𝑜𝑔 𝐿 = 1.76 + 6.02 𝑛
F N𝑞 𝑑𝐵 ... (6-10)

Bandwidth Requirement of PCM

The information rate of PCM channel is 𝑛𝑓𝑠 bits/sec, if message bandwidth is 𝑓𝑚𝑎𝑥

and the sampling rate is 𝑓𝑠 (≥ 2𝑓𝑚𝑎𝑥 ) then 𝑛𝑓𝑠 binary pulses must be transmitted per

second.

Assuming the PCM signal is a low-pass signal of bandwidth 𝐵𝑊𝑃𝐶𝑀 , the required

minimum sampling rate is 2𝐵𝑊𝑃𝐶𝑀 . Thus:

2𝐵𝑊𝑃𝐶𝑀 = 𝑛𝑓𝑠

𝑛 Hz ... (6-11)
𝐵𝑊𝑃𝐶𝑀 = 𝑓 ≥ 𝑛𝑓𝑚𝑎𝑥
2 𝑠

𝐵𝑊𝑃𝐶𝑀𝑚𝑖𝑛𝑖𝑚𝑢𝑚 = 𝑛𝑓𝑚𝑎𝑥
Hz ... (6-12)

149
(CH 6) Pulse and Digital Modulation

Ex 6-5:

In a binary PCM system, the output signal-to-quantization ratio is to be hold to a

minimum of 40 dB. If the message is a single tone with fm=4 kHz. Determine:
1- The number of required levels, and the corresponding output signal-to-quantizing noise ratio.

2- Minimum required system bandwidth.

Solution:

1) 𝐿 = 2𝑛
S𝑜
= 10000 = 40 𝑑𝐵
N𝑞

S𝑜 3𝐿2
= (S.T)
N𝑞 2

2
∴ 𝐿 = √ ∗ 10000 = [81.6] = 82
3

𝑛 = log 2 82 = [6.36] = 7

∴ 𝐿 = 27 = 128

2) Minimum system bandwidth = 𝑛𝑓𝑚𝑎𝑥 =7*4 kHz=28 kHz

H.W:

Consider a single tone signal of frequency 3300 Hz. A PCM is generated with a sampling

rate of 8000 sample/sec. the required output signal-to-quantizing noise ratio is 30 dB.

1) What the minimum number of uniform quantizing levels needed?. And what the

minimum number of bits per sample needed?

2) Calculate minimum system bandwidth required.

Ans: (a) 26.5 (b) 20 kHz

150
(CH 6) Pulse and Digital Modulation

2- Delta Modulation:

It is a sampling way to convert analog signal into digital with reduced bandwidth
Comparator
+ e(t) V
f(t) d(t)
- - V
Sampler
fs

f(t)
Integrator

f(t)&f(t) f(t)
Slope overload

f(t)

t
Ts
d(t)

e(t) Larger error

Its produces information about the difference between successive samples.

𝑒(𝑡) = 𝑓(𝑡) − 𝑓̃(𝑡) , where 𝑓̃(𝑡) is a stair case approximation of 𝑓(𝑡)

151
(CH 6) Pulse and Digital Modulation

The sampler with rate (𝑓𝑠 ≫ 𝑁𝑦𝑞𝑢𝑖𝑠𝑡 𝑟𝑎𝑡𝑒) produces pulse train 𝑑(𝑡) where:

∆𝑉 𝑒(𝑡) > 0
𝑑(𝑡) = ∆𝑠𝑔𝑛[𝑒(𝑡)] = {
−∆𝑉 𝑒(𝑡) < 0

𝑑(𝑡) represents the derivative of 𝑓(𝑡)

The demodulator will integrate 𝑑(𝑡) to produce 𝑓𝑠 (𝑘𝑇𝑠 ) smoothed by LPF with 𝐵𝑊 of

𝑓𝑚𝑎𝑥

Integrator
f(t) Recovered
d(t) LPF
f(t)

Slope overload problem:

Due to finite step size ∆𝑉 of integrator and if the slope of 𝑓(𝑡) is Larger than𝑓̃𝑠 (𝑡)

will not track 𝑓(𝑡) in its value [(𝑓̃𝑠 (𝑡)) and 𝑓(𝑡) will diverge from each other]. This will

produce distortion at Rx side when 𝑑(𝑡) is used to construct 𝑓̃𝑠 (𝑡).

To avoid slope overload, the step size must be kept such that:

F
𝑑𝑓(𝑡)
| | < ∆𝑉. 𝑓𝑠
𝑑𝑡 𝑚𝑎𝑥
... (6-13)

For single tone case 𝑓(𝑡) = 𝐴𝑚 𝑐𝑜𝑠𝜔𝑚 𝑡

𝑑𝑓(𝑡)
| | = 𝐴𝑚 𝜔𝑚 , therefore
𝑑𝑡 𝑚𝑎𝑥

𝐴𝑚 𝜔𝑚
∆𝑉𝑚𝑖𝑛 = F ... (6-14)
𝑓𝑠

152
(CH 6) Pulse and Digital Modulation

• For speech signal, the typical frequency analysis show that about 70% of total

energy lies between 600 and 1000 Hz indicating that peak energy is located that

almost at frequency of 800 Hz called response frequency 𝑓𝑟 =800 Hz, then we could

assume ∆𝑉𝑚𝑖𝑛 for speech to be:F

2𝜋(800)𝐴𝑚
∆𝑉𝑚𝑖𝑛 =
𝑓𝑠
Edeqd ... (6-15)

where 𝑓𝑝 in the maximum amplitude of the speech signal.

x(f)

70% of the
total area

f (Hz)

Quantizing Error:

Assuming quantizing error is equally likely in the interval (-∆V,∆V)

𝐵 (ΔV)2
𝑁𝑞 = .
𝑓𝑠 3
Edeqd ... (6-16)

Where B is the preconstruction filter bandwidth

Output Signal to Noise Ratio:

Edeqd ... (6-17)

𝑆𝑜 3𝑓𝑠 ̅̅̅̅̅̅̅
𝑓 2 (𝑡)
=
𝑁𝑞 (Δ𝑉)2 𝐵

153
(CH 6) Pulse and Digital Modulation

For single tone message𝑓(𝑡) = 𝐴𝑚 𝑐𝑜𝑠𝜔𝑚 𝑡

𝑆𝑜 3𝑓𝑠 2
=
𝑁𝑞 8𝜋 2 𝑓𝑚 2 𝐵
Edeqd ... (6-18)

Ex 6-6:

A DM has sampling frequency of 64 kHz is used to encode speech signal of ±1 volt:

1- Find minimum step size to avoid step overloading.

2- Find 𝑆𝑁𝑅𝑞 assuming speech has uniform probability density function (PDF) over

the interval [-1, 1] volt.

Solution:

2𝜋(800)𝑓𝑝
1. For speech signal ∆𝑉𝑚𝑖𝑛 =
𝑓𝑠

2𝜋(800)(1)
∆𝑉𝑚𝑖𝑛 = ≅ 78 𝑚𝑉
64000

𝑆𝑜 3𝑓𝑠 ̅̅̅̅̅̅̅̅
𝑓2 (𝑡)
2. = (Δ𝑉)2 𝐵
𝑁𝑞

1 P(f)
̅̅̅̅̅̅̅
𝑓 2 (𝑡) = ∫ 𝑓 2 𝑝(𝑓)𝑑𝑓
−1
1/2
1
1 2 1
= 2∫ 𝑓 𝑑𝑓 =
0 2 3 f(volt)
-1 1

-fp fp
𝑆𝑜 (3)(64000)(1/3)
= ≅ 35 𝑑𝐵
𝑁𝑞 (0.078)2 (3400)

Note:

Compare this result of 35 dB with PCM at 64000 bps (𝑓𝑠 = 8 𝑘𝐻𝑧, 𝑛 =

8 𝑏𝑖𝑡𝑠⁄𝑠𝑎𝑚𝑝𝑙𝑒) then 𝑆𝑁𝑅𝑞 ≅ 48 𝑑𝐵. i.e PCM is better than DM for the same bit rate.

154
(CH 6) Pulse and Digital Modulation

H.W:

A DM system is designed to operate at 3 times the Nyquist rate for the signal with a 3

kHz bandwidth. The quantization step size is 250 mV. Determine:

a) Maximum amplitude of a 1 kHz input sinusoid for which the delta modulator does not

show slop over load.

b) The post filter output signal-to-quantizing noise ratio for the signal in part a.

155
 (CH 6) Pulse and Digital Modulation

Bit Multiplexing:

Serial o/p of PCM or the o/p of DM PCM

or
f1(t) DM
for N messages had multiplexed and coder

transmitted over one channel. PCM

or Rate= NRb bits/sec
f2(t)
If Rb is the information rate of each DM channel
coder
source coder (nfs for PCM & fs for
PCM
DM) then the bit rate of channel will be f3(t) or
DM
coder
NRb if all messages have the same bit

rate.

Note:

For none equal bit rate messages, and then sub-group multiplexing has done first before

final TDM multiplexing.

Ex 6-7

Setup multiplexing scheme using TDM for 3 speech messages, each sampled at 8 kHz

and PCM quantized into 8 bits/sample identifying the bit rate at each part of the

multiplexing processes.

Solution:

157
(CH 6) Pulse and Digital Modulation

PCM 64 kbps
f1 n=8 bits
fs=8 kHz

PCM 64 kbps
Speech

rate rate
f2 n=8 bits
fs=8 kHz 192 kbps 384 kbps

PCM
f3 n=8 bits
fs=8 kHz 64 kbps

PCM
rate
f4 n=8 bits
192 kbps
Music fs=24 kHz

Transmission of Digital signals:

There are two ways to transmit the digital signals, these are:

a) Baseband transmission and,

b) Sinusoidal modulation (digital carrier system).

For binary data, it is assumed that logic “0” is transmitted as the waveform 𝑆𝑜 (𝑡)

and logic “1” as the waveform 𝑆1 (𝑡) over a bit duration 𝑇𝑏 = 1/𝑅𝑏 , where Rb is the

bitrate (bps). Also it is assumed that 𝑃(0𝑇 ) = 𝑃(1𝑇 ) = 1⁄2.

a) Baseband Transmission:

These are used for short or medium distance communication. The signals are

transmitted without carrier modulation (frequency shifting).

158
(CH 6) Pulse and Digital Modulation

1- Unipolar Transmission:

The waveform and power spectral density are summarized below:

RZ = Returned to Zero

NRZ= Non Returned to Zero

1 0 1 1 0 1 0 1 NRZ
Tb
2

1 0 1 1 0 1 0 1 RZ
Tb

Notes:

1- The unipolar Return to zero (RZ) format increases the power at data rate, but

double the bandwidth.

2- The spectrum of the unipolar RZ contains line spectrum at 0, ∓𝑅𝑏 , ∓2𝑅𝑏 , ∓3𝑅𝑏 , ….

in addition to continuous spectrum due to nonzero dc power

3- When large number of zeros or ones exists, synchronization problem will occur.

4- The error probability of unipolar NRZ

159
(CH 6) Pulse and Digital Modulation

𝑆𝑁𝑅
𝑃𝑒 = 𝑄(√ ) in AWGN.
2

• 𝑄(𝑥) is called Marcum function or Error Function. It is tabulated for x<3 or

can be approximated for large x as:

𝑥2
1 0.7 −
𝑄(𝑥) ≈ (1 − )𝑒 2 ,𝑥 > 2
√2𝜋𝑥 𝑥2

Ex 6-8

Calculate average power a signal has unipolar NRZ format

Solution

The average power for unipolar NRZ is:

H.W

Repeat the previous example for unipolar RZ format

2- Bipolar transmission:

The waveform and power spectral density has summarized below:

So(t) S1(t) P(f)

0
NRZ

-A f
-1 0 1
Tb Tb Tb
Tb

Tb Tb
RZ

0 +A
-A 0 f
-2 0 2
1 1
Tb Tb Tb Tb
Tb Tb
2 2

160
(CH 6) Pulse and Digital Modulation

1 0 1 1 0 1 0 1
-A

Tb
Tb
2
A
1 1 1 1 1
0
-A 0 0 0

Tb
2

Notes:

1- BW of bipolar RZ is greater than BW of bipolar NRZ.

𝟏
2- Bipolar RZ has significant frequency content at 𝒇 = 𝑹𝒃 = = clock of the data.
𝑻𝒃

3- The spectrum has almost zero DC power (exactly zero when P(0)=P (1) )

4- The synchronization problem is minimized in bipolar RZ format.

5- Error probability of NRZ bipolar 𝑃𝑒 = 𝑄(√𝑆𝑁𝑅) for AWGN channel.

Ex 6-9

Calculate average power of a signal has bipolar RZ format

Solution

The average power for bipolar RZ is:

H.W

Repeat the previous example for bipolar NRZ format

161
(CH 6) Pulse and Digital Modulation

3- Biphase and Differential Manchester (Split phase)

Notes:

• The duration of the bit is divided into two halves. The voltage remains at one level

during the first half and moves to the other level in the second half. The transition at

the middle of the bit provides good synchronization.

• The Manchester code has no dc power, high power at data rate, less bandwidth

than bipolar RZ.

• It is less complex than Bipolar RZ since it uses only two voltage levels ±A.

162
(CH 6) Pulse and Digital Modulation

4- Alternate Mark Inversion (AMI) RZ code

• Positive and negative pulses (of equal magnitude) are used for symbol 1, and no

pulse is used for symbol 0. In either case the pulse retunes to0 before the end of bit

interval.

• The BW here is almost less than Manchester code and the DC power is almost

zero. However, the power is max. at half the data rate.

Ex 6-10
Encode the binary data 10110001 using the following transmission codes:
Unipolar (RZ/NRZ), Bipolar (RZ/NRZ), AMI and Manchester.

Solution

163
Other Types of Digitizing

1- Adaptive DM:

Previous DM is called linear DM since the step size is fixed. This has the
disadvantage that the dynamic range of speech amplitude level large so that ∆V
chosen is not the best, if more or less levels of 𝑓𝑓(𝑡𝑡) occur. In adaptive DM, the
step size ∆V is changed (adapted) according to the slope of 𝑓𝑓(𝑡𝑡).

So, if 𝑓𝑓(𝑡𝑡) has small

slope, ∆V is made small to
reduce quantization noise, but
if ∆V has large slope, ∆V is
made large to avoid slope
overload.

If the step size adaptation is made in continuous form with slope of the
input, then the name CVSDM (continues variable slope DM) is used, or it is made
in discrete form with finite number of step size to form discrete variable slope
DM. In the CVSDM, peak detector circuit are used to control ∆V, while in
discrete VSDM logic circuit are used to control ∆V.

2- Differential PCM (DPCM)

The difference between sampled signal 𝑓𝑓(𝑘𝑘𝑇𝑇𝑠𝑠 ) and quantized signal 𝑓𝑓̂(𝑘𝑘𝑇𝑇𝑠𝑠 )
[error] is quantized into to n bits (recall that this is the same as DM with n=1, i.e
𝐷𝐷𝐷𝐷 ≡ 1𝑏𝑏𝑏𝑏𝑏𝑏 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷). So the derivative of the signal is quantized into n bits.

ADC
N-bit
f(kTS) + error n bit/sample
quantizer
M

+
- encoder
bit rate
f(kTS) = nfs

Accumulator
DAC
Acc.
At the Rx, the accumulator (Acc.) will integrate DPCM signal to produce 𝑓𝑓̂(𝑘𝑘𝑇𝑇𝑠𝑠 )
to DAC output then smoothed using LPF.

recived
n-bits Acc. DAC LPF f(t)
DPCM

Typical value of n is less than that used for PCM (n=3 or 4 bits) since n is
used here not to code actual f(t) samples but to encode the derivative of f(t) which
has less dynamic range.
(CH 6) Pulse and Digital Modulation

b) Sinusoidal Modulation: (Digital Carrier Systems):

1- ASK: Amplitude Shift Keying:

∅0 (𝑡𝑡) = 𝐴𝐴1 cos 𝜔𝜔𝑐𝑐 𝑡𝑡

∅1 (𝑡𝑡) = 𝐴𝐴2 cos 𝜔𝜔𝑐𝑐 𝑡𝑡 Over bit duration Tb ... (6-22)

1 0 1 0
A21
𝜔𝜔𝑐𝑐 : carrier angular frequency A1 t
Special case from ASK if A1=0, then A2
Tb Tb Tb Tb

∅0 (𝑡𝑡) = 0

∅1 (𝑡𝑡) = 𝐴𝐴 cos 𝜔𝜔𝑐𝑐 𝑡𝑡 Over bit duration Tb ... (6-23)

1 0 1 0
This is called OOK A

t
(ON-OFF Keying)

Tb Tb Tb Tb

The spectrum of OOK is

ΦASK(f)

f
fc+Rb

-fc fc
fc-Rb

𝐵𝐵𝐵𝐵𝑃𝑃𝑃𝑃𝑃𝑃 ≈ 2𝑅𝑅𝑏𝑏 Hz ... (6-11)

(First null bandwidth)

164
(CH 6) Pulse and Digital Modulation

Ex. 6-11
Find the minimum bandwidth for an ASK signal transmitting at 2000 bps. The
transmission mode is half-duplex.

Solution:
BW ≈ 2Rb = 2*2000 = 4kHz

Ex. 6-12
Given a bandwidth of 10,000 Hz (1000 to 11,000 Hz), draw the full-duplex ASK
diagram of the system. Find the carriers and the bandwidths in each direction.
Assume there is no gap between the bands in the two directions.

Solution:
For full-duplex ASK, the bandwidth for each direction is
BW = 10000 / 2 = 5000 Hz
The carrier frequencies can be chosen at the middle of each band
fc (backward) = 1000 + 5000/2 = 3500 Hz
fc (forward) = 11000 – 5000/2 = 8500 Hz

165
(CH 6) Pulse and Digital Modulation

Modulator of ASK:

OOK Power
data Amplifier

Or using Mixer
Mixer ASK
data (OOK) Power
(unipolar) Amplifier

Acos ωc(t)

Demodulator of ASK:

1) Non coherent (no need for carrier recovery):

This has done using envelope detector.

Detector
OOK
o/p 1 0 1 + data
-
vth

t
Tb

2) Coherent detector:

𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝜔𝜔𝑐𝑐 𝑡𝑡 Mixer
𝑂𝑂𝑂𝑂𝑂𝑂 = � LPF
0
@Rb + comp data
-
Carrier
Vth=A2/4
recovery

166
(CH 6) Pulse and Digital Modulation

Here, a carrier recovery circuit is required to generate the carrier signal cos
ωct then, the LPF will cutoff of Rb is used to give the high and low states of the
data.

2 2 𝐴𝐴2 𝐴𝐴2
𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑜𝑜/𝑝𝑝 = �𝐴𝐴 𝑐𝑐𝑐𝑐𝑐𝑐 (𝜔𝜔𝑐𝑐 𝑡𝑡) = 2
+
2
𝑐𝑐𝑐𝑐𝑐𝑐2𝜔𝜔𝑐𝑐 𝑡𝑡 rejected by LPF
0

Then LPF output will be either A2/2 or “0”and with Vth= A2/4, data are
obtained by a threshold comparator.

2- Frequency Shift Keying (FSK):

∅0 (𝑡𝑡) = 𝐴𝐴 cos 𝜔𝜔1 𝑡𝑡
∅1 (𝑡𝑡) = 𝐴𝐴 cos 𝜔𝜔2 𝑡𝑡 Over bit duration Tb ... (6-25)
Φ(t)
For 𝜔𝜔2 > 𝜔𝜔1 , then timing of FSK
1 0 1
is as shown. It is clear than FSK
is the sum of two (OOK) one for t
logic “0” and the other for logic
“1” And hence the spectrum of FSK
Tb Tb Tb
consists of two sinc functions, one S1(t)
at f1 and the other at f2 as shown. f2 f2

∅(𝑡𝑡) = 𝑆𝑆1 (𝑡𝑡) + 𝑆𝑆2 (𝑡𝑡)

FSK (OOK) (OOK) S2(t) f1

167
(CH 6) Pulse and Digital Modulation

ΦFSK(f)

f
F1-Rb F1 F2 F2+Rb

∆f

If ∆𝑓𝑓 = 𝑓𝑓2 − 𝑓𝑓1 then the null BW will be:

𝐵𝐵𝐵𝐵 = Δ𝑓𝑓 + 2𝑅𝑅𝑏𝑏 Hz ... (6-26)

Ex 6-13
Find the minimum bandwidth for an FSK signal transmitting at 2000 bps.
Transmission is in half-duplex mode, and the carriers are separated by 3000 Hz.
Solution:
BW = ∆f + 2Rb = 3000 + 2*2000 = 7kHz

Modulator of FSK:

R1 FSK f2
data VCO
o/p
f1
Vin
R2 adsjust Vin
V1 VCO charecteristics

A voltage-controlled oscillator is usually used for FSK modulation. A VCO

is an oscillator with an internal tuning capacitor, which controlled by an external
voltage Vin. This capacitor is called varicap ( ).

168
(CH 6) Pulse and Digital Modulation

Demodulator of FSK:

1) Noncoherent detector: (no need for carrier recovery)

Two BPF are used, one tuned at f2 and the other at f1. A high output will
appear at the corresponding arm for logic “0” or logic “1”.

BPF
@f2 Envelpe
detector

+ comp data
data
-

BPF
@f1 Envelpe
detector

2) Coherent PLL (Phase Locked Loop) detector:

FSK Loop Filter

(LPF) + comp data
A phase locked loop (PLL) -

consists of a phase detector (PD), Vth

VCO
loop filter (LPF) and a VCO at its
LPF
feedback. The characteristic of this o/p

PLL against frequency charging of V2

the input is shown besides of fo being

f1
the free running frequency of the fo f2
Freq
V1
VCO (its frequency o/p when i/p d.c.
is zero).

169
(CH 6) Pulse and Digital Modulation

𝑉𝑉1 + 𝑉𝑉2
𝑉𝑉𝑡𝑡ℎ =
2 ... (6-27)
𝑉𝑉1 +𝑉𝑉2
A threshold comparator with 𝑉𝑉𝑡𝑡ℎ = is adjusted to give the received data using
2

a threshold comparator.

3) Zero Crossing Frequency Discriminator:

1 2 3 4
ZDC Monostable
Averger 5
FSK τ=Timeconstant
+ comp data
-

Vth
The zero crossing detector (ZDC) changes sinewave into rectangular waves
with frequencies f1 and f2.

1 f2 f1 f2
t

3 τ τ τ τ

4
Vth
t

1 0 0 1 t

170
(CH 6) Pulse and Digital Modulation

The positive going edge monostable gives fixed time constant τ chosen such
1
that 𝜏𝜏 ≤ ( ). The average value of the monostable output is high level for
𝑓𝑓2

frequency f2 and is low level for frequency f1. A threshold comparator with a
threshold Vth chosen midway between the high & low value will give data output.

3- Phase Shift Keying (PSK):

∅0 (𝑡𝑡) = 𝐴𝐴 cos 𝜔𝜔𝑐𝑐 𝑡𝑡

∅1 (𝑡𝑡) = 𝐴𝐴 cos(𝜔𝜔𝑐𝑐 𝑡𝑡 + 𝜋𝜋) = −𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝜔𝜔𝑐𝑐 𝑡𝑡 Over bit duration Tb ... (6-25)

When there is only two phases to describe data values. The modulation is called
Binary Phase Shift Keying (BPSK).

171
(CH 6) Pulse and Digital Modulation

Ex 6-14
Find the minimum bandwidth for a BPSK signal transmitting at 2000 bps. The
transmission mode is half-duplex.

Solution:
BW = 2Rb = 2*2000 = 4kHz

PSK demodulator:

Since information is the phase of the carrier, then only coherent PSK is
possible, i.e. coherent carrier phase must be generated at the receiver.

172
(CH 6) Pulse and Digital Modulation

Carrier Recovery (Squaring Loop):

f LPF g

@Rb
BPSK a

T-FF Vth
2ωc ωc
2 BPF
(.) ZDC *1/2

+
@2ωc

I
b c d e

h
data

a Tb Tb

t
c

t
d
t
e

t
f
t
g

t
g

1 0
t

173
(CH 6) Pulse and Digital Modulation

The ±𝐴𝐴 cos 𝜔𝜔𝑐𝑐 𝑡𝑡 PSK modulation is first canceled using squarer to give
𝐴𝐴2
(1 + cos 𝜔𝜔𝑐𝑐 𝑡𝑡), the double carrier frequency is filtered using BPF tuned at 2ωc,
2

then T-FF is used to generate carrier at ωc with correct phase.

Differential PSK (DPSK)

There is possibility in PSK detection that the receiver will receive “0” as “1”
or “1” as “0” (i.e. data complement) due to he initial phase of the carrier (phase
ambiguity). To solve this problem, a technique called differential PSK (DPSK) is
used. The encoder of the DPSK is shown below:

Coded o/p
data x(n)
SET PSK DPSK
d(n)
D Q modulator o/p

CLR
Q
FF Carrier
fc

The coded output x(n) is obtained from transmitted data using:

� �(��𝑛𝑛�)���
𝑥𝑥 (𝑛𝑛) = 𝑏𝑏 +��𝑥𝑥��(�𝑛𝑛���−���1�)�, where +≡ 𝐸𝐸𝑥𝑥 − 𝑂𝑂𝑅𝑅

i.e exclusive-NORing previous output code with present data. x(n) then transmitted
using conventional PSK modulator with 0o and 180o phase shift for logic “0”
and logic “1”. The extra requirement for DPSK noncoherent detection is the
analogue delay liner Tb and then multiply with itself, LPF then a threshold
comparator.

174
(CH 6) Pulse and Digital Modulation

Ex 6-15

175
(CH 6) Pulse and Digital Modulation

Ex 6-15:

With the help of data of previous example, explain the operation of DPSK
detector.

Solution:

−𝐴𝐴 cos 𝜔𝜔𝜔𝜔 &+ 𝐴𝐴 cos 𝜔𝜔𝜔𝜔will be used for 180o& 0o phase shift then;

H.W:

Repeat encoder and decoder operations when x(0)=0, and comment to the results.

176
(CH 6) Pulse and Digital Modulation By Dr. Hikmat Al-Shamary & Dr. Tariq M. Salman

Advantages of DPSK:

1- Noncoherent detection (no carrier generation)

2- No phase ambiguity (possible data complement) since the decoded data is
independent of random initial choice of the state of D-FF (x(0)).

Disadvantages of DPSK:

The only disadvantage is the need of analogue delay line by Tb time. This is
usually implemented using charge coupled devices (CCD) as analogue delay line.
(CH 6) Pulse and Digital Modulation

Multilevel keying Technique (M-ary Signalling):

These are usually used for bandwidth economical systems (such as

telephone channels) to increase data rate transmission over bandlimited channels.

Concept:

The concept of M-ary signaling in to group binary bits and deal with them as
symbols. Hence, M is taken as 2r (2,4,8,16,… ). For example M=4 means r=2.

data 1 0 0 1 1 1 0 0 0 1 1 1 1 0

grouping of 2-bits (r=2, M=4) i.e. 4 level

Then each successive 2-bits has grouped and considered as symbol. Hence,
these will be four possible signals ∅𝑜 (𝑡), ∅1 (𝑡), ∅2 (𝑡) 𝑎𝑛𝑑 ∅3 (𝑡)for the possible
groups of “00”, “01”,“10” and“11”. And so on for M=8, r=3, …

In general:

177
(CH 6) Pulse and Digital Modulation

𝑟 = log 2 𝑀 bits/symbol ... (6-30)

𝑀 = 2𝑟 no. of channel states ... (6-31)

1
𝑇𝑠 = =(log 2 𝑀). 𝑇𝑏 symbol duration ... (6-32)
𝑅𝑠

𝑅𝑏
𝑅𝑠 = symbol rate (baud rate) ... (6-33)
log 2 𝑀

M-ary ASK (MASK):

M carrier level are used for M=4 then:

∅0 (𝑡) = 0

∅1 (𝑡) = 𝐴 cos 𝜔𝑐 𝑡

∅2 (𝑡) = 2𝐴 cos 𝜔𝑐 𝑡

∅3 (𝑡) = 3𝐴 cos 𝜔𝑐 𝑡 … (6-34)

For data
MASK Φ3(t)
0 0 0 1 1 1 1 0 Φ2(t)
Φ1(t)
∅0 ∅1 ∅3 ∅2
Φ0(t)
t
Then transmitted waveforms will
be:∅0 (𝑡), ∅1 (𝑡), ∅3 (𝑡), ∅2 (𝑡) with
symbol duration Ts=2Tb Ts Ts Ts Ts

178
(CH 6) Pulse and Digital Modulation

Signal State Diagram imag

This is a representation of carrier Φ0 Φ1 Φ2 Φ3

signals in complex plane (x-axis ≡ real part
≡ cos ωct) & (y-axis ≡ imaginary part ≡ real
0 A 2A 3A
sin ωct) of ejωct.

To draw the signal state diagram of 4 level MASK then on the real
axis ∅0 , ∅1 , ∅2 & ∅3 will be found since modulation is done in amplitude of the
inphase component (cos ωct) of the carrier.

Simple MASK modulator:

Symbols mixer
Serial/
data Digital/
parallel
analogue MASK
shift
converter
register
r-bit DAC
cosωct

Simple MASK demodulator (noncoherent):

Z (M-1)
Envelope
MASK
detector
threshold (M-1) Logic data
components
bits
C1
C2

CM-1

179
(CH 6) Pulse and Digital Modulation

𝐴 3𝐴 5𝐴
For M=4 then 3 comparator with𝑉𝑡ℎ = ( ) , ( ) & ( ) are used and logic truth
2 2 2

table will be:

C1 C2 C3 data
C1, C2, C3 are comparators outputs.
0 0 0 00
𝐴
If envelope detector 𝑍 < then C1=C2=C3=0 and ∅0 (𝑡) is
2
1 0 0 01 3𝐴 𝐴
detected. If >𝑍> then C1=1, C2=C3=0 and ∅1 (𝑡) is
2 2
1 1 0 10
detected and so on.
1 1 1 11

Application of MASK:

MASK has little application on telephone or space channels (sensitive to

amplitude variation).

Multilevel PSK (MPSK):

(a) 4- Level PSK≡ QPSK≡ Quadrature PSK:

Here M=4, r=2 with constant carrier amplitude and frequency.

Data 0 0 0 1 1 0 1 1 ∅0 (𝑡), ∅1 (𝑡), ∅3 (𝑡) 𝑎𝑛𝑑 ∅2 (𝑡)…,

∅0 ∅1 ∅2 ∅3 are assignments with four phases

𝜋 3𝜋 5𝜋 7𝜋
( , , , )
4 4 4 4

∅3 ∅1 ∅0 ∅2

180
(CH 6) Pulse and Digital Modulation

Note that the gray coding is used (symbols Im ag

used at adjacent phases differ by one bit).

Φ1 Φ3
Hence: A A
(-,+) 0 1 1 1 (+, +)
QPSK o/p=±𝐴𝑐𝑜𝑠𝜔𝑐 𝑡 ± 𝐴𝑠𝑖𝑛𝜔𝑐 𝑡 Re
A A
i.e QPSK is the sum of two carriers in the (-,-) 0 0 1 0 (+,-)

phase Quadrature, i.e: Φ0 Φ2

∅3 (𝑡) = 𝐴𝑐𝑜𝑠𝜔𝑐 𝑡 + 𝐴𝑠𝑖𝑛𝜔𝑐 𝑡

∅1 (𝑡) = −𝐴𝑐𝑜𝑠𝜔𝑐 𝑡 + 𝐴𝑠𝑖𝑛𝜔𝑐 𝑡

∅0 (𝑡) = −𝐴𝑐𝑜𝑠𝜔𝑐 𝑡 − 𝐴𝑠𝑖𝑛𝜔𝑐 𝑡

∅2 (𝑡) = 𝐴𝑐𝑜𝑠𝜔𝑐 𝑡 − 𝐴𝑠𝑖𝑛𝜔𝑐 𝑡 … (6-35)

𝑅𝑏
𝑅𝑠 = 𝑇𝑠 = 2𝑇𝑏
2 ... (6-36)

181
(CH 6) Pulse and Digital Modulation

Ex 6-16
Find the symbol rate (baud rate) and bandwidth and for a 4-PSK signal
transmitting at 2000 bps. Transmission is half-duplex mode.
Solution
Rs = Rb / 2 = 1000 symbol/sec or 1000 baud
BW = 2Rs = 2000 Hz

QPSK Modulator:

±A ±A cosωct

I
2-bit `

M
data Serial/ QPSK
parallel π/2
Q
(I,Q)
cosωct
(+,+) Φ3
±A ±A sinωct
(+,-) Φ2
(-, -) Φ0
(-,+) Φ1

Tb
data
1 0 1 1 0 0 0 1 Ts=2Tb

Ts
I (I,Q)
1 1 0 0 11 S3
Q
10 S2
0 1 0 1 00 S0
01 S1
S2 S3 S0 S1

182
(CH 6) Pulse and Digital Modulation

QPSK Detector:

Z1(t) I
LPF +
-
Parallel/
QPSK data
serial
π/2
2-bit
Q
LPF +
Z2(t)
-

𝑍1 (𝑡) = ±𝐴𝑐𝑜𝑠 2 𝜔𝑐 𝑡 ± 𝐴 𝑠𝑖𝑛𝜔𝑐 𝑡 𝑐𝑜𝑠𝜔𝑐 𝑡 ⟹ 𝐼 = ±𝐴

Since the LPF cancel
𝑍2 (𝑡) = ±𝐴 𝑐𝑜𝑠𝜔𝑐 𝑡 𝑠𝑖𝑛𝜔𝑐 𝑡 ± 𝐴𝑠𝑖𝑛2 𝜔𝑐 𝑡 ⟹ 𝑄 = ±𝐴 the 𝑐𝑜𝑠2𝜔𝑐 𝑡 and

Note that the detector needs carrier phase 𝑠𝑖𝑛𝜔𝑐 𝑡𝑐𝑜𝑠𝜔𝑐 𝑡 terms

Recovery 𝑐𝑜𝑠𝜔𝑐 𝑡 at the Rx (coherent detection).

(b) MPSK in general M=8 16, …:

Im ag
For M=8, the MPSK state diagram
is shown Here: 011
010
001
Rs=Rb/3 & Ts=3Tb
110 000 Re
Note that again gray coding is used
111
for assigning symbols such that 100
101
adjacent phases bits differ in one bit
only.

183
(CH 6) Pulse and Digital Modulation

Ex 6-17
Given a bandwidth of 6000 Hz for an 8-PSK signal, what are the baud rate and bit
rate?
Solution
M=8
Rb = BW/2 = 3000 bps
𝑅𝑏 3000
𝑅𝑠 = = = 1000 𝑠𝑦𝑚𝑏𝑜𝑙/𝑠𝑒𝑐.
𝑙𝑜𝑔2 𝑀 3

H.W 6-6
Draw the signal constellation for 16-PSK identifying the angles between each two
levels.

(c) QAM Quadrature Amplitude Modulation:

Im ag
Here two or more amplitude are
used with 2, 4, 8, phases, for 1011
1010 1001
example 16 QAM may have two 0011
0010 0001
amplitudes and 8 phases. For 1110 1000
0110 0000
Re
M=16, Rs=Rb/4 and Ts=4 Tb. 0111
0100
0101
With gray level for each amplitude 1111
1100
1101
Application:

Telephone (digital transmission)

184
(CH 6) Pulse and Digital Modulation

MFSK: Multi-level FSK:

For M=4, then for frequencies are used to assign ∅0 , ∅1 , ∅2 &∅3 . The
spectrum of 4-FSK is as shown with the equal spacing ∆f.

Φ(f)
∅0 (𝑡) = 𝐴𝑐𝑜𝑠𝜔1 𝑡

∅1(𝑡) = 𝐴𝑐𝑜𝑠𝜔2𝑡

∅2(𝑡) = 𝐴𝑐𝑜𝑠𝜔3𝑡
f
∅3(𝑡) = 𝐴𝑐𝑜𝑠𝜔4𝑡 … (6-37) f1 f2 f3 f4

f f f
𝑅𝑏
𝑅𝑆 = , 𝑇𝑆 = 2𝑇𝑏
2

MFSK modulator:

Fr.
Serial/ MFSK
data Digital/ f4
parallel
analogue VCO f3
shift
converter f2
register
r-bit f1
DAC

Vo
v1 v2 v3
VCO charecteristics
Vo
v3
v2
v1 M-Levels

185
(CH 6) Pulse and Digital Modulation

Noncoherent MFSK Detector:

C1 C2 C3 C4 symbol
1 0 0 0 00
0 1 0 0 01 Logic Truth Table
0 0 1 0 10
0 0 0 1 11

Envelope
detector Symbol
2-bits
BPF f1 C1
+
-

BPF f2 C2
+
-
MFSK
M=4 Parallel/
Logic data
serial
BPF f3 C3
+
-

BPF f4 C4
+
-

Note that only once of the envelope detectors is high during Ts time,
corresponding to one of the received frequencies 𝑓1 , 𝑓2 , 𝑓3 &𝑓4 .

186
(CH 6) Pulse and Digital Modulation

Coherent MFSK (PLL detectors):

Vo C1
LPF
MFSK C2 Parallel data
M-1
threshold Logic to
comparators
serial
VCO
CM-1 r-bit

Vo
VCO free running frequency is
chosen midway between f2& f3
+3A
(center of the band). Suppose
that the liner region of PLL +A
f1 f2
gives Vo= -3A, -A, +A & +3A, f
-A f3 f4
corresponding to f1, f2, f3& f4
-3A fo=VCO free
respectively. running frequency

(M-1) threshold comparators (3 comparator here) will give C1, C2& C3output
and in a similar schematic as in MASK these outputs will be decoded into the
received data (Vth are -2A, 0, +2A).

Application of MFSK:

Main important application is in HF (High frequency) transmission to

combat multipath fading on ionospheric layer. Note that not all frequencies will be
faded.

187
 Detection of Digital Signals in Noise
Binary Signals:
Let x(t) be a binary signal having two waveforms shapes (one for logic "0" and the
other for logic "1"). n(t) is the noise component added due to channel such that the
Probability Density Function (PDF) of n(t) is f(n) with zero mean.

y(Tb)
x(t)+n(t) Detector
+ data
-

Vth

y(Tb) is the detector output at t=Tb (bit duration for both "0" and "1" signals). A
comparator with threshold voltage Vth is used to decide if the
received x(t) is "0" or "1" according to the decision rules:
if y(Tb) > Vth then decide the received x(t) is a "1".
if y(Tb) < Vth then decide the received x(t) is a "0".
The problem is to find the best choice of Vth such that the overall net error
probability (average error prob) of "1" & "0" is minimum. This optimum Vth
depends on type of the binary signals and the PDF of n(t).
In this course, we will assume that the noise is AWGN with PDF:
𝑛 2
1 − 2
𝑓(𝑛) = 𝑒 2𝜎 and the binary signals “0” and “1”
2𝜋 𝜎

are equiprobable. First, we will take the simplest case of unipolar signals
 Unipolar binary signals:
First we discuss the simplest case if x(t)=0 volt for logic"0" and x(t)=+A
volt for logic"1" over the bit duration Tb.
+A
x(t(
0
Tb

P(0T)
P(1T)
0.5

0 +A

Note that for noiseless case (y(t)=x(t)) then the PDF of y(t) is discrete similar to the
PDF of x(t). When n(t)≠0, the PDF of y(t) will be the same as the PDF of n(t) (f(n)) but
d.c. shifted by 0 volt or +A volts corresponding to x(t). Let f1(y) be the PDF of y when
"1" is transmitted and fo(y) be the PDF of y when "0" is transmitted.
 area=p(0R/1T) area=p(1R/0T)
If we assume equiprobable then the best(optimum) threshold is the
mid-way between the “0” volt and +A volt which is +A/2 as shown.
Vth 
p(0 R / 1T ) =  f 1( y )dy and p(1R / 0T ) =  fo( y )dy
− Vth
Note that since Vth is in the middle, then the areas representing the
error prob. p(0R/1T) or p(1R/0T) are equal.
To find pe= p(0R/1T) = p(1R/0T), then, we find the area under the
curve:
2
  y
1 −
pe =  fo( y )dy =  e 2 dy
2

Vth A / 2 2 

Note that this definite integral can not be evaluated analytically. The
following substitution will help to evaluate this integral numerically:
Let z=y/, then dy=dz.
For y=A/2, then z=A/(2), and for y→, then z→. Changing the variable and the
limits of this integral reduces it into:
 z2
1 − A
pe = A 2
e 2
dz = Q( )
2
2

where Q(x) is a function called Marcum function which gives the area under the
standard Gaussian curve (=1, =0) from x=(A/2) to .

Red area = Q(x)

  −x
Q( x) =  p( z )dz =  p( z )dz
x −
This Q(x) is numerically evaluated and tabulated for x<3, or can be
approximated for large x ( say x>2) as:

1 0.7 −0.5 x 2
Q(x)≈ (1 − 2 )e
2 x x

Example: Equiprobable binary signals are represented by 0 & 2 volts.

For AWGN with 2 = 0.16 volts2, find the optimum threshold setting
and the net error prob.
Solution: For equiprobable case with unipolar signals affected by
AWGN, then Vth=A/2(mid-way) =2/2=1 volt,
and 2=0.16, then =0.4 hence pe=Q(A/2)=Q(2/2*0.4)=Q(2.5)
1 0.7 −0.5∗6.25
(1 − )𝑒
Pe=Q(2.5)  2𝜋 2.5 6.25  6.2*10-3

Note: In general, for equiprobable signals A1 volts and A2 volts and for
AWGN, then:
Vth|op=(A1+A2)/2 (mid-way between A1 and A2)
and p(0R/1T)=p(1R/0T)=pe=Q[d/(2)], where d=A2-A1
Where d is the absolute distance between A1 and A2
 red area=pe
Example: Bipolar equiprobable binary signals ± A are affected by AWGN
with variance 2, find the threshold and the error prob.
solution: Here, we have A1=-A and A2=+A then vth=(+A+(-A))/2=0 volt
and pe=Q[d/(2)], where d= +A-(-A)=2A,hence:pe=Q[2A/(2)]=Q[A/ ],
Example: The three equiprobable signals +3A, 0, -2A are affected by AWGN
with =0.4A. Find the values of the thresholds and the overall net error prob.
Solution:
Here, we have two
thresholds, one Vth1 that
Separates -2A from 0 volts
And Vth2 that separates
+3A from 0 volts.
Since equprobable, then Vth1=(-2A+0)/2=-A volts and
Vth2=(+3A+0)/2=1.5A .
To find the error prob, we find d1=2A ( absolute distance between -2A and 0
volts) and d2= 3A ( absolute distance between +3A and 0 volts)
also we will assume -2A volt for “0” logic state , 0 volt for “1” logic
state and +3A volts for “2” logic state, then:
pe1=p(0R/1T)=p(1R/0T)= errors between -2A and 0 volts, then:
pe1=Q[d1/(2)]=Q[2A/(2*0.4*A)]=Q(2.5)=6.2*10-3
pe2=p(2R/1T)=p(1R/2T)= errors between +3A and 0 volts, then:
pe2=Q[d2/(2)]=Q[3A/(2*0.4*A)]=Q(3.75)=8.9*10-5
net error prob=p(0T) p(1R/0T)+ p(1T) p(0R/1T)+p(1T) p(2R/1T)+ p(2T) p(1R/2T)
=(1/3)*pe1+(1/3)*pe1+(1/3)*pe2+(1/3)*pe2
=(2/3)*pe1+(2/3)*pe24.19*10-3

Where p(0T)=p(1T)=p(2T)=1/3 (equiprobable)

 Matched Filter
Detection problem: Detection of the signal x(t) embedded in
AWGN noise n(t) will produce the output y(Tb) which was the
signal used at the decision block (comparator block). The device
that carries out the function of detection is called a Matched Filter .
This matched filter is simply a linear system with impulse response
h(t) (or frequency response H()) 2that acts as a filter. The job of this
filter is to maximize the ratio | y(Tb
2
)| .
n o (t )

Matched filter
y(t)+no
x(t)+ni H(w)
h(t)
This filter is called “Matched “ since for a certain signal x(t), there exists
a filter with impulse response h(t) matched to it (maximizes the above
ratio).
After long derivation, the following two equations are obtained:
1- h(t)=x(Tb-t).
Which gives the impulse response h(t) of a matched filter matched to
the signal x(t). Hence the impulse response of a matched filter matched
to the signal x(t) is the negative time of x(t) shifted by Tb.
2- | y(Tb) |2 E
[ ] max =
2
n o (t ) (o / 2)

Which gives Max SNR at matched filter output. where

o=one-sided AWGN spectral density (Watt/Hz) and
𝑇𝑏
𝑎𝑛𝑑 𝐸 = ‫׬‬0 [𝑥(𝑡)]2 dt= energy of the signal over bit duration Tb.
Practical Implementation of the Matched filter:
The practical implementation of a matched filter can be deduced as the
product of the input x(t)+n(t) and x(t) then integrate the result over t=0 to
t=Tb. The figure below shows the practical implementation of the matched
filter:
t=Tb
Tb
x(t)+n(t)
 ( )dt
0
Y(Tb(

x(t)

Error probability of binary signal detection using matched filter:

Assume that the matched filter is used to detect the binary signal waveforms
so(t) and s1(t) embedded in AWGN. Here, we need two matched filters, one
matched to so(t) with impulse response ho(t)=so(Tb-t) and the other matched
to s1(t) with impulse response h1(t)=s1(Tb-t)
 Matched
yo(Tb)
filter
h0(t)

s(t)+n(t)
data
Matched
filter y1(Tb)
h1(t)

Tb
Where: E o =  s o2 (t )dt = energy of so(t) =o/p of ho(t) if s(t)= so(t) and
0

Tb
E1 =  s12 (t )dt = energy of s1(t) =o/p of h1(t) if s(t)= s1(t) also:
0

Tb
Eo1 =  so (t ) s1 (t )dt =cross energy between so(t) and s1(t)
0
=o/p of ho(t) if s(t)= s1(t)
= o/p of h1(t) if s(t)= so(t)
For equiprobable p(0T)=p(1T)=0.5, and for Gaussian case:

Tb

 − 2
[ s (t ) s (t )] dt
Eo + E1 − 2 E01
o 1

pe = Q ( 0
) = Q( ) -------------(1)
2 o 2 o

Error prob for digital carrier systems using matched filter (coherent detection):
1- ASK (OOK) (ON-OFF) keying:
Here the two signals are so(t)=0, s1(t)=Acost over bit duration Tb. We use
the general equation:
Eo + E1 − 2 E01
pe= Q (
2
)
o
where for OOK, Eo=E01=0 (since So(t)=0) , and:
E1=(A2Tb)/2 (s1(t) is a sinusoid whose normalized power=A2/2), putting
Eo, E01 and E1 then:
A2Tb
pe = Q( ) and express in terms of average signal
4 o
power, then:
S S Eb
pe = Q ( ) = = = energy / bit
S=[0.5Eo+0.5E1]/Tb=A2/4, then:  o Rb and if  o Rb  o
Then: pe = Q(  )
2-BPSK signals (Binary Phase Shift Keying):
Here the two signals are so(t)=-Acost, s1(t)=Acost over bit duration
Tb. Again, using the same general equation, then:
Eo=E1=[A2Tb]/2 since both so(t) and s1(t) are two sinusoids, also:
Tb
E01=  − A2 cos2  t dt =-[A2Tb]/2.
0
Also S=[0.5Eo+0.5E1]/Tb=A2/2=average signal power, then one can
show that after putting Eo, E1 and E01 in the general equation:
2S which is better than OOK
pe = Q( ) = Q( 2 )
 o Rb
3-FSK (frequency shift keying):
Here the two signals are so(t)=Acos1t, s1(t)=Acos2t over bit duration
Tb. Note that pe depends also on
= d = 2- 1
After a similar longer derivation, one can show that:
sin  d Tb
pe = Q(  (1 − )
 d Tbwhich is a general formula used for matched
filter detection of FSK signals.( where  = S = Eb = energy / bit as
before)  o Rb  o
Special cases in FSK:
1-if dTb=, 2, 3,.., or fd=0.5Rb, Rb, 1.5Rb,... this gives Eo1=0
(orthogonal FSK), and pe = Q(  ) (similar to OOK)
sin d Tb
2-if dTb=4.49rad, or fd=0.715Rb, then: = −0.217
d Tb
And this gives pe = Q( 1.217  ) which is the best
(optimum)performance of FSK. sin  d Tb
pe = Q(  (1 − )
( Note: Use the general equation  d Tb for FSK in
solving problems, do not mix with above special cases.)
Example: Matched filter detection is used to detect BPSK signals at a
rate of 600bps. If transmitted power is 5KW over an HF channel having
estimated path losses of 150dB, find the error prob if the noise at
detector input has one sided spectral density of 10-15 Watt/Hz.
 • Solution:
PT =trans power=5000 W , PT = 10 Log10(5000) = 36.989 dB
then, PR=S= PT - PLoss =36.989 – 150 = -113.01 dB = 10(-113.01/10)
S=5*10-12 W=average signal power.

S 5 *10 −12
= = −15 = 8.333 then pe = Q( 2 ) = Q( 2 * 8.3333 ) = Q(4.08)  2 *10 −5
 o Rb 10 * 600
Example: Repeat previous example for FSK signals with f1=700Hz, f2=2000Hz.
Solution:
Here d=2(2000-700)=2600, and d Tb=2600/600=4.3333, then and for the same
=8.333:
sin d Tb sin 4.333
pe = Q(  (1 − ) = Q( 8.333(1 − ) = Q(2.84)  2.3 * 10 −3
d Tb 4.333
Which is worst than BPSK for the same .(remember that both BPSK and FSK have the
same average power of A2 /2, but the performance of BPSK is better)