Chapter 2

Modulation and Detection

Abstract  In this chapter, we present a number of topics surrounding modulation

and demodulation. We begin by introducing system-level block diagrams of AM and
FM/PM modulators and demodulators, and explaining their respective principles of
operation. In particular, under the topic of AM Modulator/Demodulator, we intro-
duce the full carrier modulator, the single sideband suppressed carrier modulator,
the double sideband suppressed carrier modulator, the envelope detector, and the
synchronous detector. Under the topic of FM and PM Modulator/Demodulator, we
introduce the VCO as FM modulator, the indirect FM modulator, the PM modulator,
the balanced discriminator FM demodulator, the quadrature FM detector, the PLL-
based FM detector, the zero-crossing FM detector, and the PM demodulator. Then,
under the topic of digital modulation, we introduce the concepts of Nyquist Limit,
data rate, Shannon Limit, information capacity, and bandwidth efficiency, as well
as specific modulation schemes, such as, binary modulation, amplitude-shift keying
(BASK), frequency-shift keying (BFSK), and phase-shift keying (BPSK), differen-
tial binary phase-shift keying (DBPSK), quadrature phase-shift keying (QPSK), π/4
shifted QPSK, minimum shift keying (MSK), M-ary quadrature amplitude modula-
tion (QAM), orthogonal frequency division multiplexing (OFDM), direct sequence
spread spectrum (DS/SS), and frequency hopping spread spectrum (FH/SS). We
also introduce the geometric representation of digital modulation schemes and the
complex envelope form of a modulation signal.

2.1 AM Modulators

An AM modulator, which implements the equation,

v AM (t ) = a0 [1 + mg (t ) ] cos ω 0 t  (2.1)

may be represented as shown in Fig. 2.1. It is seen that the magnitude of the modu-
lation index, m, if too large may cause distortion of the resulting modulated signal.
The corresponding frequency spectrum for the AM signal is given by [1],

[2π δ(ω − ω0 ) + 2πδ(ω + ω0 ) + FB (ω − ω0 ) + FB (ω + ω0 )]  (2.2)

a0
FM (ω) =
2
which is depicted in Fig. 2.2. As can be seen, due to symmetry with respect to the
origin, the same information is carried in both the upper (centered at + ω0) and lower

H. J. De Los Santos et al., Radio Systems Engineering, 21

DOI 10.1007/978-3-319-07326-2_2, © The Authors 2015
22 2  Modulation and Detection

Carrier
a0 cosω 0 t
v AM (t)
x
g(t) m + AM Output
Baseband Analog Multiplier
Signal (Mixer)
1
a DC Bias

g(t) g(t)

v AM (t)
v AM (t)

b c
Fig. 2.1   a Depiction of system-level block diagram of AM modulator. b Depiction of modulating
function, g( t), and resulting modulated function, vAM( t) for the case of a correctly modulated AM
signal, in which (1 + mg (t )) > 0. c Depection of modulating function, g( t), and resulting modulated
function, vAM( t) for the case of an incorrectly modulated AM signal, in which (1+ mg (t) <0) . In
this case the modulation index, m, is too large and it causes distortions in the amplitude envelope.
After [1]

FM (ω)
Spectra of
Baseband Signal
a 0 πδ (ω + ω 0) π a 0 mδ (ω + ω 0)
FB (ω + ω 0 ) FB() FB (ω − ω 0 )

0 BW=2 ωm
ω
−ω 0 −ω m ωm +ω0

Fig. 2.2   Spectra of AM signal. The spectra of the baseband signal, centered at zero frequency,
possess a maximum frequency ωm

(centered at − ω0) sidebands. Therefore, not only one of them is redundant, but the
transmitted power is split between the two sidebands. Finally, the modulated signal,
centered on ω0, occupies twice ωm, i.e.,
BWAM = 2ωm  (2.3)
The excess, unnecessary, power and bandwidth accompanying the AM signal rep-
resented by (2.2), motivated another scheme to eliminate one of the sidebands,
namely, single sideband AM or SSB modulation.
2.1  AM Modulators 23

Fig. 2.3   Depiction of filter-

ing method for producing a
mg(t) BPF vSSB(t)
single sideband (SSB) AM X (Filters one
of the side
signal. After [1]
bands)

a0 cosω0 t

Fig. 2.4   a, b Block diagram mg(t) vAM1

depicting the balanced AM
modulator method to create + vAM (t)
cos ω0 t +
a single sideband AM signal.
c Modulation signal, g( t), -
resulting modulated signal
AM
vAM2
with full carrier, vAM( t), and a -mg(t)
resulting modulated signal vDSB(t)
with suppressed carrier, mg(t)
X
vSSB( t). After [1]
mg(t) Carrier cosω 0 t
Oscillator vSSB(t)
ω0
-90° Σ
sin ω0 t +/-
-90°
Phase Shift X
b vDSB (t)

g(t)

vAM(t)

vSSB(t)

2.1.1 Methods to Create Single Sideband (SSB) AM Signals

Two main methods to create an SSB signal are presented, namely, the filtering meth-
od and the balanced modulation method. In the filtering method, whose block dia-
gram is presented in Fig. 2.3, one of the sidebands is eliminated by direct filtering.
In the balanced modulator method, the carrier is modulated with both + g( t), and
− g( t), and then subtracted, as depicted in Fig. 2.4a. This operation is implemented
by the block diagram shown in Fig. 2.4b (for the case in which there is no DC added
to g( t)), and results in the waveforms vSSB( t) shown in Fig. 2.4c [1].
24 2  Modulation and Detection

vAM(t) D Baseband(BB)
Out
LPF
R
C

BW < 1/RC < f0

a

BB Out

AM Carrier
b
Fig. 2.5   a Envelope detector circuit plus lowpass filter. b Modulated AM signal, vAM( t). The high
frequency carrier signal is filtered by the RC circuit. The modulating (baseband) envelope signal,
g( t), is extracted out of all its harmonics, by the lowpass filter and delivered to the output as BB.
After [1]

2.2 AM Demodulators

A variety of schemes are available to demodulate, i.e., to extract, the information-bear-

ing signal, g( t), from the received signal vAM( t) or VSSB( t); these are discussed next.

2.2.1 Envelope Detector

The Envelope Detector followed by a lowpass filter is perhaps the most funda-
mental AM demodulation system, Fig. 2.5. In this circuit, the received input signal
is rectified by a diode, D, and passed by an RC filter which, with an inverse time
constant greater than the modulating signal bandwidth, but smaller than the carrier
frequency, is not responsive to the high-frequency carrier, but only to the modulat-
ing signal, the signal envelope and its harmonics. The harmonics are then attenu-
ated, by a lowpass filter which selects the modulating signal [1].
2.3  FM Modulators 25

Fig. 2.6   Product detector BB Out

AM
plus lowpass filter. When the
locally generated carrier is x LPF
synchronized to the transmit- SSB/
DSBSC
ter, the circuit is called syn-
chronous detector. After [1]
a0cos ω0t

2.2.2 Product Detector

In the product detector, the incoming signal is multiplied by a locally generated

carrier, and the resulting signal is passed through a lowpass filter, Fig. 2.6 [1]. In
particular, the product of the incoming signal with the locally generated carrier
yields,
1
K [1 + mg (t ) ] cos 2 ω0 t = K [1 + mg (t ) ] ⋅ (1 + cos 2 ω0 t )
2
1 1
→ K [1 + mg (t ) ] ⋅ + K [1 + mg (t ) ] ⋅ ⋅ cos 2 ω0 t  (2.4)
2 2

which contains a first term whose spectral content is that of the baseband signal, and
a second term whose spectrum is centered at twice the carrier frequency, i.e., the
second term yields a double-sideband suppressed carrier (DSBC) signal. Therefore,
at the output of the lowpass filter, only the baseband signal remains.

2.3 FM Modulators

Two FM demodulators will be presented, namely, the direct and the indirect
methods.

2.3.1 Direct FM Modulator

The direct FM modulator uses a voltage controlled oscillator, Fig. 2.7.

In this circuit, the VCO has a free-running frequency, f0, given by,

1 
f0 = (2.5)
2π L0C0

This frequency is modulated by changing the value of frequency-determining ca-

pacitors (varactors) through the application of the baseband signal. In particular, the
VCO output frequency becomes a time-dependent function given by,

f (t ) = f 0 + kg (t ) (2.6)
26 2  Modulation and Detection

High-QLC Tank or Crystal

mg(t)
ω0

FM Out
Frequency
Amplifier
Multiplier
L

kg(t)+Vbias
Cvar Varactor Diode

C = C 0 + m eq g (t)
Fig. 2.7   Voltage-controlled oscillator (VCO) as FM modulator. After [1]

Fig. 2.8   Baseband sig-

nal ( top) and frequency
mg(t)
modulated carrier ( bottom).
After [1]

f<f0 vFM (t) f>f0

As the amplitude of the baseband signal increases and decreases, the carrier fre-
quency follows, Fig. 2.8 [1].

2.3.2 Indirect FM Modulator

Rather than modulating the varactor voltage of a VCO, in the indirect FM method
the modulating signal is first integrated, then used to phase-modulate a crystal-
controlled carrier frequency. The signal is then multiplied for producing a wideband
FM signal. The block diagram of an indirect FM modulator is shown in Fig. 2.9.

2.3.3 PM Modulator

To effect phase modulation, the constant carrier signal is generated independently,

and then passed through an LC circuit whose capacitance is modulated by the
2.4  FM Demodulators 27

a0 cosω 0 t

mg(t) FM Out
Phase Frequency
Integrator
Modulator Multiplier Large Mp
Mp<0.5

Fig. 2.9   Indirect FM modulator. After [1]

Fig. 2.10   Phase modulator Carrier

circuit. After [1] a0 cosω 0 t

PM Out

φ = φ 0 +kg(t)
L mg(t)+Vbias
Cvar Baseband
Varactor Signal
Diode

baseband signal, Fig.  2.10. For small varactor voltage swings, the LC circuit is
linear and small phase shifts (a few degrees) are obtained. Larger deviations may
be obtained using a frequency multiplier. The time-domain waveform of PM is
similar to that of FM, and can only be distinguished from FM from knowledge of
the source.

2.4 FM Demodulators

The various methods for effecting FM demodulation may be surmised from exam-
ining the FM waveform, namely [1],

vFM (t ) = A cos ϕ (t ) = Acos ω 0 t + K ∫ g (t )dt  (2.7)

In the first place, differentiating the instantaneous frequency,

dϕ(t)
ωi (t) = = ω0 + Kg(t) (2.8)
dt 
it is seen that we get the desired output g(t), if we cancel ω0.
On the other hand, differentiating vFM(t), we obtain:
dvFM (t ) dϕ
= − A sin ϕ (t )
dt dt
= − A [ ω 0 + Kg (t ) ] sin ϕ (t ) (2.9)
28 2  Modulation and Detection

v FM (t)  Kg ( t ) 
d Envelope A ω 0 1 +
Limiter
dt Detector  ω 0 

 Kg ( t ) 
Aω 0 1 +  sin ϕ ( t )
a  ω0 

i(t) H (ω )

+
L v 0 (t)
G C δ
-
ω
ω0
b c 1 LC

H (ω ) H (ω )
Operating
Region

∆ω

ω ω0 ω
ω0
1 LC 1 LC
δ δ

1
1 = ω 0 + δ
= ω − δ LC
d LC
0

Fig. 2.11   a FM detector block diagram. b Circuit implementing differentiation (d/dt) block.
c Transfer function of GLC circuit in b. d Relation of frequency deviation to operating region.

which looks like the an AM signal with envelope given by,

 Kg (t ) 
A [ ω 0 + Kg (t ) ] = Aω 0 1 +  (2.10)
 ω0 

Therefore, we may obtain g( t) by the envelope-detecting operation dvFM( t)/dt.

2.4.1 FM Balanced Demodulator

The balanced demodulator implements FM demodulation as shown in Fig. 2.11a.

In this block diagram, the limiter is inserted, prior to differentiation, to ensure that
no terms coming from A result when the derivative is taken (otherwise these varia-
tions in A may obscure the true amplitude variations due to g( t)) [2].
Then, middle block in Fig. 2.11a, differentiation, dvFM( t)/dt, is effected, which
results in a signal whose amplitude is modulated by g( t); this is like an AM signal
with high-frequency carrier ϕ( t). The differentiation is implemented by a parallel
GLC circuit, Fig. 2.11b, with transfer function,
2.4  FM Demodulators 29

Output Amplitude

ω0 ω

Amplitude Linear
Deviation

Frequency
a 2∆f Deviation
Slope Detector Envelope Detector

+ v 0 ( t ) ∝ ∆ ω = Kg ( t )
v FM (t)
-

-
b
Fig. 2.12   a Characteristic of differentiation circuit, where a frequency deviation is translated into
an amplitude deviation. b Balanced FM demodulator circuit differentiates and then detects FM
signal, extracting baseband signal

 V0 (ω)  1/G
 
|H (ω)| =  =  (2.11)
I (ω)  1 + (ωC − 1/ωL)2 /G2

The circuit has a damping α = G/2C, and resonance frequency 1 LC , which is

a distance δ from the carrier frequency, ω0. If the carrier frequency, ω0, is set at the
middle of the positive slope of the transfer curve, Fig. 2.11d, then it will deviate a
1 1
distance Δω, so that = ω 0 + δ on this operating region, and = ω0 - δ
LC LC
on negative slope operating region. Then, assuming ∆ω << ω0 , δ << α, and δ << ω0,
it can be shown [2] that the difference between H evaluated at these points is given
by,

2δ
H −H = ∆ω  (2.12)
ω0 + δ ω0 − δ
Gα2
It will be recalled that, differentiation of a function of time with respect to time, in
the time domain, is proportional to multiplication by frequency in the frequency
domain, i.e., d/dt → j ω . Therefore, (2.12) is the frequency domain representation
of differentiation [2].
Within a small deviation about the high carrier frequency ω0, the slope of the
transfer curve, H, is linear, Fig. 2.12a. While this differentiated signal is now an
30 2  Modulation and Detection

Phase
Detector
v FM (t) v 0 (t ) ∝ g (t )
x LPF
FM in

Phase v φ (t)
f i (t ) = f 0 + k f g (t )
Shift
π
( f ) = − + 2π K ( f − f 0 )
2

−π
Phase Shift

π
−
2
FM Signal
0
f0
Frequency
Fig. 2.13   FM demodulation by phase detector. After [1]

AM signal with its amplitude dictated by g( t), but still at a high frequency, it must
be passed by an envelope detector to remove the high-frequency carrier and finally
extract g( t). FM demodulation is, thus, completed by adding an envelope detector,
Fig. 2.12b [1, 2], producing an output voltage given by,

v0 (t ) ∝ ∆ω = Kg (t ) (2.13)

2.4.2 FM Quadrature (Detector) Demodulator

Examination of an FM signal reveals that its phase is shifted by an amount propor-

tional to its instantaneous frequency (2.14).
vF M (t) = A cos (ω0 t + ϕ(t))  (2.14)

This fact is exploited by the product detector, Fig. 2.13 [1], to detect the phase dif-
ference between an incoming FM signal and the signal at the output of a phase shift
network (2.15),
  
(2.15)   
vϕ (t) = A cos 2πf0 t + 2πkf g(t )dt + ϕ(fi (t)
2.4  FM Demodulators 31

Fig. 2.14   Phase-locked loop Phase

as FM demodulator. After [1]
Detector
FM in BB Out
x LPF
v FM (t) v LPF (t)
ωi (t ) = ω0 + Kg(t ) VCO
v 1 (t)

Since the phase shift introduced by the phase shift network is proportional to the
instantaneous frequency of the input FM signal, the output voltage of the phase
detector, after removing the high-frequency components by a lowpass filter, will
also be proportional to the instantaneous frequency of the input FM signal, thus
reproducing the original modulating signal, (2.16).

π
v0 (t ) = A′ 2 cos (φ( f i (t )) = A′ 2 cos ( − + 2π K ( f i (t ) − f 0 ))
2
= A′ 2 sin  2π Kk f g (t )  ∝ g (t ) (2.16)

The latter conclusions is clearly seen when one considers the series expansion of the
sine function, which would contain the fundamental, which is proportional to g( t),
the third power of g( t), etc. Thus, only the fundamental term, proportional to g( t),
would survive the lowpass filter, hence, (2.16).

2.4.3 Phase-Locked Loop (PLL)-Based FM Demodulator

The PLL, Fig. 2.14 [1], is a closed loop control system that tracks the variations in
the received signal phase and frequency (2.17).
vF M (t) = A cos (ω0 + ω)t  (2.17)

In this scheme, the VCO output,

v1 (t) = A1 cos (ω0 t + ϕ1 (t))t (2.18)

where,
(2.19) dϕ1 (t)
ω1 = ω0 +
dt
is compared with the input signal using a phase comparator (detector). The phase
comparator produces an output voltage proportional to phase difference. The phase
difference, in turn, is low-pass-filtered, producing VLPF( t), and fed back to the VCO
to control its output frequency. The control of the VCO, VLPF( t), is itself the de-
modulated FM signal, g( t) [1].
32 2  Modulation and Detection

2.4.4 Zero-Crossing Detector FM Demodulator

As shown in the previous FM demodulation methods, they typically involve a

frequency-to-amplitude (FM-to-AM) conversion. In the zero-crossing detec-
tor approach, the frequency-to-amplitude conversion operation is performed by
counting the number of zero-crossings in the input FM signal; its principle of
operation is as follows [3]. Given that the instantaneous frequency deviation in
the FM signal is,
dϕi (t)
(2.20)
ωi = ω0 +
dt

then,
 t2 t2
dϕ(t) = 2πfi (t)dt (2.21)
t1 t1

from where, upon integration between adjacent zero-crossing times t1 and t2, where
t2 > t1, one obtains,
 t2
ϕ(t2 ) − ϕ(t1 ) = [2πf0 + mg(t)]dt (2.22)
t1

But, also, because t1 and t2 are adjacent zero crossings, we must have,

ϕ(t2 ) − ϕ(t1 ) = π (2.23)

or
 [ 2π f 0 + mg (t ) ] ⋅ (t2 − t1 ) ≈ π (2.24)

which, using 2πfi (t), may be written as,

π ≈ 2πfi (t) · (t2 − t1 )  (2.25)

from where we obtain,


1
fi (t) = (2.26)
2π(t2 − t1 )
Now, since,
 mg(t)
fi (t) = f0 + (2.27)
2π
2.4  FM Demodulators 33

vFM (t) v1(t) v2(t) v3(t) vOut(t)

Monostable Low Pass
Limiter Differentiator Multivibrator Filter
a

v FM (t)

v 1 (t)

v 2 (t)

v 3 (t)

v Out (t)

b
Fig. 2.15   a Block diagram of zero-crossing detector for FM demodulation. b Waveforms through-
out the demodulator

Fig. 2.16   a FM demodula- v0

tor transfer curve. b PM
demodulator transfer curve

∆ω ∆φ
ω0 ω0t

a b

g( t) may be found by measuring the spacing between zero-crossings in the interval
t2 − t1.
The implementation of this method is depicted by the block diagram in Fig. 2.15
[3].
Following Fig.  2.15 (from top to bottom), the input FM signal, vFM( t), is ap-
plied to a limiter which, upon clipping the signal renders a rectangular pulse train,
v1( t), reflecting the FM time variation. This FM pulse train, in turn, is applied to a
differentiator, producing sharp pulses v2( t). These sharp pulses, coming out of the
differentiator, are then employed as triggers to a monostable multivibrator (a “one
shot”), which produces a pulse train v3( t), such that its average duration is propor-
tional to the desired (modulating) baseband signal. Upon passing this pulse train by
a lowpass filter, its average, a slowly varying DC component, vOut( t), is extracted;
this is the desired demodulated signal, g( t) [3].
34 2  Modulation and Detection

dϕ
∆φ = ∆ω
PM dt
Detector Differentiator

K 4 g(t)
FM
Waveform FM
Detector K 3 g(t)
a

PM K 1 g(t)
Detector

PM ∆ϕ = ∫ ∆ ω dt
Waveform FM ∆ω
Integrator
Detector
K 2 g(t)
b
Fig. 2.17   a FM demodulator. b PM demodulator

2.5 PM Demodulators

As described in the last few sections, an FM demodulator consists of a circuit whose

output voltage is proportional to the instantaneous frequency deviation, Fig. 2.16a.
A PM demodulator, on the other hand, consists of a circuit whose output voltage is
proportional to the instantaneous phase deviation, Fig. 2.16b [4].
Since the phase is the integral of the frequency, one should be able to perform the
demodulation function for either type of demodulation, FM or PM, with a single de-
tector, FM or PM, and an integrator or differentiator; this is shown in Fig. 2.17 [4].

2.6 Digital Modulation

In digital modulation, the carrier amplitude, frequency, or phase are made to vary
(i.e., are modulated) according to a digital representation of the baseband signal.
The first step in digital modulation, therefore, is to convert the baseband signal into
a digital stream of bits. This entails sampling the baseband signal at a frequency fs,
and representing each sample by N quantization bits, Fig. 2.18 [1].
As is known, to enable recovery of the original signal from the sampled signal,
the original signal must be sampled at a frequency greater than twice its maximum
frequency, Bm. This is the so-called Nyquist sampling rate limit. If the original
signal is sampled every tS = 1/fS seconds, and every sample contains N bits,
then the rate of creation of samples is N/tS = fS N bits/s. This is the sampled
data rate, R. It is said, then, that the original signal is represented by a data rate
of R bits/s [1].
2.6  Digital Modulation 35

S in

Sampling
fS > 2Bm
S Out 1/ f S

Fig. 2.18   Digitization of input baseband signal, Sin, via sampling at time intervals 1/fS, to produce
a sampled output representation, SOut, where each sampling point is in turn represented by N bits.
After [1]

The data rate, R, utilized in transmitting a given signal, may not be increased
arbitrarily by increasing fS, but it is limited by the bandwidth of the channel through
which the signal travels. This capacity, in particular, is given by [1],
C = 2 BLog 2 M [bits / sec]  (2.28)
where B is the channel bandwidth and M is the number of bits utilized to represent
a signal level being transmitted. For example, if we use two bits to represent a sig-
nal level, then M = 2. If three bits, M = 3, see Fig. 2.19. C is known as the Shannon
limit. and represents the maximum bit rate allowed by a channel. It is evident from
Fig. 2.19 that, while the resolution with which a given signal level may be repre-
sented increases with the number of bits utilized, the fact that the step-size decreas-
es as M increases results in an increase in the likelihood that noise might "push" one
level into the next, thus contributing to an error, i.e., it is easier for noise to confuse
the true quantized level. A standardized parameter that may be used to compare
channel capacity is the so-called bandwitdh efficieny, which is given by [1],

C
= 2 Log 2 M [bits / sec /Hz ]  (2.29)
B
A plot of bandwidth efficiency versus M is shown in Fig. 2.19.
A standardized parameter that may be used to compare channels is the so-called
bandwitdh efficieny, which is given by,
C
= 2 Log 2 M [bits / sec /Hz ]  (2.30)
B
36 2  Modulation and Detection

100 1000
0111
011 0110
0101
010 0100
0011
001 0010
∆1 0001 ∆1/2
000 0000
a b
Bandwidth Efficiency [bits/sec/Hz] 8

2
2 4 6 8 10 12 14 16

M
c
Fig. 2.19   Signal quantization. a Signal levels represented by 3 bits, i.e., M = 3 bits. b Signal levels
represented by 4 bits, i.e., M = 4 bits. c Bandwidth efficiency versus number of bits representing a
signal level. After [1]

A plot of bandwidth efficiency versus M is shown in Fig. For M = 3, the maximum
bandwidth efficiency would be 3.17.
A measure of channel capacity, that doesn’t require that M be specified, is in
terms of the signal-to-noise ratio (SNR). The SNR is defined as,

Eb R
SNR =  (2.31)
N0 B

where Eb is the average energy per bit, and N0 is the noise power spectral density. In
practice, C < R, therefore, one obtains,

C  E R
= Log 2 1 + b ⋅  (2.32)
B  N0 B 

What this formula indicates is that, for a given noise level, N0, channel capacity is a
function of the average energy per bit, Eb, expended [1–3].

2.6.1 Binary Modulation—Amplitude-Shift Keying (ASK)

When the baseband signal is quantized into two levels, N = 2, it is represented by a
binary stream of 1’s and 0’s. Binary modulation of the carrier amplitude is called
2.6  Digital Modulation 37

0 1 0 1 0 0 1 0

T T T T T t
a
0 1 0 1 0 0 1 0
ASK Signal

t
Binary ASK  BASK
2T
b
Fig. 2.20   Amplitude-Shift-Keying. a Binary signal. b Modulated signal. After [7]

S(f) 1
T

−
3 3 f
T T

a −
2
T
0 2
T

S(f)
S(f) fc +
1
T

3 3
−
T T
f
b fc −
2
T
fc fc +
2
T

Fig. 2.21   a Spectrum of baseband signal. b Spectrum of ASK Wave (positive frequencies only).
After [7]

ON-OFF Keying (OOK) or Amplitude-Shift-Keying (ASK), Fig. 2.20, whose cor-

responding spectrum is given in Fig. 2.21.

2.6.2 Binary Modulation—Frequency-Shift Keying (FSK)

Binary modulation of the carrier frequency is called Frequency-Shift Keying

(FSK). In this scheme, as depicted in Fig. 2.22, the transmitted frequency is dic-
tated by a binary sequence of 1’s and 0’s, according to which frequencies f1 and f0
are transmitted over intervals of time T; the corresponding spectrum is as depicted
in Fig. 2.23.

2.6.3 Binary Modulation—Phase-Shift Keying (PSK)

Binary modulation of the carrier phase is called Phase-Shift Keying (PSK). This
is illustrated in Fig. 2.24. In this scheme, the phase of the carrier is shifted 180° in
38 2  Modulation and Detection

g 0 (t) 0 1 0 1
1
t

g 1 (t)
1
t
a T 2T 3T

0 1 0 1 0
A
t Frequency - Shift Keying

-A Binary FSK  BFSK

f1 f0
b
Fig. 2.22   Frequency-shift-keying. a Modulating binary signal. b FSK wave. After [7]

g0(f) 1
T
or g1(t)
Envelope
3 3
−
T T
f
2 0 2
a −
T T

S(f)
∆f
2∆f
2
1 1
T T
Envelope
3 3 3 3
− 2B − 2B
T T T T f
b f0 −
2
f0 f 0 +
2
f1 −
2
f1 f1 +
2
T T T T

Fig. 2.23   a Spectrum of modulating baseband signal. b Spectrum of FSK Wave (positive frequen-
cies only). After [7]
2.7  Complex Envelope Form of Modulation Signals 39

Fig. 2.24   a Modulating g(t) 0 1 0 1

binary signal. b Phase-shift
1 t
keying (Binary PSK
BPSK). After [7]
-1 …
T
a

1 0 1 0
A
t

-A
b

G(f) 1
2T

Envelope
3 3
−
T T
f

a −
2 0 2
T T

S(f)
1 1
2T 2T

Envelope Envelope
3 3 3 3
− 2B − 2B
T T T T
f
-fc fc
2 2 0 fc −
2
fc +
2
− fc − − fc − T T
b T T

Fig. 2.25   a Spectrum of modulating baseband signal. b Spectrum of PSK wave. After [7]

accord with whether the modulating signal is a 1 or a 0. The pertinent spectrum is

shown in Fig. 2.25.

2.7 Complex Envelope Form of Modulation Signals

Waveforms that are employed for digital modulation may be expressed in complex
envelope form. [3, 5] i.e.

M(t) = I (t) + j Q(t) = A(t)ej ϕ(t) (2.33)

40 2  Modulation and Detection

where I( t) and Q( t) represent in-phase and quadrature envelope waveforms, respec-
tively, and have the following forms,

I (t ) = ∑ I k pI (t − kTs − τ ) (2.34)
k


Q(t ) = ∑ Qk pQ (t − kTs − τ ) (2.35)
k

where Ik and Qk represent sequences of discrete variables mapped from the base-
band (information) data with a symbol rate of 1/Ts, pI( t) and pQ( t) represent finite
energy pulses, such as rectangular or Gaussian, τ is a possible delay, and A( t and
ϕ(t) ) are the envelope amplitude and phase, respectively [3, 5]. The amplitude and
phase in (2.33) are given by,

A(t) = I 2 (t) + Q2 (t) (2.36)



Q(t )
φ(t ) = tan −1 (2.37)
I (t )

The symbol duration, Ts, of an M-ary keying modulation is related to the bit dura-
tion Tb of the originally bynary data stream as,

Ts = log 2 M ⋅ Tb (2.38)

Digital modulation beyond binary (i.e., with more than two levels) usually exploits
the I and Q representation.

2.7.1 M-ary Modulation—MPSK

The representation of these modulation schemes is better visualized via the concept
of constellations. We begin this subject by introducing the geometric representation
of modulation signals.
If there are M possible signals, the modulation signal set S can be represented
as [5]:
(2.39)
S = {s1 (t), s2 (t), ..., sM (t)}

While for binary modulation schemes, a binary information bit is mapped directly
to a signal and S will contain only two signals, for higher-level modulation schemes
( M-ary keying) the signal set will contain more than two signals; each signal (or
symbol) will represent more than a single bit of information.
With a signal set of size M, it is possible to transmit a maximum of log2M bits of
information per symbol. Since the elements of S may be viewed as points in a vec-
tor space, the representations are called constellations. When the baseband signal is
2.7  Complex Envelope Form of Modulation Signals 41

quantized into M levels, N = M, it is represented by an M-ary stream of 1’s and 0’s.
Now, from the fact that any finite set of physically realizable waveforms in a vector
space can be expressed as a linear combination of N orthonormal waveforms, which
form the basis of that vector space, representing the modulation signals on a vector
space, requires finding a set of signals that form a basis for that vector space. Once
a basis is determined, any point in that vector space can be represented as a linear
combination of the basis signals [3, 5],

 {ϕ (t ) j = 1, 2,..., N }
j
(2.40)

such that,
N
 si (t ) = ∑ sij φ j (t ) (2.41)
j =1

The basis signals are orthogonal to one another such that,

∞
φi (t)φj (t)dt = 0 i  = j  (2.42)
−∞

and each of the basis signals is normalized to have unit energy, i.e.,
∞
 E=
-
∫ϕ
∞
i
2
(t )dt = 1 (2.43)

2.7.2 Binary Phase Shift Keying Modulation—BPSK

In this digital modulation scheme, the phase of a constant envelope carrier is

switched between two signals m1 (bit 1) and m2 (bit 0), the two phases being sepa-
rated by 180°. The BPSK signal is [3],

2 Eb
sBPSK (t ) = cos (2π f c t + θ c ) 0 ≤ t ≤ Tb (binary 1)  (2.44)
Tb

or
2 Eb
sBPSK (t ) = − cos (2π f c t + θ c ) 0 ≤ t ≤ Tb (binary 0)  (2.45)
Tb

which may be written as,

 2 Eb
sBPSK (t ) = m(t ) cos (2π f c t + θ c ) (2.46)
Tb
42 2  Modulation and Detection

Q I: In-phase
Q: Quadrature
I
Eb
− Eb

Fig. 2.26   BPSK constellation diagram

This is equivalent to double sideband suppressed carrier modulated waveform,

where m( t) is the modulating signal and cos( 2πfct) is the carrier; it can be generated
using a balance modulator.
Assuming a rectangular pulse ( p( t) = rect(( t-Tb/2)Tb), for BPSK signals s1( t) and
s2( t) given by,

2 Eb
s1 (t ) = cos (2π f c t ) 0 ≤ t ≤ Tb  (2.47)
Tb

and
2 Eb
 s2 (t ) = − cos (2π f c t ) 0 ≤ t ≤ Tb (2.48)
Tb

Eb and Tb are the energy and period per bit, respectively, ϕi( t) consists of a single
waveform, namely,

2
ϕ1 (t ) = cos (2π f c t ) 0 ≤ t ≤ Tb  (2.49)
Tb

Using this basis signal, the BPSK signal set can be represented as,

S BPSK = { Eb ϕ1 (t ), − Eb ϕ1 (t )  } (2.50)

This signal set can be shown geometrically as in Fig. 2.26. This, so-called constella-
tion diagram, provides a graphical representation of the complex envelope of each
possible symbol state.
It should be noticed that [3, 5]: (1) The number of basis signals will always be
less than or equal to the number of signals in the set. (In BPSK we have one basis
signal, but two signals in the set); (2) The number of basis signals required to repre-
sent the complex modulation signal set is called the dimension of the vector space;
(3) If there are as many basis signals as there are signals in the modulation signal
set, then all the signals in the set are orthogonal to one another.
The properties of a modulation scheme that can be inferred from its constellation
are as follows: (1) The bandwidth occupied by the modulation signals decreases as
2.7  Complex Envelope Form of Modulation Signals 43

PBSK (f) (dB)

f c -3Rb f c -2Rb f c -R b fc f c + R b f c +2Rb f c +3Rb

Fig. 2.27   Power spectral density for BPSK. The bandwidth, BW = 2 Rb = 2 / Tb

the number of signals/dimension increases, in particular, the more densely packed

a constellation, the more bandwidth-efficient it is; (2) The bandwidth occupied by
a modulated signal increases with the dimension N of the constellation; (3) The
probability of bit error is proportional to the distance between the closest points in
the constellation [3, 5].
It can be shown, that the typical power spectral density for BPSK is given by [3],

E  sinπ( f − f ) T  2  sinπ( − f − f ) T  2 
PBPSK ( f ) = b  c b
 +
c b
  (2.51)
2  π( f − f c ) Tb   π( − f − f c ) Tb  
 
This is plotted in Fig. 2.27. Further, the bandwidth containing the signal energy is a
function of the shape of the pulse [3]. For a rectangular pulse, 90 % of signal energy
is contained within a bandwidth BW~1.6Rb, whereas for a raised cosine filtering
pulse with α = 0.5, 100 % of signal is contained within a BW = 1.5Rb [3].

2.7.2.1 Binary Phase Shift Keying Detection

The detection of a BPSK employs a coherent or synchronous demodulation ap-

proach to extract the modulated signal, Fig. 2.28. Accordingly, information about
the carrier phase and frequency must be available at the receiver [5].
In particular, the received signal is first squared to generate a DC signal and a
varying sinusoid at twice carrier frequency. The DC is filtered out and a frequency
divider recreates the carrier which, multiplied by itself and lowpass-filtered results
in the modulation signal m( t) [3].
44 2  Modulation and Detection

cos 2(2πfct+θ) cos(4πfct+2θ) cos(2πfct+θ)

m(t)cos(2πfct+θ) Square Bandpass Frequency m(t)
Lowpass
Law Filter Divider
2fc Filter
Device f/2

m(t)cos(2πfct+θ) m(t)cos 2(2πfct+θ)

Carrier recovery circuits

Fig. 2.28   BPSK demodulation circuit

{mk} 1 0 0 1 0 1 1 0
(dk-1} 1 1 0 0 1 1 0 0 0 dk = mk ⊕ dk −1
(dk} 1 1 0 1 1 0 0 0 1
Fig. 2.29   Illustration of differential BPSK. dk is unchanged from previous symbol if mk = 1, and
dk is toggled if mk =  0

Assuming an additive white Gaussian noise channel (AWGN) with noise spec-
tral density N0, the average probability of bit error for BPSK is a function of the
energy per bit, Eb is given by [3],
 2 Eb 
Pe , BPSK = Q   (2.52)
 N0 

where,
∞
1 
Q( x) = ∫ exp ( − x 2 / 2)dx (2.53)
x 2π

2.7.3 Differential Binary Phase Shift Keying—DBPSK

In the DBPSK modulation approach, the BPSK technique is employed in a

non-coherent fashion to circumvent the need for a coherent reference signal at the
receiver. In this scheme, the input binary sequence is first differentially encoded and
then modulated using a BPSK modulator. In particular, the input binary sequence
{mk} is used to generate a differentially encoded sequence {dk} by complementing
the modulo-2 sum of mk and dk − 1. Figure 2.29 illustrates the differential encoding
process [3].

2.7.3.1 DPSK Modulator

To implement DBPSK modulation, a one-bit delay element and logic circuit gener-
ate a differentially encoded sequence, and then the differentially-encoded bit stream
2.7  Complex Envelope Form of Modulation Signals 45

DBPSK
Input Data { d k} Signal
Logic Product
{ m k} Circuit Modulator

{ d k - 1}
cos (2πf ct)
Delay
Tb

Fig. 2.30   DBPSK modulator

DPBSK Demodulated
Signal Bandpass Logic Integrate Threshold Signal
Filter Circuit and Dump Device

Delay
Tb

Fig. 2.31   DBPSK demodulator

is passed through a product modulator. This is depicted in the block diagram of

Fig. 2.30 [3].

2.7.3.2 DPSK Demodulator

Implementation of the DPSK demodulator involves recovering the received sequence

recovered through a process complementary to the modulation process, see Fig. 2.31
[3]. The approach exhibits a reduced receiver complexity, but at the expense of a 3dB
inferior energy efficiency. The probability of bit errors is given by [3], [5],

1  E 
Pe , DPBSK = exp  − b   (2.54)
2  N0 

2.7.4 Quadrature Phase Shift Keying Modulation—QPSK

In this modulation scheme, the phase of the carrier takes on four equally-spaced val-
ues, namely, 0, π/2, π, and 3π/2, where each value of phase corresponds to a unique
pair of baseband (message) bits given by [3, 5],
46 2  Modulation and Detection

π/2 Q Q
Es 3π/4 π/4
π 0
I I
Es
5π/4 7π/4
a 3π/2 b
Fig. 2.32   QPSK Constellations

2 Eb π
sQPSK (t ) = cos (2π f c t + (i − 1) ) 0 ≤ t ≤ Ts i = 1, 2, 3, 4 (2.55)
Tb 2


2 Eb  π 2 Eb  π
sQPSK (t ) = cos (i − 1)  cos (2π f c t ) − sin  (i − 1)  sin(2π f c t (2.56)
Tb  2 Tb  2

With the basis functions:


2
ϕ1 (t ) = cos (2π f c t ) (2.57)
Ts

and

2
ϕ 2 (t ) = sin(2π f c t ) (2.58)
Ts

the QPSK set is,

  
 π  π
SQPSK =  Es cos (i − 1)  ϕ1 (t ), − Es sin (i − 1)  ϕ 2 (t ) i = 1, 2, 3, 4 (2.59)
  2   2  

and its geometrical constellation is as shown in Fig.  2.32. The constellations are
equivalent; they differ by a 45-degree rotation.
It can be shown that the power spectral density for QPSK, see Fig. 2.33, is given
by [3],
 sin 2π( f − f ) T  2  sin 2π( − f − f ) T  2 
PQPSK ( f ) = Eb   
c b c b
 + (2.60)
 π( f − f c ) Tb   π( − f − f c ) Tb  
 

The null-to-null BW = Rb, which is half of that of a BPSK signal [3, 5].
2.7  Complex Envelope Form of Modulation Signals 47

Fig. 2.33   Power spectral density for QBPSK. The bandwidth, BW = Rb

Lowpass
Filter

Input
Data QPSK
Rb Serial to Local Bandpass Signal
Parallel Oscillator + Filter
Converter
90 0

Lowpass
Filter

Fig. 2.34   QPSK modulator

2.7.4.1 Quadrature Phase Shift Keying Modulator

In the QPSK modulator, a unipolar bit stream m( t) carrying the information sig-
nal at the rate Rb is first converted into a bipolar non-return-to-zero (NRZ) se-
quence. This sequence is then split into in-phase and quadrature components,
mI( t) and mQ( t), each having a bit rate Rb/2. These two binary sequences, in turn,
are separately modulated by two carriers, so that each is considered BPSK, and
summed to produce the QPSK signal. In order to confine the power spectrum
of the QPSK signal to its designated band, it is subsequently passed through a
bandpass filter (BPF). A block diagram implementing the modulator is shown in
Fig. 2.34 [3, 5].
48 2  Modulation and Detection

Lowpass Decision
Filter Circuit

Received Recovered
Signal Carrier Signal
Bandpass Symbol
Recovery Multiplexer
Filter Timing
Circuit
Recovery

900

Lowpass Decision
Filter Circuit

Fig. 2.35   QPSK demodulator

Fig. 2.36   π/4 QPSK constel- Qk Qk

lations: a Possible states
when θk-1 =  nπ/2. b Possible
states when θk -1 = nπ / 4 Ik Ik

a b

2.7.4.2 Quadrature Phase Shift Keying Demodulator

The demodulator system for a QPSK-modulated signal is depicted by the block

diagram in Fig.  2.35. The incoming QPSK signal is first filtered by a front-end
bandpass filter to remove the out-of-band noise and adjacent channel interference
accompanying it. Subsequently, the filtered output is split into two parts, each of
which is then coherently demodulated using I and Q carriers. The coherent carriers
used are recovered from the received signal using carrier recovery circuits as before
(see BPSK). Next, the outputs of the demodulators are passed through decision
circuits which generate I and Q streams. Finally, the two stream components are
multiplexed to reproduce the original binary sequence [3, 5].

2.7.5  π
 /4 QPSK—Shifted Quadrature Phase-Shift Keying
Modulation

In this scheme, two QPSK constellations, shifted by π/4 with respect to each other,
are sampled to constitute the modulated signal. Selecting every successive bit by
alternating between the two constellations, ensures that there is at least a shift which
is an integer of π/4 radians between successive symbols. This guarantees that there
2.7  Complex Envelope Form of Modulation Signals 49

cos2πfct

mIk Lowpass I(t)

Ik
Filter π/4 QPSK
mk Serial to Signal Signal
Parallel
Converter
Mapping + Amplifier

mQk Qk Q(t)
Lowpass
N −1 N −1 Filter
I( t ) = ∑I
k =1
k p( t − kTs − Ts / 2 ) = ∑ cosθ
k =1
k p( t − kTs − Ts / 2 )

N −1 N −1 -sin2πfct
Q( t ) = ∑Q
k =1
k p( t − kTs − Ts / 2 ) = ∑ sinθ
k =1
k p( t − kTs − Ts / 2 )

Fig. 2.37   π/4 QPSK modulator. Definition of variables: I k = cosθ k = I k -1cosφ k - Q k -1sinφk ;
Q k sin θ k = I k -1sinφ k + Q k -1cosφ k ; Phases of kth and k–1st symbols θ k = θ k -1 + φk

is a phase transition for every symbol, which enables a receiver to perform timing
recovery and synchronization [3, 5]. Figure 2.36 depicts the two constellations.

2.7.5.1  π/4 QPSK—Shifted Quadrature Phase-Shift Keying Modulator

In the circuit implementation of this modulation scheme, Fig.  2.37, the input bit
stream is first partitioned, by a serial-to-parallel converter, into mI, k and mQ, k par-
allel bit streams, where each symbol has a rate equal to half that of the incom-
ing bit rate. Subsequently, each of the streams is applied to a signal mapping cir-
cuit which generates the k-th in-phase and quadrature pulses, Ik and Qk, over the
kT < t < (k + 1)T time interval. This is followed by the separate modulation of Ik
and Qk by two carriers which are then added to produce the π/4 QPSK signal [3].
The block diagram implementing this is shown in Fig. 2.37.

2.7.5.2  π/4 QPSK—Shifted Quadrature Phase-Shift Keying Demodulator

Demodulation of a π/4 QPSK signal proceeds as follows (the numbers correspond

to those in diagram of Fig. 2.38): (1) The incoming π/4 QPSK signal is first quadra-
ture-demodulated using two LO signals with the frequency identical to the transmit-
ted carrier; (2) The output product is applied to lowpass filters, whose output is sam-
pled at the maximum amplitude every Ts seconds; (3) The two sequences are then
passed through differential decoders; (4) The output of differential decoder may be
expressed as xk = cos (φ k − φ k −1 ), yk = sin(φ k − φ k −1 ); (5) The output of the differ-
ential decoder is applied to the decision circuit, which uses the table in Fig. 2.36(b)
to determine the output bit as follows: SI = 1, if x k > 0 or SI = 0, if x k < 0
SQ = 1, if y k > 0 or SQ = 0 if y k < 0.
50 2  Modulation and Detection

 
ZN
/RZSDVV 'HFLVLRQ
)LOWHU ZNZ N]N]N 'HYLFH

6DPSOHDWPD[ [N
2XWSXWIRUHYHU\7V 6,
FRVπIFW 

π436. 
5),QSXW
%DQGSDVV 2XWSXW
)LOWHU 0XOWLSOH[HU

VLQπIFW 6DPSOHDWPD[
2XWSXWIRUHYHU\7V 64

/RZSDVV 'HFLVLRQ
]NZ NZN]N
)LOWHU \N 'HYLFH
a ]N

,QIRUPDWLRQELWVP,NP4N 3KDVH6KLIWφN
 π
 π
 π

b
 π

Fig. 2.38   a π/4 QPSK demodulator. Definition of variables: In-Phase: w k = cos(φ k - γ ); In

Quadrature: z k = sin (φ k - γ ); φ k = tan -1 (Q k / I k ); γ due to noise. b Carrier phase shift corresponding
to various input bit pairs

2.7.5.3  π/4 Shifted QPSK—Demodulator with IF Differential Detector

This scheme circumvents the need for local oscillators (LO), see Fig. 2.38, by us-
ing a delay line and 90° phase shifter to implement the quadrature, and two phase
detectors, see Fig.  2.39. The operation is as follows (the numbers correspond to
those in diagram of Fig. 2.39): (1) The received signal is converted to intermediate
frequency (IF) and band-pass filtered; the bandpass filter is designed to match the
transmitted pulse shape to preserve carrier phase and minimize noise power; (2) The
received IF is differentially decoded using delay line and two mixers; (3) Similar to
baseband differential detector already discussed [3].

2.7.5.4  π/4 QPSK Demodulator: FM Discriminator

This scheme operates as follows (the numbers correspond to those in diagram of

Fig. 2.40): (1) The input signal is first filtered using a bandpass filter (BPF) that
2.7  Complex Envelope Form of Modulation Signals 51

Lowpass Decision
Filter Device
Sample at max.
2 Output for every Ts
Modulated
IF input signal
Demodulated
Bandpass Output
Ts
Filter Multiplexer
1
900
Sample at max.
Output for every Ts

Lowpass Decision
Filter Device
3

Fig. 2.39   π/4 QPSK Demodulator IF Differential Detector

3 4
π /4 QPSK 2
signal Integrate
Bandpass FM and
Limiter
Filter Discriminator Dump
1

Demodulated
Four- level
Four Parallel to Waveform
Threshold Serial
Detector Converter

Fig. 2.40   π/4 QPSK Demodulator: FM Discriminator

is matched to the transmitted signal; (2) The filtered signal is passed by a hard-
limiter to minimize fluctuations in its envelope; (3) The FM discriminator extracts
the instantaneous frequency; (4) The extracted signal is integrated over each period
giving the phase difference between two sampling instants [3].

2.7.6 Binary Frequency-Shift Keying—BFSK

In this scheme, the frequency of a constant-amplitude carrier is switched between two

values (bits 1 and 0). Then, depending on how frequency changes, there may be dis-
continuous phase or continuous phase between bits. The BFSK signal is given by [3],

 2 Eb
sBFSK (t ) = cos (2π f c + 2π∆f c )t 0 ≤ t ≤ Tb (binary 1) (2.61)
Tb
52 2  Modulation and Detection

Correlator
Mixer
T

∫ 0
dt
3
2
+ Output
r(t) cos2 πf c t
1 + Decision
sin2πf c t Circuit
-
T

∫
0
dt

Mixer Correlator

Fig. 2.41   BFSK demodulator: Coherent detector

or

2 Eb
sBFSK ( t ) = cos (2π f c − 2π∆ f )t 0 ≤ t ≤ Tb (binary 0) (2.62)
Tb

which is synthesized by,

2 Eb ∞

s BFSK (t ) = cos (2π f c t + 2π k f ∫ m(ξ )d ξ ) (2.63)
Tb −∞

It is noticed that, even if m( t) is discontinuous, due to the integration, the phase is
continuous [3].

2.7.6.1 BFSK Modulator

The most common way to effect BPSK modulation is direct FM, that is, modulating
the frequency of an oscillator with the message signal [3].

2.7.6.2  BFSK Demodulator: Coherent Detector

This demodulation scheme is implemented as shown in Fig. 2.41, and operates as

follows (the numbers correspond to those in diagram): (1) Two correlators are ap-
plied to reference signals that are generated locally and that are coherent; (2) The
difference of the correlator outputs is then compared in a threshold comparator;
(3) The output signal is the result of the comparison, in particular, if the difference
signal has a value greater than the threshold, it is classified as “1”, otherwise it is
classified as a “0.” The probability of bit error is given by [3],
2.7  Complex Envelope Form of Modulation Signals 53

Matched Envelope
Filter at Detector
fL
2 3
FSK RF input
1
+ Output
+ Decision
Circuit
-
Matched
Envelope
Filter at
Detector
fH

Fig. 2.42   BFSK demodulator: Noncoherent detector

 Eb 
Pe , BPSK = Q   (2.64)
 N0 

2.7.6.3  BFSK Demodulator: Noncoherent Detector

To demodulate the BFSK signal, the block diagram shown in Fig. 2.42 is employed;
it operates as follows (the numbers correspond to those in diagram): (1) The in-
coming signal is first split into two paths, namely, a lower path where it is passed
through a filter that is matched to the frequency fH, and an upper path where it is
passed by a filter matched to the frequency fL; (2) The matched filters function as
bandpass filters centered at fH and fL, and are passed through envelop detectors; (3)
The difference of the outputs of the envelope detectors is then sampled at times
t = kTb; (4) Depending on the magnitude of the sampled value, the decision circuit
decides the bit classification as 1 or 0. The probability of bit error rate is [3],

1  −E 
Pe, FSK , NC = exp  b   (2.65)
2  2 N0 

2.7.7  Minimum Shift Keying—MSK

The MSK modulation approach entails producing a peak deviation in the frequency,
equal to one-quarter the bit rate; since it is a special type of continuous phase-fre-
quency shift keying (CPFSK), MSK is also denoted continuous FSK, with modula-
tion index of 0.5. In general, the modulation index for this scheme is defined as:
kFSK = ( 2ΔF)/Rb, where ΔF is the peak deviation; Rb is bit rate. Mobile radio systems
frequently use MSK due to its spectrally efficient modulation scheme. MSK pos-
sesses: (1) Constant envelope; (2) Good BER performance; (3) Self-synchronizing
capability [3].
54 2  Modulation and Detection

Bandpass 2 x(t)
fc+1/4T - + 3
+
mI(t) 4
cos2πfct 1
sMSK(t)
+
mQ(t)
+
Bandpass + y(t)
cosπt/2Τ Fc-1/4T +

Fig. 2.43   MSK modulator

The MSK signal is given by,

N -1 N -1
sMSK (t ) = ∑ mIi (t ) p (t - 2iTb ) cos 2 π f c t + ∑ mQi (t ) p (t - 2iTb - Tb ) sin 2 π f c t  (2.66)
i=0 i=0

where,
  πt 
 cos  2T  0 ≤ t ≤ 2Tb
p (t ) =   b (2.67)
 
 0 elsewhere

where mI and mQ are the even and odd bits of a bipolar serial data stream with values
+/− 1; they are fed at the rate Rb/2 [3]. The MSK waveform may be represented by,

2 Eb πt 
sMSK (t ) = cos (2π f c t − mIi (t )mQi (t ) + ϕk ) (2.68)
Tb 2Tb

MSK may be visualized as a special type of continuous phase FSK, in which ϕk is 0

or π depending on whether mI( t) is 1 or −1 [3].

2.7.7.1  MSK Modulator

The block diagram of the MSK modulator is shown in Fig.  2.43, and its opera-
tion is as follows (the numbers correspond to those in diagram): (1) The carrier is
multiplied with cos( πt/2T) to produce two-phase-coherent signals at fc + 1/4T and
fc − 1/4T; (2) The two FSK signals are separated using two bandpass filters centered
at the respective frequencies; (3) And combined to form in-phase and quadrature
carrier components, x( t) and y( t), respectively; (4) These carriers are multiplied
with even and odd bit streams, mI( t) and mQ( t) to produce sMSK( t).
2.7  Complex Envelope Form of Modulation Signals 55

2 3 4 mI(t)
Integrator 3 Threshold
Device
4
t=2(k+1)T

SMSK(t) 1 x(t)

y(t)
t=2(k+2)T
Threshold mQ(t)
Integrator
Device

Fig. 2.44   MSK demodulator

2.7.7.2 MSK Demodulator

The block diagram of the MSK demodulator is shown in Fig. 2.44, and its operation
is as follows (the numbers correspond to those in diagram): (1) First, the received
incoming signal is multiplied by the respective in-phase and quadrature carriers x( t)
and y( t); (2) The output of the multipliers are integrated over two bit periods and
dumped to a decision circuit at the end of the period; (3) Based on the level of the
signal, a threshold detector decides its classification as 1 or 0; (4) The output data
streams correspond to mI( t) and mQ( t), which are offset to obtain the demodulated
signal.

2.7.8 M-ary QAM—Quadrature Amplitude Modulation

In this scheme, the carrier amplitude is allowed to vary together with the phase. The
signal is given by,

2 Emin 2 Emin
si (t) = ai cos (2 π f c t) + bi sin(2 π f c t)
Ts Ts

0 ≤ t ≤ T i = 1, 2,...., M (2.69)

where Emin is the energy of the signal with the lowest amplitude, and ai and bi are a
pair of independent integers according to location of a signal point. The pertinent
constellation for, e.g., M = 16, is as given in Fig. 2.45.

2.7.9  OFDM—Orthogonal Frequency Multiplexing

The OFDM approach to signal transmission provides power-efficient signaling for

a large number of users in the same channel. In this scheme, multiple sub-carriers
56 2  Modulation and Detection

Fig. 2.45   Constellation of φ2(t)

M-ary QAM (M = 16)

φ1(t)

Fig. 2.46   a OFDM modula-

tor block diagram. b Carriers fc1
modulated by rectangularly-
shaped data pulses are Serial to Parallel
OFDM Modulated
densely spaced and orthogo- Modulating fc2
Signal

Combiner
nal in frequency. After [1] Data
…
fN

a
Power

fc1 fc2 fc3

b Frequency

are employed to transmit signal bits in parallel for very high throughput. The spec-
tra of different sub-channels can partly overlap and pulse shaping is not necessary.
OFDM is usually combined with QAM or M-PSK [1] (Fig. 2.46).

2.7.10 Direct Sequence Spread Spectrum Modulation

The idea behind the direct sequence (DS) spread spectrum (SS) scheme, depicted
in Fig. 2.47, is to greatly expand or spread the carrier spectrum relative to the infor-
mation rate [6]. The spread spectrum signal attains an anti-jamming capability by
forcing the jammer to deploy the transmitted power over a much wider bandwidth
than would be necessary for a conventional system [6]. In other words, for a given
jammer power, the jamming power spectral density is reduced in proportion to the
ratio of the spread bandwidth Bs to the un-spread bandwidth B.
2.7  Complex Envelope Form of Modulation Signals 57

RX/TX
Undesired
Signals
Recovered
Data B C Data
Spreading Spreading
A D
PSD
PSD
RF Bandwidth

A B

B frequency fc frequency

Noise + Other Users

PSD + Jammers PSD

C D

fc frequency fc frequency

Fig. 2.47   Direct sequence spread spectrum modulation concept

2.7.10.1  Direct Sequence Spread Spectrum Modulation/Demodulation

DS/SS modulator and demodulator block diagrams are shown in Fig.  2.48. To
produce an SS signal, the input data is passed through a data encoder and a data
modulator, which results in the signal message m( t). m( t) is subsequently applied
to a mixer, where it is multiplied by the product of the carrier and a pseudorandom
sequence (a random-like signal of ±1 p( t)) to produce [6],

2 Es
VSS (t ) = m(t ) p (t ) cos (2π f c t + θ ) (2.70)
Ts

where,
∞ 
(2.71)

m(t) = am pm (t − mTb )
m=−∞

and
∞
cn p  (t − nTc )  (2.72)

p(t) =
n=−∞

The DS/SS system is employed in Code Division Multiple Access (CDMA), where
each user has pseudo-noise (PN) sequence, and multiple users share the same BW;
the signals of one user appearing as noise to others. A feedback shift register with
58 2  Modulation and Detection

Mixer
Data In
Encoder Data
Modulator
m(t) V SS (t)

p(t) cosωc t
a
Mixer
Data De- Decoder
Modulator
VIF(t)+Noise Data Out

p(t)cosωc t
b

Feedback Logic

1 2 3 … k
PN Sequence Output
Clock
c
Fig. 2.48   Direct sequence spread spectrum: a Modulator. b Demodulator. c PN sequence genera-
tor diagram

k stages is employed for generating the PN sequence, (2.72), Fig. 2.48c. For a shift
register of k stages, the PN sequence length is N = 2k − 1 [6]. The measure of
interference rejection of a DS/SS system is given by the so-called processing gain,
defined by,
Tb Rc 
PGDS = = (2.73)
Tc Rb

2.7.11 Frequency Hopping Spread Spectrum

Modulation/Demodulation

The frequency hopping (FH) spread spectrum scheme, Fig. 2.49, involves the peri-
odic change of the transmission frequency [6]. In this scheme a sequence of modu-
lated data bursts with time-varying, pseudo-random carrier frequencies, is transmit-
ted. In FH the processing gain is given by,
2.7  Complex Envelope Form of Modulation Signals 59

Mixer
Data VSS(t)
Modulator
Data In

Frequency Hopped Carrier

Synthesizer

a PN
Mixer
Data De
De-
VSS(t)+Noise
Modulator
Data Out

Hopped Carrier Frequency

Synthesizer

b PN
Frequency

Tc
c Time

Fig. 2.49   Frequency hopping spread spectrum: a Modulator. b Demodulator. c Frequency hop-
ping versus time. Tc is the duration of time in a given frequency band

W
PGFHS =  (2.74)
B

where, W is the frequency range through hopping can occur, and B is the bandwidth
of the band being hopped [6].
60 2  Modulation and Detection

2.8 Summary

In this chapter, we have addressed a number of topics surrounding modulation and

demodulation, with emphasis in familiarizing the reader with their system-level
block diagrams. We began by introducing system-level block diagrams of AM and
FM/PM modulators and demodulators, and explaining their respective principles of
operation. In particular, under the topic of AM Modulator/Demodulator, we intro-
duced the full carrier modulator, the single sideband suppressed carrier modulator,
the double sideband suppressed carrier modulator, the envelope detector, and the
synchronous detector. Under the topic of FM and PM Modulator/Demodulator, we
introduced theVCO as FM modulator, the indirect FM modulator, the PM modula-
tor, the balanced discriminator FM demodulator, the quadrature FM detector, the
PLL-based FM detector, the zero-crossing FM detector, and the PM demodulator.
Then, under the topic of digital modulation, we introduced the concepts of Nyquist
Limit, data rate, Shannon Limit, information capacity, and bandwidth efficiency, as
well as specific modulation schemes, such as, binary modulation, amplitude-shift
keying (BASK), frequency-shift keying (BFSK), and phase-shift keying (BPSK),
differential binary phase-shift keying (DBPSK), quadrature phase-shift keying
(QPSK), π/4 shifted QPSK, minimum shift keying (MSK), M-ary quadrature am-
plitude modulation (QAM), orthogonal frequency division multiplexing (OFDM),
direct sequence spread spectrum (DS/SS), and frequency hopping spread spectrum
(FH/SS). We also introduced the geometric representation of digital modulation
schemes and the complex envelope form of a modulation signal.

References

1. J. Dąbrowski, Course “Introduction to RF Electronics,” Lecture Notes; Division of Electronic

Devices, Department of Electrical Engineering (ISY), Linköping University, 2006.
2. Mischa Schwartz, Information Transmission, Modulation, and Noise, McGraw-Hill, 1970.
3. T. S. Rappaport, Wireless Communications: Principles and Practice, Second Ed., Prentice-
Hall, Inc. 2002.
4. H. J. De Los Santos, Lecture Notes, “Communications Circuits,” Course EE115D, Department
of Electrical Engineering, University of California, Los Angeles (UCLA), Fall Quarter, 1990.
5. Q. Gu, RF System Design of Transceivers for Wireless Communications, Springer, 2005.
6. R. C. Dixon, Spread Spectrum Systems with Commercial Applications, Third Edition,
John Wiley & Sons, Inc., New York, 1994.
7. Available: [Online]: http://www.clee.freehomepage.com/teaching.html/
http://www.springer.com/978-3-319-07325-5