DigitalModulation

CI.IAPTER
OUTLINE
l-l Introduction 2-7 BandwidthEfficiencv
'-l Information Capacity,Bits, Bit Rate, Baud, and 2-8 Carrier Recovery
.V[-aryEncoding 2 9 Clock Recovery
l-_1 Amplitude-Shift Keying 2-10 Differential Phase-ShiftKeying
lJ Frequency-ShiitKeying 2- 11 Trellis Code Modulation
2--i Phase-Shift Keying 2-12 Probability of Error and Bit Error Rate
l-6 Quadrature-Amplitude Modulation 2-13 Enor Performance

OBJECTIVES
I Define electonic communications I Describe8- and 16-PSK
I Define digital modulation and digital radio I Describequadrature-amplitudemodulation
I Define digital communications I Explain 8-QAM
I Define information capacity I Explain I6-QAM
I Define bit, bit rate, baud, and minimum bandwidth I Define bandwidth fficiency
I Explain Shannon'slimit for information capacity I Explaincarrierrecovery
I Explain M-ary encoding I Explain clock recovery
I Define and describedigital amplitude modulation I Define and describedifferential phase-shiftkeying
I Define and describefrequency-shiftkeying I Define and explain trellis-code modulation
I Describe continuous-phasefrequency-shiftkeying I Define probability of error and bit error rqte
I Define phase-shift keying I Develop error performance equationsfor FSK,
I Explain binary phase-shiftkeying PSK, and QAM
I Explain quaternaryphase-shiftkeying

47
2-1 INTRODUCTION
In essence,electroniccommunicationsis the transmission,reception,and processingof in-
formation with the use of electronic circuits. Information is defined as knowledge or intel-
ligence that is communicated (i.e., transmitted or received) between two or more points.
Digital modulation is the transmittal of digitally modulated analog signals (carriers) be-
tween two or more points in a communications system. Digital modulation is sometimes
called digital radiobecausedigitally modulated signals can be propagatedthough Earth's
atmosphereand used in wireless communications systems.Traditional electronic corrunu-
nications systemsthat use conventional analog modulation, such as amplitude modulation
(AM),frequency modulation (FM), andphase modulation (PM), are rapidly being replaced
with more modern digital moduluation systemsthat offer several outstanding advantages
over traditionalanalogsystems,suchaseaseof processing,easeof multiplexing,and noise
immunity.
Digital communicationsis arather ambiguous terrn that could have entirely different
meanings to different people. In the context of this book, digital communications include
systemswhere relatively high-frequency analog carriers are modulated by relatively low-
frequency digital information signals (digital radio) and systemsinvolving the transmis-
sion of digital pulses(digital transmission).Digital transmissionsystemstransponinfor-
mation in digital form and, therefore,require a physical facility betweenthe transmitter and
receiver, such as a metallic wire pair, a coaxial cable, or an optical fiber cable. Digital
transmissionis coveredin Chapters10 and 11.In digital radio systems,the carrierfacility
could be a physicalcable,or it could be free space.
The property that distinguishes digital radio systems from conventional analog-
modulation communications systemsis the nature of the modulating signal. Both analog
and digital modulation systemsuse analog carriers to transportthe information through the
system. However, with analog modulation systems,the information signal is also analog,
whereaswith digital modulation, the information signal is digital, which could be computer-
generateddata or digitally encodedanalog signals.
Referring to Equation 2-1, if the information signal is digital and the amplitude (l/)
of the carrier is varied proportional to the information signal, a digitally modulated signal
called amplitude shifi keying (ASK) is produced. If the frequency A is varied proportional
to the information signal,frequencyshift keying (FSK) is produced, and if the phaseof the
carrier (0) is varied proportional to the information signal, phase shift keying (PSK) is pro-
duced. If both the amplitude and the phaseare varied proportional to the information sig-
nal, quadrature amplitude modulation (QAM) results.ASK, FSK, PSK, and QAM are all
forms of dieital modulation:
v(t)= Vsin (2n' ft + 0)
tt l
ASK FSK PSK (2-1\
\, /
t qR Mr

Digital modulation is ideally suited to a multitude of communications applications,

including both cable and wireless systems.Applications include the following: (1) rela-
tively low-speed voice-band data communications modems, such as those found in most
personal computers; (2) high-speed data transmission systems,such as broadbanddigital
subscriber/ines (DSL); (3) digital microwave and satellitecommunicationssystems;and
(4) cellular telephonePersonal CommunicationsSystems(PCS).
Figure 2-1 shows a simplified block diagram for a digital modulation system.In the
transmitter, the precoder performs level conversion and then encodesthe incoming data
into groups of bits that modulate an analog carrier. The modulated carrier is shaped(fil-

48 Chapter 2
 Transmitter Receiver
Transmission

Output
data

FIGURE2-1 Simplifiedblock diagnam of a digital nadiosystem

tered), amplified, and then transmitted through the transmission medium to the receiver.
The transmission medium can be a metallic cable, optical fiber cable, Earth's atmosphere,
or a combination of two or more types of transmission systems. In the receiver, the in-
coming signals are filtered, amplified, and then applied to the demodulator and decoder
circuits, which extracts the original source information from the modulated carrier. The
clock and carrier recovery circuits recover the analog carrier and digital timing (clock)
signals from the incoming modulated wave since they are necessaryto perform the de-
modulationprocess.

2-2 TNFORMATION
CApACtry BITS,BtT RATE,BAUD,
AND MARY ENCODING

2-2-1 InformationGapacity,Bits, and Bit Rate

Information theory is a highly theoretical study of the efficient use of bandwidth to propa-
gate information tfrough electronic communications systems.Information theory can be
used to determine the information capacity of a data communications system.Information
capacity is a measureof how much information can be propagatedthrough a communica-
tions system and is a function of bandwidth and ffansmission time.
Information capacity representsthe number of independentsymbols that can be car-
ried through a system in a given unit of time. The most basic digital symbol used to repre-
sent information is the binary digit, or bit.Therefore, it is often convenient to expressthe
information capacity of a system as a bit rate. Bit rate is simply the number of bits trans-
mitted during one secondand is expressedin bits per second (bps).
In 1928, R. Hartley of Bell TelephoneLaboratories developed a useful relationship
among bandwidth, transmission time, and information capacity.Simply stated, Hartley's
law is

I *B Xt (2-2)
where I: information capacity (bits per second)
B : bandwidth (hertz)
L_ transmissiontime (seconds)

Digital Modulation 49
 From EquationZ-2, it can be seen that information capacity is a linear function of
bandwidth and transmission time and is directly proportional to both. If either the band-
width or the transmission time changes,a directly proportional changeoccurs in the infor-
mation capacity.
In 1948, mathematician Claude E. Shannon (also of Bell Telephone Laboratories)
published a paper in the Bell SystemTechnicalJournal relating the information capacity of
a communications channel to bandwidth and signal+o-noise ratio. The higher the signal-
to-noise ratio, the better the performance and the higher the information capacity. Mathe-
matically stated,the Shannonlimitfor information capaciry is

r: ar's,(r+ $) (2-3)

I : 3.32Bfog,o(f (2-4)

where I - information capacity (bps)

B _ bandwidth (hertz)
q_ signal-to-noisepower ratio (unitless)
N
For a standardtelephone circuit with a signal-to-noisepower ratio of 1000 (30 dB)
and a bandwidth of 2.7 kHz, the Shannonlimit for information capacity is

I : (3.32)(2700)logls (1 + 1000)
: 26.9 kbps

Shannon'sformula is often misunderstood.The results of the preceding example in-

dicate that 26.9 kbps can be propagatedthrough a2.7-Wlz communications channel. This
may be true, but it cannot be done with a binary system. To achieve an information trans-
mission rate of 26.9 kbps through a2.7-WIz channel,each symbol transmittedmust contain
more than one bit.

2-2-P Mary Encoding

M-ary is a term derived from the word binary. M simply representsa digit that corresponds
to the number of conditions, levels, or combinations possible for a given number of binary
variables. It is often advantageousto encode at a level higher than binary (sometimesre-
ferred to as beyond binary or higher-than-binary encoding) where there are more than two
conditions possible. For example, a digital signal with four possible conditions (voltage
levels, frequencies,phases,and so on) is an M-ary system where M : 4. If there are eight
possible conditions, M : 8 and so forth. The number ofbits necessaryto produce a given
number of conditions is expressedmathematically as

N: l o g 2M (2-s)
where N : number of bits necessary
M : number of conditions, levels, or combinations possible with N bits
Equation 2-5 can be simplified and rearrangedto expressthe number of conditions possi-
ble with Nbits as

2N : M (2-6)
For example, with one bit, only 21 : 2 conditions are possible.With two bits,Z2 : 4 con-
ditions are possible, with three bits,23 : 8 conditions are possible, and so on.

50 Chapter 2
 2-2-3 Baud and MinimumBandwidth
Baad is a term that is often misunderstoodand commonly confusedwith bit rate (bps). Bit
rate refers to the rate of change of a digital information signal, which is usually binary.
Baud, like bit rate, is also a rate of change;however, baud refers to the rate of changeof a
signal on the transmissionmedium after encoding and modulation have occurred. Hence,
baud is a unit of ffansmissionrate, modulation rate, or symbol rate and, therefore,the terms
symbols per second and baud are often used interchangeably. Mathematically, baud is the
reciprocal of the time of one output signaling element,and a signaling element may repre-
sent severalinformation bits. Baud is expressedas

UaoO: I (2-7)
ts

where baud : symbol rate (baud per second)

/" : time of one signaling element (seconds)
A signaling element is sometimescalled a symbol and could be encodedas a changein the
amplitude, frequency, or phase. For example, binary signals are generally encoded and
transmitted one bit at a time in the form of discrete voltage levels representing logic ls
(highs) and logic 0s (lows). A baud is also transmitted one at a time; howeveq a baud may
representmore than one information bit. Thus, the baud of a data communications system
may be considerably less than the bit rate. In binary systems(such as binary FSK and bi-
nary PSK), baud and bits per secondare equal. However, in higher-level systems(such as
QPSK and 8-PSK), bps is always greaterthan baud.
According to H. Nyquist, binary digital signals can be propagatedthrough an ideal
noiselesstransmissionmedium at arate equal to two times the bandwidth of the medium.
The minimum theoreticalbandwidth necessaryto propagatea signal is called the minimum
Nyquist bandwidth or sometimesthe minimum Nyquistf'requency.Thus,f6: 28, wheref6
is the bit rate in bps and B is the ideal Nyquist bandwidth. The actual bandwidth necessary
to propagate a given bit rate depends on several factors, including the type of encoding and
modulation used,the types of filters used,systemnoise, and desirederror performance.The
ideal bandwidth is generally used for comparison purposesonly.
The relationship betweenbandwidth and bit rate also appliesto the opposite situation. For
a given bandwidth (B), the highest theoretical bit r ateis 28. For example, a standardtelephone
circuit has a bandwidth of approximately 2700 Hz, which has the capacity to propagate
5400 bps through it. However, if more than two levels areusedfor signaling (higher-than-binary
encoding), more than one bit may be transmitted at a time, and it is possible to propagatea bit
rate that exceeds28. Using multilevel signaling, the Nyquist formulation for channelcapacity is

f o: 2 8l og 2M (2-8)
where /6 : channel capacity (bps)
B : minimum Nyquist bandwidth (hertz)
M : number of discrete signal or voltage levels
Equation 2-8 can be rearrangedto solve for the minimum bandwidth necessaryto
passM-ary digitally modulated carriers

\
(2-e)
': (J\logrM/
If N is substituted for log2 M, Equation 2-9 reduces to

IL
B :" : A/
(2-r0)

where N is the number of bits encodedinto each signaling element.

Digital Modulation 51
 If information bits are encoded (grouped) and then converted to signals with more
than two levels, transmission rates in excessof 28 are possible, as will be seen in subse-
quent sectionsofthis chapter.In addition, since baud is the encodedrate ofchange, it also
equalsthe bit rate divided by the number of bits encodedinto one signaling element.Thus,

:
Uaua (2-1r)
f;
By comparing Equation 2-10 with Equation2-Ll, it can be seenthat with digital mod-
ulation, the baud and the ideal minimum Nyquist bandwidth have the same value and are
equal to the bit rate divided by the number of bits encoded.This statementholds true for all
forms of digital modulation except frequency-shift keying.

2-3 AMPLITUDE-SHIFT
KEYING
The simplest digital modulation techniqueis amplitude-shift keying (ASK), where a binary
information signal directly modulates the amplitude of an analog carrier. ASK is similar
to standard amplitude modulation except there are only two output amplitudes possible.
Amplitude-shift keying is sometimes called,digital amplitude modulation (DAM). Mathe-
matically, amplitude-shift keying is

v1"s(r): [1 + ,.1,y1[f,"",1,;l] (2-r2)

where vo"r(t) : amplitude-shift keying wave

v^(t) : digital information (modulating) signal (volts)
Al2 : unmodulatedcarrier amplitude (volts)
ro. : analog carrier radian frequency (radians per second,2nf"t)
In Equation 2-12, the modulating signal (v-[l]) is a normalized binary waveform, where
+1 V: logic 1 and -1V: logic0. Therefore, foralogic 1 input,v^(t): +1V, Equation2-12
reducesto

: II
u1'"1y(t) +1l[+*,(,.')]
: A cos(o.r)

andfor a logic 0 input,v-(t) : - I V, Equation2-L2reducesto

v(,'r(r): 11 - rfi*rt,"rl]
:0
Thus, the modulatedwave vo,o(t),is eitherA cos(ol.r)or 0. Hence,the carrier is either"on" or
"off" which is why amplitude-shiftkeying is sometimesreferredto as on-offkey,r?g(OOK).
Figure 2-2 shows the input and output waveforms from an ASK modulator. From the
figure, it can be seen that for every change in the input binary data stream, there is one
changein theASK waveform, and the time of one bit (lu) equalsthe time of one analog sig-
naling element(/"). It is also important to note that for the entire time the binary input is high,
the output is a constant-amplitude,constant-frequencysignal, and for the entire time the bi-
nary input is low, the carrier is off. The bit time is the reciprocal of the bit rate and the time
of one signaling element is the reciprocal of the baud. Therefore, the rate of changeof the

52 Chapter 2
 Binary
input

DAM
output
FIGURE2-2 Digitalamplitudemodu-
lation:[aJinputbinary;[b] ourpur
DAM wavefonm

ASK waveform (baud) is the sameas the rate of change

of the binary input (bps); thus, the
bit rate equalsthe baud. with ASK, the bit rate is also equal
to the minimum Nyquist band-
width' This can be verified by substituting into Equations 2-
l0 and 2-11 andsettinsN to 1:
t

s :l! _I
baua:fr: f .
1 I

Example 2-1
Determinethebaudandminimumbandwidthnecessary
to passa 10kbpsbinarysignalusingampli
tudeshift keying.
Solution For ASK, N : 1, andthebaudandminimumbandwidth
aredeterminedfrom Equations
2-ll and2-10,respectively:
_ 10,000
I

baud:frp: 1o,ooo
The use of amplitude-modulatedanalog carriers to transport
- digital information is a relatively
low-quality, low-cost type of digital modulati,onand, therefore, is seldom
used except tbr very low_
speedtelemetry circuits.

2-4 FREOUENCY-SHIFT
KEYING
Frequency-shiftkeying (FSK) is anotherrelatively simple,
row-performancetype of digital
modulation' FSK is a form of constant-amplitudeangle modulaiion
similar to standardfre-
quencyrnodulation (FM) except the modulating signal is a
binary signal that variesbetween
two discrete voltage levels rather than a continuously changing
waveform. conse_
quently, FSK is sometimes called binary f's/( (BFSIi). ThJgeieral
-utog
e*Ipr"ssionfor FSK is
vy"r(t)= V, cosl2nff" + v*(t) Aflt] (2-13)
where vq,*(t): binary FSK waveform
% : peak analog carrier amplitude (volts)
f,: anaTogcarrier center frequency (hertz)
A/: peak change (shift) in the analog carrier frequency (hertz)
v^(t) : binary input (modulating) signal (volts)
From Equation2-r3, it can be seenthat the peak shift in
the carrier frequency (A/) is
proportional to the amplitude of the binary input signal (v^[t]),and
the direction of the shift

Digital Modulation
53
 Binaryinput FIGURE2-3 FSKin the frequency
signal domain

is determined by the polarity. The modulating signal is a normalized binary waveform

whe rea logic1: * l Va nd a l og i c0 : -l V .Th us,fo ral o g i c l i np ut ,v* (t): * l ,Eq u a -
tion 2-13 can be rewritten as

v1"r,Q): V" cosf2n(f. + Lfltl

For a logic 0 input, v*(t) = - 1, Equation2-13 becomes

vq,*(t) : V" cosf2n(f, - Lfltl

With binary FSK, the can'ier center frequency (f.) is shifted (deviated) up and down
in the frequency domain by the binary input signal as shown in Figure 2-3. As the binary
input signal changesfrom a logic 0 to a logic 1 and vice versa, the output frequency shifts
between two frequencies: a mark, or logic I frequency (f^), and a space,or logic 0 fre-
quency (f,). The mark and spacefrequenciesare separatedfrom the carrier frequency by the
peak frequency deviation (A/) and from each otherby 2 Lf.
With FSK, frequency deviation is defined as the difference between either the mark
or spacefrequency and the center frequency, or half the difference between the mark and
spacefrequencies.Frequency deviation is illustrated in Figure 2-3 and expressedmathe-
matically as

Lf =
lf^ - f,l (2-r4\

where A/: frequency deviation (hertz)

lf^ - f"l: absolutedifference between the mark and spacefrequencies(hertz)
Figure 2-4a shows in the time domain the binary input to an FSK modulator and the
correspondingFSK output. As the figure shows, when the binary input ff6) changesfrom a
logic I to a logic 0 and vice versa, the FSK output frequency shifts from a mark (f) to a space
ffi) frequency and vice versa. In Figure 2-4a,the mark frequency is the higher frequency
(f" + Lfl, and the spacefrequency is the lower frequency (f, - A/), although this relation-
ship could be just the opposite.Figure 2-4b shows the truth table for a binary FSK modula-
tor. The truth table shows the input and output possibilities for a given digital modulation
scheme.

2-4-'l FSK Bit Flate,Baud, and Bandwidth

In Figure 2-4a, it can be seenthat the time of one bit (t ) is the same as the time the FSK
output is a mark of spacefrequency (r"). Thus, the bit time equals the time of an FSK sig-
naling element, and the bit rate equalsthe baud.

54 Chapter 2
 t^
Binary
input

binary frequency
input output

0 space(f,)
I mark ff-)
/" spacefrequency
/^, marktrequency;

(a) (b)

FIGURE2-4 FSK in the time domain: [aJ wavefonm;[b] tnuth table

The baudfor binary FSK can alsobe determinedby substitutingN : I in Equa-

tion2-11:
f_
baud: f,
l:
FSK is the exception to the rule for digital modulation, as the minimum bandwidth is
not determinedfrom Equation 2-10. The minimum bandwidth for FSK is given as
B: l(f ,-f)-(f* -f)l
: lf" - f^l + 2f6
and since V" - f*l equals2Al the minimum bandwidth can be approximated as
B: 2(L f+ fo ) (2-rs)
where B : minimum Nyquist bandwidth (hertz)
A/: frequency deviation (V^ - f,l) (hertz)
f6: inplutbit rate (bps)
Note how closely Equation 2-15 resembles Carson's rule for determining the ap-
proximate bandwidth for an FM wave. The only difference in the two equationsis that, for
FSK, the bit rate ffi) is substitutedfor the modulating-signal frequency ffi).

Example2-2
Determine (a) the peak frequency deviation, (b) minimum bandwidth, and (c) baud for a binary FSK
signal with a mark frequency of 49 Wlz, a spacefrbquency of 51 kHz, and an input bit rate of 2 kbps.
Solution a. The peak frequency deviation is determined from Equation 2-14:

l49kl1z - 5rkHzl
Af :
: I kIIz
b. The minimumbandwidthis determinedfrom Equation2-15:
B: 2(1000+ 2000)
:6 WI z
:
c. For FSK,N 1, andthe baud is determined from Equation2-ll as

2000
oauo : = 2000
I

Digital Modulation 55
 Bessel functions can also be used to determine the approximate bandwidth for an
FSK wave. As shown in Figure 2-5, the fastest rate of change (highest fundamental fre-
quency) in a nonreturn-to-zero (NRZ) binary signal occurs when alternating ls and 0s are
occurring (i.e., a squarewave). Since it takes a high and a low to produce a cycle, the high-
est fundamental frequency presentin a squarewave equals the repetition rate ofthe square
wave, which with a binary signal is equal to half the bit rate. Therefore,

"t- -f
- u- (2-16)
2
where = highest fundamental frequency of the binary input signal (hertz)
"f"
/6 = inputbitrate (bps)
The formula used for modulation index in FM is also valid for FSK: thus,

Lf
tt - - : (unitless) (2-r7)
fo

where h = FM modulation index called the h-factor in FSK

fo = fundamental frequency of the binary modulating signal (hertz)
A/ : peak frequency deviation (hertz)
The worst-case modulation index (deviation ratio) is that which yields the widest band-
width. The worst-caseor widest bandwidth occurs when both the frequency deviation and
the modulating-signal frequency are at their maximum values. As described earlier, the
peak frequency deviation in FSK is constant and always at its maximum value, and the
highest fundamental frequency is equal to half the incoming bit rate. Thus,

lf^ - f"l
2
I^-
(unitless)
fa
2

(2-18)

f= tvt2
I

Output fm iG fm

FfGURE2-5 FSK modulaton,t6, time of one bit = 1,/fo; f-, mankfrequency; f",
space fnequencyiTr, period of shontest cycle;1/f',, fundamentalfnequencyof
binanysquanewave; f6, inpuLbit rate [bps]

56 Chapter 2
 where h: h-factor (unitless)
f-: mark frequency (hertz)
f, : spacefrequency (hertz)
fo: bit rate (bits per second)

Example2-3
Using a Besseltable, determinethe minimum bandwidth for the sameFSK signal describedin Exam-
ple 2-1 with a mark frequency of 49 kHz, a spacefrequency of 5 1 kHz, and an input bit rate of 2 kbps.
Solution The modulation index is found by substituting into Equation 2-17:

t^-
l49Wlz - 51 kHzl
2 kbps
_ 2kllz
2 kbps
:1
From a Besseltable,threesetsof significantsidebandsareproducedfor a modulationindex of
one.Therefore,thebandwidthcanbe determinedasfollows:
B:2 (3 x 10 0 0)
: 6000Hz
Thebandwidthdetermined in Example2-3 usingtheBesseltableis identicalto thebandwidth
determinedin Example2-2.

2-4-2 FSK Transmitter

Figure 2-6 shows a simplified binary FSK modulator, which is very similar to a conven-
tional FM modulator and is very often a voltage-controlled oscillator (VCO). The center
frequency (f,) is chosensuch that it falls halfway between the mark and spacefrequencies.
A logic 1 input shifts the VCO output to the mark frequency, and a logic 0 input shifts the
VCO output to the spacefrequency.Consequently,as the binary input signal changesback
and forth between logic I and logic 0 conditions, the VCO output shifts or deviatesback
and forth between the mark and spacefrequencies.
In a binary FSK modulator, A/is the peak frequency deviation of the carrier and is
equal to the difference between the carrier rest frequency and either the mark or the space
frequency (or half the difference between the carrier rest frequency) and either the mark or
the spacefrequency(or half the differencebetweenthe markand spacefrequencies).AVCO-
FSK modulator can be operatedin the sweep mode where the peak frequency deviation is

NRZ FSKoutput
binaryJ1-[1-[T- FSK Modulator
input -------------> (vco) ----->
kt = Hzlv

FIGURE2-6 FSKmodulaton

Digital Modulation 57
 Analogmarkor
spacefrequency Rectifiedsignal
I

FSKinput Dataouipul

FIGURE2-7 NoncohenentFSK demodulator

Multiplier

FSKinput

FIGURE2-8 CohenentFSK demodulaLon

simply the product of the binary input voltage andthe deviation sensitivity of the VCO. With
the sweep mode of modulation, the frequency deviation is expressedmathematically as

Af : v*(t)k1 (2-re)
where Arf : peak frequency deviation (hertz)
v^(t) : peak binary modulating-signal voltage (volts)
ft, : deviation sensitivity (hertz per volt).

With binary FSK, the amplitude of the input signal can only be one of two values, one
for a logic 1 condition and one for a logic 0 condition. Therefore, the peak frequency devi-
ation is constant and always at its maximum value. Frequency deviation is simply plus or
minus the peak voltage of the binary signal times the deviation sensitivity of the VCO.
Since the peak voltage is the same for a logic 1 as it is for a logic 0, the magnitude of the
frequency deviation is also the same for a logic 1 as it is for a logic 0.

2-4-3 FSK Fleceiver

FSK demodulation is quite simple with a circuit such as the one shown in Figure 2-7.The
FSK input signal is simultaneously applied to the inputs of both bandpassfilters (BPFs)
through a power splitter. The respectivefilter passesonly the mark or only the spacefre-
quency on to its respectiveenvelope detector.The envelope detectors,in turn, indicate the
total power in each passband,and the comparator respondsto the largest of the two pow-
ers.This type ofFSK detectionis referred to asnoncoherentdetection;there is no frequency
involved in the demodulation process that is synchronized either in phase, frequency, or
both with the incoming FSK signal.
Figure 2-8 shows the block diagram for a coherentFSK receiver.The incoming FSK
signal is multiplied by a recoveredcarrier signal that has the exact samefrequencyand phase
as the transmitter reference. However, the two transmitted frequencies (the mark and space
frequencies) are not generally continuous; it is not practical to reproduce a local referencethat
is coherent with both of them. Consequently, coherent FSK detection is seldom used.
5S Chapter 2
 Blnary
data out

J TL

FIGURE2-g PLL-FSKdemodulaton

Phmodiacontinultc

FIGURE2-1O Noncontinuous
FSK
wavefonm

The most common circuit used for demodulating binary FSK signals is the phase-
locked loop (PLL), which is shown in block diagram form in Figure 2-9. APLL-FSK de-
modulator works similarly to a PLL-FM demodulator.As the input to the PLL shifts be-
tween the mark and space frequencies, the dc error voltage at the output of the phase
comparatorfollows the frequency shift. Becausethere are only two input frequencies(mark
and space),there are also only two output error voltages. One representsa logic 1 and the
other a logic 0. Therefore, the output is a two-level (binary) representationof the FSK in-
put. Generally, the nafural frequency of the PLL is made equal to the center frequency of
the FSK modulator. As a result, the changesin the dc error voltage follow the changesin
the analog input frequency and are symmetrical around 0 V.
Binary FSK has a poorer error performance than PSK or QAM and, consequently,is sel-
dom usedfor high-performancedigital radio systems.Its useis restrictedto low-perforrnance,
low-cost, asynchronousdata modems that are used for data communications over analog,
voice-band telephonelines.

24-4 Frequency-ShifE
Continuous-Phase Keying
Continuous-phasefrequency-shift keying (CP-FSK) is binary FSK except the mark and
spacefrequenciesare synchronizedwith the input binary bit rate. Synchronoussimply im-
plies that there is a precisetime relationshipbetweenthe two; it doesnot meanthey areequal.
With CP-FSK, the mark and spacefrequenciesare selectedsuchthat they are separatedfrom
the centerfrequency by an exact multiple of one-half the bit rate (f^ndf" : nlfu/21), wherc
n: any integer). This ensuresa smoothphasetransition in the analog output signal when it
changesfrom a mark to a spacefrequency or vice versa.Figure 2- 10 shows a noncontinuous
FSK waveform. It can be seenthat when the input changesfrom a logic 1 to a logic 0 and
vice versa,there is an abrupt phasediscontinuity in the analog signal.When this occurs,the
demodulator has trouble following the frequency shift; consequently,an error may occur.
Figure 2-11 shows a continuous phaseFSK waveform. Notice that when the output
frequency changes,it is a smooth, continuous transition. Consequently,there are no phase
discontinuities. CP-FSK has a better bit-error performance than conventional binary FSK
for a given signal-to-noiseratio. The disadvantageof CP-FSK is that it requires synchro-
nization circuits and is, therefore, more expensiveto implement.
Digital Modulation 59
 f5

Mark ftoquenoy,f. Spaccftoquencry,fg

FIGUHEe-11 Continuous-ohase
MSKwaveform

2-5 PHASE-SHIFT
KEYING
Phase-shift keying (PSK) is another form of angle-modulated,constant-amplitudedigital
modulation. PSK is an M-ary digital modulation scheme similar to conventional phase
modulation except with PSK the input is a binary digital signal and there are a limited num-
ber ofoutput phasespossible. The input binary information is encodedinto groups ofbits
before modulating the carrier. The number of bits in a group rangesfrom 1 to 12 or more.
The number of output phasesis defined by M as describedin Equation 2-6 and determined
by the number of bits in the group (n).

2-5-1 Binary Phase-ShiftKeying

keying(BPSK),whereN : 1 andM : 2.
formof PSKis binaryphase-shifi
Thesimplest
Therefore, with BPSK, two phases(2r : 2) are possible for the carrier. One phaserepre-
sentsa logic 1, and the other phaserepresentsa logic 0. As the input digital signal changes
' state (i.e., from a I to a0 or from a 0 to a 1), the phaseof the output carrier shifts between
two anglesthat are separatedby 180o.Hence, other namesfor BPSK arephase reversalkey-
ing (PRK) and biphase modulation. BPSK is a form of square-wave modulation of a
continuous wave (CW) signal.

2-5-l-1. BPSK transmitter. Figure 2-12 shows a simplified block diagram of a

BPSK transmitter.The balancedmodulator acts as a phasereversing switch. Depending on

Ol L -J
f-l L- lJr-n
Modulated
PSK output

FIGUBE2-12 BPSKtnansmitter

60 Chapter 2
 Binarydata In

Dl (onl

sin obt

+V(Binary1l
tb)

-V (Binary0)
G}

FIGURE2-13 [a) Balancedring modulator;[b] logic1 input;(cJlogicO inpw

the logic condition ofthe digital input, the carrier is transferredto the output either in phase
or 180oout ofphase with the referencecarrier oscillator.
Figtre 2-13 showsthe schematicdiagram of a balancedring modulator.The balanced
modulator has two inputs: a carrier that is in phasewith the referenceoscillator and the bi-
nary digital data. For the balancedmodulator to operateproperly, the digital input voltage
must be much greater than the peak carrier voltage. This ensuresthat the digital input con-
ftols the on/off stateof diodes Dl to D4. If the binary input is a logic 1 (positive voltage),
diodes Dl and D2 arc forward biased and on, while diodes D3 and D4 are reversebiased
and off (Figure 2-l3b). With the polarities shown, the carrier voltage is developed across

Digital Modulation 61
 (+90'l
cos (t t

-sin oht 3in obt

(18r1 (o'l
Binaryinput n'
'thIi Logic 0 Logic 1

Logic 0 18ff
Logic I

(a)
-cos(tt
Fqr)
(b)

cos olct

t1 g) 'G -- -- -- - *- F - - - --- ---a 0oReference

Logic0 Logic1

-cos obt
(cl

FIGURE2-14 BPSKmodulaton: [a] tr"uthtable;[bJphason

diagnam;[cJconstellation
diagnam

transformerT2 in phasewith the carrier voltage acrossT1. Consequently,the output signal

is in phasewith the referenceoscillator.
If the binary input is a logic 0 (negativevoltage), diodes Dl and D2 arereversebiased
and off, while diodes D3 and D4 arc forward biased and on (Figure 2-I3c).As a result, the
carrier voltage is developedacrosstransformerT2 180" out of phasewith the carrier voltage
acrossTl. Consequently,the output signal is 180oout of phasewith the referenceoscillator.
Figure 2-14 shows the truth table, phasor diagram, and constellation diagram for a BPSK
modulator. A constellation diagram, which is sometimescalled a signal state-spacedia-
gram, is similar to a phasordiagram except that the entire phasor is not drawn. In a constel-
lation diagram, only the relative positions of the peaksof the phasorsare shown.

2-5-l-2 Bandwidth considerations of BPSK. A balancedmodulator is a product

modulator; the output signal is the product of the two input signals.In a BPSK modulator,
the carrier input signal is multiplied by the binary data.If * 1 V is assignedto a logic 1 and
- 1 V is assignedto a logic 0, the input carrier (sin ro.r) is multiplied by either a * or - 1.
Consequently,the output signal is either * 1 sin co"/or - I sin @ct;thefirst representsa sig-
nal that is in phase with the referenceoscilla0or,the latter a signal that is I 80' out of phase
with the referenceoscillator. Each time the input logic condition changes,the output phase
changes.Consequently,for BPSK, the output rate of change(baud) is equal to the input rate
of change(bps), and the widest output bandwidth occurs when the input binary data are an
alternating 1/0 sequence.The fundamentalfrequency (i) of an alternative 1/0 bit sequence
is equal to one-half of the bit rate (f612).Mathematically, the output of a BPSK modulator
is proportional to

BPSK output : fsin(2nf"t)l x [sin(hEfct)] (2-20)

62 Chapter 2
 Binary
Input

EPSK
orrtpW

sin rqt -sin (tt sin oct -sin oct sinqt -sin qt

o 180 0 180 0 1&) Dcgccr

0 fr 0 ft. 0 r Rrdlem

FIGUHE2-15 Outputphase-vensus-time
relationship
fona BPSKmodulator

where ,t : maximum fundamental frequency of binary input (hertz)

f": rcferencecarrier frequency (hertz)
Solving for the trig identity for the product of two sine functions,
1l
- - +
;cosl2n(f, f")rl lcosl2n(, f")rl
Thus, the minimum double-sidedNyquist bandwidth (B) is

f , +f " f"+f"
-(f, + f") -f" + f"
or
,h
andbecausef : fblz, wheref6: input bit rate,
2f,
B :;

where B is the minimum double-sidedNyquist bandwidth.

Figure 2-15 shows the output phase-versus-timerelationship for a BPSK waveform.
As the figure shows, a logic 1 input produces an analog output signal with a 0" phase an-
gle, and a logic 0 input produces an analog output signal with a 180" phase angle. As the
binary input shifts betweena logic 1 and a logic 0 condition and vice versa,the phaseofthe
BPSK waveform shifts between 0' and 180o,respectively.For simplicity, only one cycle of
the analog carrier is shown in each signaling element, although there may be anywhere
between a fraction of a cycle to severalthousandcycles, dependingon the relationship be-
tween the input bit rate and the analog carrier frequency.It can also be seenthat the time of
one BPSK signaling element (r,) is equal to the time of one information bit (16),which in-
dicatesthat the bit rate equals the baud.

Example24
For a BPSK modulator with a carrier frequency of 70 MHz and an input bit rate of 10 Mbps, deter-
mine the maximum and minimum upper and lower side frequencies, draw the output spectrum, de-
termine the minimum Nyquist bandwidth, and calculate the baud.

Digital Modulation 63
 t sin(ocD
BPSK ------->
Input

FIGURE2-1 6 Block diagnamof a BPSK neceiven

Solution SubstitutingintoEquation2-20yields
=
output (sin ro"t)(sinco.t)
: [sin2ru(5MHz)r][sin2n(70MHz)tl

: 2n(70MHz- 5 MHz)r- j"o, 2n(70MHz+ 5MHz)t

J"o.
lower side frequency upper side frequency

Minimum lower side frequency (LSF):

LSF = 70 MHz - 5 MHz : 65 MHz
Maximumuppersidefrequency(USF):
U SF:7 0 MHz+ 5MH z:75 MH z
Therefore,the outputspectrumfor theworst-case
binaryinputconditionsis asfollows:
The minimumNyquistbandwidth(B) is

'<-- B= 5 MH z ----------->
I
I
I
I

67.5MHz 70 MHz 72.5MHz

(Suppressed)

B :7 5M Hz - 6 5MH z : l 0 M Hz
the
and baud:"fa 10
or megabaud.

2-5-l-3 BPSK receiyer. Figure 2- 16 shows the block diagram of a BPSK receiver.
The input signal may be * sin <rr.ror - sin o./. The coherent carrier recovery circuit detects
and regeneratesa carier signal that is both frequency and phasecoherentwith the original
transmit carrier. The balancedmodulator is a product detector; the output is the product of
the two inputs (the BPSK signal and the recoveredcarrier). The low-pass filter (LPF) sep-
aratesthe recoveredbinary data from the complex demodulatedsignal. Mathematically, the
demodulation processis as follows.
For a BPSK input signal of *sin to"/ (logic 1), the output of the balancedmodulator is

output : (sin ol.r)(sin a,t) : sin2 to.t (2-2r)

64 Ghapter2
 (filtered out)

t/
or sintrrt : - cos2a-,t): ! - ^z orct
f,(t tcos

leaving : *ir:
output losic1
It can be seenthat the output of the balancedmodulator contains a positive voltage
(+lllzlV) and a cosinewave at twice the carrier frequency (2 to.) The LPF has a cutofffre-
quency much lower than 2ro. and, thus, blocks the second harmonic of the carrier and
passesonly the positive constant component.Apositive voltage representsa demodulated
logic L
For a BPSK input signal of -sin co,/(logic 0), the output of the balancedmodulator is
output : (-sin ro./)(sin @"t) : -sin2 or"l
(filtered out)

or -sin2ro"/: -irr- cos2o.r): -:.

)A"zr",
;
leaving output : -tV : logic 0

The output of the balanced modulator contains a negative voltage (-[1/2]V) and a
cosine wave at twice the carrier frequency (2o1").Again, the LPF blocks the secondhar-
monic of the carrier and passesonly the negative constant component.A negative voltage
representsa demodulatedlogic 0.

2-5-2 QuaternaryPhase-ShiftKeying
Quaternary phase shift keylng (QPSK), or quadrature PSK as it is sometimescalled, is an-
other form of angle-modulated,constant-amplitudedigital modulation. QPSK is an M-ary
encoding schemewhere N : 2 and M : 4 (hence, the name "quaternary" meaning "4").
With QPSK, four output phasesare possible for a single carrier frequency. Becausethere
are four output phases,there must be four different input conditions. Becausethe digital in-
put to a QPSK modulator is a binary (base 2) signal, to produce four different input com-
binations, the modulator requires more than a single input bit to determine the output con-
dition. With two bits, there are four possible conditions: 00, 01, 10, and 11.Therefore, with
QPSK, the binary input data are combined into groups of two bits, calleddibits.In the mod-
ulator, each dibit code generatesone of the four possible output phases (+45", +135',
-45o, and - 135'). Therefore, for each two-bit dibit clocked into the modulator, a single
output change occurs, and the rate of change at the output (baud) is equal to one-half the
input bit rate (i,e., two input bits produce one output phasechange).

2-5-2-l QPSK transmitter. A block diagram of a QPSK modulator is shown in

Figure 2-17.Two bits (a dibit) are clocked into the bit splitter. After both bits have been se-
rially inputted, they are simultaneouslyparallel outputted. One bit is directed to the I chan-
nel and the other to the Q channel.The I bit modulatesa carrier that is in phasewith the ref-
erence oscillator (hence the name "I" for "in phase" channel), and the Q bit modulates a
carrier that is 90o out of phaseor in quadraturewith the referencecarrier (hence the name
"Q" for "quadrature" channel).
It can be seenthat once a dibit has been split into the I and Q channels,the operation
is the sameas in a BPSK modulator. Essentially,a QPSK modulator is two BPSK modula-
tors combinedinparallel.Again, for alogic I : *1V and alogic 0 : -1V twophases
are possible at the output of the I balanced modulator (*sin or./ and -sin o.t), and two

Digital Modulation 65
 Log i c l -+ 1 V
Logic0--1 V

Logi cl = +1 V
Logi c0* -1 V

FIGURE2-17 QPSKmodulator'

phasesare possible at the output of the Q balanced modulator (+cos o,t and -cos o./).
When the linear summer combines the two quadrature (90" out of phase) signals, there
are four possible resultant phasorsgiven by theseexpressions:* sin olct * cos ot"/, * sin
ocl - cos @./, -sin oct + cos ro.t, and -sin ro.t - cos oct

Example2-5
FortheQPSKmodulator thetruthtable,phasor
shownin Figure2-17,construct diagram,
andcon-
stellationdiagram.
Solution ForabinarydatainputofQ : 0 andl : 0, thetwo inputsto thelbalancedmodulatorare
- 1 and sin o./, andthe two inputs to the Q balancedmodulatorare - I and cos ctlct Consequently,
theoutputsare
I balancedmodulator: 1- l)(sin o/) : - I sin ol r
Q balanced modulator= (- l)(cosaA : -l cosoJ
andtheoutputof the linearsummeris
-l coso"/ - I sina"t : 1.414sin(ro.t- 135")
For the remainingdibit codes(01, 10,and l1), the procedureis the same.The resultsareshownin
Figure2-18a.

In Figures 2- 18b and c, it can be seenthat with QPSK each of the four possible out-
put phasorshas exactly the sameamplitude. Therefore, the binary information must be en-
coded entirely in the phase of the output signal. This constant amplitude characteristicis
the most important characteristic of PSK that distinguishes it from QAM, which is ex-
plained Iater in this chapter.Also, from Figure 2-18b, it can be seenthat the angular sepa-
ration betweenany two adjacentphasorsin QPSK is 90". Therefore, a QPSK signal can un-
dergo almost a *45" or -45o shift in phaseduring transmissionand still retain the correct
encoded information when demodulated at the receiver. Figure 2-19 shows the output
phase-versus-timerelationship for a QPSK modulator.

66 Chapter 2
 Binary OPSK
lnoui ouFut
OI phsr.

00 -135..
01 -45c
t0 +135o
tl +1l5o

(a)

co. qt -.ln qt coeqt oo qt + dnqot

10 ? tt
sin(qt+ 1351 rln (o;t + tl5'l r0 ccqt 11
a a
/
, rln o"t
-.in qct l--------- ---------1 (().rgbilro.l

-rln o5t -----------i----------- lln qt

€qt- sln qt €8qt ooo6t+dnqt

00 01
sin (qt- 135'l oin(o5t- tl6'l -oo8 dlct 01
(bl GI

[aJtruth table:[bJphasordiagnam;[cJconstellation
FIGURE2-18 QPSKmodulaton: diagram

DiHt o t o lo r o t
lnput 10 01 | I 00

OPSK
ontp{rt
phase

FIGURE2-19 Outputphase-vensus-time
relationship
fona GPSK
modulaton

2-5-2-2 Bandwidth considerations of QPSK. With QPSK, becausethe input data

are divided into fwo channels, the bit rate in either the I or the Q channel is equal to one-half
of the input datante(f612). (Essentially, the bit splitter stretchesthe I and Q bits to twice their
input bit length.) Consequently,the highest fundamentalfrequency presentat the data input
to the I or the Q balanced modulator is equal to one-fourth of the input datarate (one-half of
f/2 : f/4).As a result, the output of the I and Q balanced modulators requires a minimum
double-sided Nyquist bandwidth equal to one-half of the incoming bit rate (f* : twice f/4
: f/2). Thus, with QPSK, a bandwidth compressionis realized (the minimum bandwidth is
lessthan the incoming bit rate).Also, becausethe QPSK output signal doesnot changephase
until two bits (a dibit) have been clocked into the bit splitter, the fastest output rate of change
(baud) is also equal to one-half of the input bit rate.As with BPSK, the minimum bandwidth
and the baud are equal.This relationship is shown in Figure 2-20.

Digital Modulation 67
 to ro
Input
d.tr t

t|n.ry
lnput
dda f5 lchannd
dtratbf2

FIGURE2-2O Bandwidthconsiderationsof a GIPSKmodulator

In Figure 2-20, it can be seen that the worse-caseinput condition to the I or Q bal-
anced modulator is an alternative 1/0 pattern, which occurs when the binary input data
have a 1100 repetitive pattern. One cycle of the fastest binary transition (a 1/0 sequence)
in the I or Q channel takes the sametime as four input data bits. Consequently,the highest
fundamental frequency at the input and fastestrate ofchange at the output ofthe balanced
modulators is equal to one-fourth of the binary input bit rate.
The output of the balanced modulators can be expressed mathematically as

output: (sin <oot)(sin

co.t) (2-22)

f^
root:2n-t and a; :2nf,

where modulating cilner

signal

Thus, output = (,inzn!){"inznt",)

_2),_
)"orz n )"o,zn
+ful4tof"-ftl\,andtheminimumband-
Theoutputfrequencyspectrumextendsfromf,
width (/r) is
/ ^\ / .\
(r *{n\ - (r"-r-r.\:zru
:ro
Y'4 / (' 4/ 4 2
Example2-6
with aninputdatatate(f) equalto 10Mbpsanda carrierfrequency
For a QPSKmodulator of
7OMHz,determinetheminimumdouble-sided Nyquistbandwidth(frv)andthebaud.Also, compare
theresultswith thoseachievedwith the BPSKmodulatdrin Example2-4.Usethe QPSKblock dia-
gram shownin Figure 2-17 asthe modulatormodel.
Solution The bit ratein both theI andQ channelsis equalto one-halfof the transmissionbit rate,or

_fi _ l0Mbps_ 5 Mbps

faq: fu
22

68 Chapter 2
 The highest fundamental frequency presentedto either balancedmodulator is

. JbQ JbI 5 Mbps

ra or-: :2 .5 MH z
2 2
The output wave from each balancedmodulator is
(srn 2nf"t)(sin 2nf.t)

2n(f,- y,)t - 2n(f"+ f.)t

)cos ;".,
- 2.5)MHzlt - j"o, 2nl(70+ 2.5)MHz)t
)cosznf(to

2n(67.sMHz)r - j.os 2n(72.5MH2)t

)cos
The minimumNyquistbandwidthis
B: ('72.5- 67.5)MHz = 5 MHz
The symbol rate equals the bandwidth; thus,
symbolrate:5 megabaud
The output spectrum is as follows:

<_B = 3.333 MHz_____--->

68.333MHz 70 MHz 71.667

71 MHz
.667 MHz

(Suppressed)

B :5 MHz

It canbe seenthatfor the sameinput bit ratethe minimumbandwidthrequiredto passthe outputof

the QPSKmodulatoris equalto one-halfof that requiredfor the BPSKmodulatorin Example2-4.
Also, thebaudratefor theQPSKmodulatoris one-halfthatof theBPSKmodulator.

The minimum bandwidth for the QPSK systemdescribedin Example 2-6 canalso be
determinedby simply substituting into Equation 2-10:

l0 Mbps
B:
2
:5 MH z

2-5'2-3 QPSK receiver. The block diagram of a QPSK receiver is shown in Figure
2-21. The power splitter directs the input QPSK signal to the I and Q product detectorsand
the carrier recovery circuit. The carrier recovery circuit reproduces the original transmit
carrier oscillator signal. The recoveredcarrier must be frequency and phasecoherentwith
the transmit referencecarrier. The QPSK signal is demodulatedin the I and Q product de-
tectors, which generatethe original I and Q data bits. The outputs of the product detectors
are fed to the bit combining circuit, where they are converted from parallel I and Q data
channelsto a single binary output data stream.
The incoming QPSK signal may be any one of the four possible output phasesshown
in Figure 2-18. To illustrate the demodulation process,let the incoming QPSK signal be
-sin o,r + cos ocf. Mathematically, the demodulation processis as follows.

Digital Modulation 69
o
Unq0({lnqt+oorqtl -1/2v0@o
{in qt + coc qt
-rln qt + coe qt
Inp|tr
oPsK
rlgnal
-sin qt + cosqt
(cosqt) (-8in qt + cos oct) +112
V flo0fu1)
FIGURE2-2'l GPSKnecerver
 The receiveQPSKsignal(-sin co"r* costo.t) is oneof the inputsto the I product
detector.The otherinput is the recoveredcarrier(sin ol"r).The outputof the I productde-
tectoris

1 : (-sin toc/ + cos co"r)(sinco"t) (2-23)

QPSK input signal

: 1-sin <rl.r)(sina,t) * (cos ro"t)(sin ol"t)

: -sin2 <o"/* (cos orr)(sin <o.l)

l ,- 1 1
: -r(l - cos 2<o.r)+ sin(<o.+ o")r + - <,r.)t
f ;sin(co"

I : -; +
I I ^2*"1"".'^)"*r'
2a,t I + 0
icos ,sin2<6,t isin

: -lu (logic
o)
Again, the receiveQPSKsignal(-sin <o,r+ cos ro.r)is oneof the inputsto the Q
productdetector.Theotherinputis therecoveredcarriershifted90oin phase(cosar.r).The
outputof the Q productdetectoris
q : (-sin ocl + cos r'l"r)(cos <rl"r) (2-24)

QPsKinputsignal canier

: cos2co"/ - (sin <o"l)(cos <o"r)

11, I
: + cos}ro,t) - * <o.)r- - o")t
,(t isin(<o" isin(co"

r l:
'2
I
- +jco,
,Y';",,,f\',^,
I
-, V(logic 1)

The demodulatedI and Q bits (0 and l, respectively) correspondto the constellation

diagram and truth table for the QPSK modulator shown in Figure 2-18.

2-5-2-4 Offset QPSK. Offiet QPSK (OQPSK) is a modified form of QpSK where
the bit waveforns on the I and Q channelsare offset or shifted in phasefrom eachother by
one-half of a bit time.
Figure 2-22 shows a simplified block diagram, the bit sequencealignment, and the
constellation diagram for a OQPSK modulator. Becausechangesin the I channel occur at
the midpoints of the Q channel bits and vice versa, there is never more than a single bit
change in the dibit code and, therefore, there is never more than a 90o shift in the output
phase.ln conventional QPSK, a changein the input dibit from 00 to 11 or 01 to 10 causes
a corresponding 180'shift in the output phase.Therefore, an advantageofOQPSK is the
limited phaseshift that must be imparted during modulation. A disadvantageof OQPSK is

DigitalModulation 71
 l4htnncl
lnput&ta

Odnnnel
Input drta

o I a (tgr) @rQt
t5
tl
tl
1 0 ? lrrls.r
tl
l4hcnnel
input drta
l hl r' lb rl I
I
dn qt ---------i---------' rln o5t
I

I
O+hrnnel lb o lo tlh l b3 o
inputd.ta ll
t5 0 I o crssr-cocqt o I r rrn
(c)

[a) blockdiagram;(b) bit alignmenu[c) constellation

FIGURE2-22 OffsetkeyedIOBPSKJ:
diagnam

that changes in the output phase occur at twice the data rate in either the I or Q channels.
Consequently, with OQPSK the baud and minimum bandwidth are twice that of conven-
tional QPSK for a given transmission bit rate. OQPSK is sometimes called OKQPSK
(offset-keyedQPSIO.

2-5-3 8-PSK
With 8-PS1(, three bits are encoded, forming tribits and producing eight dffierent output
phases.With 8-PSK,n : 3, M :8, andthereareeight possibleoutput phases.To encodeeight
different phases,the incoming bits are encoded in groups of three, called tribits (2' : 8).

2-5-3-18-PSK transmitter. A block diagram of an 8-PSK modulator is shown in

Figure 2-23. The incoming serial bit stream enters the bit splitter, where it is converted
to a parallel, three-channeloutput (the I or in-phase channel, the Q or in-quadrature chan-
nel, and the C or control channel). Consequently, the bit rate in each ofthe tlree chan-
nels is/6/3. The bits in the I and C channels enter the I channel2-to- -level converter,
and the bits in the Q and C channels enter the Q channel 2-to-4-level converter. Essen-
tially, the 2-to- -level converters are parallel-input digital-to-analog converters
(DACs). With two input bits, four output voltages are possible. The algorithm for the
DACs is quite simple. The I or Q bit determines the polarity of the output analog sig-
nal (logic I : *V and logic 0 : -V), whereas the C or C bit determines the magni-

72 Chapter 2
 8-PSt(
output

FIGURE2-23 8-PSKmodulator

+1.4)7V
I c Output oc Output
00 -0.541V 01 -1.307V +0.541V
01 -1.307V 00 -0.541V 0v
10 +0.541V 11 +1.3)7V 0tl1 v
11 +1.307V 10 {-0.541V
-1.307V
(a) tb)

FIGURE2-24 l- and G-channelZ-to4level conventers:[a] l-channeltnuth table;

[bJ Gl-channeltruth table; [cJ PAM levels

tude (logic I : 1.307 V and logic 0 : 0.541 V). Consequently,with two magnitudes and
two polarities, four different output conditions are possible.
Figure 2-24 shows the truth table and correspondingoutput conditions for the 2-to-
4-level converters.Becausethe C and C bits can never be the sameloglc state,the outputs
from the I and Q 2-to-4-level converters can never have the same magnitude, although they
can have the same polarity. The output of a2-to-4-level converter is an M-ary, pulse-
amplitude-modulated(PAM) signal where M : 4.

Example2-7
Foratribitinputof Q = 0, I = 0, andC:0(000), determinetheoutputphaseforthe
8-PSKmod-
ulator shown in Figare 2-23.

Sof ution The inputs to the I channel2-to-4-Levelconverter are I : 0 and C : 0. From Figure2-24
the output is -0.541 V. The inputs to the Q channel z-to- -level converter are Q : 0 and C = l.
Again from Fi gxe 2-24, the output is - I .307 V.
Thus, the two inputs to the I channelproduct modulators are -0.541and sin co"r.The output is
1 = 1-0.5a1)(sin co"t)= -0.541 sin rrl./
The two inputs to the Q channel product modulator are -1.307 V and cos trlct The oulput is
q : 1-1.307)(cos or.r) : -1307 cos oc/

Digital Modulation 73
 The outputsof the I andQ channelproductmodulatorsarecombinedin the linearsummerandpro-
ducea modulatedoutputof
summeroutput: -0.541 sin trt r - 1.307cosro.r
: 1.41sin(co.t- 112.5")
Fortheremaining tribitcodes(001,010,011, is thesame.The
100,101,110,andlll), theprocedure
resultsareshownin Figtxe2-25.

+oSlt rinqt+ t.3tc6St

orc o rc
10 0 1t0

-1,9)7 Binqt + 0,541co3qt +1307 Ein oot + 0.f*1 @. qt

orc o rc
101 111

-sln act I 5 -----f -------------l si nq t

-1,9)7 sanqt - 0,541oo! qct +1.q17lln o6t-0.54f co.St

orc o rc
00 1 011

-0.541sin obt- 1.307cog +0,541sln qt - 1.3t7 co. qt

o tc J o rc
0 00 -c6qt 0' I 0
8kr&y 6-PSK (b,
l|put qrtptjt
or c phate
00 0 -1 12.5. co3qt
00 1 -157,6. a a
01 0 -dr.50 100 110
01 1 -22.5"
r00 +112.5'
t01 +1d1.5'
11 0 +617.6o
t11 +22.5"
a a
(at 101 11 r

-sln oct sln qt

a o
001 0t 1

o a
m0 {o8(tt 010

(c)

FIGUHE2-25 BfSK modulaton:

[a) tnuthtable;[bJphasondiagram;
{c) constellation
diagnam

74 Ghapter 2
 Tribit (uc
input 0(x)
8.PSK
output
phtse I
tl I
-112.5. | -1s7.5"1 -6-r.5. | +211.5'

FIGURE2-26 Outputphase{ersus-time
nelationship
for an 8-PSKmodulator

From Figure 2-25, it can be seenthat the angular separationbetween any two adja-
cent phasorsis 45', half what it is with QPSK. Therefore, an 8-PSK signal can undergo al-
most a -122.5" phaseshift during transmissionand still retain its integrity. Also, each pha-
sor is of equal magnitude; the tribit condition (actual information) is again contained only
in the phaseof the signal.The PAM levels of 1.307and 0.541 are relative values.Any lev-
els may be usedas long as their ratio is 0.54111.301and their arc tangentis equalto 22.5o.
For example, if their values were doubled to 2.614 and 1.082, the resulting phase angles
would not change,although the magnitude of the phasor would increaseproportionally.
It should also be noted that the tribit code between any two adjacentphaseschanges
by only one bit. This type of code is called the Gray code or, sometimes,the maximum dis-
tance code. This code is used to reduce the number of transmissionerrors. If a signal were
to undergo a phaseshift during transmission,it would most likely be shifted to an adjacent
phasor.Using the Gray code results in only a single bit being received in error.
Figure 2-26 showsthe output phase-versus-timerelationship of an S-PSK modulator.

2-5-3-2 Bandwidth considerations of 8-PSK. With 8-PSK, becausethe data are

divided into three channels,the bit rate in the I, Q, or C channel is equal to one-third of the
binary input datarate (f613).(Thebit splitter stretchesthe I, Q, and C bits to three times their
input bit length.) Becausethe I, Q, and C bits are outputted simultaneouslyand in parallel,
the 2-to-4-level convertersalso seea changein their inputs (and consequentlytheir outputs)
at arate equalto fr/3.
Figttre 2-27 showsthe bit timing relationship between the binary input data; the I, Q,
and C channeldata; and the I and Q PAM signals.It can be seenthat the highestfundamental
frequencyin the I, Q, or C channelis equalto one-sixththe bit rate of the binary input (onecy-
cle in the I, Q, or C channeltakesthe sameamountof time as six input bits). Also, the highest
fundamentalfrequencyin either PAM signal is equal to one-sixth of the binary input bit rate.
With an 8-PSK modulator, there is one changein phaseat the output for every three
datainputbits. Consequently,thebaudfor 8PSKequalsS/3, the sameastheminimumband-
width. Again, the balancedmodulators are product modulators; their outputs are the product of
the carrier and the PAM signal. Mathematically, the output of the balancedmodulators is

0 : (X sin o"r)(sin o.t) (2-2s)

t

where clot: 2n"j I and a; :2nf,t

_n
modulating signal carrier

and X: +1.307or +0.541

Thus, r: (" sinzn!)Gr^rnr,,)

: l*"r^Q, X
,c o s
2n

Digital Modulation 75
 2phrn2voftrgrlil*
l'iner
fb/3 1O rl.gtTv
1o.541

Bin.ry
Inglt
drta f5

Input
data 16
Highst
fundamenta!
Cc-trannel fiequenoy
dda fb/3

t+hennel
dlta fbl3

+ 1A J7 V
--- +0.541V
F
I

+1.307 -0.541 +0.541 +1.307

tln obt sin o6t sin oct 6in obt

2-27 Bandwidth
FIGURE considenations
of an &PSKmodulator

Theoutputfrequencyspectrumextendsfromf, + f ul6 tof" - f il6, andtheminimumband-

width (fij is

v,.i)-(,- ;):T:;
/ a\ / a \2 f u f u

Example2-8
For an 8-PSK modulator with an input data tate (f) equal to l0 Mbps and a carrier frequency of
70MHz, determine the minimum double-sided Nyquist bandwidth (fiv) and the baud. Also, compare
the results with those achieved with the BPSK and QPSK modulators in Examples2-4 and 2-6. Use
the 8-PSK block diagram shown in Figvre 2-23 as the modulator model.
Solution The bit rate in the I, Q, and C channelsis equal to one-third of the input bit rate, or

{ -€ l0 Mbps
-{
Jbc - JbQ - Jbr -
- = 3.33Mbps

7E Chapter 2
 Therefore, the fastestrate of change and highest fundamental frequency presentedto either balanced
modulator is
3.33Mbps
f":+o,$o,f!:
zzz
: 1.667Mbps

The outout wave from the balance modulators is

(sin2nf"t)(sin2nf"t)

1l
2n(f, - f,)t - 2n(f" + f")t
;cos rcos
ll
I2- Zcos2n[(70 - r.667)MHz]r - | cos2n[(70 + r.667)MHz]t

1r MHz)t - cos2n(71.667MHz)t
,cos2n(68.333 |
The minimum Nyquist bandwidthis
3 = (7r.667- 68.333) MHz : 3.333MH2
The minimumbandwidthfor the 8-PSKcan alsobe determinedby simply substitutinginto
Equation2-10:
l0 Mbps
B:
:3 .3 3M H2
Again,thebaudequalsthebandwidth;thus,
baud: 3.333megabaud
The outputspectrumis asfollows:
< -B = 2. 5 M H z +

68.75MHz 70 MHz 71.25MHz

71.25MH2

(Suppressed)
B:3 .3 33 MH 2
It can be seenthat for the sameinput bit rate the minimum bandwidthrequiredto passthe outputof
an 8-PSKmodulatoris equalto one-thirdthat of the BPSK modulatorin Example2-4 and507oless
thanthatrequiredfor theQPSKmodulatorin Example2-6.Also, in eachcasethe baudhasbeenre-
ducedby the sameproportions.

2-5-3-3 8-PSK receiver. Figure 2-28 shows a block diagram of an S-PSK receiver.
The power splitter directs the input 8-PSK signal to the I and Q product detectorsand the
carrier recovery circuit. The carier recovery circuit reproducesthe original reference os-
cillator signal. The incoming 8-PSK signal is mixed with the recoveredcarrier in the I prod-
uct detector and with a quadrature carrier in the Q product detector. The outputs of the prod-
uct detectors are 4-level PAM signals that are fed to the 4-to-2-level analog-to-digital
converters (ADCs). The outputs from the I channel -to-2-level converter are the I and
C bits, whereas the outputs from the Q channel 4-to-Z-level converter are the Q and
C bits. The parallel-to-serial logic circuit converts the UC and Q/C bit pairs to serial I,
Q, and C ouput data sffeams.

2-5-4 16-PSK
16-PSKis an M-ary encoding technique whereM : 16; there are 16 different output phases
possible. With 16-PSK, four bits (called quadbits) are combined, producing 16 different
output phases.With 16-PSK, ft : 4 and M : 16; therefore, the minimum bandwidth and

Digital Modulation 77
o
Analog4e
dUtrl
@nrrwttr
Analog-to-
digital
@nwn€r
FIGURE2-28 B-PSKneceiver
. .- .- aJ ' ''- '.1 :
 cos (&t
01mo . 0011
0101o . 0010

0 110. . u)01

011 1. . 00q)
Bitcode Phase Bitcode Phas6
1000o o 1111
(xno 11.250 rm0 19r.25"
0001 :x].7r 1001 213.76"
0010 ffi.250 1010 236^25" 10 01. . 11 1 0
q)11 78,75" 10 11 258.750
0100 101.25' 1100 ?€1.25"
0101 123,75" 11 01 3ft1.750 10 10. . 1 101
0110 146.25' 11 10 326.26. 1011 0 o 11(X)
0111 168.75" 11 11 348.75' -cos oct

(a) {b)

2-29 16-PSK:[aJtnuthtable;[b] constellation

FIGURE diagram

baud equal one-fourth the bit nte (f614). Figure 2-29 shows the truth table and constella-
tion diagramfor 16-PSK,respectively.ComparingFigures2-18,2-25, and2-29 showsthat
as the level ofencoding increases(i.e., the values ofn andM increase),more output phases
are possible and the closer each point on the constellation diagram is to an adjacentpoint.
With 16-PSK, the angular separationbetween adjacentoutput phasesis only 22.5".There-
fore, 16-PSK can undergo only a 11.25" phase shift during transmission and still retain its
integrity. For an M-ary PSK system with 64 output phases(n : 6), the angular separation
between adjacentphasesis only 5.6'. This is an obvious limitation in the level of encoding
(and bit rates) possible with PSK, as a point is eventually reachedwhere receivers cannot
discern the phaseof the received signaling element. In addition, phaseimpairments inher-
ent on communicationslines have a tendencyto shift the phaseof the PSK signal, destroy-
ing its integrity and producing errors.

MODULATION
2-6 QUADRATURE-AMPLITUDE
Quadrature-amplitude modulation (QAM) is a form of digital modulation similar to PSK
except the digital information is containedin both the amplitude and the phaseof the trans-
mitted carrier. With QAM, amplitude and phase-shift keying are combined in such a way
that the positions of the signaling elementson the constellation diagrams are optimized to
achievethe greatestdistancebetweenelements,thus reducing the likelihood of one element
being misinterpreted as another element. Obviously, this reduces the likelihood of errors
occumng.

2-6-1 8-OAM
8-QAM is an M-ary encodingtechniquewhere M : 8. Unlike 8-PSK, the output signal from
an 8-QAM modulator is not a constant-amplitudesignal.

2-6-l-l8-QAM transmitter. Figure 2-30a shows the block diagram of an 8-QAM

transmitter.As you can see,the only difference between the 8-QAM transmitter and the 8-
PSK transmitter shown in Figure 2-23 is the omission of the inverter between the C chan-
nel and the Q product modulator.As with 8-PSK, the incoming data are divided into groups
of three bits (tribits): the I, Q, and C bit streams,each with a bit rate equal to one-third of

Digital Modulation 79
 &OAM
output

t/o c Output
00 -0.54rv
01 V
-1,307
10 +0.5t11V
11 + 1. 30 7V
(a) (b)

FIGURE2-3O B-GAMtnansmitten:[a] block diagnam;[b] tnuth table 24 levelconventens

the incoming data rate. Again, the I and Q bits determine the polarity of the PAM signal at
the output of the 2-to-4-level converters,and the C channel determinesthe magnitude. Be-
causethe C bit is fed uninverted to both the I and the Q channel 2-to-4-level converters,
the magnitudes of the I and Q PAM signals are always equal. Their polarities depend on
the logic condition of the I and Q bits and, therefore, may be different. Figure 2-30b shows
the truth table for the I and Q channel Z-to-4-level converters; they are identical.

Example2-9
F ora tri biti np uto fQ :0 ,I : 0, and C : 0( 000 ), d et er mi net he out pu t am pl i t u dean dphas ef or t h e
8-QAM transmitter shown in Figure 2-30a.
Sofution The inputs to the I channel 2-to-4level converter are I : 0 and C : 0. From Figure 2-30b,
the output is -0.541 V. The inputs to the Q channel2-to-4-level converterare Q : 0 and C : 0. Again
from Figure 2-30b, the output is -0.541 V.
Thus, the two inputs to the I channel product modulator are -0.541 and sin ro.L The output is
I : (-0.541)(sin o.t) = -0.541 sin ro.r
The two inputs to the Q channel product modulator are -0.541 and cos oct. The output is
Q : (-0.541)(cosco.l): -0.541 cos rrlc/
The outputs from the I and Q channel product modulators are combined in the linear summer and pro-
duce a modulated output of
mer output : --0.54rsin or"l -0.541 cos o,c,
: 0.765 sin(trr.r- 135")
F ort he rem ain in gff ibi t c ode s (0 01, 0 10, 01100,
1, 101 ,1 10, an d111 ), t hepr oc e dur ei s t he s am e . Th e
results are shown in Figure 2-3 I .
Figrne 2-32 shows the output phase-versus-timerelationship for an 8-QAM modulator. Note
that there are two output amplitudes, and only four phasesare possible.

2-6-l-2 Bandwidth considerationsof 8-QAM. In 8-QAM, the bit rate in the I and
Q channelsis one-thirdof the input binary rate,the sameas in S-PSK.As a result,the high-
est fundamental modulating frequency and fastest output rate of changein 8-QAM are the
same as with 8-PSK. Therefore, the minimum bandwidth required for S-QAM is f ul3, the
sameas in 8-PSK.

80 Chapter 2
 Einary
input S{IAM output
o tc AmplltndG Ph!r.

0 00 0.765V -135.
001 13t8V -135.
010 0.7€5V {5r
0lr lAr8V -45'
100 0.tGEV +135'
r 01 l.8rgv +1s.
t 10 0.76V +4f
1lt 1"8a8V +45'

{rl

cos (tl
llf (tr,l8Vf

110

a o
fito 0 10

001 -cos oBt 0 11

(cl

FIGURE2-31 8-OAMmodulator:[aJtnuthtable;[b] phasondiagnam;[cJconstellation

diagnam

(uc i olc l orc

1 01! 1 10111 1

0.765V i r.aa8V 0.765V i 1.qasy 0.765V 1 .8 4 8V i 0 . 76 5 V i 1 .8 4 8V

-135" I -135' -"{5o | -+r' +135' +1 3 5 'l +4 5 "1 + rtS '

FIGURE2-32 Outputphaseand amplitude-versus-time for'8-CIAM

relationship

2-6-l-3 8-QAM receiver. An S-QAM receiver is almost identical to the 8-PSK re-
ceiver shown in Figure 2-28.The differences are the PAM levels at the output of the prod-
uct detectors and the birlary signals at the output of the analog-to-digital converters. Be-
causethere are two transmit amplitudespossible with 8-QAM that are different from those
achievable with 8-PSK, the four demodulated PAM levels in S-QAM are different from
those in 8-PSK. Therefore, the conversion factor for the analog-to-digital convertersmust
also be different. Also, with 8-QAM the binary output signals from the I channel analog-
to-digital converter are the I and C bits, and the binary output signals from the Q channel
analog-to-digital converter are the Q and C bits.

Digital Modulation 81
 2-6-2 16-OAM
L6-QAMis anM-Ty systemwhereM : 16.Theinpurdataareacted
As with the16-PSK,
on in groups of four (24 : 16). As with 8-QAM, both the phase and the amplitude of the
transmit carrier are varied.

2-6-2-l QAM transmitter. The block diagram for a 16-QAM transmitter is shown
in Figure 2-33.T\e input binary data are divided into four channels:I, I', Q, and Q'. The
bit rate in each channel is equal to one-fourth of the input bit rate (f bl4).Fow bits are seri-
ally clocked into the bit splitter; then they are outputted simultaneouslyand in parallel with
the I, I', Q, and Q' channels.The I and Q bits determine the polarity at the output of the
2-to-4-level converters(a logic I : positive and a logic 0 : negative). The I' and Q' bits
determine the magnitude (a logic I : 0.821 V and a logic 0 : 0.22 V). Consequently,the
Z-to-4-level convertersgeneratea 4-level PAM signal. Two polarities and two magnitudes
are possible at the output of each 2-to- -level converter.They are 0.22V and +0.821 V.
The PAM signals modulate the in-phase and quadraturecarriers in the product mod-
ulators. Four outputs are possible for eachproduct modulator. For the I product modulator,
they are +0.821 sin o"/, -0.821 sin ol,r, -10.22 sin tD"t,and -0.22 sin oc/. For the Q prod-
uct modulator, they are +0.821 cos oct, -10.22 cosaJ, -0.821cos ocl, and,-0.22 cos ocr.
The linear summer combines the outputs from the I and Q channelproduct modulators and
producesthe 16 output conditions necessaryfor 16-QAM. Figure 2-34 shows the truth table
for the I and Q channel 2-to-4-level converters.

FIGURE2-33 16-QAMtransmitterblockdiagram

tr Output oc r Oubut
00 4.22V 00 4.22V
01 -0.821V 0l -0.821V
10 +O,22V 10 +O.22V
1l +O.E21V It +O.&11V
Irl (bt

FIGURE2-34 Trwh tablesfonthe F and O-channelZ-to{-

(a) | channel;[b) Glchannel
levelconvertens:

a2 Chapter 2
 Example2-1O
Foraquadbitinpu tof l :0 ,I' :0 ,Q:O ,an de ': 0(00 00 ),de ten ni n eth eou tpu tamp lit udean d
phasefor the I6-QAM modulatorshownin Figure2-33.
Sofution The inputsto theI channel2-to-4-levelconverterareI : 0 andI, : 0. FromFigure2-34,
theoutputis -0.22Y.Theinputsto thee channel2-to-4-levelconverterare : 0 and : 0. Again
e e,
fromFigure2-34,theoutputis -0.22V.
Thus,the two inputsto the I channelproductmodulatorare -0.22Y andsino.r. The outputis
1 : 1-0.22)(sinoct) : -0.22 sinaS
The two inputsto theQ channelproductmodulatorare -0.22v andcosroctThe outputis
Q : (-0.22xcosoct) : -0.22 cosa"t
The outputsfrom theI andQ channelproductmodulatorsarecombinedin thelinear summerandpro-
ducea modulatedoutputof
summeroutput: _0.22 sin rol _ 0.22 cosaS
= 0.311sin(ro;- 135")
For theremainingquadbitcodes,theprocedureis the same.The resultsareshownin Figwe 2-35.

Elnary
input
1&oAM
ocv I r ouFut
0 00 0 0.311V -136.
0 00 1 0"850v -165"
0 01 0 0.311v -a5.
0 01 1 0.850v -15"
0 10 0 0.E5ov -1(}5.
0101 1.101V -135.
0 11 0 0"850v -750
0111 1.16rV -+50
1000 0.311V 1350
1001 0.850v 165c
1010 0,311V a50
1011 0.860v 150
1100 0.60v 1050
1 10 1 1.161V 135"
1 11 0 0.850v 75.
1 1t l 1.161V a50

(a)

1. 161 . 1101 11 q) . T . 1110 lltl.

o 1001 1000. + . r01o tofl.

I
I
F - - -- -- -- - J----i ----F ---------l
I
o flDl flXD e * . OO1OOOfi.

. 0 1 0 1 0 1 0 0. r . 0 1 1 0 0 11 1 .

(b)

FIGURE2-35 16{l,AM modulator:[a] truth table;[bJphasordiagram;[cJconstellation

diagnam

Digital Modulation
83
 2 phucd4 vohagalcvclr
i.lnqt
r 0.22V
*0.821v

To O+hannel
2-to-&lewl
convorter

r r cv orrc ro rro o
lnput
datf fb
High€Et

frh!nnel
data fbl4

l-ehrnnel
d.ra fb/4

+0.&1 V
+o'?:2v
t.ch.nrct
PAMout -0.22V
-0.E41v

kh6nn€l
bahnced
modulatot
out

-0,82r sln olct +0,22sin or"t -0.821sln olt

FIGURE2€6 Bandwidthconsiderations
of a 16{lAM modulator

2-6-2-2 Bandwidth considerations of 16-QAM. With a 16-QAlt4, becausethe in-

put data are divided into four channels,the bit rate in the I, I', Q, or Q' channel is equal to
one-fourth of the binary input data rate (ftlD. (The bit splitter stretchesthe f, I', Q, and Q'
bits to four times their input bit length.) Also, becausethe I, I', Q, and Q' bits are outputtec
simultaneouslyand in parallel, the 2-to-4-level converterssee a changein their inputs and
outputs at a rate equal to one-fourth of the input data rate.
Figure 2-36 shows the bit timing relationshirpbetween the binary input data; the I, I'.
Q, and Q' channel data; and the I PAM signal. It can be seenthat the highest fundamenta
frequency in the I, I', Q, or Q' channelis equal to one-eighthof the bit rate of the binary in-
put data (one cycle in the I, I', Q, or Q' channel takes the sameamount of time as eight in
put bits). Also, the highest fundamental frequency of either PAM signal is equal to one-
eighth of the binary input bit rate.
With a 16-QAM modulator, there is one changein the output signal (either its phase,
amplitude, or both) for every four input data bits. Consequently, the baud equalsf614, the
same as the minimum bandwidth.

84 Chapter 2
 Again,thebalancedmodulatorsareproductmodulatorsandtheiroutputscanbe rep-
resentedmathematicallyas
output: (X sin <o"t)(sin
o"r) (2-26)

where @,t: znf:t and @ct:zrEfJ

6-

modulating signal ciuner

and X:+0.22 or +0.821

/ rc \
Thus, output: (Xsin zn,ft\sn2nfS)
\ o, /

x (" v / ,c\
: ^rr$ - .fr\ = cos2np.
cos
i fl, ; ilt
-lful8tof"-falS,andtheminimumband-
Theoutputfrequencyspectrumextendsfromf,
width (/r) is

/r ,\/t\2fifu _ r!l : _= _
l f + " _!l _l f
(' 8/ (' 8/ 8 4

Example
2-11
withaninputdatarate(f) equalto l0 Mbpsanda carrierfrequency
Fora 16-QAMmodulator of
70 MHz, determinethe minimum double-sidedNyquist frequency(f,v)andthe baud.Also, compare
theresultswith thoseachievedwith theBPSK,QPSK,and8-PSKmodulatorsin Examples2-4,2-6,
and2-8.Usethe 16-QAMblock diagramshownin Figure2-33asthemodulatormodel.
Solution Thebit ratein theI, I', Q, andQ' channelsis equalto one-fourthof the inputbit rate,or

fn : fu' : fn = f'o : *: 19P = 2'5MbPs

Therefore, the fastest rate of change and highest fundamental frequency presented to either balanced
modulator is

r":+ o,rf o,+ - +: r# : t.25MHz

The outputwavefrom thebalancedmodulatoris
(sin2nf"t)(sin2rcfS)

| ^ ,^ - I
- +
tcos2rE(fc f")t rcos2n(f, f,)t
ll
f cos2n[(70- r.2s)MHz]t - ,cos2nl(70 + 1.25)MHzlr

ll
2n(68.75 MHz)r - 2n(71.25MHz)t
,cos rcos
The minimum Nyquist bandwidthis
B : (7r.25- 68.75)MHz : 2.5MHz
The minimum bandwidthfor the 16-QAM can alsobe determinedby simply substitutinginto
Equation2-10:
_ 10Mbps
B :-
o
:2 .5MH z

Digital Modulation s5
 The symbol rate equals the bandwidth; thus,
symbol rate : 2.5 megabaud
The output spectrumis as follows:

Input
ttttttl
irriiiii
l^ I -.1- Hi.h..t |
-*-
B : 2.5 MH z
For the sameinput bit rate,the minimumbandwidthrequiredto passthe outputof a 16-QAMmod-
ulatoris equalto one-fourththat of the BPSKmodulator,one-halfthat of QPSK, and25Valessthan
with 8-PSK.For eachmodulationtechnique,the baudis alsoreducedby the sameproportions.

Example2-12
For the following modulation schemes,construct a table showing the number of bits encoded,num-
ber of output ccinditions, minimum bandwidth, and baud for an information data rate of 12 kbps:
QPSK, 8-PSK, 8-QAM, 16-PSK,and 16-QAM.
Solution
Modulation n M B (Hz) baud

QPSK 2 4 6000 6000

8.PSK -l 8 4000 4000
8-QAM 3 8 4000 4000
I6-PSK 4 16 3000 3000
16-QAM 4 16 3000 3000

From Example 2-12, it can be seenthat a l2-kbps data stream can be propagatedthrough a n:urower
bandwidth using either 16-PSK or 16-QAM than with the lower levels of encoding.

Table2-I summarizesthe relationship betweenthe number of bits encoded,the num-

ber of output conditions possible, the minimum bandwidth, and the baud for ASK, FsK,
PSK, and QAM. Note that with the three binary modulation schemes(ASK, FSK, and

Table2-1 ASK,FSK,PSK,and GAMSummarv

Modulation Encoding Scheme OutputsPossible Minimum Bandwidth Baud

ASK Single bit 2 ft fb

FSK Single bit 2 f.
BPSK Single bit 2 fb fo
QPSK Dibits 4 h t2 fo 12
8-PSK Tribits 8 h /3 fi / 3
8-QAM Tribits 8 fu 13 fo13
l6-QAM Quadbits 16 fi /4 fo /4
16-PSK Quadbits 16 fo 14 fa 14
32-PSK Five bits 32 fu /5 fa /5
32-QAM Five bits 32
64-PSK Six bits 64 fu 16 fu16
64-QAM Six bits 64 fa /6 fo 16
128-PSK Seven bits 128 fu /7 f i /7
128-QAM Seven bits t28 fo /7 f i /7
Note:/6 indicates a magnitude equal to the input bit rate.

86 Chapter 2
 BPSK), n : l, M : 2, only two output conditions are possible,and the baud is equal to the
bit rate. However, for values of n > l, the number of output conditions increases,and the
minimum bandwidth and baud decrease.Therefore, digital modulation schemeswhere n > 1
achieve bandwidth compression(i.e., less bandwidth is required to propagate a given bit
rate). When data compressionis performed, higher data transmissionrates are possible for
a given bandwidth.

2.7 BANDWIDTHEFFICIENCY
Bandwidth fficiency (sometimes called information density or spectral fficiency) is often
used to compare the performance of one digital modulation technique to another. In
essence,bandwidth efficiency is the ratio of the transmissionbit rate to the minimum band-
width required for a particular modulation scheme.Bandwidth efficiency is generally nor-
malized to al-Hz bandwidth and, thus, indicates the number of bits that can be propagated
through a transmission medium for each hertz of bandwidth. Mathematically, bandwidth
efficiency is

bit rate(bps)
or.,_ transmission (2-27)
-'' minimumbandwidth (Hz)

: bits/s : bits/s : bits

hertz cycles/s cycle
where Bq = bandwidth efficiencY

Bandwidth efficiency can also be given as a percentageby simply multiplying Bq by 100.

Example2-13
For an 8-PSK system, operating with an information bit rate of 24 kbps, determine (a) baud, (b) min-
imum bandwidth, and (c) bandwidth efFrciency.
Solution a. Baud is determined bv substitutins into Equation 2-10:

24.000
baud = =8000
,

b. Bandwidth is determined by substituting into Equation 2-11:

24,000
B: = 8000
J

c. Bandwidth effrciency is calculated from Equation 2-27:

24,000bps
Bq:
8000 Hz
= 3 bits per secondper cycle of bandwidth

Example2-14
For I6-PSK and a transmission system with a 10 kHz bandwidth, determine the maximum bit rate.
Solution The bandwidth efficiency for 16-PSK is 4, which meansthat four bits can be propagated
through the system for each hertz of bandwidth. Therefore, the maximum bit rate is simply the prod-
uct of the bandwidth and the bandwidth efficiency, or
bitrate:4 X 10.000
: 40,000bps

Digital Modulation a7
 Table2-2 ASK,FSK,PSK,and oAM Summany

Modulation Encoding Scheme Outputs Possible Minimum Bandwidth Baud Bn

ASK Single bit 2 fa fr
FSK Single bit 2 fb Jt
BPSK Single bit 2 fb fo
QPSK Dibits 4 .fr/2 fu/2 2
8-PSK Tribits 8 fa /3 h t3 3
8-QAM Tribits 8 fu /3 fo 13 3
16-PSK A
Quadbits 16 fi /4 fu 14
l6-QAM Quadbits l6 fo14 fu/4
/
32-PSK Five bits 32 fu15 5
64-QAM Six bits 64 fa /6 h /6 6

Note:/6 indicates a magnitude equal to the input bit rate.

2-7-1 DigitalModulationSummary
The properties of several digital modulation schemesare summarized in Table 2-2.

2-A CARRIERRECOVERY
Carrier recovery is the processof extracting a phase-coherentreferenceczrrier from a re-
ceiver signal. This is sometimes calledphase referencing.
In the phasemodulation techniquesdescribedthus far, the binary data were encoded
as a precisephaseof the transmittedcarrier. (This is referred to as absolutephase encoding.)
Dependingon the encodingmethod, the angular separationbetweenadjacentphasorsvaried
between 30o and 180'. To correctly demodulatethe data, a phase-coherentcarrier was re-
covered and comparedwith the received carrier in a product detector.To determine the ab-
solute phaseof the received carrier, it is necessaryto produce a carrier at the receiver that is
phasecoherentwith the transmit referenceoscillator. This is the function of the carrier re-
covery circuit.
With PSK and QAM, the carrier is suppressedin the balancedmodulators and, there-
fore, is not transmitted. Consequently,at the receiver the carrier cannot simply be tracked
with a standardphase-lockedloop (PLL). With suppressed-carriersystems,such as PSK
and QAM, sophisticatedmethods of carrier recovery are required, such as a squaring loop,
a Costas loop, or a remodulator

2-A-1 SquaringLoop
A common method of achievingcarrier recovery for BPSK is the squaring loop. Figure 2-37
shows the block diagram of a squaring loop. The received BPSK waveform is filtered and
then squared.The filtering reducesthe spectral width of the received noise. The squaring
circuit removesthe modulation and generatesthe secondharmonic of the carrier frequency.
This harmonic is phasetracked by the PLL. The VCO output frequency from the PLLthen
is divided by 2 and used as the phasereferencefor the product detectors.

vco
BPSK Rocovcrad
input carriar

FIGURE2-37 Squaningloop canniennecoverycincuitfor a BPSK necetven

88 Chapter 2
 With BPSK, only two output phasesare possible: f sin ro,/ and -sin to"/. Mathe-
matically, the operation of the squaring circuit can be described as follows. For a receive
signal of * sin co"r,the output of the squaring circuit is
output : (*sin co.t)(*sin o"l) : +sin2 o"t
(filtered out)

//
: - cos2o.r) : - lcos2.;-^t
|ft ; 2

For a received signal of -sin ro./, the output of the squaring circuit is
output : (-sin crl.r)(-sin o.l) : +sin2 (D.t

'T*)',
lll^
:t (l - cos 2o" t):t -
,co s2a ct

It can be seenthat in both cases,the output from the squaringcircuit containeda con-
stant voltage (+ll2V) and a signal at twice the carrier frequency (cos 2al.t). The constant
voltage is removed by filtering, leaving only cos 2ro./.

2-A-2 Gostas Loop

Asecond method of carrierrecovery is the Costas,orquadrature,loop shown in Figute2-38.
The Costasloop producesthe sameresultsas a squaringcircuit followed by an ordinary PLL
in place of the BPF. This recovery schemeusestwo parallel tracking loops (I and Q) simul-
taneouslyto derive the product of the I and Q compohentsof the signal that drives the VCO.
The in-phase (I) loop uses the vco as in a PLL, and the quadrature(Q) loop uses a 90o
shifted VCO signal. Once the frequency of the VCO is equal to the suppressed-carrier

Balanced
product
detector

)igital Modulation 89
 frequency, the product of the I and Q signals will produce an error voltage proportional to
any phaseerror in the VCO. The error voltage controls the phaseand,thus, the frequency of
the VCO.

2-8-3 Flemodulator
A thfud method of achieving recovery of a phaseand frequency coherentcarrier is the re-
modulator, shown in Figure 2-39.The remodulator produces a loop error voltage that is
proportional to twice the phaseerror betweenthe incoming signal and the VCO signal. The
remodulator has a faster acquisition time than either the squaring or the Costasloops.
Carrierrecoverycircuits forhigher-than-binaryencodingtechniquesaresimilarto BPSK
exceptthat circuitsthat raisethereceivesignalto the fourth, eighth,andhigherpowersareused.

2-9 CLOCKHECOVERY
As with any digital system, digital radio requires precise timing or clock synchronization
between the transmit and the receive circuitry. Becauseof this, it is necessaryto regenerate
clocks at the receiver that are synchronouswith those at the transmitter.
Figure 2-40a shows a simple circuit that is commonly used to recover clocking in-
formation from the received data.The recovereddata are delayedby one-half a bit time and
then compared with the original data in an XOR circuit. The frequency of the clock that is
recovered with this method is equal to the received data rate (fr). Figure 2-40b shows the
relationship between the data and the recoveredclock timing. From Figure 2-40b,it can be
seen that as long as the receive data contain a substantial number of transitions (1/0 se-
quences),the recovered clock is maintained. If the receive data were to undergo an ex-
tendedperiod of successiveI s or 0s, the recoveredclock would be lost. To prevent this from
occurring, the data are scrambled at the transmit end and descrambledat the receive end.
Scrambling introduces transitions (pulses) into the binary signal using a prescribed algo-
rithm. and the descramblerusesthe samealsorithm to remove the transitions.

90 Chapter2
 Data

Delayed
d,rt8

Rccovered
clock

(a) (b)

FIGURE2-4O [a) Clockrecoveny

circuit;[b] timingdiagr"am

1 0 1 1 1 00 0 1 1 01

0 -/l /l /+/+/+/+/t /l /+/+/+/+

(referencebitl
18e r8{r. 180' e

tbt

FIGURE241 DBPSKmodulator':
[a) blockdiagnam;[bJtimingdiagram

2-1O DIFFERENTIAL KEYING

PHASE-SHIFT
Dffirential phase-shift keyins (DPSK) is an alternative form of digital modulation where
the binary input information is contained in the difference between two successivesig-
naling elementsrather than the absolute phase.With DPSK, it is not necessaryto recover
a phase-coherentcarrier. Instead, a received signaling element is delayed by one signal-
ing element time slot and then compared with the next received signaling element. The
difference in the phase of the two signaling elements determines the logic condition of
the data.

2-1o-1 DifferentialBPSK
2-10-1-lDBPSKtransmitter. Frgxe2-4Iashowsa simplifiedblockdiagramof a
keying(DBPSK)transmitter.
differentialbinaryphase-shifi An incominginformationbit is

Digital Modulation 91
 Recovered
data

Bdlancod modul6toi output

(+sin or6tl(+sin obtl + - cos 2oct
- 11 |
(-*in <o6tl(-*in oLt) 2oct
-
-r-icos
(-sin <r1t)(+sin or"tl = - * cos Z<o"t
] |
( al

/n n /n,/-\. /a,t---\,/-\_/-\,/-\ /ra /^n

OBFSK 1$' 1800 00 00 0' 0' 180' 00 1800 180. 180" (r 0e
input phase (referencephasel
r rr l l l l l l l l l
Recovered
blt strbam
ttttrtttitii
10 1110 00110 1

(b)

FIGURE
242 DBPSKdemodulaton:
[a) blockdiagnam;
[b] timingsequence

XNORed with the precedingbit prior to enteringthe BPSK modulator (balancedmodulator).

For the fust databit, thereis no precedingbit with which to compareit. Therefore,an initial ref-
erencebit is assumed.Figure 2-4lb showsthe relatiortshipbetweenthe input data,the XNOR
output data,and the phaseat the output of the balancedmodulator.If the initial referencebit is
assumeda logic 1, the output from the XNOR circuit is simply the complementof that shown.
In Figure 2-41b, the first data bit is XNORed with the reference bit. If they are the
same,the XNOR output is a logic I ; if they are different, the XNOR output is a logic 0. The
balanced modulator operatesthe same as a conventional BPSK modulator; a logic I pro-
duces *sin rcl'ctatthe output, and a logic 0 produces -sin or./ at the output.

2-10-l-2 DBPSK receiver. Figure 2-42 shows the block diagram and timing se-
quence for d DBPSK feceiver. The received signal is delayed by one bit time, then com-
pared with the next signaling element in the balanced modulator. If they are the same, a
logic I (* voltage) is generated.Ifthey are different, a logic 0 (- voltage) is generated.If
the referencephaseis incorrectly assumed,only the hrst demodulatedbit is in error. Dif-
ferential encoding can be implemented with higher-than-binary digital modulation
schemes,although the differential algorithms are much more complicatedthan for DBPSK.
The primary advantage of DBPSK is the simplicity with which it can be imple-
mented.With DBPSK, no carrier recovery circuit is needed.A disadvantageof DBPSK is
that it requires between I dB and 3 dB more signal-to-noiseratio to achieve the same bit
error rate as that of absolutePSK.

2-11 TRELLISCODEMODULATION
Achieving data transmissionrates in excessof 9600 bps over standardtelephonelines with
approximately a3-WIz bandwidth obviously requires an encoding schemewell beyond the
quadbitsusedwith 16-PSKor I6-QAM (i.e.,M must be significantly greaterthan 16).As
might be expected,higher encoding schemesrequire higher signal-to-noise ratios. Using
the Shannonlimit for information capacity (Equation 2-4), a data transmissionrate of 28.8
kbps through a32OO-Hzbandwidth requires a signal-to-noiseratio of
(bps) : (3.32x B) loe(l + SN)

92 Chapter 2
 therefore. 28.8kbps: (3.32)(3200)
log(l + S/N)
28,800: 10,624log(1 + SA.{)
28.800
= log(l + S/N)
n,6U

2. 71: log( 1 +SA {)

thus, -1+s/ N
702' 71
513:1+S AI
512: SAI

in dB, S/I{iae): l0log 512

:2'7 dB

Transmissionrates of 56 kbps require a signal-to-noise ratio of 53 dB, which is virtually

impossible to achieve over a standardtelephonecircuit.
Data transmissionratesin excessof 56 kbps canbe achieved,however,over standard
telephonecircuits using an encoding technique called trellis code modulation (TCM).
Dr. Ungerboeck at IBM Zueich ResearchLaboratory developedTCM, which involves us-
ing convolutional (tree) codes, which combines encoding and modulation to reduce the
probability of error, thus improving the bit error performance.The fundamentalidea behind
TCM is introducing controlled redundancy in the bit stream with a convolutional code,
which reducesthe likelihood of transmission errors. What sets TCM apart from standard
encoding schemesis the introduction of redundancy by doubling the number of signal
points in a given PSK or QAM constellation,
Tiellis codemodulationis sometimesthoughtof asa magicalmethodof increasingtrans-
mission bit ratesover communicationssystemsusing QAM or PSK with fixed bandwidths.
Few peoplefully understandthis concept,asmodemmanufacturersdo not seemwilling to share
information on TCM. Therefore,the following explanationis intendednot to fully describethe
processofTCM but ratherto introducethe topic and give the readera basic understandingof
how TCM works and the advantageit hasover conventionaldigital modulation techniques.
M-ary QAM and PSK utilize a signal set of 2N : M, whereN equals the number of
bits encoded into M different conditions. Therefore, N : 2 producesa standardPSK con-
stellation with four signal points (i.e., QPSK) as shown in Figure 2-43a. Using TCM,
the number of signal points increasesto two times M possible symbols for the samefactor-
of-M reduction in bandwidth while transmitting each signal during the sametime interval.
TCM-encodedQPSK is shown in Figure 2-43b.
Trellis coding also defines the manner in which signal-statetransitionsare allowed to
occur, and transitions that do not follow this pattern are interpretedin the receiver as trans-
mission errors. Therefore, TCM can improve error performanceby restricting the manner
in which signals are allowed to transition. For values of N greater thanZ, QAM is the mod-
ulation schemeof choice for TCM; however, for simplification purposes,the following ex-
planationusesPSK as it is easierto illustrate.
Figure 2-44 shows a TCM schemeusing two-state8-PSK, which is essentiallytwo
QPSK constellationsoffset by 45". One four-stateconstellationis labeled0-4-2-6,and the
other is labeled l-5-3-7. For this explanation, the signal point labels 0 through 7 are meant
not to representthe actual dataconditions but rather to simply indicate a convenientmethod
of labeling the various signal points. Each digit representsone of four signal points per-
mitted within each of the two QPSK constellations.When in the 0-4-2-6 constellation and
a0 or 4 is transmitted,the systemremains in the sameconstellation. However, when either
a2 or 6 is transmitted,the systemswitchesto the l-5-3-7 constellation.Once in the l-5-3-7

Digital Modulation 93
 o
AI
I
I
I
I
| \. ono

450
i
I t:_ _--> -a
- o - -- -- -- ->a o<-
'l

I I

i
a
t
a
(a) (b)

FIGURE243 GPSKconstellations:
[a) standardencodingfopmat;[b) trellisencodingfonmat

FIGURE244 8-PSKTCM constellations

constellation and a 3 or 7 is transmitted,the systemremains in the sameconstellation, and

if a 1 or 5 is transmitted, the system switches totJnle0-4-2-6constellation. Remember that
each symbol representstwo bits, so the system undergoes a 45o phase shift whenever it
switches between the two constellations. A complete error analysis of standard QPSK
compared with TCM QPSK would reveal a coding gain for TCM of 2-to-l I or 3 dB.
Table 2-3lists the coding gains achieved for TCM coding schemeswith several different
trellis states.
The maximum data rate achievable using a given bandwidth can be determined by re-
arranging Equation 2- 10:

N X B :f6

94 Chapter 2
 Table 2-3 TnellisCodino Gain

Number of Trellis States Coding Gain (dB)

2 3.0
I
5.5
8 6.0
16 6.5
JZ 7.1
64 73
r28
256 '7.4

where N: number of bits encoded(bits)

B : bandwidth (hertz)
/6 : transmissionbit rate (bits per second)
Remember that with M-ary QAM or PSK systems,the baud equals the minimum re-
quired bandwidth. Therefore, a 3200-Hz bandwidth using a nine-bit trellis code pro-
ducesa 3200 baud signal with eachbaud carrying nine bits. Therefore,the transmission
ratef6: 9 x 3200: 28.8 kbps.
TCM is thought of as a coding schemethat improves on standardQAM by increasing
the distancebetween symbols on the constellation(known as the Euclidean distance).The
first TCM systemused a five-bit code, which included four QAM bits (a quadbit) and a fifth
bit used to help decodethe quadbit. Transmitting five bits within a single signaling element
requiresproducing 32 discerniblesignals.Figure 2-45 showsa 128-pointQAM constallation.

a
'11000( 00 10 1 (
a
1001
001 00 1 1 0 0010111
a a a
10111 r 100001
1 101 1C 11 0001
a a
0 11 0101
1 01 0001 1( 0 1 0
0 11
0001 0 0 0 11 1 1|
rA
a a
II a
00 10 1 1
11 0C 11 1t 0 10
111 0 1 1 0 01 )1 0 | 1 1 0 0 1
a a a a
0 01
) 1 0 10
0010001 1 011011 10
01 001001
01100
c 100 0(
o a a a
101'10001000 ) 1 101
00000
)00001 101 1 1 0)0 100001
10 t1 |
a a a
| 10 (
00 01
1 10 01 1 111 1 1 11 11 1 00 1 1 0 1 1 011 1 0 1 1 1 0 1 0 11 0 1 1 11 ' 1 11 0 1 11
aaaalaaaa
010100001010000101000010100001010010101000010100001010000101000

000000001100110000100011001000001100111010000001001110110000001

0 1 01 1 0 10 01 1 0 1 10 10 1 1 0 000 1 1 1 110 1 0 0 10 00 0 1 11 0 10 1 0 0 10 1

0101000 270 0100000

FIGURE245 1Z8-Point QAM TCM constellation

Digital Modulation 95
 236 224 216 212 2't8 228
aa a a oa
234 205 185 173 164 162 170 181 197 220
aaaa a o o a aa
226 193 16s 146 133 123 121 125 137 154 179 207
a a a aa o a a a aa a
229 189 1s6 131 110 96 87 33 92 100 117 140 172 208
aaaaaa a
a a a aa a a
2 0 1 1 6 0 126 98 79 64 c6 54 62 71 90 112 141 180 221
aaaaaa a a a a oa a a a
222 177 135 102 77 55 41 JC 3 1 37 48 65 9 1 1 18 15 5 1 9 8
aaoaaaa a a a a aa o a o
203 158 119 84 60 39 24 1 7 15 20 30 49 72 101 138 182 230
o a aa aa a a a a a ao a a aa
194 148 108 75 50 28 13 b 4 8 21 38 63 93 127 171 219
a a aa aa a o a a a a aa a a a
238 186 142 103 69 43 22 9 1 0 5 16 3 2 5 6 8 5 12 2 1 6 3 2 1 3
aa aa aa a o o a o o a aa a a a
1 90 14 4 10 6 73 25 11 3
45 2 7 1 8 36 59 88 124 166 217
a aa aa o oa oa a a a a aa a
1 99 15 2 11 3 80 52 3 3 19 1 2 1 0 14 2 6 4 2 66 97 134 174 225
a aa aa a aa a oa a a a aa a
210 167 128 94 67 47 34 27 23 29 40 57 81 111 147 187 237
a aa ao oaa OO.. a a ao a
232 183 149 115 89 68 53 46 44 51 61 78 99 132 168 209
a aa aa aaa a aa a a a aa
21 4 17 5 1 3 9 1 1 6 95 82 74 70 76 86 104 129 157 195 235
aa aa aoa a ao a a a aa
205 176 150 1 3 0 1 1 4 1 07 1 05 1 09 1 2 0 1 36 161 191 227
aaa aaa aaaa aaa
215 184 169 153 145 143 151 159 178 202 231
aoaao a a a aa a
233 211 200 192 188 196 204 223
ooaa aaaa

FIGURE246 One-founthof a 96GPoint GAM TCM constellation

A 3200-baudsignal using nine-bit TCM encoding produces512 different codes.

The nine data bits plus a redundant bit for TCM requires a 960-point constellation. Figure
2-46 shows one-fourth of the 960-point superconstellation showing 240 signal points.
The full superconstellation can be obtained by rotating the 240 points shown by 90',
180o,and 270".

2-12 PROBABILITY
OF ERRORAND BIT ERRORRATE
Probability of error P(e) and bit error rate (BER) are often used interchangeably, al-
though in practice they do have slightly different meanings. P(e) is a theoretical (math-
ematical) expectation of the bit error rate_for a given system. BER is an empirical (his-
torical) record of a system's actual bit error performance. For example, if a system has a
P(e) of 10-s, this means that mathematically you can expect one bit error in every
100,000 bits transmitted (l/10-s : l/100,000). If a system has a BER of 10-5, this
means that in past performance there was one bit error for every 100,000 bits transmit-
ted. A bit error rate is measured and then compared with the expected probability of er-
ror to evaluate a system's performance.

96 Chapter 2
 power ratio (or,morespecif-
Probabilityof erroris a functionof thecarrier-to-noise
ically,the averageenergyper bit-to-noisepowerdensityratio) andthenumberof possible
encodingconditionsused(M-ary).Carrier-to-noise powerratio is the ratio of the average
carrier power (the combined power of the carrier and its associatedsidebands)to the
thermalnoisepowen Carrierpower can be statedin wattsor dBm, where

lOlogm
C14r^): (2-28)

Thermalnoisepoweris expressed
mathematically
as
N: KTB (watts) (2-2e)
where N : thermal noise power (watts)
K : Boltzmann's proportionality constant (1.38 X 10-23 joules per kelvin)
7 = temperature(kelvin: 0K: -273 C, room temperature: 290 K)
B : bandwidth (hertz)

Stated in dBm, N14n.):

fOfoSffi (2-30)

Mathematically, powerratio is
the carrier-to-noise
C C=
(unitlessratio) (2-3r)
N KTB
where C : carrierpower (watts)
N: noisepower(watts)

Stated in dB.
f,ru"l:10loef,
: C(as-)- N<*-l (2-32)
Energyperbit is simplytheenergyof a singlebit of information.Mathematically,
en-
ergyper bit is
Eb: CTbglbit) (2-33)
where E6: energ! of a single bit (joules per bit)
T6 : time of a single bit (seconds)
C : carrier power (watts)

Statedin dBJ. Eo@"o: 10 log E6 (2-34)

andbecauseTt : tlft, wheref6 is thebit ratein bits per second,E6 canbe rewrittenas

(2-3s)
",:;(rybit)
Staredin dBJ. Et4et):tOrcCl (2-36)

: 10logC- l 0l ogf6 (2-37)

Noise power density is the thermal noise power normalized to a 1-Hz bandwidth (i.e.,
the noise power presentina l-Hz bandwidth). Mathematically, noise power density is

DigitalModulation 97
 Nr: + (w/H") (2-3s)
" B'

where Ne : noise power density (watts per hertz)

N: thermal noise power (watts)
B : bandwidth (hertz)

N
Statedin dBm, Nqao-;: l0log - 10log B (2-3e)
,r,
: Nras-) - 10 log B (2-40)
CombiningEquations2-29 and2-38yields
KTB :
&: KT(WlHz) (2-4r)
"
Stated in dBm. : lolog#.
Nqanm) lologT (2-42)

Energy per bit-to-noise power density ratio is used to compare two or more digital
modulation systems that use different transmission rates (bit rates), modulation schemes
(FSK, PSK, QAM), or encoding techniques (M-ary). The energy per bit-to-noise power
density ratio is simply the ratio of the energy of a single bit to the noise power present in
1 Hz of bandwidth. Thus, E6lNs normalizesall multiphase modulation schemesto a com-
mon noise bandwidth, allowing for a simpler and more accurate comparison of their error
performance.Mathematically, E6lN s is

Eu
-C / fo - C B (2-43)
No NIB Nft
where E6l,lsis the energy per bit-to-noise power density ratio. RearrangingEquation 2-43
yields the following expression:
E"CB
-- - -::- x- (2-44)
NoNf u
where E/No: energyperbit-to-noisepowerdensityratio
Cft,{: carier-to-noisepowerratio
B/f6: noisebandwidth-to-bit rateratio

: 10los * ror"tt- (2-4s)

ft to"l
Statedin dB.
fi
or : l0log Eb- lDlogNe Q'46)
From Equation2-44, it can be seenthat the ErlNsratro is simply the product of the carrier-
to-noise power ratio and the noise bandwidth{o-bit rate ratio. Also, from Equation 2-44,it
can be seenthat when the bandwidth equals the bit rate, E/No: 9111J.
In general, the minimum carrier-to-noisepower ratio required for QAM systemsis
less than that required for comparablePSK systems,Also, the higher the level of encoding
used (the higher the value of trf), the higher the minimum carrier-to-noise power ratio. In
Chapter 24, sevetal examples are shown for determining the minimum carrier-to-noise
power and energy per bit-to-noise density ratios for a given M-ary systemand desiredP(e).

Example 2-15
For a QPSK systemand the given parameters,determine
a. Carrierpowerin dBm.
b. Noisepowerin dBm.

9B Chapter 2
 c. Noisepowerdensityin dBm.
d. Energyperbit in dBJ.
e. Carrier-to-noise
powerratio in dB.
f. ErlNoratio.
C : l0-r2 w ,fa= 60 kbps
N :t.2 x10 -14 W B : l 2}k H z
Solution a. The carrierpowerin dBm is determinedby substitutinginto Equation2-28:

10- 12
C = 10 toe : -90 dBm
0;01

b. The noise power in dBm is determined by substituting into Equation 2-30:

I t v 1n-14
N = l0 toe:;oo- = - 10e.2dBm

c. The noisepowerdensityis determinedby substitutinginto Equation2-40:

No = -109.2 dBm - l0 log 120kHz : - 160dBm
d. The energyperbit is determinedby substitutinginto Equation2-36:
10 -1 2
4 = l0 toe : - 167.8dBJ
60 kbp.
powerratio is determinedby substitutinginto Equation2-34:
e. The carrier-to-noise
c = l 0 -r2
lo tot : 19.2dB
t ,, , ,0."
f. The energyperbit-to-noisedensityratiois determined
by substitutinginto Equation2-45:
Eu
1 9 '2 + rorog = 2z.2dB
No= ffi

2-13 ERRORPERFORMANCE
2-13-'l PSK Error Performance
The bit error performancefor the various multiphase digital modulation systemsis directly re-
lated to the distancebenveenpoints on a signal state-spacediagram. For example, on the siglal
state-spacediagram for BPSK shown in Figare 2-47a, it can be seenthat the two signal points
(logic I and logic 0) have maximum separation(d; for a given power level (D). In essence,one
BPSK signal state is the exact negative of the other. As the figure shows, a noise vector (V,y),
when combined with the siglal vector (V"), effectively shifts the phaseof the signaling element
(VsD alpha degrees.If the phase sffi exceeds +90o, the signal element is shifted beyond the
threshold points into the error region. For BPSK, it would require a noise vector of suffrcient
amplitude and phaseto produce more than a -f 90o phaseshift in the signaling element to pro-
duce an enor. For PSK systems,the generalformula for the threshold points is

T P: + a (2-47)
M
where M is the number of signal states.
The phaserelationship between signaling elementsfor BPSK (i.e., 180" out of phase)
is the optimum signaling format, referred to as antipodal signaling, and occurs only when
two binary signal levels are allowed and when one signal is the exact negative of the other.
Becauseno other birby-bit signaling schemeis any better, antipodal performance is often
used as a referencefor comparison.
The error performance of the other multiphase PSK systemscan be compared with
that of BPSK simply by determining the relative decreasein error distancebetween points

Digital Modulation 99
 +Thr-lpld
Polnt

togts0lS vs ebe|cl

-ff t+rm*rotCeotlo

(bl

FIGURE247 PSKemor region:tal BPSK;(b) OPSK

on a signal state-spacediagram. For PSK, the general forrrula for the maximum distance
I between signaling points is grven by

I
. 360" dl2
s rnu =s n M= T Q-48)
where d: errordistance
M: numberofphases
D: pe*sipalamplitude
RearangingEquation2-48 aad,solving for d yields '

o -- (rrio $, ) r a (24s)
\ hI /
T Figure247b showsthe signalstate-spaoe diagramfor QPSK.From Figurp247 mdfuile-
tion 248, it canbe seenthatQPSKcantolerafeonly a t45' phaseshift FromEquation247,

100 Ghapter 2
 the maximum phaseshift for 8-PSK and 16-PSK is -r 22.5oand -r 11.25", respectively.Con-
sequently,the higher levels of modulation (i.e., the greaterthe value of M) require a greater
energy per bit-to-noise power density ratio to reducethe effect ofnoise interference.Hence,
the higher the level of modulation, the smaller the angular separation between signal points
and the smaller the error distance.
The general expressionfor the bit error probability of an M-phase PSK system is

P(r) : ff i*" ' (2-s0)

where erf : error function

z : sin(n
f M) (t/-toeru) O/8, t nl
By substituting into Equation2-50, it can be shown that QPSK provides the sameer-
ror performance as BPSK. This is becausethe 3-dB reduction in error distance for QPSK
is offset by the 3-dB decreasein its bandwidth (in addition to the error distance,the rela-
tive widths of the noise bandwidths must also be considered).Thus, both systemsprovide
optimumperformance.Figure2-48 showstheerrorperformancefor2-,4-,8-,16-, and 32-PSK
systemsas a function of E6lNs.

1 2 13 14 15
Eu/No (dB)

FIGURE248 En'onrates of PSKmodulation

systems

Digital Modulation 101

 Example2-16
Determine the minimum bandwidth required to achieve a P(e) of 10-7 for an 8-PSK system operat-
ing at 10 Mbps with a carrier-to-noisepower ratio of 11.7 dB.
Sof ution From Figure 2-48, the minimum EulN6ratio to achieve a P(e) of l0-7 for an 8-PSK sys-
tem is 14.7 dB. The minimum bandwidth is found by rearranging F-qtation2-44;
B _ -E , _C
fr N oN
: 14.7dB - 11.7dB : 3 dB
R
i :a nti l og3 = 2
It
B :2 x 1 0Mbp s:2 0MH z

2-13-2 QAM Error Performance

For a large number of signal points (i.e., M-ary systemsgreaterthan 4), QAM outperforms
PSK. This is becausethe distance between signaling points in a PSK system is smaller
than the distance between points in a comparable QAM system. The general expression
for the distancebetween adjacent signaling points for a QAM systemwith L levels on each
axis is
Y2 ,o ( z - sr \
a:
L- |
where d: error distance
I : number oflevels on each axis
2 : peak signal amplitude

In comparing Equation 2-49 to Equation 2-5l,it canbe seenthat QAM systemshave

an advantageover PSK systemswith the samepeak signal power level,
The general expression for the bit error probability of an L-level QAM system is

I l L- t\
P(,):,"r/\ Q'sz\
. )ertc(z)
where erfc(z) is the complementary error function.

Ylog2L lEu
t:
t- l lNo
Figure 2-49 shows the error performance for 4-, 16-,32-, and 64-QAM systemsas a func-
tion of E6lN6.
Table 2-4lists the minimum carrier-to-noisepower ratios and energy per bit-to-noise
power density ratios required for a probability of error 10-6 for several PSK and QAM
modulation schemes.

Example2-16
Which systemrequiresthe highest E6lNsratiofor a probability of error of 10-6. a tour-levelQAM
system or an 8-PSK system?
Solution From Figure 2-49, the minimum E6lNs fttio required for a four-level QAM system is
10.6 dB. From Figure 2-48, the minimum E6lNs ratro required for an 8-PSK system is 14 dB.
Therefore, to achieve a P(e) of 10-6, a fourJevel QAM system would require 3.4 dB less E6lNe ratio.

1OP Chapter 2
 12 13 la 15 16 17 t8 19
Eb/ttoldB)

FIGUHE249 Epnornatesof GIAMmodulationsystems

of VariousDigitalModulation
Table24 PerfonmanceCompanison
Schemes(BEH- 1O-61

BPSK 10.6 10.6

QPSK r3.6 10.6
+QAM 13.6 10.6
8-QAM r7.6 10.6
8-PSK lE.5 t4
I6PSK 24.3 18.3
lGQlv m.5 14.5
32-QAM 24.4 17.4
64-QAM 26.6 18.8

'l03
 to-2

to-3 Noncoherent

Coherent

P ''o'

to-5

to+
11 1 3 14

FIGURE2-5O Ennon
natesfor FSKmodulation
svstems

2-13-3 FSK Error Performance

The error probability for FSK systems is evaluated in a somewhat different manner than
PSK and QAM. There are essentiallyonly two typesof FSK systems:noncoherent(asyn-
chronous) and coherent(synchronous).With noncoherentFSK, the transmitter and receiver
are not frequency or phasesynchronized.With coherentFSK, local receiver referencesig-
nals are in frequency and phaselock with the transmitted signals. The probability of error
for noncoherentFSK is

P(,)::"-r(-r+) (2-s3)

The probability of error for coherentFSK is

tE^
P( e): e rfc VNo (2-s4)

Figure 2-50 shows probability of error curves for both coherent and noncoherentFSK for
severalvalues of E /No. From Equations 2-53 and 2-54, it can be determinedthat the prob-
ability oferror for noncoherentFSK is greater than that ofcoherent FSK for equal energy
per bit-to-noise power density ratios.

OUES TIONS
2-1. Explain digital transmission and digital radio.
2-2. Defrne information capacity.
2-3. What are the three most predominant modulation schemesused in digital radio systems?

1o,4 Chapter 2
 2-4. Explain the relationship between bits per secondand baud for an FSK system.
2-5. Define the following terms for FSK modulation: frequency deviation, modulation index, and
deviation ratio.
2-6. Explain the relationship between (a) the minimum bandwidth required for an FSK system and
the bit rate and (b) the mark and space frequencres.
2-7. What is the difference between standardFSK and MSK? What is the advantaeeof MSK?
2-8. Define PSK.
2-9. Explain the relationship between bits per secondand baud for a BPSK system.
2-10. What is a constellation diagram, and how is it used with PSK?
2-11. Explain the relationship between the minimum bandwidth required for a BPSK systemand the
bit rate.
2-12. ExpIainM-ary.
2-13. Explain the relationship between bits per secondand baud for a QPSK system.
2-14. Explain the significance of the I and Q channelsin a QPSK modulator.
2-15. Defrne dibit.
2-16. Explain the relationship between the minimum bandwidth required for a QPSK system and the
bit rate.
2-17. 'V'lhatis a coherent demodulator?
2-18. WhatadvantagedoesOQPSKhaveoverconventionalQPSK?WhatisadisadvantageofOQPSK?
2-19. Explain the relationship between bits per secondand baud for an 8-PSK system.
2-20. Define tribit.
2-21. Explwn the relationship between the minimum bandwidth required for an 8-PSK system and
the bit rate.
2-22, Explain the relationship between bits per secondand baud for a l6-PSK system.
2-23. Defrne quadbit.
2-24. Defne Q{14.
2-25. Explun the relationship between the minimum bandwidth required for a 16-QAM system and
the bit rate.
2-26. What is the difference between PSK and QAM?
2-27. Defne bandw idth effi ciency.
2-2E. Define carier recovery.
2-29. Explain the differencesbetween absolutePSK and differential PSK.
2-30. What is the purpose of a clock recovery circuit? When is it used?
2-31. What is the difference between orobabilitv of error and bit error rate?

PROB LE M S
2-1. Determine the bandwidth and baud for an FSK signal with a mark frequency of 32 Wlz, a
spacefrequency of 24 kHz, and a bit rate of 4 kbps.
2-2. Determine the maximum bit rate for an FSK signal with a mark frequency of 48 kHz, a space
frequency of 52kHz, and an available bandwidth of l0 kHz.
2-3. Determine the bandwidth and baud for an FSK signal with a mark frequency of 99 kHz, a
spacefrequency of 101 kHz, and a bit rate of 10 kbps.
2-4. Determine the maximum bit rate for an FSK signal with a mark frequency of lo2kllz, a space
frequency of 104 kHz, and an available bandwidth of & kHz.
2-5. Determine the minimum bandwidth and baud for a BPSK modulator with a carrier frequency
of 40 MHz and an input bit rate of 500 kbps. Sketch the output spectrum.
2-6. For the QPSK modulator shown in Figrrez-t7, changethe *90' phase-shiftnetwork to -90"
and sketch the new constellation diagram.
2-7. For the QPSK demodulator shown in Fi gure2-21, determinethe I and Q bits for an input signal
of sin or"r - cos oct

Digital Modulation 1O5

 2-8. For an 8-PSK modulator with an input data rate (f) equalto 20 Mbps and a carrier frequency
of 100 MHz, determine the minimum double-sided Nyquist bandwidth (f,,) and the baud.
Sketch the output spectrum.
2-9. For the 8-PSK modulator shown in Figure 2-23, changethe referenceoscillator to cos roctand
sketch the new constellation diagram.
2-10. For a 16-QAM modulator with an input bit rate (f) equal to 20 Mbps and a carrier frequency
of 100 MHz, determine the minimum double-sided Nyquist bandwidth Qlv) and the baud.
Sketch the output spectrum.
2-11' For the 16-QAM modulator shown in Figure 2-33, change the reference oscillator to cos oct
and determine the output expressionsfor the following I, I', Q, and Q' input conditions: 0000,
11 11 ,10 10 ,a nd0 10 1 .
2-12. Determine the bandwidth efficiency for the following modulators:
a. QPSK,/6 : l0 MbPs
b. 8 - PSK, 6 :21 M b ps
c. 16-QAM,/6 : 20 Mbps
2'13. For the DBPSK modulator shown in Figure2-4}a, determinethe output phasesequencefor the
followinginputbitsequence:00110011010101(assumethatthereferencebit: l).
2-14. For a QPSK system and the given parameters,determine
a. Carrier power in dBm.
b. Noise power in dBm.
c. Noise power density in dBm.
d. Energy per bit in dBJ.
e. Carrier-to-noise power ratio.
f. EulN6rauo.
c : 1o-r3 w ,fa : 3o kbps
N: 0 .0 6 x 1 0- rs w B =6 0 k H z
2-15. Determine the minimum bandwidth required to achieve a P(e) of 10-6 for an 8-PSK system
operating at 20 Mbps with a carrier-to-noisepower ratio of 11 dB.
2-16. Determine the minimum bandwidth and baud for a BPSK modulator with a carrier frequency
of 80 MHz and an input bitratef6: I Mbps. Sketch the output spectrum.
2-17. For the QPSK modulator shown in Figure2-17, change the reference oscillator to cos toct and
sketch the new constellation diagram.
2-18. For the QPSK demodulator shown in Figtne2-21, determine the I and Q bits for an input sig-
nal -sin oct + cos oc/.
2-19. For an 8-PSK modulator with an input bit l.latefb : 10 Mbps and a carrier frequencyf, : 80
MHz, determine the minimum Nyquist bandwidth and the baud. Sketch the output spectrum.
2-20. For the 8-PSK modulator shown in Figure 2-23, change the *90o phase-shift network to a
-90o phase shifter and sketch the new constellation diagram.
2-21. Foral6-QAMmodulatorwithaninputbitratef = l0Mbpsandacarrierfrequencyf":60MHz,
determine the minimum double-sided Nyquist frequency and the baud. Sketch the output spectrum.
2-22. For the 16-QAM modulator shown in Figxe2-33, change the 90" phase shift network to a - 90o
phaseshifter and determine the output expressionsfor the following I, I', Q, and Q' input con-
ditions:0000,1111,1010,and 0101.
2-23. Determine the bandwidth effrciency for the following modulators:
a. QPSK,/6 : 20 MbPs
b. 8-PSK,' : 28 Mbps
c. I6-PSK,/6 : 40 Mbps
2-24, FortheDBPSKmodulatorshowninFigure2-4la,determinetheoutputphasesequenceforthe
following input bit sequence:11001100101010(assumethat the referencebit is a logic 1).

106 Chapter 2