0% found this document useful (0 votes)
20 viewsdigital modulation unit 3
Chapter 8 Digital Modulation
Uploaded by
ssg8348Copyright
© © All Rights Reserved
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
20 viewsdigital modulation unit 3
Chapter 8 Digital Modulation
Uploaded by
ssg8348Copyright
© © All Rights Reserved
Available Formats
Download as PDF or read online on Scribd
/ 58
Digital Modulation
We have discussed the baseband digital data transmission schemes in the previous chapter,
where signals were transmitted directly without any frequency translation. Baychand
signals have frequency components and power that are significant at low frequencies.
Such signals are suitable for transmitting through a pair of wires or coaxial cables.
However, baseband signals cannot be transmitted over micorwave link or satellite link.
With transmitting antennas of practical size, radiation of low-frequency spectrum will be
very much inefficient. Telephone channel, even if it is available for data transmission,
is primarily designed for transmission of speech signal which has spectral content from
about 200 Hz to about 3.2 kHz. Since baseband data signals have significant low-frequency
components below 200 Hz, a telephone line also cannot be used for transmission of signals
without modulation. Advantages of optical fibre communication can be exploited if large
number of data signals can be multiplexed forming a combined signal of larger bandwidth,
Inall these cases, itis necessary to shift the spectrum of baseband signal to high-frequency
range and that can be done by modulating the carrier of appropriate frequency by the
baseband signal.
The different types of basic modulation schemes used in transmission of digital data
are (i) varying the amplitude of carrier in accordance with data signal, (ii) varying the
phase of the carrier in accordance with data signal, and (ii) varying the frequency of the
aarrier in accordance with the data signal. These schemes are generally called as amplitude
modulation (AM), phase modulation (PM) and frequency modulation (FM), respectively.
When the data isa binary sequence, the amplitude in AM or phase in PM or frequency in
FM switches between two distinct values corresponding to symbol 0 and symbol 1. These
special cases are called as amplitude shift keying (ASK), phase shift keying (PSK) and
frequency shift keying (FSK), respectively.
For data transmission at higher rates through band-limited channels, combined
modulation schemes are also used. Examples are amplitude phase keying (APK) and
{uadrature amplitude modulation (QAM). | in
Using signal constellation diagrams, the performance of these modulation schemes
the presence of noise can be studied and probability of symbol error can be computed.
311 312 Digital Communications
8.1. Digital Modulation Formats
In digital communications, the digital data is usually in the form of binary: It can also
be a block of n binary bits encoded into M = 2" possible discrete amplitudes, or ay
discrete phases or M discrete frequencies. Such schemes are generally called as M-ary
signalling. M = 2 corresponds to binary with symbols being 0 and 1. As discussed earlier,
some characteristics of carrier, like amplitude, phase or frequency, is varied in accordance
with the modulating signal, With a sinusoidal carrier, the modulation produces discrete
changes in amplitude, phase or frequency of the carrier Sometimes, combination of both
amplitude and phase of the carrier may also be changed.
The three basic binary modulation schemes are ASK, PSK and FSK. In ASK the
amplitude of carrier will have a value say A volts for binary symbol 1 and a value zero
for symbol 0. It is equivalent to switching a sinusoidal oscillator on and off and therefore
this modulation process is also called as on-off keying.
On the other hand, if the carrier amplitude and frequency are fixed but the phase has
value zero radians for symbol 1 and x radians for symbol 0, the modulation process is
called PSK or binary PSK (BPSK). Therefore in BPSK, the phase of the carri
between 0 radians for symbol 1 and 7 radians for symbol 0.
In FSK, the carrier amplitude is fixed but the frequency has a value fo Hz. for symbol 0
and a different value fy Hz for symbol 1. The instantaneous frequency of the carrier switches
between fy Hz and fy Hz corresponding to symbols 0 and 1. Figure 8.1 gives ASK, PSK and
FSK waveforms.
PSK and FSK waveforms have constant amplitudes and hence the information resides
only in the phase or frequency of the carrier, respectively. Hence, in these modulation
schemes, the information in the transmitted signals are not affected by amplitude
nonlinearities of the channel, which are normally present in microwave or satellite links.
Demodulation is the inverse process of recovering the original binary waveform from
the received modulated signal whereas detection refers to the process of symbol decision.
In digital communication, the terms demodulation and detection are used interchangably.
Demodulation can be done using either a coherent receiver or a noncoherent receiver. The
demodulation process then respectively called as coherent detection and noncoherent detection.
In the case of coherent detection, synchronized local reference of the transmitted signals
must be available. That is, the receiver should have the exact knowledge of the carrier
wave’s phase reference. The receiver is then said to be phase locked to the transmitter. In
coherent detection, the received signal is crosscorrelated with each one of the reference
signals and the decision is based on the correlator output.
In noncoherent detection, the knowledge of phase of carrier wave is not needed. Tht:
the complexity of receiver is reduced. However, as we see later, the performance of not"
coherent receivers in the presence of noise is inferior as compared to coherent recieve?
Out of many modulation schemes available, a particular choice is made based on te
following desirable requirements that are me
er switches
(i) Maximum data rate.
(i) Minimum transmitted power. 8. Digital Modulation 313
Binary
data’
'
@
“1
@)
© t
0
sure 8.1 Waveform for (a) amplitude-shif eying (b) phase-shift keying and (c) frequency-shift
keying.
(ii) Minimun
() Maximu:
() Minimu:
(i) Minima
m channel bandwidth.
in immunity to interfering signals
m circuit complexity and hey
nce minimum cost.
m probability
of symbol error,
All these requirements cannot be simultaneously met,
other, However, transmitte
‘d power and bandwidth, w
Communi
Some of these could conflict each
ation system, should be as small we possible,
vhich are primary resources of any
8.2. Coherent Binary Modulation Techniques
one}, binary modulation Schemes are ASK, PSK, and FSK andl we shall discuss them
’Y one,
¥ —— ASK signal =
O00 = [Freox2nf.t
Figure 8.2 ASK modulator.
Fiery, yt,
Simba,
NRZ binary data levels
8.2.1. ASK or ON-OFF Keying
An ASK waveform can be represented by a single basis function defined by
2
Hi) = | zeoseafd, O JEF/2.
Denoting P.(1) as probability of error when symbol 1 is transmitted, we have
Pe) =P fn <
Symbol 1 transmit] (8.16)
FAD is the conditional probability that xy < VE;/2 and that symbol 1 is transmitted.
PA) can be computed using the conditional probability density fumetion f¢, (e1/1) given
equation (8.13). Hence,
iT VEZ
Pel) = Sx, Idx)
No Joo
2 2
_ L re op |- SV (8.17)
VINO J—oo 0
Pu
x1 — VES (8.18)
No 818 Digital Communications
then
dx = Nody (8.19)
Consequently, equation (8.17) becomes
1 pc MavETN A
no=zf expl—y"Idy (8.20)
as even symmetry about y = 0, the integral given by
Since exponential function exp[=y"] hi
(8.20) can be written as
1% 2
PA -z/ exp(—)") dy (8.2
= FB Javea? a 2)
One can easily verify by referring Figure 8.6 that the shaded areas on the left and right
sides denote the integrals in equations (8.20) and (8.21) respectively.
Comparing equation (8.21) with the definition of error function complement given in
Appendix A, we have
1 1 [&
Pe) = gerfe E 2]
Similarly, we can compute the probability of error P,(0) when symbol 0 is transmitted 2s,
(8.23)
P.(0) =P [x > = symbol 0 is transmited]
P,(0) is the conditional probability that x; > /E;/2 and that symbol 0 is transmitted. This
conditional probability can be computed using the conditional probability density function
fx, £110) given in equation (8.14). That
-
P.O) = f «Ge 10)d 2)
Vane ert dey
Equation. (8.20) Equation, (8.21)
——-y
Figure 8.6 A plot of e exp(-y"). ting for fx, (110) from equation (8.14), we get
Substituun
8. Digital Modulation 319
1 00 2
P.(0) = f at |
VIN J vriyy ox al dey (8.25)
Substituting,
A 8.26)
No" (8.
‘equation (8.25) gets transformed as
P.) = [*
Oe Pd) 8.27)
V8 Sayy yen PY (8.27)
Equation (8.27) is identical to equation (8.21). Hence,
Pe(1) = P,(0) =
Weymbol I occurs with a probability P(1) and symbol 0 occurs
with a probability P(0) (we
note P(1) + P(0) = 1), then average probability of symbol en
tor P- is given by
Pe= P(1)Pe(1) + P(O)P.(0) (8.29)
With P() = PC) = 1/2 and Vy, = VE5/2, the average probability of symbol error P, is
E
P, "(3 =| (830)
Thus, itis seen that the average probability of symbol error is a function of ratio of symbol
snergy per bit to twice noise power spectral density. We note that it is independent of
transmitted signal waveform, Referring Figure 8.6, we find that as /E,/No increases the
shaded area under the curve decreases. This in turn means that as -VEs/No increases, the
average Probability of error decreases,
8.2.2. Coherent Binary PSK
'n this Section, the signals for binary PSK, signal space, signal constellation diagram,
‘ecsion regions, Seneration of PSK waveforms, their detection and noise analysis are
Studied, :
inetitary PSK, the signatssy(®)andsp()that ate used to represent binary symbols | and
"SPectively are defined as
2Ey
sil) cos(2rft), O 1/T), and f- is equal to some integer multiple of 1/7,
From equations (8.31) and (8.32), we see that s(Q and s20 are out of phase by 180°, 4
pair of signals which differ only in phase by 180° are referred to as antipodal signals, tps
seen that signals s(1) and so(f) ean be represented in terms of only single basis function
$10 having unit energy. Let us define this basis function by
os(2rfct), O! Region Z, ——>
& i 5
aa —$1
~\E, o Ey
The
point
+VEj and message
VEj. Thus, the si
the decision region Z
Point sy (at Ej). Similarly the deci
of points closest to message point sy (at —/F;).,
‘The signal constellation diagram and the
Poin aing decision regions are shown in Figure 8.8. The distance beteacn message
Points is 2/E;,.
iitision Rule: "Choose signal sy () or binary symbol 1, ifthe received signal point x) falls
inthe region Z). That is, choose symbol 1, if is greater than zero. On the other hand,
Thx Signal s2@ or binary symbol 0, if the received signal point x; falls in region Zo,
‘tis, choose symbol 0, if is less than zero. If xy =, the decision fs arbitrary; we may
tiPa lair coin and take a decision in favour of either symbol | or symbol 0 by preassigning:
head for symbol | and tail for symbol 0.
Generation of Bi :
Shown mn of Binary PSK:
PSK wave can be generated by using a product modulator as
Tepre
n Figure 8,9(a).. The input binary sequence is in NRZ-polar form y
Sented by a rectangular pulse of constant amplitude + /E; and symbol
“*Tecangular pulse of constant amplitude — JE. The other input t
with symbol 1
O represented
(0 the product
a al 322 Digital Communications
Product
/ modulator
5 Vix) ——. X ——* Binary PSK wave
® 40 =F Freos2aht
Sample and reset
ateveryT, Decision
ee ef a ae NS 2 2 0 9 1 —~ Binary wave
x) i Sia pena
oO
®) Vin=0
Figure 8.9 (a) Binary PSK generator. (b) Coherent binary PSK receiver.
modulator is the carrier $1(). The carrier and the timing pulses used to represent the
binary data are derived froma common master clock. The output of the product modulator
is the required binary PSK waveform.
Detection of Binary PSK: Since the information resides in the phase of the carrier, phase
reference must be available at the receiver. The detection process is hence called coherent
detection. The coherent binary PSK receiver is shown in Figure 8,9(b).
Let x(t) be the received binary PSK signal corrupted by AWGN. Also, let $1(0) be the
local coherent reference signal which is synchronized with respect to the frequency and
phase of the carrier used at the transmitter. The correlator output is sampled every 7
seconds and reset. The sample value x) is compared with the threshold, Vi, = OVolts: If
x1 > 0, the receiver decides in favour of symbol 1. On the other hand, if, < 0 the receive
decides in favour of symbol 0. Ifx1 = 0, the decision i trary. .
Decision Error: Decision error can be o
to poise the received signal point x falls on region Zs, and a clecision is made in [2804 ©
sx or eyo Oa i Sisvals2(0 is transmitted but the noise is such that the rece
Bion) Zi,and vectiver widlee Aen A o
symbol 1. 1 ane receiver makes a decision in favour of signal si!
f two types: (i) signal s1 (0) is transmitted but due,
Probability of Error Calculation: ‘The re
w(t) with the assumption that symbol 0
ceived signal x(0) in the presence of AGWN
or s2(0 is transmitted is given by
ADS 2H +UY; 0 Sts, igital Modulation 323 g
= This
elator output all = Ts
th tN
nef sonod= if 620 + wep (Od
hn
n n
= [sodas f wD i(de
0
=x ssa ty
—VEj, the above equation becomes
VE; + w
where 2 is the sample value of the random variable W
with mean zero and vi
variable,
(8.40)
having a Gaussian distribution
ple value of a Gaussian random ’
ble X; is
a
iance No/2. Hence, x, is also a sam
X1. The expected value of this random varial
FIX) = E[-JE, + W,)
(8.41)
Since the expected value of a constant is the constant itself; we can write
FIX] = — VE; + EU)
=-VE+0
=-VE, (8.42)
Abo, the variance of X1 is found as follows.
VarlX1] = Var[—/E, + Ww) (8.43)
Since, the variance of a constant is zero, we can write
Var[X1] = 0+ *
= % (8.44)
AGaussian tandom variable is com
pletely specified by its mean value
the conditional probability density function of random v,
and variance. Hence,
“smitted is given by
ariable X, given that symbol 0 is
1 = + VB)
Sx, 0110) = ex, (8.45)
When on vinyl Ny
ain ante! O is transmitted, an error will occur, if +1 > 0 (that is, xy lies in the region
"ch case a decision is ma
de in favour of symbol 1.
Pe(0) = P(x > O|Symbol 0 is transmitted) (8-46)
Dm |
324 Digital Communications
(0) can be computed by integrating conditional probability density function fy).
given below.
P.(0) = f Sx; 110) de (847
Substituting for fy (e110) from equation (8.45) into equation (8.47), we have,
1 tr _ [ 1 SED a
p Lf exp|-—
PO = eRe fy OP No 648)
Substituting
, x + VE en
1
JNo i
and
dx, = /Nodu (8.50)
we get,
P,(0) = z f exp(w2) du (85)
VEGING
“The right-hand side of equation (8.51) represents the shaded area shown in the right halfof
Figure 8.10. Thus, P.(0) is the area under the tail of Sad exp(—u?) curve from u = JEN
tou = 00,
‘The integral in equation (8.51), which gives P,(0) can be expressed in terms of
complementary error function denoted by erfe (Appendix A). Thus,
1 Ey :
P,(0) = =! (8.52)
oO) zee ( #)
Similarly, if symbol | is transmitted, then the received signal would be
xO =H) +0) 8)
Area equal to
Pay Area equal to
P.O)
Figure 8.10. A plot of Je exp(-u?). a
= This
Coherent detector output at (= Ti
a= [sooo
lo
Z 4 74
i 7
= [o0 +wperoa = fsoH0 + f wba
Sxpssn tu)
8. Digital Modulation 325
(8.54)
(8.55)
From equation (8.38), we have s11 = Ej. Consequently, equation (8.55) becomes
0 38),
= VE +uy
(8.56)
ee a tr
Mean of the random variable X; is
FIX] = ELVE; + 4)
= VE; + EUW)
= VE, +0
=VvEy (8.57)
and the variance of X, is
VarlX1] = Vart VE, + Wy)
= VarlVEy) + Vartiv]
=0+No/2 = Ny/2 (8.38)
fatssian random variable X; is com
pletely described j
hence, the conditional Probability density function of X,
n terms of its mean and variance,
isgiven by
given that symbol 1 is transmitted
; el G1 ~ VE)?
Ji, Gl) = Fam |
N
ansmitted, an error will occur if x; <0 (that is xy li
" MWwhich case a decision is made in favour of symbol 0. The probabilit
transmitted is
(8.59)
ies in the region
ty of error when
Pel) =
K .
cngitlitonal Probability given in equation (8.60) can be expressed in terms of
Nl probability density function fi (x41), That is,
Pix, < O|Symbol 1 transmitted] (8.60)
. 8.61)
P= [found ‘
be
ae 326 Digital Communications
Substituting for fx, (111) from equation (8.59), we have
0 xy — VE)?
PQ) = exp[- Je (8.62)
TaN Jv No
Substituting
(8.63)
and
dxy = /Nodu (8.64)
The integral in equation (8.62) gets transformed as
1 p7vEIR : :
PD = Fe exp(—0?) du (8.65)
00
Thus, P,(1) is the area (shown shaded on the left side) under R exp(—u?) curve bounded
by the limits x = —oo and « = —/E;/No. Since the exponential function, Je Pl")
is symmetrical with respect to u = 0, the two shaded areas in Figure 8.10 are equal. This
allows us to write equation (8.65) as
1 se
P.(1) = =! exp(—u?)du (8.66)
Gf eae
Thus, it follows that
P.(0) = PA) = serfe (367)
If symbols 0 and 1 are equiprobable, that is, P(0) = P(1) = 1/2, we may write the
average probability of symbol error as
1
Pes glPe(O) + Pe] (8.68)
Since, P,(0) = Pe(1), we can write
P_ = P.(0) = P(A) = dente, | # 8.69)
(1) = 5enfey (
It is of interest to note that, whenever the symbols occur with equal probabilities, #N€
aus space gets partitioned in a symmetric manner, Consequently, the condition®
symbol error probabilities and average probability of symbol error will all have the 5"
value. We also note that the average ability of
p note that the average probability of error d ratio of sig
energy per bit to twice noise powe y lepends only on the ratio of sit
¥ spectral density. Also for a gi 1, as si
eigy /E; increases, avera Sig of 7 for a given channel, 4
energy VE; increases, average probability of symbol error P, Gennes
G ases.
ign 8. Digital Modulation 327
hing equation (8.30), we have
Recalling
I
Peask’= ee
1g the above equation with equation (8.69),
te performance, the signal energy per bit in case
rate performs ‘
compared to that of PSK.
we conclude that for the same error
of ASK must be increased 4 times
8.3. Coherent Binary FSk
ction, the signals for binary FSK, sign
generation of FSK signals, coherent detection
| “In binary FSK, the signals s1(0) and s(t),
al constellation diagram, decision regions,
OfFSK signals and noise analysis are studied,
which represent symbols 1
and 0 respectively
soidal signals having two distinct frequencies. That is, for
2E,
Symbol 1: 51(0) VF, tah O
