In this lecture, we evaluate the performance of the detector in

terms of the probability of error in AWGN.

First, we consider binary modulation and, then, M-ary types of
modulation.

Probability of Error for Binary PAM

As discussed
binary PAM
Signals
are
represented as .

of tab
Si ( t ) =
Jeb d(t)

Szct ) =
-

Jeb Alt) of tea

dimensional
or
geometrically as one

vectors ) signal points

:

S, =
Jes Sz
=
-

JEB
,

K
- d ,z= 2ps -01
-

fees Efb
→
Sz
we assume that silt) & Salt) are

likely
equally Sitt was
.

transmitted the
anthem ) ,

demoed
signal
the
received from
.

thatched fillet )
( cohhelation type or
-

is TET
s sit=

TX (Ef Nola)
n =

-1%410 NE) ,

,
 transmitted
similarly
the demod
,
when salt)
output is
was

⑨ th Flo NE )
.

→r sun
-
-
=
-

NC Eb Nok)
-

in two conditional pdfs of rare

Me peelsd
←
* HEE

In this case, the decision rule based on the correlation metric

T

compares r with the threshold zero. If r > 0, the decision is made

in favor of s1(t) and if r < 0, the decision is made that s2(t) was
transmitted.

Given that s1(t) was transmitted, the probability of error is simply

the probability that r < 0; i.e.,

P ( else ) -

foot Ddr
 that set ) was
Similarly if
transmitted
,
we

the
assume

probability of error
,

probability that r > O,

-

the
equals
pigessay
which
=

foo f ( Hsa) dr

QUIET) .

the probability of error

Nino ,
average
Pb = PCs e) Pleb ) . + Pls) P ( 454
since ,
s, It ) & Salt)
Pls, ) =P ( S2) =
are

1/2
equally probable ,

&
Ps = I P Cels , ) +
{ Peeks )
 Ps = a ( 5¥)
We should observe two important characteristics of this
performance measure.

First, we note that the probability of error depends only on the

ratio Eb/N0 and not on any other detailed characteristics of the
signals and the noise.

Second, we note that 2Eb/N0 is also the output SNR from the
matched filter (and correlation-type) demodulator. The ratio Eb/N0
is usually called the signal-to-noise ratio (SNR), or SNR/bit.

PROBABILITY OF ERROR FOR B PPM

two vectors I
en this case
and I are two
the
signal
dimensional and

expressed as

I =
( Eis o
) ,
I =
@ Eef)
Es =

energy for
each
of the waveforms .

.
To evaluate the error
probability
transmitted Then
,
we

assume that I was .

vector at the MP
the received of the

demodulator is
Ti =
( fest hi , nz ]
now ,
we
compute correlation metrics
( ( ti I )
.
and C ( t.SI )

c crisp
-

[ Btn , na .
(9)
=
Ef (Test n ) ,

c. ( ti si )-
-

faith , niffy )
=

ffbhz
Then the peels ,) is the probability
that C ( ti .SI) > CCR I ) .
.

.
te .

P ( els
)
,
=
p[ Ctr -547 P ( EST ) )
=
PUTIN fastest >

pleb ) ,
=P [ Nz -

ni ) > ⑨ ]
Since n1 and n2 are zero-mean, statistically independent Gaussian
I

random variables, each with variance N0/2, the random variable x =

n2 − n1 is zero mean Gaussian with variance N0. Hence,
 P ( nz n > B)
¥* /egg %No
-
= '
i
E dn

at fa÷ EM '
d n

=
Q ( Tmo )

=
Due to symmetry, the same error probability is obtained when
we assume that s2 is transmitted.

Consequently, the average error probability for binary

orthogonal signals is

Pb=Qf1aE#
If we compare the probability of error for binary antipodal
.

signals with that for binary orthogonal signals, we find that

orthogonal signals require a factor of two more energy to
achieve the same error probability as antipodal signals.

Since 10 log10 2 = 3 dB, we say that orthogonal signals are 3-dB

poorer than antipodal signals.

The difference of 3 dB is simply due to the distance between the

two signal points, which is (d12)^2= 2Eb for orthogonal signals,
whereas (d12)^2= 4Eb for antipodal signals.
The error probability versus 10log10 Eb/N0 for these two types of
signals is shown below.

As observed from this figure, at any given error probability, the Eb/
N0 required for orthogonal signals is 3-dB more than that for
antipodal signals.

Fig .

Probability of
Error
for BPA Mf BPPM
the of
probability two error
for binary
commune

scheme with equiprobable messages

be
expressed in terms of the
may
distance b In the two signals If I

P*
-

-
Q
(TIE)
this expression is valid for any binary
comma scheme with two equiprobable signals .

For BPAM d, 2 = 2
@ &
Pb = Q
=Q(T¥o)
for BPPM diz -

JI
& Pb Q ( JIE) .
-

QUIET)
- x -
 PROBABILITY OF ERROR FORM
Arf
-

PAM
h
,

Smit ) = Am
Tey dit) o.at 'T

M 1,2 ng
Q (t )
-

Sm
-
-
-

Sm =
Am
Egf . Am
-

Ym -
I -
M
)
& Em =

fotsmetsdt =

ltmeg
Eg Energy of
-

pulse basic .

Distance bln two adjacentpoints 29g

=

is calculated
Average signal
as
energy
Eav -
Ig I. Em =

MII Eg
-

The average probability of error for M-ary PAM can be determined

from the decision rule that maximizes the correlation metrics.

Equivalently, the detector compares the demodulator output r with a

set of M − 1 thresholds, which are placed at the midpoints of
successive amplitude levels, as shown below.

Thus, a decision is made in favor of the amplitude level that is closest

to r .
threshold
←

-÷'¥.?n .
Em
-

Sm smh Sm -12 Sm -13 Smite

signalpoints k¥0
← .
If the mth amplitude level is transmitted, the demodulator output is

S =
Sm th = Am Egf the
[ N( O Nok )
,

On the basis that all amplitude levels are equally likely a priori,
the average probability of a symbol error is simply the probability
that the noise variable n exceeds in magnitude one-half of the
distance between levels.

However, when either one of the two outside levels ±( M − 1) is

transmitted, an error can occur in one direction only.

fyefjfhfsecantwoaurymib.ms
any
direction

- ÷
#T
C M 1)
-

T t CM -

)
t

For these
occurs
end
in one
signals
direction
,
error

Thus
ave.ra.ge probability of symbol erroris

Piya MMI p [ Ir sent fog ] -

>

M 2h42 .iq
egg
=
.
2 du

m-tf.gg#eH2dn
(M )
2
-
t
=
114=24%-015,9777
;t÷ )
Eav = K Eban
.

=CogzM ) Ebav

pm =
24ft a
/
Es,%-
SNR

bill
=
average

,
We also observe that
the SNR/bit increases
by over 4 dB for
every factor of two
increase in M.

For large M, the

additional SNR/bit
required to increase
M by a factor of two
approaches 6 dB
PROBABILITY OF FOR M ORTHOGONAL
ERROR
any
-

SIGNALS :

the vector space representation of May

signals given by
is
ppm

ST -

[Ef O - - -
O
]
5
[
-

o Rs O -
-
o )
?
.

Im =
CO .
O .
-
-
-

. @]

in
signals have equal energy Es .

To evaluate the probability of error, let us suppose that the signal s1

-

is transmitted. Then the received signal vector is

Mia )
I =
( ⑨ + hi ,
Nz -
-
-

i id oils
,
NCO , Nok )
For equal energy orthogonal signals, the optimum detector selects
:

the signal resulting in the largest crosscorrelation between the

received vector r and each of the M possible transmitted signal
vectors {s_m}, i.e.
ta
I Im m 1,2
-
-

Tism )
-

C(
-

= ,

thus
cct.si/=Es(Estni)
 C. ( I SI ),
=
Esf Nz

:
'

C Ct , In ) = Esf nM
be eliminated
Ef may
the common
factor
correlation metrics After
from all
.

correlate Hp
this ,
the pdf of the first
( H = Test ni ) is (n '
-

⑤ Halo
e-
to ,
Chi ) =

f M I
Costella
pdf of other
-

and the

ftp.elnml/xioM--
outputs are

frm ( nm ) 2,3 M
-

.
-
y

correct decision is
Now the probability of r is than
larger
,

that
the probability other M -
,

I outputs
the
each of is
" 2
Mz nm .
This probability
Iri ) frit)dr
-
-

Eg ( nun Amar , ,
Pc =
, nor , .
.
-

-
Es Kees K
logs M
-
-

,
-
 K 77 I
g

Tl
Probability of bit error for coherent detection of orthogonal signals.
The graphs of the probability of a binary digit error as a function of
the SNR/bit, Eb/N0, are shown in Figure for M = 2, 4, 8, 16, 32,
64.

This figure illustrates that by increasing the number M of waveforms,

one can reduce the SNR/bit required to achieve a given probability
of a bit error.

For example, to achieve a Pb = 10^5, the required SNR/bit is a little

more than 12 dB for M = 2, but if M is increased to 64 signal
waveforms (k = 6 bits/symbol), the required SNR/bit is
approximately 6 dB.

Thus, a savings of over 6 dB (a factor of four reduction) is realized

in transmitter power (or energy) required to achieve a Pb = 10^5 by
increasing M from M = 2 to M = 64.