Iynai ommunications 3- 36 Baseband Digital Transmission -|

Examples for Understanding

Example 3.5.2 A bipolar binary signal is applied as an input to an integrator and dump
receier such that s1 (0) = +A and s2 (t) = -A volts, for the duration of 0 to T sec. If

P(s)= P(s2) =0.5, =10 W/Hz, A = 10 mV and transmission data rate of

2
10 kbits/sec, ) find Probability of error P.
i) lf the bit rate is increased to 10 Mbits/sec, what value of A is needed to attain the same
value of Pe as in part i).
Solution: Given data:

psd of white noise, O = 10W/Hz

2

amplitude, A 10x10 v
1
data rate =10x10°. Hence T, =-
10x103
Probability of occurrence of both the symbols is equal, i.e. 0.5.
i) To obtain probability of error Pe
Equation 3.5.22 give probability of error of integrate and dump receiver as,

P, A2T
No
Putting values in above equation,

(10x103) 1
P2erte2x10 x10x10 , Here T=-
10x103

erfc 5
ii) To obtain 'A' for bit rate of 10 Mbits/sec
The probability of error is to be maintained same. i.e,
1 A2
it v-
2 2x10-9 x10x10
Here T = .
10x106
A2
2x10x10x106
A V0.1 =0.3162 volts
Thus the amplitude must be increased to 0.3162 volts
to maintain same
error.
probability

TECHNICAL PUBLICATIONS An up thrust

for knowledge
Cquation
equatiOn 3.7.7I
3.7.7J gn
give the
response of the matched filter. impulse

Example for UnderstandingJ

Example 3.7.1 A
polar NRZ waveform is to be received by a matched filter.
Binary 1 is
Tepresented by rectangular positive pulse and binary zero is
a
represented by a rectangular
negative pulse. Find out the impulse response of the matched filter and sketch it.
Solution: Let xj (t) represent the
positive rectangular pulse whose duration is T as
shown in Fig. 3.7.1 (a). Let x» (t)
represent a negative rectangular pulse whose duration
isalso T shown in 3.7.1 (b). i.e.,
as Fig.
1 (t) = +A for 0StsT
.. (3.7.8)
and (t) =
-A for 0stsT
2

L e t us calculate the difference signal x (t). i.e,

x (t) = X1 (t) - X2 ( ) for 0stST

x (t) =A -(-A)
x (t) = 2A for 0stST (3.7.9)

- An up thrust for knowledge

TECHNICAL PUBLICATIONS
Fig. 3.7.1 (a) The signal pulse (t)
representing binary '1'. It is given by
equation 3.7.8

X(t)
Fig. 3.7.1 (b) The signal pulse x2 (t)
representing binary '0'. It is given by
equation 3.7.8

x(t)=x,(t)-x2(1)
2A

Fig. 3.7.1 (c) The difference of

x (t) and x2 (t). It has the amplitude of
2A. It is given by equation 3.7.9

t
T

x(-t)
2A
Fig. 3.7.1 (d) The difference signal fold=
A in time. It also has amplitude 2A. t is
given by equation 3.7.10
t

A X(T-t)
2A Fig. 3.7.1 (e) The difference signal
x (-t) delayed by T. It is same as thee
input signal x (t) of Fig. 3.7.1 (c). This
S G iespuuse oi ne matcned filter

t
DIgitar C o m u n a I O I S ' * *

This difference signal is shown in Fig. 3.7.1 (c). The time reversed version of r(t) wil
wil
be

a(-1)=21 for -T sI<0 (3.7.10

Note that in the above equation, the time reference is inverted. This signal is show

in Fig. 3.7.1 (d).

N o w let us delay x(1} by "T seconds.
r(T - 1) = 2A for 0 IST ... (3.7.11

The signal is shown in Fig. 3.7.1 (e).

Impulse response of a matched filter is given by equation 3.7.6 as,
2k
h (t) =
x(T -

t)
No
Putting x (T - t) from equation 3.7.11 we get,

h (t) = 2A for 0stsT

No
4 Ak
for 0s t<T
No
Fig
Fig. 3.7.2 shows the sketch of this h(t)
impulse response. 4Ak
No
Thus with = 1 , the above figure will
No
be same as Fig. 3.7.1 (e) and Fig. 3.7.1 (c).
This shows that the shape of the impulse
response of the matched filter is similar (or
matched) to the shape of the input signal Fig. 3.7.2 Impulse response of a matched filter
for rectangular pulse input
x (t). Hence it is called matched filter.
Example 3.7.2 Show that the maximum signal to noise ratio of the matched filter is,
2E
Pmax
No May-11, Set-3, Mar

TECHNICAL PUBLICATIONS - An up thrust for knowledge

3 55 Baseband Digilal iransmission-1
Drgital Communications

Solution: To derive this equation, consider the signal to noise power ratio of optimum

filter is given by equation 3.6.17 as,

P ma d . (3.7.21)
Si ()
e have seen that when white noise is present, then optimum filter is caled matched

N
tilher. I'ower spectral density of white noise is S,i (f)=Putting this value in above

quation we get,

Pma dt
NO
2

. (3.7.22)

Rayleigh's energy theorem states that,

IxF a =f*0d=E (energy) . (3.7.23)

With this result

equation 3.7.22 becomes,

Pmax 2E
No
E
Thus maximum
signal to noise power ratio of matched tilter is,

2E
P max
No . (3.7.24)

The above equation can be rearranged as follows

E
Pmax
No

Here E is energy of signal x() andis power spectral density (psd) of white noise.
Hence we can write,
Energy of the signal x (t)
Pmax psd of white noise .(3.7.25)

TECHNICAL PUBLICATIONS An up thrust for knowledge

3 Digitei Communications
Baseband Digital Transmission- 3- 57 Baseband Digital Transmission - i

Example 3.7.3 Shorw that for a matched

filter the maximum
Le. signal component occurs ats T
x T) N
sampling instant) and has magnituie E, i.e.
energy of the signal x (1).
Solution: The tourier
transtorm of the 2k
output signal Ay () is given as, Maximum value of
Xo (T) will result when
No 1
(3.7.26) Hence, Xo (T) = E
Here, X () is fourier transtorm of
input signal x(t) and whent= T
H) is transfer function of matched filter Example 3.7.4 Show that the
output signal of a matched filter is proportional to a
version of the auto-correlation shifted
function of the input signal to which the filter is matched.
Putting value of H () from equation 3.7.3 get, Solution: The output signal
we

fourier transform. ie.,

xo () is obtained from its spectrum Xo f) by taking inverse
No
Xo ) =

Xo ()e/2t df
2k x
No Putting Xo ) from equation 3.7.27 we
get,
We know that X(f)X (f)=|X(IE. Hence above equation becomes,
Xo (f) = X j2T
o )= X e -125T .ej2ntdr
.(3.7.27)7
From this relation we can obtain xo (t) by taking inverse arier transtorm. i.e.,
NIX(oP/20-T)df (3.7.28)
The
Xo ) = Xo ()el2t af By definition of FT energy spectral densíty (esd) is defined as,
esd
(f) |X(DI =

By definition
X 2ej2nT.e2t df Hence
equation 3.7.28 becomes,
Putting Xo f) from equation 3.7.27
o ) ve2f1-T) d
-co
R -T) . (3.7.29)

One of the
property of esd states that auto-correlation fumction R
at f =
T,Xo (t) becomes, Spectral density y (N form a fourier transform (T) and energy
pair. i.e.,
Xo (T) =N X( e/2f(T-T) dr R(T) v ) By property of esd
(3.7.30
That is, R(T) =

v(e|25fT df
(3.7.31)
ar since e i2/ (T-T) =e" = 1
and
w(f= R(t)ei2nft dt
By Rayleigh's energy theorem we know that . (3.7.32)

X O I df = x () dt =E, hence above equation becomes,

TECHNICAL PUBLICATIONS An up thrust for knowledge TECHNICAL PUBLICATIONS An up thrust

lor knowledge
Digital Communications 3- 58 Baseband Digital Transmission

3.7.31 to RHS of equation 3.7.29 we get,

Applying equation

(t) =
2k R(t-T)
(3.7.33
No

This proves that Xo (1) is proportional to auto-correlation function of x(t) shifted to

2k
right by T. Here is a constant term and it depends upon psd of noise.
No
82
Baseband Digital Transmission-1 Digital Communications 3-83 Baseband Digital Transmission - I
Example 3.9.2 A
polar binary signal s;0) is +1 V or
-

Additive white noise with 1 V

pulse during the interval 10, T]. It is that No
Determine the maximum bit
power spectral
density= 10 W/Hz is added to the signal. given 2
=
10 watts/Hz. Hence No =
2x10. Therefore above equation
rate that can be sent with
bit error becomes,
probability of Pe S 10
Take Q13.71] 10, AT
= iwhere Q) =

.e*/2dh 2x2x10-5
= 13.764

Solution: Given data A =

t1. Hence A< =1. Therefore,
This is
binary PCM transmission. PSD of 1XT = 13.764
white noise
is,= =10-watts/Hz 4x10-5
Amplitude of the pulse, A = t 1V T = 0.55 msec.
error
probability, P, s 10
For binary PCM, error Bit rate = 1.816 kbps.
probability is given as, T 0.55x10
055x103
A2T O00
P AT |
3N 2

The relation between the error function and Q-function is,

With the help of above result we can write

as,
error
probability in terms of
Q-function

2N
10 2N
It is given that Q [3.71] =
10. Hence from above
equation,
A 2T
1 3.71

AT
2No
13.764

TECHNICAL PUBLICATIONS An up thrust for TECHNICAL PUBLICATIONS An up thrust for krowledge

knowledge
Examples. for Understandingg
T Is represented by
Example 3.9.1 In a binary PCM system using on-off signaling symbol
the pulse
A for 0sts Tp
s(t)=
elsewhere

Arndsymbol 0' is represented by svitching off the pulse. The receiver uses matched filter.
Assume that the receiver noise is white gaussian, with Zero mean and PSD of No
2
Determine the average probability of and 0
error, when s1ymbols 1 occur with equal
probability.
oiulion: We have obtained the
expression for probability of error for binary PCM in
Sec 3.9 The
amplitude of the pulse is A. The expression for error probability is
equation 3.9.5 as given by

A 2T
P erfc4N
nere = 1h 1s ne dil
perioa. ience aDOve
equation can ve written as,
Pe
P erfeA
4 N4T,

TECHNICAL PUBLICATIONS- An up thrust for

knowledge