Chapter 40

Random Variables To

There are a type signals

of
to Deterministic signals Signals whose

analytical andgraphical description is

known at all time instant
2 Random signals thesignal which is

unpredictable which are uncertain

to
Topic to concept of probability

1 what is a random Experiment

Ied
e
A sample space S is collection of all possible
Separately identifiable outcome of random

experiment for example in tossofcoin

Sample space is 1,21314516

Event is set of outcome meeting some

specification for example odd number or throw

0 Even include 1,315 as possibleoutcome

Each outcome of an experiment is element

or sample point

Null event to Event with No samplepoint

Complement of any event A AC is an event

that contain all the sample points that are not

 present in Aa

Union of events A and B AUB E

Unionof two events is a new event AUB

which contain all the sample point in A and B

Intersection events AaB simply A B

of two
or

This is a new event that contain all the samples

points which are common to both A and B

Note
if AaB p then Aand B events are

called disjoint event 09 Mutually exclusive

event
Relative frequency and probability

Probability of any event A is PA NGAI

1e Number of sample points in A Total
number
ofsamplepointsin
samplespace

P AUB NCA t N 1B NCAA B MAHDI B

PCAA B
N
 PLAY I PLA

Q1

Q2

Wewill study more and complete probabity in

Mathematics Here we are Just looking Basics

Conditional probability and Independent
Events

Conditional Probability s The fituatemin

which one ofthe event has already occurred
and you havetofind probability occurrence
of of
another event
Written as PLAIB Given that event B has
already occurred thenwho
is
probabilitythat event A
will occur

we can write that

PCA B PIA X PCBIA
um
probability that
I
both even occur probability that probability
that Bwi
event Aoccer occur
A
given
hasalready
occurred
 OR
PIAB Pl B X PCHB

fo PLAB PLAI.pro AJ
PCB

and PIBIA PLAIBIRB

PLA

4
Independent Event
When two Events are independent then PCAIB
will not depend on Bgthus NAIB PLA gthus

Insuch cases PCAB PIA MB

Topic 2.0 Random Variable

TABOO

n Icn
05
Probability ofacointoss
experiment
s
s

Topic
Discrete Random arlable

Random Valuable is discrete RV

if the variable
take only distinct number ni
thus RVcanbe Ni ican be any value
between too
Ricki is called probability Mass
function
 we can
easily see that
Puni L

funof Probabilities
of
all possible

Values
of Rv is 1

For 2 RV
Nandy
Pay Mos yj ie Probability that

y yj
A ni and

Bey ni I
yj

also Palydeity I

Type Yjtri I
also Paylaig yj Palyckily Pylyj
type yjpei Puni

also Pay Cai yj Palai

 Cumulative Distribution function CCDD

a
CDF Fx se Px xen Generally
Haircare
function
Thus Fx Coo Px Koo I

FXC a PxCxc o O

Fx n is always an increasingfunction

Q
Topic 40 Continuous Random Variable

The RV can take

any Value in a certain Interval
Puni show the probability that Rikki

PDF z n Rani

Probabilitydensity
function

I e I
Mr
Ni
µ
fine Rtx canhave Value in
range ki n

thus 92Cai da I le the

area under PDF is always 1

Cumulative distribution function CDF

Fx n Ric Xen f Randa

 Fx a
Ig FxC D Og
the CDF is always an increasing

function

Thus Pdf Blah akaFica

also P MERE 4 Randa

Construct density function for above

discrete RV
11 Ran
Hu
42

0
purr
I 2
 nPxcn

ke

10 10

ofthe given PMF

CDF
find Krog
Plus.io
find

PMF of a axis
given
FX n _O NEO
KHL O l k f lo

look H 10

fendK PC 2 cues
Q for Rv Pmies
ae bn x o
fun o otherwise

find relation btw a b

Prob PC tent 3

plot fun

Topic 5.0 Gaussian Random Pariable 7

The PDF
of Gaussian Rv is
L e Mz
Pxen
VIT

Fun
14 e an

kda
Icy Syd
e I fwy
 h au Yoo
Thegeneral Gaussian PDF is pW L e
JIA T
n

Lo Fx a J Pxcusdu

e en
a 4202dm
Funk
fo
r

Let F Kiyo
died
Fx a
f 4 e ok

Fun I Quote

Topic 60 Joint Distribution

we know that

Pf Xen og
Yey Fxy my
thejoint PDF is pxycn.gl 2
xayyH4D

also Pxylum Indy I

also Marginal densities pxcn5IPxyln.ci dy

and
Pyly If Pxy my dn

Z
alf.gg

EfgxggdnfxCnI
foFxyCn.y1dyM
arginalPPF
Topic to flakslical Averages

let Ki is a discrete Random Variable then

f outof N trials NI comes Ni times

Hz h Nr h
Hz n
Nb n

Then Mean u I Hint Nuh 1143ns

of
Nnnn
N
n
I I I ni Pulse
i I

Thus Expected Value a lie meanof U RVis

of
Ecu ki Ricki

also Ecu Pxcn dm

Mean of a function of a Random Variable
then
if gas is afunction
of n
ElginD gCais Rani

and agentRani du
ElginD

Mean
of the
sum

If gluey g gamey gun g grin 4 are

the Rv then
t
filthy gun y 1gun y t tgnlu.us

g.cn DtgunytgscTDt gnlnTD

Thus Tty Tty

Meanoftheproductoftwofunct

I
 the two random variable
if my are

independent then

Pxycu g Ran Pycy

Thus
E Ny JmyPxy my dndy
Else ECD

when the two Variables

Ty I
g are independent

Moments to

nth moment Rvn is defined the

of as mean

often thus
get Ten
p Cn du

The nth central moment

ofRvn is
CEEJ IJPxenidu
L
 and central moment of Riles called flandaed
deviation

Taz EFF k 165 EEE

2
ET E

Ela Ela

Variance
of a sum of independent Rv

g
MY
Ty's E

Tz Z Felt Ty2
as when are independent then
Randy
D and Cy g are also independent
 V
This result can be extended to
any number of
Variables

Topic 80 Central limit theorem

Under the certain condition the sum large

of
numberOf independent RV lends to begaussian
Rv IndependentOf the PDF of the Variables
being added

It will be shown that if I k th t b Nn

And N H Nz Ny Nn are independent RV

then PDFof 2 will be convolutionof

PDFOf Mi gNr Nz Nn
Topic 90 Correlation't

when we want to find the dependence

between two 124 nandy then we we

compute Covariance Tay two Rv's ie

of

Tey eesexy D

Variance our nFtTeF

 The covariance

Fey G Dey D
sty sty sty try
Try sty IT
Thus when K and are independent
y
ny IT
then and thus

O
guy

Thus when are independent then

Randy
their covariance is Zero

Note for independent Rv we know that

gingold gin gets

Note two Rv independent then
if are

they are uncorrelated but not Vice versa

Qand Solution

Q
Q

FewRandom Variables

1 Binomial RV if we conduct an

experiment where the experiment has

only 2 outcomes
Success paobab p
failure probab 9 GD

then an experiment is lenducted

if n

probability of Success Ktemes is

ternes their

Pate Mcu Ch h
pkg
 MeanValueOf Binomial RV is Np
Variance of a n
Npg

2 Poisson RV E
If a is a RV and
f is
Poisson RV the parameter ofPoisson ru
is d
Shen
Plans e
dq g
Meanof Poisson Rv L
u d
Variance of u
3 Rayleigh density
Ma
flu if
e u cry

Mean aJFK
Variance G AID are
Few important Point Impfor GATE like
exam

FLAIM P xen M Paterno

PCM
Topic 100

functionof one Random Variable

fuppose n is a Ril and y is another RVcreated

by the relation
g goes
We know that Eg is then KCG and
any number
NG are also numbers
g
Fyly P yay P
glassy
Example axtb
10 y
Now Fyly P ntb
Ey
P see
Ha
Fyly Fx Fba if a o

Fyly I Fx Fda a 20
if
Example 2 o a then
g

n
iii I't
Example 30
Example 40

Example 50
 Example 60

How to fend fycy le PDF

My

We want to find PDF of interms of ng

y
 We know that
goes thus
ya
g Cy My

Thus
fyup f xlustfully
fycy filthy At
doff

The above formulae is used when

ginfunctein
is Montonic Increasing or decreasing function g
thus n is one to one g
y g

but
if y gen is not one to one then
thus
y gcn will have roots Kw
g gin grey glad

I
 t
fylyt Zfxcudklg.cm

xfyyr fxgfIn tfxgHuIytfg'tiff

Example

T.yt antbfycyt fqfxly.ba

2

3 ang a o their
g
fyly Tayy a fxLTy1a tfxfFla
 4

6
Topic 11 Two Random Variables 0

we know that

Joint PDF is Fcu y PEX Eng Yey

L FC o o
y
Else 05 0
Les
f carat 1

DOO

ooo

Incase independent Rv case

of
Flu y Fun FYLD g fxyln.gg fxcmfyly
Topic 12 Onefunction of two Rv

F gluey q Lis Rv which is afunctear of

Randy
FeG Paez P gcse y EE

Examples Katy
Faz Pf Hy S2
AY
H
Thus integrating over the
Nty 2

Stay we get
mmmm
f2121 1 fo fcng dudy
Z
fztD ddzFzlD

fzcyzfffk y.y dy
Thus
fun fo ft yay dy

Now if ney are independent then

f212 If f K y fly dy
0Th's PDF PDF at PDFy
Thus when we add independent RV then
the PDF Of resultant RV is corn of PDFof all
RV

Example2
if Ryu and y are Independent
a e an Ucu
fun
e B Uy
fyly p

If 2 Nty find PDF 92

Example3 Two independent 124

Nandy are

Uniformly distributed in Interval g1 gjnd

PDF Of L kty o
Example4 o Two Independent
124 are uniformlydestabu
Nandy
InSame Interval 1 find P ay o

Examples 0 X.gl2gXz are three Independent 1211g Uniformly

distributed 1 P Nith EU
on
find
1 I 1
Nice Method 1
Xitxz EK is
I
Probab 1542
0
t
PLXitxzEXD fokfdk 3
kg gg

nfxin
Example G O HIT

f A
ITI The

tann find fyly

g
Example 70 A zero mean Gaussian Rv is passed through a

HWRfindPDF ofoutput

8.0
Example o
2 axtb nd PDF Of 2 Take Randy as
independent
o
ye axtby

Example 90 The Joint PDF

ca Y
fxycu.gg Say e U o y 0

o otherwise

PDF
find marginal fxing fycy
find fuly HD
Q

to Explain MAP detection Rule't

I
 Chapter 50
Random Process

How will
fo we define Random
process

Tospecify Run we
repeat the experiment a large dumb

times to butto define random process

of find Pyu
wedogame thing valve of t
foreach
 lunate's
Ensemble

We canbuythat Random process is also outcome

of
random experiment here the outcome
ofeach trial is
a waveform that is a function of t
Wecan see that t t we fample
the waveforms weget Runt

Note Wecan Random process is combination

say that of
infinite number Random Variables
of
Rampling of Random process results in RV
Collection all waveforms ofRPis called
of
ensemble
 fowhat is nth order PDF

Topic 5.1 Autocorrelationfunction of Random

process

Rxftp.gt L kltHNtD HIM

Here ke and n are 1211 createdby Sampling Random

process at tf and tu

P N

Rx te te
N
f Nth pain Negus du du
N

Topic 52 Classificationof Random process

 Firstorder densityof
stationaryRPis independge

Strict sense stationary

1st 2ndorderPDF
f only and
 Mtk A cos
Question ett
if dis
RV which is Uniformly distributed in
range 0102A
is ut Wss Ausyes

Process
Topic 5.3 Ergodic

The Random process has two

typeof Statistical
averages 1 Ensemble
Aug
2 Time
Aug

Ensemble The Valve found averageof

Aug avg by
all Sample points at one

particulartime
Time Aug Thetime average of random process
 Ensemble Averages Time Averages

Htm
IF JaiPadre
c0
deign
t.fiInctidt
d T runner
RxxCtsgtD Htt Utd NtitectD
murmur
HAD Nctitt
417 I ftp.huttutthde

Now Ergodic Process When the Ensemble

avg and teme avg

are
fimilar to each other
Verifythat RP NHK Acos otto goes Rv
Uniformly distributed Otoag find that
whether Rpis Wide Sense
flationary or not

Aus Ho

ARP Aws wot g Ae's

given by Kitt a

Random Variable Uniformly distributed between

0 1 find mean and ACF of Ntl ta Coswot

wot
IzcosCoswote
 Topic 54 PSD Of Random Process 3

ThePSD Random Process is founia

of any
transformof ACF ofthe Random process
Question 3
Question

Question
Quested
Question to
Topic5 5 Multiple Random Process

Cross Correlation function

mm
Rxy test sects yay
I
point flationary Ran
to
RxyCtigtDsRxyLtI tD RxylD

RxyLti.tD HHT yTtD I

unconellated
Few Important Properties
I
ACF i RxxCt
RxxH E Nlt Htt 4
RxxCt is even function

IRxx lol ER xxCD

2 Cross correlation Rxy E

Rxy E E Xlt Htt 4

Rxytt Ry
IRxy t ETR xx b RyyLD
Rxy to l E
ta Rxxlo t Ryy o

3 Auto Covariance Cxxle 2

CxxH E
Htt uD xftu
g
Cxx It Rxx It tex

4 Cross covariance Cxy It

Cxyk E
f Ct Ux
yate Uy
 Cxyth Rxyt Ux

their uncorrelated
If Cxy 0

Rxy H o then orthogonal

If
5 PSD Of random Process

Sxx PSD
of Rxx E
w

Thus Sxx Is ve and

w
always
even
Sexy w and Frog
RxyC

a
Syx w A FT Of Ryne

Since Rxy t
Ryne
x Sxy w
Sync W