Random Processes 539

Gii) p) = p6). p = |20, 14.0.27 0 28S |

T h e game lasts more than 7 plays is (he stops playing only when
he has 0 or 6)
- P(X, = 1) + P(X, 2 ) + P(X, = 3) + P(X, = 4) + POX, = 5)

+0+2+0+ 13 = =0.4218
54
64 128 128128

6.3 POISSON RANDOM PROCESS

Let X) represent the number of occurrences of a certain event in the interval

(0, ), then {XO} is said to be a discrete Poisson random process if it satisfies
the following postulates:
1. P[l occurrence in the interval (1, 1 + A/)] = At +0(Ar).

2. P[0 occurrence in the interval (t, t+A)] =1- At + O(A).

3. P[2 or more occurrences in ( , t + Ar)] = O(A).

4. X() is independent of the number of occurrences of the event in the

interval prior and after (0, 1).
5. Probability that the event occurs a specified number of times in the
interval ( , 1 + Ar) depends only on Ar, but na

6.3.1 Probability Law of Poisson Process

Let be the number of occurrences per unit time.

Let P,() = P[X() = r] = probability that there are n occurrences in (0, ).

P,. 1)= PLX) =n + 1]

A ) = P[n - 1) occurrence in (0, 1) and I occurrence in ( + A)

n occurrence in (0, 1) and 0 occurrence in (t, t+ Ar)|

Pt + A) =P, - 10(2A) +
P,) ( -

Aar)
P - 1 )(AAI) + P , ) - P ^ ) AA
P + Ar)- P , ) = AA|P, - 1) - P,OI

P,(t+A)-P) =alP, )- P,OI

Ar

Aking the limit as Al -> 0, we get

lim
P.( +A)-0= ALP. ,)- P,O1
Ar0 A
540
ELx= ) Pnlt) shaz
Probability and Random Processes Random Processes 541
Pot) e a
[P. =

AP,- 1)- P,0

n=ol,2
d
2 - y"
P)= P,- 1)- P,O1 n=0
n! n
(n-1)! = 0

Let the solution of

Eq. (6.1) be

Aty"
2 1
(A)"A- Are- e" ( - 1

P,) s0 n-)!
n=l
n!

IP,)-ATf0+
fo) + (A) -Aue1 =Ate-*i
n! (n-1)!
and - 1 ) = G y - 1

(n 1)!
Substituting in Eq. (6.1), we get
ELX- P,)=
" n
S ) + - f ' ) = a l Ay"= )- fo n =l
nl
(n-1)!* 0)= T1f)- n! |
Cancelling
2""-1
11 throughout, we get
Inn-1)+n
n=l
n

S0+r = f)-f)
n n!
n =l

f') = f ) » s)=-åf)
n =

f') - 2 )(n -2)!

ft)
Integrating with respect to , we get A , , (A)
2
log V)] = - l +A
+A
=e*(AN) e +

S ) =e" =e A ELXO=(A) + u 3)
S ) = ke", where k = ei
Var[X()| = EX() - {E[XO? = (i)* + A - (A¥
Now, 0 ) = P[number of occurrence in (0,0)] = Po(0) = 1
Var[X)= u
S) =ke S0) =k =1 k =1|
J) =e 3,3 Autocorrelation of Poisson Process
Substituting in Eq. (6.2), we gct
P() = (A"

E{XU) 1KU,) - AU)+X°u)}

6.3.2 Mean of Poisson Process =

ELXG)I EXG,) XU)] ELXo,1 -

Poicenn random thon

EJNU,N |ENU) -ENU, FLN°,M
Iet X he a nrnrnce
 t a - n t ) +E (AL)+MM

- o E 7 - nt
e= |+
542 *t4A
Probability and andomProceser
Random Pfocesses
Tom T
++ equ63)
6)
EC =at ohicl, isa judi Ppes54
1 t , n o F , a sbionai PrO@
Sun of two independent Poisson processes is also a Poisson process
=

R1,4, + 2 |min (,. 1, Proof Let (X,O} and {X,O} be two independent Poisson random processes
6.3.4 aith mean armrival rate and 2, respcctively.
Autocovariance of Poisson Process et
XC) = X1) + X2)

Sunco X(0} and {X2)} are Poisson processes, we have

Ca12)=Rl. 1)-ELXU) ELXUI ELX,O=
and EX) = t + A. E[X}U)]= i t + t
,>'
Caa'12)=A min(, 1, E X = ELX,) + X,O

6.3.5 Correlation Coefficient ELX,O +EX,O1

of Poisson Process =

A + 1t =(2 + 2)
ELX= ELX,0) + X,OP

CU,1) Cxx) E[X(t) + X;(0) +2X, )X,)

Cxx . ) Cxx (l2.12) =
E[Xj(1+ E[X +2E[X,t)X,(t)]
j + At + +t + 2E[X,)] ETX,0
r ? + At +Ait +lt +241d
Cu 1)- = (2 + A,)P +(A, + 2)r

X+X) is also a Poisson process with mean arrival rate A +

6.3.6 Properties of Poisson Process

3 Dilference of two indcpendent Poisson processes is not a Poisson process.
U1. Poisson process is not a stationary
process. Proof Let {X,O} and {X,O} be two independent Poisson random processcs
Let {X()} be a Poisson random process, then with mean arrival rate , and , respectivcly.
LetX() = X10) - X2)
e (A We know that
P,)= n = 0, 1, 2, ...

ELXO = n , )
n = 0 ELXOI = EX,O - X,01

= E}X,O1 - ELX,) = A - = (A, 2,M

ELXO= E;Y,) - X,OF
n = 0
n! n = 0
(n-1)!
=
E[X ) + Xit)- 2X,) X,t1
n - 1 ) d u e a (Auyr-1

(n-1)!
=
E[X+ E]Xa1-2E1X,|E]X,)
n=| n = l

= ( - A,PP+ ( , + A,
Ae
 Since h e Parameler not AA2, n2-ni n3-n2 n3
cajferece o too mdependan beisson PrO Cesses e Cta-b) (t3-t) C),ns M2h
544. Probability and Random Processes
yno a besson prO h . 2 -n1) (n3-n1) RandomProcesses545
4Poisson process is a Markov process (memoryless propertv). Nowto prove Poisson process is a Markov Process:
P10of Let {X,)} and {X,)} be two indcpendent Poisson proccsses wih P[X (3) =lg, X((,) =n, X(G) =nl
same parameter A. Then PX)=n/X(1,) =
n2. X(G,) =n] P[X() = M. X(1) =n]

PLXG) = n,l= e (A By substituting Eqs. (6.4) and (6.5), we get

e 2-1* * " ((s - I 2 ) " 3 (A

and
PLXU) =n,= e(A1"2 2

The second-order
probability function of a homogcncous Poisson process, t
PLXO)=n,, X() n,1 P[Xu,) n/XG,) n,l PXG,) n,l (6
= = = = =

P(BIA) =P(AnB)
P A n B ) = P(A) P(BIA)
P(A)
=

P[X0) =

n,] P[n, - n,) number of occurrences in ((,-4}

(2-)

(2-1) (-)

(n2 - n)

=
P[(n, -

n,) number of occurrences in ( -

1)|
The third-order probability density function of a Poisson random process is =

PLXU) =

n/X(0,) n =

PLXO)=n XU,) =
X0,) n,l
n,
=

Tho probability of thc occurrence of thc event Nu)

Aheprevious valuc X(/,) = n. but not on XT,)= n
=

n, depcnds only on

P[XU,) n, XU,) n,1 PlX) n,/X0,) n,l

= = = = =

Poisson proccss is a Markov process.

2
(A1"(0 -4)2-"A-|2,-4,) The interarrival (interval between two successive occurrences) tume ol a
(2 ) (12) with parameter A follows an exponcntial distribution with
Poisson process
"2
mean
(nn2) Proof Let X) denote the number of occurences in a time interv al r.
and , .
-12 " (2" (, -1,)"-" LetTbe the interval betwcen two successive occurrences E,
random variable.
s acontinuous
Let the event E, occurs at the time instant
 PT>t) PCE =
did hoB occu In PE e PEga=
546 Probability and Random Processes (ti,ti t+t)] Random Processes 547
P no ereut oCcuss in (t
=

P[X) 0= =
e pt (p+q=l p=l-q)
0!
P(T> ) = e n!

Cumulative distribution function, shich is the PDF of the Poisson process with parameter 4p.
NOfollows Poisson process with parameter Ap.
F) =

P(Ts) =
1 P(T> 1) 1 -e
-

bank
Probability density function, EKAMPLE 6.41 Supposc that customers arrive at a
according to a

Poussor process with a mean rate 3 per minute. Find the probability that during
S) =

FO) =
0
(-A)e =
ieà, cume 1ntcrval of two minutes
which is the PDF of
Hence proved.
exponential function. exactly 4 customers arrive,
(n) more than 4 customers arrive.
6. If the number of
occurrences of an cvent E in an Soiution Given: mean arrival rate = 3
Poisson process {X()} with interval of length
paramcter 2 and if cach occurrences of E /$ 2h0 =2 minutes
constant probability p of
being recorded and the recordings are has a N know that
cach other, then the number
N) of the recorded independent of
occurrenees in f is als0 1
Poisson process with
paramcter p. Ay" 6"
PLX() =n]=+ n=0. T. 2....
Proof PNO) n] P(the
= =

event E occurs (n t r) times in t and n of

=6)
them being recorded) 0.PM2)=41 4! =0.1338

n PpX(2) 4] = 1 - PlLN2) 4
P[the event E occurs (n t r) times| x Pln of them
r = 0 =1 -

{P[X(2) =

o PLA(2) =
1] PLK2) =
2
PX2) = 31 PLX(2) = 4|}
being recorded oul of (n t r) occurrencs
1-6)_(6) (6) 6)' 6)
(n +r)Ut /,p'q.(q=1-p) 0! 1! 2 3! 41
r= 0

(6)
p A y y ) f (ntr)!
66 0.7298
2! 4
t r)! r!n!
EXAMPLE 6.42 If thc customers arrive at a bank according to a Poisson
n proceKswith mean rate 2 per minutc, find the probability that during a l-minute
Sincen, picrval no customer arrives
(n-ri
Solhrion Given: I=| a t 2

2
r) We know that

dpt e(A)
n = 0. 1. 2.
 ^=I
P IE) =n] =
e (2), n:Ol,2 t:2,
Seu Cien A t =2 Random Processes 5 4 9
548 X(I) =o
Probabílity and : Processes
Random - 0 13S
EXAMPLE 6.43 A radioactive source cmits particles al a rate of S e(Ar) n = 0, T, 2, ...
minute in accordancc with Poisson process. Each particle emitted has a PLX)= n]
probability of 0.6 of bcing recorded. Find the probability that 10 particles ate
recordcd in a 4-minute period.
PLX(2) =
0] =
2 = 0.135
0!
Solution We know that
within 10 weeks maximum 5 failures
can

Smc there are only 5 spare parts, 10.

ea (Apt" be out of order. For
=

the machine will not

P[X) =
n] =
,n =
0, 1, 2,. Ocur such that
n

PLX(10)0]+ P[X(10) =1]+ P[X(10) 2]

=

Given: P=0.6,
Ppx(10) s SI \+ P[X(10) = 3] + P[X(10) = 4]+ P{X(10) =51)
n= 10,
= 4
and A= 5 e(10 10(10 e 10) e10
.. ap 3 0! 1! 2! 3!

(10) (10
e (3)"
PLX) =

n] = n= 0, 1, 2, . 4! 5

PX(4) =
10] =*
4 (12)10 = 0.104 =01 (10) 10) +
1! 2!
(10(10),0=o.067
3! 4 : J
10!
EXAMPLE 6.46 Qucries presented in a computer database are following a
EXAMPLE 6.44 The particles are emitted from a radioactive source al te
Passom process of rate A =6 queries per minutc. An experiment consists of
rate of 20 per hour. Find the probability that exactly 5 particles are
during a 15-minute pcriod.
emitod oatoring the database for m minutes and recording N(m) the number of
gacTies presented.
Solution Given: 20 per hour, What is the probability that no queries arriving in a onc-minute interval"
i) What is the probability that cxactly 6 qucries arriving in a onc-minute
= 15 minutes = hour
4
interval?
querics arriving in hall-
and n=5
( What is the probability that less than 3 a

minute interval? AU May '07|

X) =
nl =
A) n = 0. I. 2, ..
Solution Given: Nm) =
number of qucries prescnted in m minutes
6

e5 = 0.175
5! 5! (6r)
P[N) =
r] = -0, 1.2,.
EXAMPLE 6,45 A machinc gocs out of order whencver a componcnt fads
The failure ofthis part follows a Poisson process with a mcan ratel per wek
Find the probability that 2 wecks have cllapsed since last failurc. Ifthereare
5 sparc parts of this componcnt in an invcntory and that the ncxt supply is
not
PIN(0) =0] =
e°

6
=0.00248
i P N)3
duc in 10 wecks, find the probability that the machinc ill not be out of ordat R PIN(1) 6] =

6!
= 0.1607
in ncxt O wecks. 1/2, 1. 2. have
Forf x
=0, we

O.u231
 e1+5+ 12:S +20 8333 +26.042
S50Probability and Random Processes =0.u40S Random Processes 551

EXAMPLE 6.47 Customers arrive at the complaint

department of a EXAMPLE 6.48 Assume that the number of messagcs input to a
the rate of 5 per hour for male
store at
customers and 10 per hour for female oommunication channel in an interval of duration seconds is a Poisson proccss
If arrivals in each case follow
customers ntb mean rate 2 = 0.3. Compute
Poisson process, calculate the probabilities that
(1) at most 4 male customers, the probability that exactly hrec messages will arrive during a

(11) at most 4 female customers will arrive in a 30-minute period 10-secondinterval,

Solution Given: (i) the probability that the number of message arrivals in an interval of
mean amival rate =
2 =5 per hour (malc)
and A = 10 per hour (female)
duration 5 seconds is between 3 and 7.

By Poisson process. probability that n customers arrive in a tume inerval ot Solution Given:A = 0.3
hours is given ByPoisson process, probability of n messages arrive inI seconds
by
e-(r" e(Ar"
PLX() n = =

n!
n =0, 1, 2,.. PLX) =
n]l =

n!
n = 0. 1. 2, ...

(1) Mean arrival rate for male customers is

A =3 ) P(exactly 3 messages arrive during a l10-second interval)

(51
PLX) =n]1 n = 0, 1, 2,..
P[X(10) = 31= 0.3)10(0.3)101
3!
Since is given in hours = 30 minutes =
30x 60 hour
2
Plat most 4 male customers arrive in a 30-minute period] 6
= 0.2240

(1.5
) PI3 X(5) 7 =

)-o-r)-1.9- 1 5)(1.5,(1.5
3! 4 5! 6
.5) .5)
7!
+P =
0.22313[0.5625 4+ 0.2109 + 0.0633
+0.0158 +00034|
= 0.1910
1+2.5+2.5), (25) (2.5]
2! 3! 4! J 6.49 Ifcustomers arrive at a counter in accordance with
AAMPLE
with
a Poisson
=
e 2'(3.5 + 3.125 + 2.60421.6279) OCESS a mean rate of 2 per minute. find the probability that the interval
= 0.8912 bchrecn 2 consccutive arrivals is
(ii) Mcan arrival ratc for fcmale customcrs is A
10 =
0) more than I minute
P l a t most 4 female custonmers arrive in a 30-minute penad) (0) betwcen I and 2 minutes. and
(u) 4minules or less.
tion The interval 7'betneen 2 consecutive arin als tollons an enponcntial
istrnbution with parameter a = 2 (given)

(101" n = 0, 1, 2, ..
where PX) =n| =
)= e A>0,t 0
J)=2e A0. 0

el+5tssf
 552 Probability and Random Processes Random Processes 553
C9 PLT >1) =, J, fLYAt= , 2e dte 0135
(i) P(1 PIX,0) = knX,0)=n-k]
< T< 2) =

)dt | 2«-2dt e = = -

e =0.117 PLX,() + X,r) = n]

(ii) P(T S 4) |f) dt PLX,)=k] PIX,0)=n-k]

e
=
=
1 -

=0.999
=

PLX,+ X, () = n]

EXAMPLE 6.50 A fisherman catchcs fish SnceX)} and {X20)} are two independent Poisson processes. {X) +
at a Poisson rate of 2
from a
large
lake with lots of fish. If per hour E0) 15 also a Poisson process with parameter A t A2
he starts fishing at 10:00 a.m.,
the probability that he
catches one fish by 10:30 a.m. and three whal
fishes by On
Solution Let X() be the total
number of fishes caught at or before time k! (n-k)!
(21 -4tA" |( +
PLX) =
n] =
n 0 , 1, 2, n!
n!
Given:2=2
a - *yn-kn! 4A"-*n!
(+ r"k!(n - k)! (4 + " k ! n - k)!
I= 10 a.m.-10.30 a.m. =
30 minutes ==

hour, n =1 ish
I= 10 a.m.-12 n!
noon =
2 hours, n = 3 fishes
(2 +A (A + k (n -k)!
PxJ-land X(2)=3- Px)
4 4+ =nC pq"-k

-)-)- where P + and +

Hence proved.
XAMPLE 6.52 If {X)} is a Poisson proccss prove that
= 0,082

EXAMPLE 6.51 If {X,)} and {X20)} are two independent Poisson PIX()= r/X () =nl=nC given s <I

proccsscs with parametcr 41 and A, respcclively, show that P{X,) =k4Kt0,

PIX(s) =roX)=n
+X0]- n} nPtq" '. where
hion PfX(s) r/X()
= =
n] =

p= and + PIX)=nl1
PLX(s) = r o X t - s ) = n - r]
TAU November 6
PX()=n
Solution We know that Sac XTs) and X(f - s) are indcpcndent

PA n B)= PAJPB) (if Aand B are indcpendent)

PEX)r/X) = n] =
PIX(s) = r] PlX(- s) =n - r]

and
PAIB)= AnB) PX()=n
P(B)
e (As) eA- [A( - s)l'
PIX0)= koX,)+ X,0)=al r! (n-r)!
PX,) = k/NX,(0) X,0)] +

=nj= PX,(0)+ X,()=n] "(NEY/

 Var(x)= E(x)
-

E]= £ ue24-1s
554
Probability and Random Processes Random Processes 555

eA(Asee (A"-"(1 - s)"-"n! 4ETX O1- 162 =

4(21 + 4/) -

162
(Ay'r'rn- r)! 8
involved in accidents per week with the given
kacefor the number of people
s -s"-"n! nC, mbutton
"r(n - r)! 28
Mcan 4/ =
4x 7 =

Var =8t =
8x 7 =
56
- ncGJ-) Poisson process. then prove that correlation
RAMPLE 6.54 If (X)} is a

Hence proved.
dicient betiween X() and X( + s) is
EXAMPLE 6.53 Thc number of accidents in Coimbatore follows a V+
process with mean of 2 per
Poissa
a
day and the number X, of the pcople inkeliod aon 1X()} follows Poisson process.
ELX)] =
in the th accident has the distribution
(independent) PX, k) = =

EX( +s) =a( + s)

Find the mcan and variance of the number of pcople involved in accidents p
autocorrelation function is
weck
R,1+ s) =E[XU) Xt + s)]
Solution Given a distribution P(X, k) -E{X) X( + s) - XD + XON
k =1, 2, 3,
= =

The mcan and variancc are 2 and 2.

=
E{X) XI + s) XO +ELY ) -

=
ELXO] ELXU + s) XO] + ELX OI -

[ M t + s) - A] + A ? + u
Mcan - ..

( A + is - ) + AP + u

Cov(l, + s) =Ryt, I + s) -

ELX)1 ELX +
s)
t s + ?+ - ( ) ( u + is)
Easily, we can show that X2) = 6.
Let us assume that the number of accidents on any day ben.
=
1s + P?+ -- ls
Lct X,. be the number
X of people
involved in the accident Sine
. 2 X, are indcpendent and identically distributed random variablswid Cnelation cocfficient between X() and X( + s) is
mcan 2 and variance 2.
X, + X, + . t X, follows a normal distribuiou Nu Cov(t.+) For Poisson proccss mean variance
mcan 2n and variancc 2n (by Central Limit Thcorem) Hence the umber
ol VanlX (0)1Varl X( +s
people involved with n accidcnts on a day is 2n.
Let n denotc the number of people involved in accidents on any day. Thca

(2/"
J2+ s) rts
1= 0. T, 2. ..
XAMPLE 6.55 If T, is the random variable denoting the timc of occurrence

Mcan ix) =

y 24r"(2*
2E X(0))= 41
dhe mth cvent in a Poisson process with parameter . show that the distribution
for t 2
- CR >0
 Find also tho Prebab1y amVng ata que ucirg
556 olen sil
Probability
funcou fnlt) o Tn
Random Processes
and
4Numbe) sh ustogndomProcesses557 Syse
Solution The probability distribution function,
2 No Telebhone.ls
arMving a'servicc station.
of cars at
amiar sio1tch boar a

3 Number at a server in a centre and many more.

F0= PT, s ) = 1 - PT, > )

Number of requcsts arriving

Here 7, is a random variable which denotcs the timc of nth MBERNOULLI RANDOM PROCESS
occurrence
P(7, > 1) = Pl(n - 1) or less number of occurrences in (0, 7)] such
outcome of the nth trial
l4X, be arandom variable which denotes the
trial is a success and PCX, 0) q. if the trial is = =

- =
1) if the
PLX, p,
outcome is a success and it
takes
That is, X, lakes the value 1 if the
-

k =0
k! kalare 1,2.3. is a Bernoullh
a failurc. Then {X,}, n
=

the is . .

alue 0if outcome

Substituting in Eq. ), we get andom Process.

AI Properties of Bernoulli Random Process

k = 0

The probability density function.

i s a discrete random process.
21is a SSS process.
I00-F0)=F0)-41-y"(
dt
BINOMMAL RANDOM PROCESS
- 0 - 2 e (an*i
sum {S,, n 1, 2. 3. ...} where S, N +N =

e discrete sequence of partial

=

k =0
t , andA7, X2, . . ,
X, are Bernoulli random variables is called a Binonmial
7-

jadom Proca
The time is assumed to be divided into unit intervals. Therclore. it is a
k =0°

discrctc time process.

2 At most one arrival can occur in any intcrval.
k! k!
k = 0 k = 0
Aivals can occur randomly and independently in cach interval with
probability p.
= 0

51 Properties of Binomial Random Process

() (A) + +
(Au"=("-|
2! (n 2)! (n -1)! ihis a Markov Process.
( )4 2)! e l Let S =

X, +
X, + .
+

+X
PS m/S, =
m)= PX, =
0) =4 1 -p
I20 P S = m/S, m - 1) = PX, 1) = p
(n-1)! (n- 1)!
depends on S Only
eaxe bunomial process is a Markov process.
6.3.7 Applications of Poisson Process