Scribd
Upload
41 views9 pages

Poisson

The document discusses several probability distributions: - The Poisson distribution describes discrete random variables and processes with independent events over time. - The negative exponential distribution describes continuous random variables like time between events. - The Poisson process models random events like customer arrivals over time as a counting process. Inter-arrival times are exponentially distributed. - The Erlang/Gamma distribution generalizes the exponential to model systems with multiple sequential steps like servers, with the Erlang a special case for integer steps.

Uploaded by

Aditya Koutharapu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
41 views9 pages

Poisson

The document discusses several probability distributions: - The Poisson distribution describes discrete random variables and processes with independent events over time. - The negative exponential distribution describes continuous random variables like time between events. - The Poisson process models random events like customer arrivals over time as a counting process. Inter-arrival times are exponentially distributed. - The Erlang/Gamma distribution generalizes the exponential to model systems with multiple sequential steps like servers, with the Erlang a special case for integer steps.

Uploaded by

Aditya Koutharapu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Poisson distribution

A discrete distribution

R. Lahdelma

14

12

10

0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
2

p(x) = e-x/x! when x0


F(x) = i=0,...x p(x)
E(x) =
Var(x) =
Very simple
Describes well many
random processes

Poisson distribution with alpha=5

Negative exponential distribution


A continuous distribution

f(x) = e-x when x0


F(x) = 1- e-x
E(x) = 1/
Var(x) = 1/2
Describes e.g. lifetime of
a component, inter-arrival
time, etc

R. Lahdelma

Exponential distribution with lambda=1


1.2
1
0.8
0.6
0.4
0.2
0
0

The Poisson process


A sequence of random events such as arrivals of
customers can be described by a counting function
N(t), t0 is the number of events that occur in time
interval [0,t]

The counting process is a Poisson process when


1. Arrivals occur one at a time
2. N(t) has stationary increments: The distribution of
the number of arrivals in interval [t, t+s] depends only
on s, not the starting point t
3. N(t) has independent increments: The number of
arrivals during non-overlapping time intervals are
independent random variables.
R. Lahdelma

The Poisson process


The probability that N(t) = n is
P(N(t)=n) = e-t (t)n/n!
This equals the Poisson distribution with =t
Thus, E[N(t)] = Var[N(t)] = t

The number of arrivals during time interval [s,t] is


due to the stationarity also Poisson distributed
with =(s-t), i.e.
P(N(t)-N(s)=n) = e-(t-s) ((t-s))n/n!
E[N(t)-N(s)] = Var[N(t)-N(s)] = (t-s)

R. Lahdelma

The Poisson process


The probability that the first arrival A1 occurs after
time t is
P(A1>t) = P(N(t) = 0) = e-0/0! = e-t

Thus, the probability that the first arrival occurs in


[0,t] is
P(A1t) = 1 - e-t

This is the cumulative distribution function for the


exponential distribution
Due to stationarity, inter-arrival times between
any two subsequent events follow the same
distribution
R. Lahdelma

The Poisson process


Random splitting
a poisson process with arrival rate is split randomly
into two with probabilities p and 1-p
this results in two Poisson processes with arrival rates
p and (1-p)

Combining (pooling)
two Poisson processes with arrival rates 1and 2 are
combined into one
this results in a Poisson process with arrival rate 1+2

R. Lahdelma

Generation of a Poisson process


The arrival process generates new events at intervals that
exponentially distributed:
f(x) = e-x

To generate such random numbers, the inverse


transformation technique is suitable
F(x) = R = 1- e-x
Generate R from a uniform distribution in range [0,1]
Solve x:
x = -ln(1-R)/
(= -ln(R)/)
is distributed exponentially
Observe that the argument of ln(.) must not be zero! If your random
number generator generates numbers in the semi-open interval [0,1),
then 1-R is in the interval (0,1] and can be used safely as argument for
ln(.)
R. Lahdelma

Erlang/Gamma distribution
Generalization of the exponential distribution

f(x) = (x)-1e-x/() when x>0


E(x) = 1/
Var(x) = 1/2
is the shape parameter
the scale parameter

Erlang distributions with beta=1,2,3,4, theta=1/beta

1.2

f(x),1
f(x),2
f(x),3

0.8

f(x),4

0.6
0.4

When =k is an integer, this


is called the Erlang distribution
x is then the sum of exponentially distributed
variables with parameter =
0.2

R. Lahdelma

0.5

1.5

2.5

3.5

Erlang distribution
A system consists of k sequential servers
service times are exponentially distributed
a customer must pass all servers before the next
customer may enter

Applications e.g. in reliability theory,


telecommunication

R. Lahdelma

You might also like