1

Poisson Process
 Outline
• Introduction to Poisson Processes
– Definition of counting process
– Definition of Poisson process

• Properties of Poisson processes

– Inter-arrival time distribution
– Waiting time distribution
– Conditional distribution of the arrival time
– Superposition and decomposition

• Non-homogeneous Poisson processes

• Compound Poisson processes

 Counting Processes

• A stochastic process {N(t), t ≥ 0} is said to be a counting

process if N(t) represents the total number of “events” that have
occurred up to time t.

• From the definition we see that for a counting process N(t)

must satisfy:

1. N(t)≥ 0.
2. N(t) is integer valued.
3. If s < t, then N(s) ≤ N(t).
4. For s < t, N(t) − N(s) equals the number of events that have
occurred in the interval (s, t].
 Increments : A
Section 2.1 The stochastic
Poisson process X(t) is
Process said to have
4
independent increments
if X(t2) - X(t1) and X(t4)-
X(t3) are independent
for any t1 < t2 < t3 < t4.
(c.f. disjoint intervals)
 Stationary
Section 2.1 The Poisson
Increments : A
stochastic process X(t) is
5
Process said to have stationary
increments if X(t2+s)-
X(t1+s) and X(t2)-X(t1)
have the same
distribution for all t1 <
t2, s > 0.
Section 2.1 The Poisson
Process
6

A counting process {N(t), t ≥ 0} is a Poisson process with

rate λ, if . . .
泊松分布有可加性，并且这个是在
一段时间内，所以参数需要乘以
（ t-s ），可以理解为从 0 到
（ t-s ）发生的总次数，相当于
P （ x= （ t-s ））

This means that the number of events (average rate of events) occurring in a time
interval of length follows a Poisson distribution.

Q: Is this process a stationary process?

yes, it is a stationary process
Section 2.1 The Poisson
Process
7

f(h) 是 h 的高阶项

Remark: This can be shown to be equivalent to Definition 1

(proof-see Theorem 2.1,1 in textbook )
当二项分布的 n 很大， p 很小
的时候，相当于泊松分布
Poisson approximation to the binomial distribution lamda 等于 np 时候的分布
Section 2.1 The Poisson
Process
8

because when we assume

that definition 2 satisfies,
then the possion ditribution is
independent
Section 2.1 The Poisson
Process
9
 Section 2.1 The Poisson
Process
10
 Section 2.1 The Poisson
Process
11

please use
mathematical
induction
 Section 2.1 The Poisson
Process
12

泊松分布指的是在一段时间内发生的次
数，这个一段时间需要自己设定，并且
在 lamda 中体现。比如这道题，原
本 lamda 是单位时间内（ 1 为单
位）中发生的平均次数，但是这里要求
的是一段时间 t 中发生的平均次数，所
以需要在原有的次数中乘以时间，也就
是 lamda*t 。新的 P(x=k) 指
的是在新的时间标准下（时间 t 为时间
上的一个单位的情况），在 t 时间内总
共发生 k 次数的概率
 Section 2.1 The Poisson
Process
13

from uni

在已知 definition 2 的情况下

相当于拓展范围
 Section 2.1 The Poisson
Process
14

The common occurence of the Poisson distribution in nature is explained

by the law of rare events. Informally, this law asserts that where a certain
event may occur in any of a large number of possibilities, but where the
probability that the event does occur in any given possibility is small, then
the total number of events that do happen should follow, approximately,
the Poisson distribution.

Example: A large number of cars may pass through a given stretch of

highway on any particular day. The probability that any specified car is in
an accident is, we hope, small. Therefore, one might expect that the
actual number of accidents on a given day along that stretch of highway
would be, at least approximately, Poisson distributed. （撞车的概率很大，但是特定数
量的车撞的概率是很小的）
 Section 2.2 Interarrival and Waiting
Times
15

For a Poisson process let X1 denote the time of the first event.
Furthermore for n>1, let Xn denote the elapsed time between the
(n-1)th and the nth event. The sequence {Xn, n=1,2,…} is called
the sequence of interarrival times( 到达间隔时间 ).
What is the distribution of the Xn?

1st event 2nd event 3rd event 4th event

0 x1 t
x2 x3 x4
 Section 2.2 Interarrival and Waiting
Times
16

Therefore, from the foregoing, we conclude that X2 is also an

exponential random variable with mean 1/ λ and, furthermore, that X2
is independent of X1. Repeat the same argument yields:
 Section 2.2 Interarrival and
Waiting Times
17

Proposition: The interarrival times X1, X2, … are

independent and identically distributed exponential
random variables with parameter λ.

Remark: The assumption of stationary and independent increments is

equivalent to asserting that ， at any point in time, the process
probabilistically restarts itself. That is ， the process from any point on
is independent of all that has previously occurred (by independent
increment ， and also has the same distribution as the original process
(by stationary increments). In other words ， the process has no
memory, (exponential distribution is also memoryless)and hence
exponential interarrival times are to be expected.( 是可以预测到的，平均时间就是
1/lamda)
 Section 2.2 Interarrival and Waiting
伽玛密度函数的基本事件的发生的密度函数是不固定的，但是我们只讨论服从泊松分布的情况，在
Times 这个情况下，等待时间服从指数分布
18

What is the distribution of Sn?

Corollary: The sum of n independent exponential random
variables, each having parameter λ, is a gamma random variable
with parameter ( n,  1 /  ).
Thus Sn, the time of the nth event of a Poisson process having
rateλ, is a gamma random variable with parameter
( n,  1 /  )

18
 Section 2.2 Interarrival and
Waiting Times
19
 Section 2.2 Interarrival and
Waiting Times
20

N(t) is the poisson distribution

 Section 2.2 Interarrival and
Waiting Times
21
 Section 2.2 Interarrival and Waiting
Times
22

Definition 3. N(t) constructed as above is a Poisson process

of rate λ.
 Section 2.2 Interarrival and
Waiting Times
23

Example 2.2.1 Suppose that people immigrate into

a territory at a Poisson rate λ = 1 per day.
(a) What is the expected time until the tenth

immigrant arrives?
(b) What is the probability that the elapsed time

between the tenth and

equals thesum
to the eleventh arrival ofexceeds
of xi, the expectation every xi is 1/lamda
two days?
 Section 2.3 Conditional
Distribution of the Arrival Times
24

This is easily checked since, for s ≤ t,

Let {Nt, t≥ 0} be a Poisson process. Then, conditional on {Nt, t≥ 0} having

exactly one event taken place in the interval [0,t], the time at which the
event occur is uniformly and independently distributed on [0,t].

Given that the number of events taken

place before t is n(t) = n, what are the
distribution of the n arrival times S1,
S2, . . . , Sn?
 Section 2.3 Conditional
Distribution of the Arrival Times
25

Order Statistics. 等于无序概率的所有可能排列的和

Let Y1, Y2, . . . , Yn be n random variables. We say that Y(1), Y(2), . . . , Y(n) are
the order statistics corresponding to Y1, Y2, . . . , Yn if Y(k) is the kth smallest value
among Y1, Y2, . . . , Yn , k = 1, 2, . . . , n. If the Yi, i = 1, . . . , n, are independent
identically distributed continuous random variables with probability density f ,
then the joint density of the order statistics Y(1), Y(2), . . . , Y(n) is given by
从小到大排序
 Section 2.3 Conditional
Distribution of the Arrival Times
26

Theorem 2.3.1 Given that n(t) = n, the n arrival times

S1, S2, . . . , Sn have the same distribution as the order
statistics corresponding to n i.i.d. uniformly distributed
random variables from (0, t).
27
 Section 2.3 Conditional
Distribution of the Arrival Times
28

Example 2.3.1 Suppose that travelers arrive at a train depot in

accordance with a Poisson process with rate λ . If the train departs
at time t, what is the expected sum of the waiting times of
travelers arriving in (0, t)?
Solution
Solution
 Section 2.3 Conditional
Distribution of the Arrival Times
31

Superposition of Independent Poisson Processes

Theorem. Superposition of independent Poisson Processes

N

(λi, i = 1, . . . ,N), is also a Poisson process with rate 

i 1
i
 Section 2.3 Conditional
Distribution of the Arrival Times
32

Prove the theorem (note that a Poisson process must satisfy

Definitions 1 or 2).
 Section 2.3 Conditional
Distribution of the Arrival Times
34

Decomposition of a Poisson Process

Given a Poisson process N = {N(t), t ≥ 0};
• If Ni(t) represents the number of type-i events that occur by time
t, i = 1, 2;
• Arrival occurring at time s is a type-1 arrival with probability
p(s), and type-2 arrival with probability 1 − p(s)
then

泊松分布服从二项分布
 N1, N2 are independent
 N1(t) is Poisson random variable having mean λtp, and
 N1(t) is Poisson random variable having mean λt(1-p), where
因为是变化的，所以用均值代替，如果不变化，那么 p 就
是实际值
Section 2.3 Conditional
Distribution of the Arrival Times

special case: If p(s) = p is constant, then

 Section 2.3 Conditional
Distribution of the Arrival Times
Proof.

Now consider an arbitrary event that occurred in the interval [0,t]. If it had occurred
at time s, then the probability that it would be a type I event would be p(s). Hence,
since this event will have occurred at some time uniformly distributed on (0,t). It
follows that

the goal if to find the pribability of s

when until time t there has been
one inccident
Proof. (cont.)

are independent Poisson

process
 Section 2.3 Conditional
Distribution of the Arrival Times
Example 2.3.1 (An Infinite Server Queue ）

Suppose that customers arrive at a

service station in accordance with
a Poisson process with rate λ.
Upon arrival the customer is
immediately served by one of an
infinite number of possible servers,
and the service times are assumed
to be independent with a common
distribution G.

type-1 customer: if it complete its service by time t

type-2 customer: if it does not complete its service by time t
 Example 2.3.1 (An Infinite Server Queue ）

N1(t)
N2(t)

N1(t) N2(t)

Any customer arrives at time s, s ≤ t, then it is a type-1 customer if its service time is
less than t – s and the probability is
Answer:

Using the decomposition of a Poisson Process, we obtain that the

distribution of N1(t)- the number of customers that have complete
service by time t – is Poisson with mean

Similarly N2(t)- the number of customers being served at time t –

is Poisson distributed with mean
 Section 2.4 Nonhomogeneous
Poisson Processes
Definition The counting process {N(t), t≥ 0} is said to be a
nonhomogeneous Poisson process with intensity function λ(t), t ≥ 0,
if

Define “integrated intensity function”

Theorem.

P[N(t+s)-N(t)=n]
Section 2.4 Nonhomogeneous
Poisson Processes
Section 2.4 Nonhomogeneous
Poisson Processes
Section 2.4 Nonhomogeneous
Poisson Processes
 Section 2.4 Nonhomogeneous
Poisson Processes
Example 2.4.1 Siegbert runs a hot dog stand that opens at 8 A.M. From 8 until 5
P.M. customers seem to arrive, on the average, at a steadily increasing rate that starts
with

 an initial rate of 5 customers per hour at 8 A.M.

 and reaches a maximum of 20 customers per hour at 11 A.M.
 From 11 A.M. until 1 P.M. the (average) rate seems to remain constant
at 20 customers per hour.
 However, the (average) arrival rate then drops steadily from 1 P.M.
until closing time at 5 P.M. at which time it has the value of 12
customers per hour.

a) If we assume that the numbers of customers arriving at Siegbert’s stand during

disjoint time periods are independent, then what is a good probability model for the
preceding?
b) What is the probability that no customers arrive between 8:30 A.M. and 9:30 A.M.
on Monday morning? What is the expected number of arrivals in this period?
Solution:
A good model for the preceding would be to assume that arrivals constitute a
nonhomogeneous Poisson process with intensity function λ(t) given by
 Section 2.4 Nonhomogeneous
Poisson Processes
Example 2.4.2. It can be shown that The “output process” of the M/G/∞ queue
(Infinite server queue having Poisson arrivals and general service distribution G )
is a non-homogeneous Poisson process having intensity function λ(t) = λG(t),
where G is the service distribution.

Queueing system notation: A / B / m

A : arrival process
B : service process
m : the number of servers
• M/G/1 queue: Markov arrivals, general service time, 1 server.
• G/M/1 queue: General inter-arrival distribution, Exponential
Service times, 1 server.
• M/G/∞ Markov arrivals are immediately attended by a server,
with general service time distribution.
 Section 2.4 Nonhomogeneous
Poisson Processes
Example 2.4.2. (cont.)

Hint: Let N(s, s + t) denote the number of service completions

in the interval (s, s + t] . If we can show that
s t
1) N(s, s+t) follows a Poisson distribution with mean  
s
G ( y )dy,
and

2) the numbers of service completions in disjoint intervals are

independent,

then we are finished by definition of a non-homogeneous Poisson

process.
Example 2.4.2. Proof

Hence, from the decomposition of a Poisson Process, the number of such

departures will be Poisson distribution with mean
 Section 2.5 Compound Poisson
Processes

The random variable X(t) is said to be a compound Poisson random variable.

E[X(t)] = E[N(t)]E[Y]=λtE[Y]
Var[X(t)] =E[N(t)]Var[Y]+Var[N(t)](E[Y])2 = λtE[Y2]
Section 2.5 Compound Poisson
Processes
Section 2.5 Compound Poisson
Processes
 Section 2.5 Compound Poisson
Processes
Examples of Compound Poisson Processes

(ii) Suppose that buses arrive at a sporting event in accordance with a Poisson
process, and suppose that the numbers of fans in each bus are assumed to be
independent and identically distributed. Then {X(t), t ≥ 0} is a compound Poisson
process where X(t) denotes the number of fans who have arrived by t. Yi
represents the number of fans in the ith bus.

(iii) Suppose customers leave a supermarket in accordance with a Poisson

process. If the Yi , the amount spent by the ith customer, i = 1, 2, . . . , are
independent and identically distributed, then {X(t), t ≥ 0} is a compound Poisson
process when X(t) denotes the total amount of money spent by time t.
Section 2.5 Compound Poisson
Processes
Example 2.5.1 Suppose that families migrate to an area at a Poisson rate λ =
2 per week. If the number of people in each family is independent and takes
on the values 1, 2, 3, 4 with respective probabilities 1/6,1/3,1/3,1/6, then what
is the expected value and variance of the number of individuals migrating to
this area during a fixed five-week period?
Solution:
Section 2.5 Conditional Poisson
Processes
Section 2.5 Conditional Poisson
Processes