Queuing Models

Download Report

Transcript Queuing Models

Queuing Models
Dr. Mahmoud Alrefaei
1
Introduction
• Each one of us has spent a great deal of
time waiting in lines.
• One example in the Cafeteria.
• Other examples of queues are
– Printer queue
– Customers in front of a cashier
– Calls waiting for answer by a technical support
2
What makes up a queue?
• The System: A collection of objects under study.
– It is important to define the system boundaries.
• The Entities: The people, organisms, or objects
that enter the system requiring some kind of service.
• The Servers: The people, organisms, or machines
that perform the service required.
• The Queue: An accumulation of entities that have
entered the system but have not been served.
3
Types of Queues
Single Stage
4
Multiple Stage - Manufacturing
Plant
5
Multi-channel Single Stage Bank
6
Parallel Single Stage Supermarket
7
Customer Discrimination – Bus
station
8
Converging Arrivals - Walk-in
and Drive-thru
9
Queue Discipline
• First Come First Served - FCFS
– Most customer queues.
• Last Come First Served - LCFS
– Packages, Elevator.
• Served in Random Order - SIRO
– Entering Buses
• Priority Service
– Multi-processing on a computer.
– Emergency room.
10
What factors affect system
performance
• The Arrivals Process.
– The time between any two successive arrivals
– Does this depend on the number of people in the
system?
– Finite populations.
• The Service Process.
– The time taken to perform the service.
– Does this depend on the number of people in the
system?
• The number of servers operating in system.
11
Measuring System Performance
• The total time an “entity” spends in the
system (Denoted by W)
• The time an “entity spends in the queue.
(Denoted by Wq)
• The number of “entities” in the system.
(Denoted by L)
• The number of “entities” in the queue.
(Denoted by Lq)
12
Measuring System Performance
• The percentage of time the servers are
busy (Utilization time)
• These quantities are variable over time.
13
Poisson Process
• Let {N(t), t>0} be the number of customers
arrive until the time t
• {N(t), is said to be a Poisson Process
having rate , for >0, if
– N(0) = 0
– N(t+s) – N(t) Does not depend on the
previous history
– N(t+s) – N(t) is independent of t.
14
Poisson Process
The number of events in any
interval of length t is Poisson
distributed with mean t. That is for
all s, t > 0 and n=0,1,2,...
(  t )  t
PN (t  s )  N ( s )  n 
e .
n!
n
15
Inter-Arrival Times
• What are inter-arrival times?
3
T3
T2
A(t)
2
T1
1
0
0
2
4
6
8
10
12
Time (t)
S1
S2
S3
• Ti is the time between the (i-1)-st and the i-th events.
• Poisson Process can be used for modeling arrival
process
16
Birth-Death Processes
• A birth-death process is used to model
populations of entities in a system
• The state of the system at time t is the number
of entities in the system at that time, often
denoted by N(t).
• Births and deaths occur at a constant rate (like
the Poisson process model)
17
Birth-Death Processes
Pi,j(t) is defined to be the probability that there
are j entities in the system at time t, given that
there were i entities in the system at time 0.
0
0
1
1
2
2
j
 j 1
1
...
j-1
j
j
j+1
...
 j 1
18
Birth-Death Processes
• The state of the system must be a nonnegative integer
• Law 1
– A birth increases the state from j to j+1
– The variable j is called the birth rate for
state j
– A birth occurs between times t and t + t with
probability jt + o(t)
19
Birth-Death Processes
• Law 2
– A death decreases the state from j to j-1
– The variable j is called the death rate for
state j (note that 0=0)
– A death occurs between times t and t + t with
probability jt + o(t )
• Law 3
– Births and Deaths are independent
20
Birth-Death Processes
• Can more than one event happen between
t and t + t?
• Why must 0=0?
• Knowledge of j and j completely
specifies a Birth-Death process.
21
Birth-Death Processes
• Birth-Death Processes can be used to
model most M/M/... queuing systems.
– An arrival is considered a “birth”.
– A service completion is considered a “death”.
• Let Pi,j(t) be the probability N(t+s)–N(s)=j
given that N(s)=i (or N(t)=j given N(0)=i
• It turns out that for many queuing systems,
Pi,j(t) will approach a limit j as t gets larger.
22
Birth-Death Processes
• This limit will be independent of the
initial state i.
• j is called the steady state or
equilibrium probability of state j.
– j can be thought of as the probability that
at some instant in the future there are j
entities in the system.
– j can also be thought of as the fraction of
time that there are j entities in the system.
23
Exponential Distribution
• The exponential distribution is
 t
characterized by f(t)  e
t
F(t)  P (T  t ) 
 s
 t

e
ds

1

e

0

E[T ] 
 tf (t )dt 
0
1

E[T 2 ]   t 2 f (t ) dt
0
Var (T )  E[T ]  E[T ] 
2
2
1
2
24
Exponential Distribution
• What is P(T>t) for an exponential
distribution with parameter ? e- t
• P(T>t+s|T>s) is the probability of
waiting a further time t after having
already waited to time s.
• What is P(T>t+s|T>s) for an
exponential distribution with
parameter ? e- (t+s) / e- s = e- t
• Answer: P(T>t+s|T>s) = P(T>t) This is
called the memoryless property
25
An Example of Memoryless
• Suppose that the amount of time one spends in a
bank is exponentially distributed with mean ten
minutes.
– What is ?
– What is the probability that a customer will spend
more a quarter of an hour in the bank?
• You have been waiting for ten minutes already. Now
what is the probability that you will spend more than
a quarter of an hour in the bank?
– What has a lack of memory, you or the
distribution?
26
Birth-Death Processes
• Consider a M/M/1 queuing system.
– Inter-arrival times are exponential with rate .
– Service times are exponential with rate .
• Suppose there are j entities in the system at
time t.
What is the probability of an arrival in the interval
(t,t + t]? Hint: use Taylor series expansion on
F(t + t)-F(t) = 1 - e- t
=
 t + o(t)
27
Birth-Death Processes
• So the arrivals and service completions of
a M/M/1 queue are a birth-death process
with the following rate diagram


0
1

2



...
j-1
j

j+1 ...

28
Birth-Death Processes
(Balance equations)
• It can be shown by induction that
 j 1 j 1   j 1 j 1   j j   j j
• These are called balance equations
• Therefore:
01   j 1
j 
0
1 2   j
29
Birth-Death Processes
• If we define the constants
01   j 1
cj 
12   j
• then
 j  c j 0

• and we know that

j 0

j
  0   c j 0  1
j 1
30
Birth-Death Processes
• So
π0 
1
1

c
j1
j
• This means that the infinite sum of the cj’s
must converge.
• If this sum is infinite then no steady-state
distribution can exist.
31