Академический Документы
Профессиональный Документы
Культура Документы
Presented by: 1. Roopam Pawar (112013) 2. Rohit Gurjar (112012) 3. Rahul Kr. Garg (112011)
Overview (I)
What is Queuing Theory Introduction History Applications Need of Queuing Theory Queuing Analysis Cost vs. Service Graph What is queuing system Block Diagram Characteristics Real world examples
Overview (II)
What is Queuing Process Components Multiple vs. Single Configuration Structure Elements Kendall Notations Operating Characteristics Queuing Models Classification Performance Measures Flow Charts Numerical Problems
cont`d
in 1953.
Objective behind its development was efforts to analyze telephone traffic congestion with the aim of satisfying the randomly arising demand for the services of Copenhagen telephone system in 1909. Erlang observed that a telephone system can be modeled by Poisson customer arrivals and exponentially distributed
4
service times.
Mathematical analysis of queues and waiting times in stochastic systems. Used extensively to analyze production and service processes exhibiting random variability in market demand (arrival times) and service times. Queues arise when the short term demand for service exceeds the capacity Most often caused by random variation in service times and the times between customer arrivals. If long term demand for service > capacity the queue will explode!
Therefore, to meet service demand Increase service capacity; and /or Raise capacity of existing capacity/facility, if possible. It is possible to build capacity to such high level that can always meet peak demand with no queues. This would become uneconomical after a stage because capacity will remain idle to
cont`d
Managers/ Management job is to balance. Inefficient/poor service excessive waiting cost in terms of:-
(a) Customer frustration (b) Loss of goodwill in the long run (c) Direct cost of idle employees (e.g. employees waiting at the store for collection of tools). Conversely stations. Goal of queuing modeling achievement of economic balance between too high a service level high set-up cost idle time for service
Prototype Example ER at County Hospital Patients arrive by ambulance or by their own accord One doctor is always on duty More and more patients seeks help longer waiting times
Customer arrivals
A queuing system can be described as follows: "customers arrive for a given service, wait if the service cannot start immediately and leave after being served" The term "customer" can be men, products, machines, ...
11
Queuing Systems can be characterized with several criteria: Customer arrival processes Service time Service discipline Service capacity Number of service stages
12
Queuing Process
Input Source
Calling Population
Jobs
Queue
Service Mechanism
cont`d
16
Multiple Queues
Servers
Single Queue
Servers
17
2. Labor specialization possible 3. Customer has more flexibility 4. Balking behavior may be deterred
Several medium-length lines are less intimidating than one very long line
2. No customer anxiety regarding choice of queue 3. Avoids cutting in problems 4. The most efficient set up for minimizing time in the queue 5. Jockeying (line switching) is avoided
18
cont`d
The Service Mechanism Can involve one or several service facilities with one or several parallel service channels (servers) - Specification is required The service provided by a server is characterized by its service time Specification is required and typically involves data gathering and statistical analysis. Most analytical queuing models are based on the assumption of exponentially distributed service times, with some generalizations. The queue discipline Specifies the order by which jobs in the queue are being served. Most commonly used principle is FIFO. Other rules are, for example, LIFO, SPT, EDD Can entail prioritization based on customer type.
19
Population
Queue
Service System
Customers leave
Arrival Process
20
1. Arrival Process.
(a) According to source (i) (ii) (b) Finite Infinite
According to numbers
(i)
(ii) (c)
Individual, as in a bank
In groups, as in a ship
According to time
(i)
(ii)
21
22
Notation of Kendall
The following is a standard notation system of queueing systems T/X/C/K/P/Z with T: probability distribution of inter-arrival times X: probability distribution of service times C: Number of servers K: Queue capacity P: Size of the population Z: service discipline
23
T can take the following values: M : markovian (i.e. exponential) G : general distribution D : deterministic Ek : Erlang distribution If the arrivals are grouped in lots, we use the notation T[X] where X is the random variable indicating the number of customers at each arrival epoch P{X=k} = P{k customers arrive at the same time} Some arriving customers can leave if the queue is too long
24
26
Capacity K
27
Size of the population T/X/C/K/P/Z The size of the population can be either finite or infinite
For a finite population, the customer arrival rate is a function of the number of customers in the system: l(n).
28
(b) This average rate is the same for all equal-sized intervals.
(c) The actual number of arrivals in one interval has no bearing on the actual number of arrivals in another interval. (d) There cannot be more than one arrival in an interval as the size of the interval approaches zero. The mean value of the arrival rate, i.e. average arrival rate, is represented by (Lambda).
30
= average arrival rate (i.e. average number of arrivals per unit interval)
e = 2.71828 or 2.7183 {# - The intervals could be time or space (length/ distance).}
(b) Occurrence or non-occurrence in any interval# is independent of the occurrence or non-occurrence in any other interval#. 31
= average service rate (i.e. average number of customers served per unit time
e = 2.7183 Expected time (mean of the exponential distribution), E(t) = 1/
Note : Mean of exponential distribution is the inverse of the mean of 33 the Poisson distribution.
Queue Arrivals
Service facility
Customers leave
Queue
Customers leave
Arrivals
36
Arrivals Arrivals
Customers leave
Arrivals
37
Arrivals
Multiple Servers in Series E.g. Machining of an item consisting of cutting, turning, grinding drilling and packaging
38
(ii) Last come-first served (LCFS) or Last in-First out (LIFO) basis (iii) Service in random order (SIRO) (iv) Priority Service E.g. Customers called according to some identifiable characteristics (say length of job), treatment of VIPs in preference to other patients in a hospital
39
involved.
(a) Jockeying i.e. going/moving from one queue to another in the hope of reducing waiting time.
(b) Balking i.e. not joining the queue at all because of the
anticipated long delay. (c) Reneging Leaving the queue and going away because they
42
An analysis of a given queuing system involves a study of its different operating characteristics. This is done using queuing models. Commonly used operating characteristics are :(a) Queue Length (Lq) Average number of customers in the queue waiting to get service. Long queues may indicate poor service performance or inadequate service facilities. Small queues may imply too much server capacity. (b) System Length (Ls) Average number of customers in the system,
Queuing Models
Assumption: Various models that we would now be seeing are based on the assumption that the system operates under equilibrium or steady state conditions and is not in the transient stage. For example, opening of a departmental store during a normal day or during a big sale day represent radically different initial conditions (small or non-existent initial queue versus long initial queue) which will affect the operating characteristics during the early part of the operating period. However, as time goes by, the system will eventually settle down to its long run or steady state tendencies.
45
Queuing Models
Another example for settling down into the steady state condition in
the long run is the concept of learning curves. When an aircraft is newly purchased, initial servicing time of the
Transient Phase
Time
46
1. If each customer or unit arrives at fixed or known intervals and the service time is known with certainty, the queuing model would be deterministic in nature.
2. Suppose customers come to a banks teller counter every 5 minutes. Interval between arrivals of any two successive customers is exactly 5 minutes. (a) Suppose the teller takes exactly 5 minutes to serve a customer here both arrival and service rates = 12 customers per hour. In such a situation, there will never be a queue and the teller will always be busy.
47
48
indefinitely; the service facility will always be busy; and the service system will eventually fail. if < or = there will be no waiting line; the proportion of
So, in other words, if < 1, the system will work and if > 1, the system will eventually fail. Note:- It has been proved mathematically that in a probabilistic model, even if = (i.e. =1) the system will eventually fail. Such deterministic models are only applicable for highly automated
50
51
f(x) = e- . x/ x!
where,
f(x) = probability of x occurrences in an interval = expected value or mean number of occurrences in an interval. Using this formula for our purpose, with average arrival rate equal to per period of time, then according to Poisson probability distribution, the probability that n customers will arrive in the system during a given time interval t is given by (a) (b) P(n arrivals in time t) = {e-t.(.t)n/n! } for n = 0, 1, 2, ..
Probability density function (pdf) of inter-arrival time ( i.e. time interval between two consecutive arrivals) = .e-t
(Note: is assumed to be constant over time and is independent of the number of units already serviced or served, queue length or any other random property of the queue).
53
Most common type of distribution used for service times is exponential distribution.
It involves the probability of completion of a service.
54
55
(b) Probability density function (pdf) of inter-service time (or time between two consecutive services) = .e-t (c) Probability that customer shall be serviced in more than time t= e-t
(d) P (no more than t time period needed to service a customer) = 1 e-t
56
(b) What is the probability that a server is busy at any particular point in time (denoted by )? It is the proportion of time that a server actually spends with the customer i.e. the fraction of time a server is busy.
(c) What is the probability of n customers being in the queuing system when it is in steady state condition? It is denoted by Pn , n = 0,1,2,3. (d) What is the probability of service denial when an arriving customer cannot enter the system because the queue is full? 4. Cost related questions. (a) What is the average cost needed to operate the system per unit of time? (b) How many servers are needed to achieve cost effectiveness?
57
Arrival Event Schedule the next arrival event Is the server busy?
Yes
No
Set delay=0 for this customer and gather statistics Add 1 to the number of customers delayed Make the server busy Schedule a departure event for this customer
Return
58
Yes
No
Subtract 1 from the number in queue Compute delay of customer entering service and gather statistics Add 1 to the number of customers delayed Schedule a departure event for this customer Move each customer in queue (if any) up one place
Return
59
Lq
Wq TH
Server utilization ratio = r Server idle ratio = P0 = 1 - r P{n > k} = Probability of more than k customers = rk+1
60
Example
On a network router, measurements show the packets arrive at a mean rate of 125 packets per second (pps) the router takes about 2 millisecs to forward a packet Assuming an M/M/1 model What is the probability of buffer overflow if the router had only 13 buffers How many buffers are needed to keep packet loss below one packet per million?
61
Example
Arrival rate = 125 pps Service rate = 1/0.002 = 500 pps Router utilization = / = 0.25 Prob. of n packets in router =
(1 - ) = 0.75 (0.25 )
n
Example
Probability of buffer overflow: = P(more than 13 packets in router) = 13 = 0.2513 = 1.49x10-8 = 15 packets per billion packets
10
n log 10
-6
/ log (0.25
= 9.96
63
Simple Problems
Problem 1 On an average, 6 customers reach a booth every hour to make calls. Determine the probability that exactly 4 customers will reach in 30minute period, assuming that arrivals follow Poisson distribution. Solution Given = 6 customers per hour t = 30 minutes = 0.5 hour n=4 t = 6*0.5 = 3 Probability of exactly 4 customers reaching in 30-minute period, = e-t.(t)n/n! = e-3.34/4! = 0.168
64
Simple Problems
Problem 2 In a bank, 20 customers on the average are served by a cashier in an hour. If the service time has exponential distribution, what is the probability that (a) It will take more than 10 minutes to serve a customer? (b) A customer shall be free within 4 minutes?
Solution
Given = 20 customers per hour t = 10 minutes = 10/60 = 1/6 hour (a) Probability that it will take more than 10 minutes to serve a customer = e-t = e-20*1/6 = 0.0357 (b) In this case, t = 4 minutes = 1/15 hour ; = 20 customers per hour Probability that a customer shall be free within 4 minutes 65 = 1 e-t = 1 e-20*1/15 = 0.736
66
(b) Expected number of customers who are in the system either waiting in the queue or being served (denoted by Ls). The data can be used for finding the mean customer time spent in the system.
67
NOTATIONS
1. n = no. of customers in the system (waiting and in service).
2. Pn = Probability of n customers in the system. 3. = Average (or expected) customer arrival rate or average arrivals per unit of time in the queuing system.
4. = Average (or expected) service rate or average number of customers served per unit time.
5. / = = Average service completion time (1/)/Average inter arrival time (1/) = Traffic utilization or server utilization factor (the expected fraction of time for which the server is busy) 6. s or c = number of service channels (service facilities or servers). 7. N = finite number of customers allowed in the system or finite population. 8. Ls = Average (expected) no. of customers in the system (waiting and in service). 9. Lq = Average (expected) no. of customers in the queue (queue length). 10. Lb = Average (expected) length of non-empty queue. 11. Ws = Average (expected) waiting time in the system (waiting and in service).
68