Sei sulla pagina 1di 14

6/4/2012

Presentation On Queuing Theory

Siddharth Bhalerao

6/4/2012 Siddharth Bhalerao

Notion of Queues are not new to us; as each one of us has spent great amount of time waiting in queues.
Queuing Theory is defined as mathematical study of queues. Queuing theory find its application in diverse fields such as communications, computers, factories, etc. Agner Erlang, a Danish engineer is considered as pioneer of queuing theory.

6/4/2012 Siddharth Bhalerao

Role of Simulation
A system designer/analyst requires analysis of system to improve performance parameters like server utilization, length, waiting times of queue, etc.
6/4/2012 Siddharth Bhalerao

A mathematical analysis is possible for simple systems, but for realistic models simulation is required.
Using the result of simulation a designer is able to get maximum performance out of the system.

Characteristics of Queuing Systems


There are 3 main characteristics of queuing systems which are: Arrival Process: It describes how entities or customers are entering in queuing system for service. Queue Discipline: It describes how next entity is going to be served in the system. Service Process: It includes service time and service mechanism
6/4/2012 Siddharth Bhalerao

Other than these, queuing systems are also characterized by:


6/4/2012 Siddharth Bhalerao

Calling Population: The population of potential customers is referred to as calling population. It could be assumed to be finite or infinite. An infinite population system sometimes makes a model easy as it assumes constant arrival rate. In finite population system arrival rate changes with customers arriving and leaving the system. System Capacity: Capacity of system is defined by number of customers who are waiting to be served. A system could be of limited or unlimited capacity.

The Arrival Pattern


Arrival process has important aspects of interarrival times of successive entities entering the system and arrival rate . Arrivals are random and are characterized by Probability Distribution Function and are modeled mostly using Poisson arrival process. For any random process to be a Poisson process with mean arrival rate ; following assumptions are made: 1. Arrivals occur one at a time. 2. Arrival process has stationary increments. 3. Arrival process has independent increments. If above assumptions are fulfilled, then probability of having n entities in a system is given by: P(n)= e-t(t)n/n!

6/4/2012 Siddharth Bhalerao

Queue Behavior and Queue Discipline


Queue behavior is the actions of customers or entities present in the queue. Balking: A customer leaving the line when they see it is too long. Reneging: A customer leaving the line on observing it is moving too slowly. Jockeying: Moving from one line to other thinking they have chosen slow line.
6/4/2012 Siddharth Bhalerao

Queue discipline refers to the order in which entities present in queue are served.
6/4/2012

Following are common queue disciplines. First -in -first -out (FIFO)

Siddharth Bhalerao

Last -in -First -out (LIFO)


Service in random order (SIRO) Shortest processing time first (STP) Priority service (PR)

Service times and Service Mechanism


The time spent by each entity at server is precisely the service time. The service time could be constant or of random duration. Random service times are characterized by PDFs.
6/4/2012 Siddharth Bhalerao

Mostly exponential, Weibull, gamma, lognormal distributions are used.

Continued.
6/4/2012

Service Mechanism involves number of server present for service. There could be a single server mechanism or parallel service mechanism. In parallel server system there could be multiple or unlimited servers present.

Siddharth Bhalerao

Queuing Notations
A widely adopted queuing notation was proposed by Kendall, and are effective for recognizing the type of queuing system. Any queuing system can be identified based on format A/B/c/N/K A or B could be M for Markov, D for deterministic, Ek for Erlang, H for Hyperexponential, G for General and GI for General Independent.
6/4/2012 Siddharth Bhalerao

Measurement Criteria
6/4/2012

It includes: L - Average number of customers in system W- Average time spent in the system Server utilization denoted by

Siddharth Bhalerao

6/4/2012 Siddharth Bhalerao

Thank You

Potrebbero piacerti anche