An Application of ADT Queue: Simulation

Tom Kelliher, CS23

Mar. 25, 1996

Simulation

A technique for modeling the behavior of natural and man-made systems. Useful for summarizing the performance of existing systems or predicting the performance of proposed systems.

• Event: ``Something'' scheduled to happen.
• The notion of time in simulation: Event driven simulation.
• Obtaining good input data.
• Correctly modeling the system.
• Interpreting the results.

A Banking Example

Customers at the First National Bank of Avarice are complaining of having to wait too long in line. What should the bank president, Mon E. Baggs, do:

• Continue serving customers with one teller?
• Use three tellers with three lines?
• Use three tellers with a single line?

What is our data?

What do we measure?

Data file is ordered pairs:

1. Arrival time.
2. Service time.

Assumptions:

1. Upon arrival, customer immediately enters (shortest) teller line.
2. When teller and customer finish:
   • Customer immediately leaves.
   • Next customer immediately begins service.

Attributes of an event:

1. Time of occurrence.
2. Activity, if any, to trigger.

What kinds of events:

1. Arrival event: Get next arrival event. Additional information: service time.
2. Departure event: Service next customer, accumulate statistics.

From arrival event to departure event: where are the waiting customers?

What we need to keep track of:

1. Time.
2. Waiting customers.
3. Events. (How many?)
4. Input data.
5. Statistics.

What ADTs can we use/do we need?

Is there a hierarchy to these ADTs?

How does it fit together?