A Simulation Study of a Class of First-Come First-Served Queues with EARMA Correlation Structure.
Abstract
The object of this thesis is to examine, via simulation, the properties of a class of single-server, first-come first-served queues in which the service times and interarrival times, while still marginally exponential, are auto-correlated and cross-correlated. The correlation is introduced by letting the service and interarrival sequences by EARMA-type processes, where EARMA stands for exponential autoregressive, moving average sequence. An extension of these ideas brings in the cross-correlation. The waiting times in the dependent queue are compared to the waiting times (with known distribution) in the independent M/M/1 queue using data analytic and formal statistical methods. Variance reduction techniques are also studied; these use distributionally known aspects of the M/M/1 queue (simulated with the same exponential sequences as the dependent queue) to control the unknown aspects of the dependent queue. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1978
- Accession Number
- ADA062165
Entities
People
- Prasert Boonsong
Organizations
- Naval Postgraduate School