Estimation of Quantile Confidence Intervals for Queueing Systems Based on the Bootstrap Methodology

Journal
Communications in Computer and Information Science
Applied Computer Sciences in Engineering
ISSN
1865-0929
1865-0937
Date Issued
2017
Author(s)
Type
text::book::book part
DOI
URL
Abstract
This paper presents a simple methodology for estimating confidence intervals of quantiles in queueing systems. The paper investigates the actual probability density function of quantile estimators resulting of independent replications. Furthermore, we present a methodology, based on the concepts of bootstrapping, i.e., re-sampling and sub-sampling, to calculate the variability of an estimator without running different independent replications. Contrary to what overlapping and non-overlapping batching procedures suggest, we propose to randomly select data points to form a sub-sample, instead of selecting time-consecutive data points. The results of this study suggest that this proposal reduces the correlation between sub-samples (or batches) and overcomes the issue of normality. © 2017, Springer International Publishing AG.