Font Size: a A A

The combined effect of marginal distributions and correlations in teletraffic

Posted on:2004-02-05Degree:M.SType:Thesis
University:University of Missouri - Kansas CityCandidate:Ouppariputtikul, PattareeyaFull Text:PDF
GTID:2465390011471314Subject:Computer Science
Abstract/Summary:
In this paper, we observe the effect of marginal distributions and correlations on auto-correlation of departures, call and time blocking probabilities, and mean number in the system. Generalized interrupted Poisson processes (GIPP), which allows convenient matching moments of the marginal distribution and correlation coefficient independently, are used as arrival processes. Two types of queueing system, the G/M/1/k and the G/M/m/m, are investigated. Also used are the correlated distributions creating from renewal processes obtained by taking the marginal distributions of the GIPP only and then applying the correlated structure without changing the first order properties. In particular, it is studied whether the higher moments or the higher correlations have a large impact on the performance measures.
Keywords/Search Tags:Marginal distributions, Correlations
Related items