Skip to main content
Technical Reports
Navigation
Technical Reports
UCSC-CRL-93-42: AN APPROACH TO THE DECOMPOSITION OF LARGE STOCHASTIC MODELS
« Previous
Next »
Authors
Published On
09/01/1993 09:00 AM
Department
Computer Engineering
Abstract
A large family of decomposition methods largely used in the solution of queueing models divides the queueing model into smaller subsystems. The latter must be solved for all values of the state description retained for the aggregate model which represents the interactions among the subsystems. For a larger queueing system, even with relatively simple decomposed subsystems, this may represent a significant computational effort. The goal of our approach is to avoid the evaluation of the subnetworks for all values of the state description vector. Instead, we propose to track the probability distribution for the aggregate state description, and to evaluate the subnetwork in the regions where this probability exhibits a local maximum. In other regions, we propose to evaluate the subnetwork at only a few points. Outside the evaluation points, we use a simple curve fitting to generate the missing points.
Download
This report is not available for download at this time.
Error | Technical Reports
Technical Reports
Error
The website encountered an unexpected error. Please try again later.