Title :
On-line derivative estimation for the multiclass single-server priority queue using perturbation analysis
Author :
Miyoshi, Naoto ; Hasegawa, T.
Author_Institution :
Div. of Appl. Syst. Sci., Kyoto Univ., Japan
fDate :
2/1/1996 12:00:00 AM
Abstract :
Gradient estimation via smoothed perturbation analysis (SPA) is considered for the multiclass single-server priority queue. Although the problem like this requires additional subpaths, generally it is shown that the estimators can be derived without any additional sample paths
Keywords :
discrete event systems; queueing theory; gradient estimation; multiclass single-server priority queue; online derivative estimation; smoothed perturbation analysis; Analytical models; Bibliographies; Communication system traffic; Discrete event simulation; Discrete event systems; Distribution functions; Performance analysis; Queueing analysis; Stochastic systems; Time sharing computer systems;
Journal_Title :
Automatic Control, IEEE Transactions on