2014 | OriginalPaper | Buchkapitel
Markov Decision Process and Linear Programming Based Control of MAP/MAP/N Queues
verfasst von : András Mészáros, Miklós Telek
Erschienen in: Computer Performance Engineering
Verlag: Springer International Publishing
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
We investigate the control problem of the optimal choice of idle server (if any) for arriving customer in order to minimize the mean system time (waiting time + service time). The considered MAP/MAP/N queue consists of a common infinite buffer and multiple identical servers with MAP service processes whose phases (internal states) are known. Customers arrive according to a MAP (whose phase is also known) and are served with work conserving policy. Idle servers preserve their phases.
We transform the obtained infinite state optimization problem to a finite state one and apply two optimization procedures, policy iteration of finite state MDP and linear programming.