Limit Theorems for Controlled Queues.
STANFORD UNIV CALIF DEPT OF OPERATIONS RESEARCH
Pagination or Media Count:
The paper studies a class of controlled queues. In particular, the decision maker may choose one of two different servers to use. The objective is to minimize the expected average cost when there is a linear holding cost per unit time per customer, a different running cost per unit time for using each of the servers, and a switching cost incurred each time the server is switched. A functional limit theorem is proved for a sequence of random functions induced by a sequence of controlled queues, each of which is governed by a policy generated from a fixed set of switching points. The limit theorem holds when the traffic intensities for both servers are approaching one that is, the sequence of queues is approaching a state of heavy traffic. Modified author abstract
- Operations Research