Optimal Control of Queueing Systems with Intermittent Service.
Abstract
The report describes models for single-server queueing systems with Poisson arrivals and general service-time distribution which are controlled by turning the server on-and-off. The objective is an operating policy which minimizes (maximizes) expected discounted cost (reward) over an infinite horizon. Four distinct models of intermittent service systems are considered. The cost structure for these models includes fixed costs for starting-up and shutting-down the service facility, a server operating cost per unit time and either a holding cost for waiting customers or a reward for serving customers. Two of the models are based on different assumptions concerning the holding cost function. The two remaining models include provisions for balking (an arriving customer chooses not to join the queue) and reneging (customers leaving the queue without being served). For each of the models, there exists an optimal policy characterized by a pair of critical numbers (N,M): turn the server on whenever the number of customers equals (or exceeds) M and turn the server off whenever the number of customers is less than or equal to N. Algorithms for computing the optimal critical numbers are described. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- May 31, 1971
- Accession Number
- AD0728006
Entities
People
- Joseph D. Blackburn
Organizations
- Stanford University