THE OPTIMAL CONTROL OF A QUEUEING PROCESS,
Abstract
Controlled queueing processes have been studied by several authors. The common model is a one-station queueing system at which customers arrive at a steady rate. The important feature is that management can vary the capacity of the service station according to the queue size, and thus, by increasing or decreasing the service intensity, control the queue size. The present paper describes the problem of determining the optimal control policy. Intervals are considered of service intensity values (not discrete sets). The arrival of customers at the service station is assumed to be a homogeneous Poisson process with a known arrival intensity. The service time is assumed to be exponentially distributed, with an intensity which can be changed and controlled. The cost structure of the system is discussed. There are three main cost components. The orientation (or set-up) cost of changing the service intensity; the service cost; and the queueing cost. The policy of optimal control is derived with an infinite horizon, which minimizes the expected total discounted future costs. This is carried out by setting and solving the proper Dynamic Programming functional equations.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 1969
- Accession Number
- AD0696990
Entities
People
- M. Yadin
- S. Zacks
Organizations
- Technion – Israel Institute of Technology