中文
Published date:2014-09-22    Provided by:School of Science
 Title: Modeling Service Networks with Time-Varying Demand-Performance Approximations and Staffing Control

Guest Speaker: Prof. Liu Yunan (North Carolina State University, USA)
Time: 2014-7-29, 15:00 - 17:00
Location: Room 7215, School of Science  


Content & Introduction:

    Motivated by the need for tools of forecasting and control in real service systems, we consider queueing networks with time-varying arrivals, customer abandonment, non-exponential service and patience distributions, and two types of routing policies: the conventional Markovian routing and the prescribed deterministic paths. We develop the many-server heavy-traffic fluid and diffusion limits. These limits have strong engineering relevance: First, they provide useful approximating formulas (in terms of the model input data) for the means, variances and distributions of important performance metrics, such as the waiting time and queue length. Second, these limits provide theoretic bases for developing efficient policies for staffing recommendations (in terms of the number of servers) so that the time-varying effect of the nonstationary demand can be reduced and controlled; we show that appropriate staffing methods can make nonstationary systems almost stationary.