【Academic Seminar】Approximating steady-state performance measures in open queueing networks
- Topic: Approximating steady-state performance measures in open queueing networks
- Speaker: Wei YOU, Columbia University
- Time and Date: 2:00 pm – 3:00 pm, January 22, 2019
- Venue: Boardroom, Dao Yuan Building
Abstract：
Modern service systems such as ride-sharing services and cloud computing networks are dramatically changing the daily lives of many people. Many of these systems can be modeled as open queueing networks. The analysis of these complex models requires accurate estimation of the system performance measures, where exact solution is usually unavailable. Moreover, it is even challenging to develop approximation methods that address the complicated dependence within and between arrival and service processes. This dependence can have a significant impact on system performance measures but is often ignored in favor of tractability in applications. In this talk, I address these challenges by developing a new non-parametric stochastic modeling approach. Our method captures the performance impact of dependence by utilizing much more distributional information to model high dimensional dependence. The new method leads to a simple and fast algorithm without cumbersome simulation or numerical steps. The approximation algorithm is supported by new heavy-traffic limit theorems and extensive simulation studies.
Biography：
Wei You is a Ph.D. candidate in the Department of Industrial Engineering and Operations Research at Columbia University, advised by Professor Ward Whitt. His primary research interest lies in the broad area of applied probability and stochastic systems, and their applications to the performance analysis and design of service systems. He received his B.S. degree in Mathematics from Nanjing University, China.
