Lefschetz Center for Dynamical Systems Seminar
Stochastic Systems Seminar
Abstract: In a variety of modern technologies, resource allocation and dynamic scheduling play a critical role in congestion control and overall system operation. Such technologies include flexible manufacturing, telecommunication networks, computing systems, as well as numerous information services that are emerging on the Internet. A common theme across these areas is the management of interdependent resources.
To capture the concept of resource allocation in a diverse service environment, we formulate a multiclass queueing model with $ Q $ parallel waiting lines and interdependent service rates. The system can be in one of $ M $ service states at any point in time. When the system is in state $ m $ the $ q-th $ queue receives service rate $ r^{q}_{m} $. The issue is how to dynamically control the service state in order to maximize the throughput and minimize various operational costs.
* e-mail: marmony@stern.nyu.edu
** This is a joint work with Nicholas Bambos (bambos@stanford.edu)
This talk will focus on implementable, scalable dynamic policies for resource allocation, which address capacity and other design concerns in various applications. Specifically, the problems of power control in wireless networks, skilled based routing in call centers, and task scheduling in manufacturing systems will all be disucssed.
<--- 1999 Index