Academics

  /  

Courses

  /  

Descriptions

  /  

Keep

IEMS 464: Advanced Queuing Theory

VIEW ALL COURSE TIMES AND SESSIONS

Prerequisites

IEMS 460-1 (Stochastic Processes 1) or equivalent, or permission of instructor.

Description

Stochastic queueing theory, from classic results all the way to the cutting edge of the field! The class also provides techniques applicable to a wide variety of topics in stochastics and applied probability.

Primary topics include:

- Scheduling theory, from single-server to multiserver. Stationary-distribution analysis and optimality results. Policies include first-come-first-served, least-attained-service, preemptive-last-come-first-served, shortest-remaining-processing-time, and the Gittins policy. Analysis methods include the tagged-job approach and the WINE method.

- Stochastic networks, including dispatching and switching networks. State-space collapse results. Policies include join-the-shortest-queue, MaxWeight, and CARD.

- Markov-modulated arrival and completion processes, both finite and infinite.

- Drift analysis, including the transform method and the basic adjoint relationship (BAR).

As this is my first time teaching this class, this is a rough plan - we may move around or alter topics to keep things cohesive and educational.

Textbooks and materials:

- "Performance Modeling and Design of Computer Systems: Queueing Theory in Action", by Mor Harchol-Balter (https://www.cs.cmu.edu/~harchol/PerformanceModeling/book.html)

- "Communication Networks: An Optimization, Control, and Stochastic Networks Perspective", by R. Srikant and Lei Ying (https://sites.google.com/view/comm-networks)

- Ziv Scully's thesis, "A New Toolbox for Scheduling Theory" (https://ziv.codes/publications/#thesis)

- Various research papers which will be distributed as part of the class.