EE 649 Course Webpage:
Stochastic Network Optimization
Instructor:
Michael J. Neely (mjneely AT usc DOT edu, 213-740-3505, EEB 520)

Textbook:
M. J. Neely. Stochastic Network Optimization with Application to Communication and Queueing Systems. Morgan & Claypool, 2010.

A PDF file for the book is available from the above web link. It is a free download from USC computers, and from any other institution that subscribes to the "Synthesis Lecture" series. A hardcopy of the book can be ordered from the same link.

Syllabus:
A syllabus for Spring 2012 is here: EE 649 Course Syllabus (2012)
Brief Course Description:
This course presents a modern theory of stochastic optimization and cross-layer control for dynamic networks. The focus is on computer and wireless networks, including networks with time varying channels, mobility, and randomly arriving traffic. Applications to operations research and economics are also considered. The general theory of Lyapunov optimization is developed for constrained optimization of time averages. This is applied to problems such as queue stability, network utility maximization, efficient energy allocation, profit maximization, inventory control, stock market trading, and other problems involving dynamic decisions. Students use the theory in a final project on a topic of their choice.

Intended Audience: Graduate students in areas of networking, communication, controls, operations research, finance, economics.

Prerequisites:
EE 464 or 465.
Familiarity with stochastic processes (such as one of the following: EE 465, 550, 562a, 562b, 549, 556) is recommended but not required. There may be some computer problems in registering due to incorrect pre-reqs existing in the USC database for this course. If needed, I will approve enrollment for any student who has the basic probability background. If this issue arises, please contact me as early as possible, with the subject of "Registering for EE649" in the email subject heading.

Learning Objectives:
  1. To introduce students to the theory of dynamic decision making for networks and other stochastic systems.
  2. To teach students how to write complex problems in the standard form of minimizing an objective subject to an additional set of constraints. This includes linear and convex programs and their stochastic counterparts.
  3. To teach the Lyapunov optimization method, a powerful tool for solving problems involving constrained optimization of time averages.
  4. To present modern, cross-layer approaches to routing, resource allocation, and flow control. To introduce backpressure, max-weight, and virtual queue techniques.
  5. To explore hot-topic problems of opportunistic scheduling, approximate scheduling, dynamic data compression, efficient energy allocation, pricing, stock trading.
  6. To enable students to apply the theory by formulating and solving their own problems that involve dynamic decisions.