4/30/02: Deadline for submission of final reports: 5/17.
2/20/02: check the updated syllabus -- all talks assigned.
2/06/02: check the updated syllabus.
1/22/02: first lecture.

About the course

Instructor: Sergio Servetto
Meeting times and place: Tu-Th, 10:10-11:25am, 407 Phillips Hall.
Office hours: Right after class, by appointment, or any time I am in the office.

Summary: Network communication problems, i.e., problems in which more than one sender and/or receiver may be present, are very different from classical point to point communication problems. For example, issues such as interference and cooperation among users, or burstiness of the sources, can have a significant impact on the design of a communications system, and arise only in the network case. As in the classical case, the holy grail of a theory of information in networks is a characterization of the ultimate limits for the performance of systems that communicate over network channels, as well as the derivation of guidelines to aid in the search for efficient codes and algorithms capable of achieving those bounds. But unlike in the classical case, where Shannon theory provides answers to almost any interesting question one could conceive (at least in many important cases), signficantly less is known about network channels. In this course we will pursue two main goals: we will study in depth whatever is known about a variety of multiuser problems, and we will work on identifying good open problems for research.

Prerequisites: Consent of instructor. ECE 562 (Information Theory) is highly recommended. A course on Coding Theory is not required, but might provide useful background/intuition. However, the main thing that you need to take this class is a taste for good research problems...