Multiuser Information Theory


Fall 2005


                    Tu Th 12:30 - 1:45 pm EST
                    Room ITV 1111


                    Prakash Narayan
                    Room AVW 2353
                    Phone: (301) 405 3661
                    Fax: (301) 314 9281

Office Hours

                    Tu Th 2:00 - 3:15 pm EST
                     Also by appointment.


As the instructor will attend ISIT 2005 (September 4 - 9), the first lecture will be held
on Tuesday, September 13, 2005. Make-up classes will follow soon thereafter.

Course Outline

This course will cover three broad areas which are of special prominence in current interest
in information theory: multiuser information theory, cryptography and network coding.

We shall begin with a development of the notions of source coding (data compression) and
channel coding for a communication situation with multiple senders or receivers; the issues
of interference, cooperation and feedback will be addressed. Particular emphasis will be
placed on: Slepian-Wolf multiterminal data compression; Wyner-Ziv rate distortion; the
multiple-access channel, with many transmitters and a single receiver; and the broadcast
channel, with a single transmitter and multiple receivers. Recent developments including
multiple-input, multiple-output (MIMO) channels will be considered. Examples of multiuser
channels with fading and Gaussian interference, along with implications for TDMA and
CDMA in wireless communication, will be studied.

We shall then turn to a treatment of an information theoretic approach to cryptography,
with the technical tools being provided by multi-user information theory. This is relevant
to security in networks. An important feature concerns the development of the notion of
"unconditional secrecy" in terms of statistical independence in lieu of the notion of
computational complexity (which dictates current practice). We shall begin with an
introduction to the notions of common randomness and secret key capacity. Next, we shall
consider problems of common randomness and secret key generation by user terminals in
the presence of an eavesdropper. A new facet of this treatment will be an analysis of the close
connections between multiterminal (Slepian-Wolf) data compression which does not involve
any secrecy constraints, and secret key generation.

The final segment of the course will be a treatment of recent developments in the field of
"network coding."


ENEE 620 (Random processes), ENEE 621 (Estimation and detection) and ENEE 627
(Information theory), or permission of the instructor.


There is no required or recommended text. The course material will be drawn largely from
a selection of books and recent journal articles.

The following two standard books will serve as the main references for preparatory material
on multiuser information theory.

T.M. Cover and J. Thomas, Elements of Information Theory, Wiley, New York, 1991.

I. Csiszár and J. Körner, Information Theory: Coding Theorems for Discrete Memoryless
, Academic, New York, 1981.

Additional material will drawn from recent articles in the IEEE Transactions on Information
Theory, IEEE Transactions on Communications, IEEE Journal on Selected Areas in
Communications, and Proceedings of Eurocrypt.

Course grade

The course grade will be determined on the basis of a student's performance in (i) a midterm
assessment of progress on a term project, and (ii) a final assessment of the completed term
project. A term project will consist of work on an assigned topic involving research on open
issues combined with a comprehensive oral presentation, and will not be of the nature of a
survey of existing literature. Problems to be addressed in a term project will be assigned at the
end of approximately four (4) weeks into the semester. Final project reports are due no later
than the last day of examinations.