[CPCC] TALK 5/11 3 PM

[Ender Ayanoglu]

		     Multiuser Multiple Antenna Systems:
            Duality and Sum Capacity of Gaussian Broadcast Channels


				Sriram Vishwanath
			Department of Electrical Engineering
			    University of Texas at Austin


			      May 11, 2004 Tuesday
				     3:00 PM
			      Engineering Tower 331


				    Abstract

Wireless systems continue to strive for ever higher data rates. This
goal is particularly challenging for systems that are power, bandwidth,
and complexity limited. However, another domain can be exploited to
significantly improve system performance - the use of multiple transmit
and receive antennas. The study of the fundamental limits of these
multiple antenna (also called multiple input multiple output) systems is
essential for understanding and exploiting the enormous potential of
these systems.

In our work, we discuss a "duality" connection between Gaussian downlink
and uplink systems. We build a mathematical connection that highlights
the synergy between the two systems. Specifically, for single antenna
systems, we show that the capacity region of one system can be calculated
from that of the other. For multiple antenna systems, we show that the
"dirty paper" achievable rate region for the broadcast system can be
calculated from the capacity region of the multiple access system and
vice-versa. Moreover, we provide explicit transformations that connect
the transmission schemes used in one system with those of the other.

This duality connection has many useful applications. First, duality leads
to the characterization of the capacity region of the multiple antenna
downlink. Using duality and bounding techniques, the optimal coding
strategy on the multiple antenna downlink is shown to be the "dirty paper
coding" scheme. As another application, the duality connection and
Karush-Kuhn-Tucker conditions from optimization theory allow us to obtain
fast iterative algorithms that make the numerical computation of the
sum capacity of broadcast channels highly efficient.

Directions: http://www.cpcc.uci.edu/Directions-CPCC-UCI.htm

CPCC Web site: http://www.cpcc.uci.edu