Michael Gastpar, Department of Electrical Engineering and Computer Sciences, University of California at Berkeley

For cooperative networks, capacity hinges on the amount of dependence between the transmitted signals. This was elegantly captured by Shannon for the case of the two-way channel, via inner and outer bounds that only differ in the amount of dependence: The inner bound does not allow for any dependence, and the outer bound allows for full dependence. It is therefore tempting to derive refined and explicit bounds that lie between these two extremes, exploiting the limits imposed by the network topology.

This talk will review some old results and recent progress. When all terminals have independent messages, a basic tool to limit dependence is a well-known "cut-set" bound due to Cover and El Gamal. For some topologies involving feedback, we show how to sharpen this bound via an extension of a "dependence-balance" argument due to Hekstra and Willems. This leads to new capacity upper bounds for multiple-access and interference channels with noiseless and noisy feedback. Then, time permitting, we will consider a scenario with dependent messages where a correlation-based argument due to Witsenhausen can be extended to calculate a limit on dependence.

Michael Gastpar (Ph.D. EPFL, 2002, M.S. UIUC, 1999, Dipl. El-Ing, ETH, 1997) has been an assistant professor at the University of California, Berkeley, since January 2003. He was also a student in electrical engineering and philosophy at the Universities of Edinburgh and Lausanne, and a summer researcher in the Mathematics of Communications Department at Bell Labs, Lucent Technologies. He won the 2002 EPFL Best Thesis Award, and an NSF CAREER award in 2004.