An MDP Approach to Computing Feedback Capacity

Speaker: Dr. Sekhar C. Tatikonda, Yale University

In this talk we examine the capacity of Markov channels with feedback. One of the main difficulties in this problem has to do with the fact that the transmitter and the receiver may have different information about the state of the channel. We show how to choose appropriate sufficient statistics at both the transmitter and receiver. We then formulate the capacity optimization problem as a Markov decision problem (MDP). The resulting Bellman equation can be viewed as a single-letter characterization of the capacity. We discuss the connections between information stability and the existence of a solution to the Bellman equation. Examples are discussed.


Sekhar C. Tatikonda is presently an associate professor of electrical engineering at Yale University. He received his PhD degree in EECS from MIT in 2000. He was a postdoctoral fellow in EECS at UC-Berkeley from 2000-2002. His research interests span topics in communications, information theory, control, and machine learning.
In particular he has worked on control with communication constraints, the role of feedback in channel coding, and inference on graphs with cycles. He received the NSF CAREER award in 2006.

Presented On: Feb 9th, 2007
Video: QuickTime streaming video