Two-way communication systems, with a transmitter/receiver at each end, are ubiquitous. The bi-directional information flow gives rise to the possibility of using some kind of feedback in order to help combating channel noise and uncertainty. It has long been recognized that feedback not only can help increase capacity when the channel has memory, but it may actually lead to decreased complexity and delay in order to meet performance targets. However, the gulf between practice and theory remains wide as far as taking advantage of the capabilities of feedback.
This is a research project on both theory-driven design and practice-driven theory of feedback communication systems. Methods of design and analysis are tightly coupled and guided by the teachings of both Shannon theory and modern coding theory. Through the synergistic combination of feedback and belief propagation decoding of sparse-graph codes, the projects studies implementable schemes for both reliable transmission and joint source-channel coding that can harness the power of even a modicum of feedback in order to improve delay-rate tradeoffs. To help bridge the gap between theory and practice, a major thrust of the project is on the analysis of technology-independent fundamental limits departing from conventional models and performance measures such as the exponential error rate decrease when transmitting at a fraction of capacity.
