Feedback is an important feature of many sensory systems. Wireless sensor networks often have a base station node that is not severely power constrained and could provide a strong feedback signal. The goal of this project is to understand the usefulness of such a feedback signal.
Our first object of study is the Gaussian multiple-access channel. When perfect (noiseless, full) feedback of the channel outputs to the transmitters is available in a causal fashion, a beautiful paper of Ozarow establishes the full capacity region for the case of two users.
Our work has focused on understanding how close one can get to this capacity in the case where the feedback is noisy. Using a novel tool, dependence-balance bounds, due to Hekstra and Willems, we were able to derive a converse bound on the usefulness of noisy feedback.
We are also studying the case of more than two users, and were able to establish a new capacity result for the case of noiseless feedback, confirming that the intuitively pleasing code of Schalkwijk and Kailath remains optimal.
- M. Gastpar and G. Kramer. On noisy feedback for interference channels. In Asilomar Conference on Signals, Systems, and Computers, Oct. 29-Nov. 1, 2006
- G. Kramer and M. Gastpar. Dependence Balance and the Gaussian Multiaccess Channel with Feedback. In IEEE Information Theory Workshop, Punta del Este, Uruguay, March 2006, p.198-202.
- M. Gastpar and G. Kramer. On Cooperation via Noisy Feedback. In International Zurich Seminar, Zurich, Switzerland, Feb. 2006, p.146-149.
- M. Gastpar. On noisy feedback in Gaussian networks. In 43rd Annual Allerton Conf. on Commun., Control and Comp., (Monticello, IL, USA), Sept. 28-30, 2005.