Gaussian MAC with Feedback and Strictly Causal State Information

Abstract

We consider a two user Gaussian multiple access channel with an additive Gaussian state process. The past values of both the state and the received symbols are strictly causally made available to the encoders at each instant. The capacity region for the noiseless case, without any feedback, was recently solved in literature. Here we study the model with noise as well as feedback. We propose a communication scheme which effectively utilizes the feedback symbols as well as the state information to enhance the achievable region. In particular, Wyner-Ziv binning on the state information and Ozarow feedback scheme for the MAC are effectively utilized, using a suitable interleaving technique. The obtained region is significantly better than the feedback capacity region with no state information.