Enabling Source Channel Separation for Communication Networks : The Uplink Case

This paper looks at the problem of enabling source-channel separation for communication networks. Source-channel separation is of critical important as it provides a great extent of modularity to system design without suffering a penalty in the process. Indeed, source coding and channel coding at the two opposite ends of the protocol stack is the prevailing approach today, and it is desirable that systems are designed in a manner that allows for such architectures not only to be elegant and simple, but also optimal from a performance perspective. In particular, the paper focuses on the Gaussian uplink network for the case when two transmitters transmit correlated messages. The paper shows that feedback can lead to source-channel separation in such a system. In doing so, the paper also derives the capacity region for such a network