A communication problem is analyzed in which two senders wish to communicate to two receivers. Encoders can cooperate over two communication links with finite capacities. Such transmitter cooperation, introduced and referred to as a conference by Willems, allows for partial message exchange between encoders. After the conference, each encoder will know a common message incompletely describing the two original messages, and its own private message containing the information that the encoders were not able to exchange.
We determine the capacity region of the compound multi-access channel with conferencing encoders when both original messages are decoded at both receivers. The obtained region is closely related to the capacity region of the multi-access channel with partially cooperating encoders as it is achieved by the same optimal conference, proposed by Willems. We show the gains due to conferencing in the Gaussian channel.
We then relax the decoding constraints. For the channel with the common message as it appears after the conference, it is assumed that a private message sent at an encoder is intended only for a corresponding decoder. The channel becomes the interference channel with common information. We derive conditions under which the capacity region of this channel coincides with the capacity region of the channel in which both private messages are required at both receivers. We show that the obtained conditions and the strong interference conditions for the interference channel with independent messages are satisfied by the same class of interference channels.
Cooperation strategies are commonly devised for information-theoretic channel models with a small number of nodes. We further apply those concepts to large scale sensor networks and show how that approach can lead to increased energy-efficiency and simple distributed algorithms.