Lossless coding of correlated sources with actions in acyclic directed networks

Abstract

This work studies the problem of distributed compression of correlated sources with an action-dependent joint distribution. This class of problems are in fact extensions of the Slepian-Wolf model, but where cost-constrained actions affect the generation of one of the sources. A network setup is investigated for the case where actions are taken at the encoder. The first source is available at a node in the network, this node can take actions which affect the generation of the other source which is available at different node in the network. Transmission occurs over a general, acyclic, directed network and both sources are required in a set of terminal nodes. The purpose of this work is to study the implications of actions on the set of achievable rates. For this network, generalized cut-set bounds are derived, and a full characterization of the set of achievable rates using single-letter expressions is provided, showing how actions affect the achievable region in a non-trivial manner. Random linear network coding is proved to be optimal in this setup, even though this is not a classical multicast problem. As a special case of this network we study a multi-user setup with two encoders and one decoder, each source is available to one encoder and transmission occurs over rate-limited link. The optimal rate region for this case is characterized, and calculated for a binary example.