Real-time communication of dependent messages

Abstract

This paper presents a source model to capture the dependency structure of messages generated by a source encoder. Under such a model, the communication problem subject to certain delay constraint may be formulated as a utility maximization problem. Over discrete memoryless channels, we derive the expression of maximal achievable utility, which appears to be governed by the channel coding theorem. We perform numerical study of two examples of source-coded messages, corresponding to layered codes and multiple description codes respectively, and show the impact of delay and channel condition on their respective achievable utilities.