Is non-unique decoding necessary?

In mutiterminal communication systems, signals carrying messages meant for different destinations are often observed together at any given destination receiver. Han and Kobayashi (1981) proposed a receiving strategy which performs a joint unique decoding of messages of interest along with a subset of messages which are not of interest. It is now well-known that this provides an achievable region which is, in general, larger than if the receiver treats all messages not of interest as noise. Nair and El Gamal (2009) and Chong, Motani, Garg, and El Gamal (2008) independently proposed a generalization called indirect or non-unique decoding where the receiver uses the codebook structure of the messages to only uniquely decode its messages of interest. Indirect (non-unique) decoding has since been used in various scenarios. The main result in this paper is to provide an interpretation and a systematic proof technique for why indirect decoding, in all known cases where it has been employed, can be replaced by a particularly designed joint unique decoding strategy, without any penalty from a rate region viewpoint¹.