IEEE International Symposium on Information Theory (ISIT)
One-shot Interference Channel Simulation
Aditya Nema, Michael X. Cao, Sreejith Sreekumar, Mario Berta
Abstract
We consider the problem of shared randomness-assisted simulation of interference channel in the one-shot setup. We derive almost-matching inner and outer bounds in terms of the partial smoothed max-divergence, with the help of auxiliary random variables. The achievability relies on a rejection-sampling algorithm to simulate two auxiliary broadcast channels between each sender and both receivers, and producing the final output based on the output of these intermediate channels, which can be thought of as gluing these auxiliary broadcast channels via multiple access channels (MACs), to preserve all the correlations between each sender-receiver pair induced by the channel. The converse follows via information-spectrum based arguments.