A Second-Order Achievable Rate Region for Gaussian Multi-Access Channels via a Central Limit Theorem for Functions

Abstract

Motivated by the growing application of multi-access networks with stringent delay constraints, we investigate the Gaussian multiple-access channel (MAC) in the finite blocklength regime. By applying central limit theorem (CLT) approximations to non-asymptotic information-spectrum inner bounds, we obtain second-order achievable rate regions for the Gaussian MAC with a positive average error probability and per-codeword power constraints. Our achievability results use spherical inputs uniformly distributed on the power shells, which lead to summations of dependent information random variables. However, we conduct the analysis through a convenient yet powerful form of the CLT, called the CLT for functions.