Stability of Bernstein's Theorem and Soft Doubling for Vector Gaussian Channels

Mohammad Mahdi Mahvari,Gerhard Kramer
2023-08-14
Abstract:The stability of Bernstein's characterization of Gaussian distributions is extended to vectors by utilizing characteristic functions. Stability is used to develop a soft doubling argument that establishes the optimality of Gaussian vectors for certain communications channels with additive Gaussian noise, including two-receiver broadcast channels. One novelty is that the argument does not require the existence of distributions that achieve capacity.
Information Theory
What problem does this paper attempt to address?