Title: New Formulas of Feedback Capacity for AGN Channels with Memory: A Time-Domain Sufficient Statistic Approach

Abstract: In the recent paper [1] it is shown, via an application example, that the Cover and Pombra [2] "characterization of the $n-$block or transmission" feedback capacity formula, of additive Gaussian noise (AGN) channels, is the subject of much confusion in the literature, with redundant incorrect results. The main objective of this paper is to clarify the main points of confusion and remove any further ambiguity. The first part of the paper applies time-domain methods, to derive for a first time, equivalent sequential characterizations of the Cover and Pombra characterization of feedback capacity of AGN channels driven by nonstationary and nonergodic Gaussian noise. The optimal channel input processes of the new equivalent sequential characterizations are expressed as functionals of a sufficient statistic and a Gaussian orthogonal innovations process. From the new representations follows that the Cover and Pombra $n-$block capacity formula is expressed as a functional of two generalized matrix difference Riccati equations (DRE) of filtering theory of Gaussian systems, contrary to results that appeared in the literature. In the second part of the paper the existence of the asymptotic limit of the $n-$block feedback capacity formula is shown to be equivalent to the convergence properties of solutions of the two generalized DREs. Further, necessary and or sufficient conditions are identified for existence of the asymptotic limits, for stable and unstable Gaussian noise, when the optimal input distributions are time-invariant, but not necessarily stationary. The paper contains an in depth analysis, with examples, of the specific technical issues, which are overlooked in past literature [3-7], that studied the AGN channel of [2], for stationary noises.