On the Cover and Pombra Gaussian Feedback Capacity: Complete Sequential Characterizations via a Sufficient Statistic

TL;DR

This paper provides a new sequential characterization of the Gaussian feedback capacity, clarifying previous confusions by expressing optimal inputs as functionals of a sufficient statistic and linking to Riccati equations.

Contribution

It introduces a novel sequential representation of the Cover-Pombra feedback capacity using sufficient statistics and Riccati equations, resolving prior ambiguities.

Findings

Feedback capacity expressed as a functional of Riccati equations

Optimal inputs as functionals of sufficient statistics

Clarification of previous literature misconceptions

Abstract

The main objective of this paper is to derive a new sequential characterization of the Cover and Pombra \cite{cover-pombra1989} characterization of the $n-$finite block or transmission feedback information ($n$-FTFI) capacity, which clarifies several issues of confusion and incorrect interpretation of results in literature. 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 characterization of the $n-$FTFI capacity is expressed as a functional of two generalized matrix difference Riccati equations (DRE) of filtering theory of Gaussian systems. This contradicts results which are redundant in the literature, and illustrates the fundamental complexity of the feedback capacity formula.