Information Masking and Amplification: The Source Coding Setting

TL;DR

This paper explores the tradeoff between amplifying one source and masking another in a two-encoder source coding scenario, providing a single-letter characterization and applying it to a fundamental entropy problem.

Contribution

It introduces a single-letter description of the amplification-masking tradeoff in two-encoder source coding, linking it to a key entropy characterization problem.

Findings

Single-letter characterization of the amplification-masking tradeoff.

Application of the tradeoff to solve a fundamental entropy problem.

Connection to recent multiterminal source coding theorems.

Abstract

The complementary problems of masking and amplifying channel state information in the Gel'fand-Pinsker channel have recently been solved by Merhav and Shamai, and Kim et al., respectively. In this paper, we study a related source coding problem. Specifically, we consider the two-encoder source coding setting where one source is to be amplified, while the other source is to be masked. In general, there is a tension between these two objectives which is characterized by the amplification-masking tradeoff. In this paper, we give a single-letter description of this tradeoff. We apply this result, together with a recent theorem by Courtade and Weissman on multiterminal source coding, to solve a fundamental entropy characterization problem.