Binary CEO Problem under Log-Loss with BSC Test-Channel Model

TL;DR

This paper introduces an efficient coding scheme for the binary CEO problem under log-loss, utilizing compound LDGM-LDPC codes to approach the theoretical rate-distortion bounds with iterative joint decoding.

Contribution

It presents a novel encoding and decoding scheme for the binary CEO problem under log-loss, achieving theoretical bounds using LDGM and LDPC codes.

Findings

Achieves the theoretical rate-distortion bound under log-loss.

Employs a joint iterative decoding algorithm based on sum-product and soft estimation.

Demonstrates improved performance over existing schemes.

Abstract

In this paper, we propose an efficient coding scheme for the two-link binary Chief Executive Officer (CEO) problem under logarithmic loss criterion. The exact rate-distortion bound for a two-link binary CEO problem under the logarithmic loss has been obtained by Courtade and Weissman. We propose an encoding scheme based on compound LDGM-LDPC codes to achieve the theoretical bounds. In the proposed encoding, a binary quantizer using LDGM codes and a syndrome-coding employing LDPC codes are applied. An iterative joint decoding is also designed as a fusion center. The proposed CEO decoder is based on the sum-product algorithm and a soft estimator.