Compress-Forward Schemes for General Networks

TL;DR

This paper advances compress-forward schemes for general networks by establishing achievable outerbounds with regular decoding, introducing layering for flow decomposition, and reducing complexity in finding optimal compression rates.

Contribution

It introduces a novel layering approach to harmonize CF schemes with flow decomposition, enabling achievable outerbounds with reduced complexity.

Findings

Outerbounds on CF rate are achievable with regular decoding.

Layering concept aligns CF with flow decomposition.

Shifting operation minimizes complexity in rate-vector optimization.

Abstract

Compress-forward (CF) schemes are studied in general networks. The CF rate for the one-relay channel defines outerbounds on both the CF rate for general networks and the compression rate-vector region supporting this rate. We show the outerbound is achievable using regular decoding with constant encoding delays, avoiding the exponential delays and restrictions on bidirectional communication in noisy network coding and backward decoding. The concept of layering is introduced to harmonize regular CF schemes with the framework of flow decomposition in the decode-forward setting. Layerings correspond to regular decoding schemes. Any desired compression rate-vector in the outerbound is achievable by some layering, which is found using the same "shift" operation in flow decomposition. In separate work, we show that "shifting" minimizes the operations needed to find layerings and thus minimizes the complexity of the compression rate-vector region.