Cache-Aided Coded Multicast for Correlated Sources

TL;DR

This paper introduces a correlation-aware caching and multicast scheme that leverages content correlation to significantly reduce network load in cache-aided networks, surpassing existing correlation-unaware methods.

Contribution

It formulates a new cache-aided multicast problem considering content correlation and proposes a novel scheme with performance bounds, improving efficiency over prior methods.

Findings

Significant load reductions achieved with correlation-aware schemes.

Proposed scheme outperforms correlation-unaware approaches.

Theoretical bounds demonstrate the effectiveness of the approach.

Abstract

The combination of edge caching and coded multicasting is a promising approach to improve the efficiency of content delivery over cache-aided networks. The global caching gain resulting from content overlap distributed across the network in current solutions is limited due to the increasingly personalized nature of the content consumed by users. In this paper, the cache-aided coded multicast problem is generalized to account for the correlation among the network content by formulating a source compression problem with distributed side information. A correlation-aware achievable scheme is proposed and an upper bound on its performance is derived. It is shown that considerable load reductions can be achieved, compared to state of the art correlation-unaware schemes, when caching and delivery phases specifically account for the correlation among the content files.