Broadcast Caching Networks with Two Receivers and Multiple Correlated Sources

TL;DR

This paper explores how to efficiently cache and deliver correlated content in a two-user, three-file broadcast network, proposing a two-step source coding scheme that approaches optimal performance.

Contribution

It introduces a novel two-step source coding approach combining Gray-Wyner compression with correlation-aware caching, and derives bounds for the worst-case demand scenario.

Findings

The proposed scheme achieves the lower bound for large cache sizes.

It is within half of the joint entropy for all other cache sizes.

The approach effectively exploits content correlation to reduce delivery load.

Abstract

The correlation among the content distributed across a cache-aided broadcast network can be exploited to reduce the delivery load on the shared wireless link. This paper considers a two-user three-file network with correlated content, and studies its fundamental limits for the worst-case demand. A class of achievable schemes based on a two-step source coding approach is proposed. Library files are first compressed using Gray-Wyner source coding, and then cached and delivered using a combination of correlation-unaware cache-aided coded multicast schemes. The second step is interesting in its own right and considers a multiple-request caching problem, whose solution requires coding in the placement phase. A lower bound on the optimal peak rate-memory trade-off is derived, which is used to evaluate the performance of the proposed scheme. It is shown that for symmetric sources the two-step strategy achieves the lower bound for large cache capacities, and it is within half of the joint entropy of two of the sources conditioned on the third source for all other cache sizes.