D2D Coded Caching Schemes for Multiaccess Networks with Combinatorial Access Topology

TL;DR

This paper introduces novel device-to-device coded caching schemes for multiaccess networks with combinatorial access topologies, improving efficiency in subpacketization and load per user over existing schemes.

Contribution

It extends multiaccess coded caching to D2D networks using $t$-design and $t$-GDD topologies, proposing schemes that outperform existing methods in key metrics.

Findings

MADCC with $t$-design topology reduces subpacketization level.

MADCC with $t$-GDD topology lowers load per user.

Proposed schemes outperform cyclic topology in key metrics.

Abstract

This paper considers wireless device-to-device (D2D) coded caching in a multiaccess network, where the users communicate with each other and each user can access multiple cache nodes. Access topologies derived from two combinatorial designs known as the $t$-design and $t$-group divisible design ($t$-GDD), referred to as the $t$-design and $t$-GDD topologies respectively, which subsume a few other known topologies, have been studied for the multiaccess coded caching (MACC) network by Cheng \textit{et al.} in \cite{MACC_des}. These access topologies are extended to a multiaccess D2D coded caching (MADCC) network and novel MADCC schemes are proposed. MADCC network has been studied so far only for the cyclic wrap-around topology. Apart from the proposed novel MADCC schemes, MADCC schemes are also derived from the existing MACC schemes in \cite{MACC_des}. To compare the performance of different MADCC schemes, the metrics of load per user and subpacketization level are used while keeping the number of caches and cache memory size same. The proposed MADCC scheme with $t$-design topology performs better in terms of subpacketization level while achieving the same load per user compared to the MADCC scheme derived from the MACC scheme with $t$-design topology in \cite{MACC_des}. The proposed MADCC scheme with $t$-GDD topology performs better in terms of load per user while achieving the same subpacketization level compared to the MADCC scheme derived from the MACC scheme with $t$-GDD topology in \cite{MACC_des} in some cases. Compared to the existing MADCC scheme with cyclic wrap-around topology, the proposed MADCC scheme with $t$-design topology performs better in terms of load per user, and the proposed MADCC scheme with $t$-GDD topology performs better in terms of subpacketization level at the expense of an increase in load per user.