Performance of spatial Multi-LRU caching under traffic with temporal locality

TL;DR

This paper introduces spatial multi-LRU caching policies for wireless networks that leverage multi-coverage to improve cache-hit probability, demonstrating superior performance under traffic with temporal locality compared to traditional methods.

Contribution

The paper proposes two novel decentralized caching policies, multi-LRU-One and -All, that adaptively improve cache performance by exploiting multi-coverage in wireless networks.

Findings

Multi-LRU policies outperform single-LRU and centralized bounds.

Passive adaptability to popularity changes enhances hit probability.

Performance varies with storage, traffic, and popularity shape.

Abstract

In this work a novel family of decentralised caching policies for wireless networks is introduced, referred to as spatial multi-LRU. These improve cache-hit probability by exploiting multi-coverage. Two variations are proposed, the multi-LRU-One and -All, which differ in the number of replicas inserted in the covering edge-caches. The evaluation is done under spatial traffic that exhibits temporal locality, with varying content catalogue and dependent demands. The performance metric is hit probability and the policies are compared to (1) the single-LRU and (2) an upper bound for all centralised policies with periodic popularity updates. Numerical results show the multi-LRU policies outperform both comparison policies. The reason is their passive adaptability to popularity changes. Between the -One and -All variation, which one is preferable strongly depends on the available storage space and on traffic characteristics. The performance also depends on the popularity shape.