A Stochastic Geometry Approach to Distributed Caching in Large Wireless Networks

Abstract

This paper introduces a stochastic geometry modelof a cellular network in which users exchange videos throughout-of-band device-to-device (D2D) communications. Users aregrouped into clusters, in which a user-based distributed cacheis formed to satisfy in-cluster video requests through D2D,avoiding the base station. This paper studies how many of theserequests could be served, considering two service metrics: theglobal relative density of served requests, and the mean ratioof served requests per cluster. The model considers random userplacement, content request statistics, and transmission schedulingand impairments, like fading and path loss. A simple communicationsstrategy to be used in the clusters, involving time-divisionmultiple access and one-hop transmissions, is introduced and themetrics are evaluated in this setup, drawing conclusions on theperformance of distributed-caching in large wireless networks.