A D2D-Enabled Cooperative Caching Strategy for Fog Radio Access Networks

Abstract

Fog radio access networks (F-RANs) is foreseen as a possible solution to provide the storage and computing capacities at the network edge. To reduce the burden on fronthaul link and provide low latency service, this paper proposes a Device-to-Device (D2D)-enabled cooperative caching strategy for fog radio access network (F-RAN). Aim at maximizing the total cache hit rate, we first formulate the cooperative caching problem as a probability-triggered combinatorial multi-armed bandit problem (CMAB). Next, an enhanced multi-agent reinforcement learning (EMARL) algorithm is designed to solve the above issue combined with user preference and content popularity prediction. Based on a real dataset from MovieLens, simulation results demonstrate that the proposed cooperative caching algorithm can improve the cache hit rate compared with existing caching schemes.