Cooperative caching strategy based mobile vehicle social‐aware in internet of vehicles

Abstract

The dramatic growth of smart in‐vehicle applications in the Internet of Vehicles and the increasing quality of experience for vehicle users have put a huge traffic load on the mobile core network. Mobile edge caching has been proposed to place content at the edge of the network to serve users in close proximity to them. However, the dynamic topology of existing edge networks and the inherent storage limitations of edge devices pose a serious challenge to vehicular edge caching. Therefore, in this article, we propose a cooperative caching strategy based on mobile prediction and social awareness that allows collaborative content decision making among edge devices. Specifically, the long short‐term memory network is used to predict vehicle trajectories, social relationships are computed using content similarity and contact rates among vehicle users to select vehicles that can serve as caching nodes, and deep reinforcement learning is employed to achieve the final caching decision. Simulation results show that the caching strategy proposed in this article achieves up to 7.7%, 24.2%, and 27.3% gains, respectively, in terms of content acquirement delay compared to the reinforcement learning algorithm, the algorithm without considering mobility, and the non‐cooperative algorithm.