A Clustering-Based Cooperative Caching and Multi-Link Bandwidth Strategy in Internet-of-Vehicles

With the emergence of the 6G network era and the exponential increase in data traffic, edge caching has emerged as a promising solution for reducing content retrieval latency and easing the backhaul burden of base stations in the context of Internet-of-Vehicles (IoV). This paper proposes a joint cooperative caching and bandwidth allocation policy (JCCBA) that aims to minimize user content retrieval latency while considering continuous task generation. To achieve this goal, the Affinity Propagation (AP) clustering algorithm is used to group vehicles, simplifying the vehicle request and content delivery process while reducing signaling overhead. The Deep Determination Policy Gradient (DDPG) algorithm is applied to obtain optimal caching and bandwidth allocation decisions based on the state information of vehicle users and road side units (RSUs). Simulation experiments show that JCCBA can effectively decrease the average acquisition latency of long-term users in both random and dynamic IoV environments, outperforming the benchmark algorithm.