A Dynamic Clustering Caching Strategy for IoT-NDN Based on User-Preferred Contents

In recent years, with the explosive growth in the number of Internet of Things (IoT) devices, the demand for content transmission over the network has increased rapidly, making Named Data Networks (NDN) a rising research focus. The in-network caching mechanism of NDN plays a significant role in enhancing network efficiency, thus making the development of efficient caching strategies particularly crucial. This paper proposes a dynamic clustering caching strategy for IoT-NDN based on user-preferred contents and the SDN-NDN architecture, aimed at maximizing cache hit rates and minimizing transmission latency. We initially designed a dynamic clustering mechanism that takes into account the preferences of individual nodes and their neighboring nodes, allowing for dynamic adjustment of node clusters. Subsequently, we constructed a Graph Attention Network-Deep Reinforcement Learning (GAT-DRL) agent, which uses node preference information to make caching decisions for each node. Simulation results indicate that compared to existing typical caching strategies, the proposed strategy significantly improves in terms of cache hit rates, transmission latency, and path stretch ratios.