Edge Collaborative Caching Based on Multi-Agent Communication Model with Gating Mechanism
Zheng Wan, Yu Zhou, Xiaogang Dong, Yuxi Pi · 2025
Collaborative edge caching in Internet of Things (IoT) networks uses Edge Servers (ESs) for localized storage and cooperation, but faces challenges from limited storage and diverse user preferences. To address this, we propose the Multi-Agent Communication Model with Gating Mechanism (MACM-GM). Its dynamic gating mechanism enables ESs to selectively share essential state information, reducing communication overhead. Additionally, Conditional Redundancy Nodes (CRNs) allow the dynamic selection of optimal ESs as auxiliary storage nodes, improving caching flexibility and resource utilization. Simulation results demonstrate that MACM-GM significantly outperforms baseline schemes in cache hit rate and backhaul traffic reduction.