Edge Caching Cooperation Method Based on Deep Reinforcement Learning

During the use of electronic licenses, IoT devices require access to data in the cloud. By introducing edge caching, we can reduce the latency and traffic consumption when massive IoT devices access data in the cloud. Cooperation between edge nodes can further improve system performance. However, how to manage edge caches effectively in cooperation scenarios remains a challenging problem. This article decomposes the cooperation problem in edge caching into two parts. Firstly, the Gate Recurrent Unit (GRU) is introduced into the Dueling Network to construct a local method for data evaluation. Then, the cooperation relationship between edge nodes is handled from two aspects: replica reduction and load balancing, and cooperation rules are designed. Experiments have shown that the approach proposed in this article can effectively improve the hit rate of edge caching systems and reduce average latency, and improve system performance.