Enhancing Video Caching Efficiency and User Experience in Cloud-Edge Networks with GRU-FedQCache
Venkateswara Reddy B, K. Bala · 2025
The high rate at which video content is being consumed and the rising trends in satisfying low latency delivery have presented serious challenges in the optimization of effective caching mechanisms in the cloud-edge networks. The issue with traditional centralized caching is that it is not always efficient, which generates serious network congestion and great user experience because of latency and bandwidth. To counter these issues, we introduce GRU-FedQCache: a federated caching framework using Deep Reinforcement Learning (DRL) to streamline content delivery in cloud-edge scenarios. The system considers changes in video popularity and network dynamics by the mechanism of scales: the Gated Recurrent Units (GRU) reflects the temporal dependencies in video content demand, and Federated Learning conserves the privacy of users. GRU-FedQCache allows caching video contents efficiently and scale to large system sizes in distributed networks, improving the cache hit rates, decreasing bandwidth costs, and guaranteeing the high quality of the user experience through local training on edge nodes and sparse aggregation at the cloud.