One of the most recently proposed 5G architectures, Fog Radio Access Network or F-RAN, has the same cloud computing-based approach of C-RANs while giving additional storage and processing power to the edge nodes so that they can address network activities. The main problem of F-RANs is the intense data traffic in the fronthaul, which is the centralized physical channel used to connect the antennas (F-APs) to the BBU pool. To minimize this bottleneck, we propose the Nuoxus model, whose objective is to cache multimedia content in the edge of the network. By taking advantage of the hierarchical structure of F-RANs, Nuoxus content replacement policy analyses the similarity of requests amongst the children nodes to define the relevance of storing a requested content in cache. It also uses this process to proactively suggest to nodes located lower in the network hierarchy the caching of relevant content. The state-of-the-art analysis reveals Nuoxus as the first model to explore the nodes’ request history to perform proactive caching without relying on external and centralized components. Furthermore, the developed proof-of-concept shows significant gains when compared to other caching strategies.