Low latency communication with end-user and knowledge of real-time network information, such as radio conditions and network statistics, are among two key advantages of Multi-access Edge Computing (MEC) technology. In this paper, we propose a hierarchical MEC architecture and present a proof-of-concept (PoC) implementation of a Radio Aware Video optimization in a fully Virtualized Network (RAVEN) use case. The PoC has been assembled in a small-scale LTE network based on OpenAirInterface and commercial terminals, representing a scenario where a real-time adaptive video streaming service is provided at a Mobile Edge platform serving multiple eNBs. Real-time radio information is provided by eNB agents which act as local controllers co-located with eNB, collecting and providing realtime access to the requested RAN data. Agents can cooperate with each other to control the network in a distributed manner or can be delegated and/or controlled by a master controller entity. Information gathered by eNB agents is thus provided to the MEC system, allowing other MEC apps to consume it. The results of the RAVEN experimental activity demonstrate the benefit of recently ETSI MEC specified RNI (Radio Network Information) service in improving the perceived user quality of experience.