Privacy Preserving Federated Learning - Year two
By virtualizing all the various appliances in the network, Network Function Virtualization (NFV) became a key enabler for the coming 5G infrastructure and nowadays a major shift is under way bringing an evolution to cloud-native NFV. In the latter operational model, applications are decomposed into microservices running inside containers to enable automated installation, configuration and scaling with the dynamic network requirements beside self-healing and automated upgrading and updating of the VNFs. Pre-provisioned rules are expected to become less important over time as Artificial Intelligence (AI) enable smarter interpretation of data and hence better reaction accordingly for efficient management of VNFs.
Centralized AI is by far the most common architecture for such analysis. However, such application has significant drawbacks ranging from high transfer cost, latency and hence slow inference to even more critical situations of diminished privacy. TO BE CONT'D