Federated learning over private 5G networks
Seungyeol Lee, Myung-Ki Shin · 2022
As the beyond 5G era approaches, the number of private 5G networks that can effectively support various vertical services continues to increase. Naturally, federated learning (FL) is in the spotlight as a learning method without data leakage on multiple private 5G networks. However, existing studies focus on algorithm-centric theoretical approaches without considering 3GPP standards and implementation. In this demonstration, we present how FL works in private 5G networks by using network data analysis function (NWDAF), a new 3GPP network function. We configured a distributed NWDAF environment and private 5G networks using Free5GC, a well-known 5G open source project.