Federated Learning Empowered End-Edge-Cloud Cooperation for 5G HetNet Security

Abstract

The distributed and heterogeneous framework in 5G heterogeneous networks (HetNet) makes it vulnerable to attacks of different kinds. Nodes for improving the network security are therefore important to eliminate such critical threats. Without cooperation or with limited cooperation, these nodes are substantially restricted in their protecting capacity due to specific characteristics such as heterogeneity, hierarchy and wide range in 5G HetNet. In this paper, we propose a federated learning empowered end-edge-cloud cooperation based framework for enhancing 5G HetNet security. In this framework, the nodes equipped with attack detection mechanisms are distributed in the end, edge and cloud of the 5G HetNet. We then design cooperative training schemes to realise the full potential of these nodes in detecting attacks. Illustrative results demonstrate the superior performance of our proposed scheme compared to three different benchmark schemes.