Federated Learning-Driven Privacy-Preserving and Security Defense for Cloud-Edge Computing: A Hierarchical Collaborative Framework

Abstract

Cloud-edge computing, integrating cloud computing’s robust computing capacity and edge computing’s low-latency response, underpins data-intensive applications like smart cities and industrial IoT. However, its distributed edge data carries severe privacy leakage risks during cloud transmission, and edge nodes’ open access makes systems vulnerable to malicious attacks. Federated Learning (FL), enabling collaborative model training without raw data sharing, offers a solution to balance data sharing and privacy protection. This study proposes a Federated Learning-Driven Hierarchical Cloud-Edge Collaborative Privacy-Preserving and Security Defense Framework (FL-HCPS), adopting a two-level FL architecture (edge horizontal federation, cloud-edge vertical federation). It designs a privacy-enhanced FL algorithm based on differential privacy and homomorphic encryption, and integrates an attack-aware adaptive defense mechanism. Experiments on EdgeIIoTset and CSE-CIC-IDS2018 datasets show FL-HCPS achieves 96.8% average attack detection accuracy, cuts privacy leakage risk by 78.3%, reduces communication overhead to 23.5% of horizontal FL, and shortens training time by 41.2%, balancing privacy, security and efficiency for cloud-edge systems.

Previous article in issue