ADAPTIVE FEDERATED LEARNING FRAMEWORK FOR PRIVACYPRESERVING EDGE INTELLIGENCE

Abstract

Federated learning (FL) has emerged as a promising paradigm for distributed model training without directly exposing raw data. However, deploying FL at the edge introduces challenges related to communication overhead, privacy leakage risks, and heterogeneous device capabilities. This paper proposes an Adaptive Federated Learning Framework (AFLF) that integrates dynamic client selection, encrypted gradient aggregation, and lightweight model optimization to achieve privacypreserving edge intelligence. The system adapts to device availability and network conditions, ensuring robust performance even in non-IID environments. Experimental results demonstrate reduced communication cost, improved model convergence, and enhanced privacy protection compared to traditional FL architectures. The proposed AFLF framework provides a scalable and secure solution for real-world edge deployments.