Trust-Centric and Privacy-Preserving Federated Learning Architectures for Ethical Data-Driven Smart City Services

The rapid growth of smart cities has led to extensive use of social sensing technologies, generating vast amounts of citizen and environmental data. While this data enables real-time insights for traffic management, public health, and energy optimization, it also raises significant privacy concerns. This chapter explores the application of Federated Learning (FL) as a privacy-preserving, decentralized approach to social sensing in urban environments. FL allows local model training on edge devices, sharing only encrypted updates, thereby reducing risks of data leakage. The chapter details FL principles, architecture, privacy mechanisms, and case studies on traffic prediction and public health monitoring, demonstrating improvements in accuracy, efficiency, and trust. It also addresses challenges such as heterogeneous data, limited device resources, and communication bottlenecks, offering governance strategies for ethical deployment. A roadmap for expanding FL applications in smart cities is presented.

Hali tarjima qilinmagan