Enhancing Efficiency and Privacy of Intelligent Public Transportation Systems Through Federated Learning and EdgeAI

In the realm of intelligent public transportation systems, deep learning (DL) techniques are widely used to extract valuable insights from mobility-related data, on top of which it is possible to realized several use-cases. However, since DL models are expensive in terms of resource and energy consumptions, they are typically deployed on third-party clouds, posing challenges such as latency, privacy, and scalability. To address these issues, Federated Learning (FL) and EdgeAI have emerged as promising solutions. Despite their potential, several open challenges persist. The application of FL in public transport systems lacks exploration, while the theoretical understanding of KD’s effectiveness remains incomplete. This research aims at addressing these challenges by leveraging FL to prevent data exchange among transport entities, and systematically employing KD for model compression. The ultimate goal is to facilitate the efficient integration of federated learning and EdgeAI to mitigate privacy and efficiency issues in distributed Intelligent Public Transportation Systems.