H-FLTN: A Privacy-Preserving Hierarchical Framework for Electric Vehicle Spatio-Temporal Charge Prediction
Robert Marlin, Raja Jurdak, Alsharif Abuadbba
TL;DR
H-FLTN introduces a privacy-preserving hierarchical Transformer-based federated learning framework to predict EV next-charge location and time across a three-tier ecosystem (EVs, DERMS, EPDC). It combines a Transformer core for spatio-temporal forecasting with privacy tools—P2P sharing, additive secret sharing, secure aggregation, TLS 1.3, and MeLSeC—alongside Dynamic Client Capping Mechanism (DCCM) and Client Rotation Management (CRM) to ensure scalable, efficient training on non-IID, mobility-driven data. The approach is validated on a large, augmented Chicago-based EV mobility dataset, showing high location accuracy (up to ~98%) and low time-prediction errors, while ablation studies quantify the impact of privacy and resource-management components. The framework supports robust, privacy-preserving energy demand forecasting and grid optimization in large-scale, smart-city deployments, with avenues for extending to additional mobility and energy metrics. The results demonstrate that integrating hierarchical FL with Transformer-based spatio-temporal modelling can deliver accurate predictions without compromising user privacy or scalability.
Abstract
The widespread adoption of Electric Vehicles (EVs) poses critical challenges for energy providers, particularly in predicting charging time (temporal prediction), ensuring user privacy, and managing resources efficiently in mobility-driven networks. This paper introduces the Hierarchical Federated Learning Transformer Network (H-FLTN) framework to address these challenges. H-FLTN employs a three-tier hierarchical architecture comprising EVs, community Distributed Energy Resource Management Systems (DERMS), and the Energy Provider Data Centre (EPDC) to enable accurate spatio-temporal predictions of EV charging needs while preserving privacy. Temporal prediction is enhanced using Transformer-based learning, capturing complex dependencies in charging behavior. Privacy is ensured through Secure Aggregation, Additive Secret Sharing, and Peer-to-Peer (P2P) Sharing with Augmentation, which allow only secret shares of model weights to be exchanged while securing all transmissions. To improve training efficiency and resource management, H-FLTN integrates Dynamic Client Capping Mechanism (DCCM) and Client Rotation Management (CRM), ensuring that training remains both computationally and temporally efficient as the number of participating EVs increases. DCCM optimises client participation by limiting excessive computational loads, while CRM balances training contributions across epochs, preventing imbalanced participation. Our simulation results based on large-scale empirical vehicle mobility data reveal that DCCM and CRM reduce the training time complexity with increasing EVs from linear to constant. Its integration into real-world smart city infrastructure enhances energy demand forecasting, resource allocation, and grid stability, ensuring reliability and sustainability in future mobility ecosystems.