Skip to content

Cross-Depot Federated Learning Scalability

Home / Open Questions / Cross-Depot Federated Learning Scalability

Background: Federated Learning (FL) models are increasingly evaluated for distributed EV charging infrastructure analytics where data heterogeneity exists across charging stations. Extending these frameworks beyond single-site deployments to collaborative multi-site environments is a critical step for scalability.

Question / Future Work: Future work is required to transition from intra-depot federated learning (where stations at a single site act as clients) to cross-depot federated learning, which involves training models across multiple geographically or operationally distinct charging sites to improve model robustness and scalability.

Why It Matters: Scaling FL from a single depot to multiple depots introduces significant challenges related to data distribution shifts, varying connectivity, and communication efficiency across heterogeneous site-level infrastructures.

Metadata & Links