A Post-Quantum Secure Federated Learning Framework for Cross-Domain V2G Authentication

Abstract

Cross-domain Vehicle-to-Grid (V2G) networks enable secure and efficient energy transactions between Electric Vehicles (EVs), Charging Stations (CS), and Electric Service Providers (ESP), forming a critical component of Edge computing-assisted Consumer devices and IoT (EACI) frameworks. However, existing authentication mechanisms often rely on centralized trusted authorities, leading to single points of failure and computational bottlenecks. Additionally, the emergence of quantum computing threatens traditional cryptographic authentication schemes, necessitating the integration of post-quantum security mechanisms to protect multi-domain V2G interactions. This paper presents a Lightweight Quantum-Resistant Authentication Protocol (LQAP) designed to address these challenges by employing a hybrid eXtended Merkle Signature Scheme (XMSS) and Lightweight Merkle Signature (LMS) system to achieve post-quantum resilience with optimized key management. Additionally, a decentralized Hierarchical Federated Learning (HFL) framework operating at the edge layer is incorporated for hardware-based secure entity identification, enabling distributed anomaly detection without centralized data aggregation. Additionally, physically unclonable functions (PUF) are incorporated for hardware-based secure entity identification. Formal security analysis using the Scyther tool confirms LQAP’s resilience against impersonation, replay, and machine learning-based inference attacks. Performance evaluations indicate that LQAP substantially reduces communication overhead by 21.45%, computational cost by 42.78%, and storage overhead by 61.95% compared to conventional schemes, thus providing an efficient and scalable post-quantum authentication solution aligned with EACI network requirements.

@article{ahmed2025post,
  title   = {A Post-Quantum Secure Federated Learning Framework for Cross-Domain V2G Authentication},
  author  = {Ahmed, Shafiq and Anisi, M. H.},
  journal = {IEEE Transactions on Consumer Electronics},
  volume  = {71},
  number  = {3},
  pages   = {8433--8440},
  year    = {2025},
  doi     = {10.1109/tce.2025.3580338}
}