Hydro-DPFLQB: A Differentially Private Federated Learning-Assisted Quantum-Safe Blockchain Method for the Green Hydrogen Supply Chain
Bildiriler > Tam Metin Bildiri
Yazarlar: Bora Buğra SEZER
2026 International Conference on Artificial Intelligence, Computer, Data Sciences and Applications (ACDSA), , , Uluslararası, Tam Metin Bildiri, ss 1-6
DOI: 10.1109/acdsa67686.2026.11468128
Özet
Recently, the conversion of excess electricity generated from renewable sources into hydrogen has emerged as a promising alternative that significantly contributes to carbonneutral targets by enabling both the storage and distribution of energy to various sectors. However, the privacy-friendly realtime traceability of large volumes of IoT data in the hydrogen supply chain, as well as the trust issue between stakeholders, are well-known challenges in the literature. This work presents a secure post-quantum blockchain-based framework enhanced by privacy-preserving federated learning. In the proposed framework, nodes in the supply chain train their local models without sharing raw data. Robust median pooling and adaptive noise scaling maintain accuracy up to 98 %, while the privacy budget increases by only$\Delta \varepsilon<0.003$. Module-Lattice-Based Key Encapsulation Mechanism (ML-KEM)-based hybrid encryption protects client models against quantum attacks, while data is securely stored in IPFS. In this study, the accuracy, privacy, and efficiency of the approach are evaluated by simulating the processes of different actors in the supply chain on a scenario-byscenario basis. The experimental results demonstrate the security and feasibility of the system by combining secure data sharing with high model performance of the proposed architecture.