Skip to main navigation Skip to search Skip to main content

Federated cross-chain trust training for distributed smart grid in Web 3.0

  • Cui Zhang
  • , Hui Yang*
  • , Chen Zhang
  • , Jie Zhang
  • , Qiuyan Yao
  • , Zhiwei Wang
  • , Athanasios V. Vasilakos
  • *Corresponding author for this work
  • State Key Laboratory of Information Photonics and Optical Communications
  • Beijing University of Posts and Telecommunications
  • Fuzhou University
  • University of Agder

Research output: Contribution to journal › Article › peer-review

9 Scopus citations

Abstract

Cross-chain federated training (FT) is a paradigm for privacy-preserving predictive scheduling models in distributed smart grids under Web 3.0. Currently, it is often operates in a mode of “off-chain training and inter-chain sharing”. However, this mode brings three major trust challenges: (1) Opaque off-chain training may lead to missing training steps; (2) Cross-chain model transmission without a verification mechanism is vulnerable to forgery and tampering attacks; (3) The isolation mechanism between different chains makes cross-chain federated aggregation unverifiable. To solve the above issues, this paper proposes a cross-chain federated trusted training scheme (CCFTTS) to ensure the trustworthiness of multi-chain FT from three aspects: off-chain training, inter-chain transmission, and cross-chain aggregation. For the first issue, we design a twin trustworthy verification algorithm (TTVA). This algorithm replicates the off-chain training process using random numbers and symbol mapping mechanisms, thereby generating evidence to verify the integrity of the training steps and ensure the trustworthiness of off-chain training. For the second issue, we use the BLS aggregate signature algorithm to ensure the trustworthiness of the model during cross-chain transmission. For the third issue, we propose a cross-chain aggregation trusted verification method (CCATV). This method combines the commitment mechanism and utilizes its homomorphic aggregation property. By comparing the aggregated commitments of the off-chain training model with the commitment of the final aggregated model, it ensures the trustworthiness of cross-chain aggregation. Experimental results show that the proposed scheme is effective in ensuring trustworthiness and is feasible in practice.

Original languageEnglish
Article number113313
JournalApplied Soft Computing
Volume180
DOIs
StatePublished - Aug 2025
Externally publishedYes

Keywords

  • Cross-chain
  • Federated learning
  • Power scheduling
  • Trust
  • Web 3.0

Fingerprint

Dive into the research topics of 'Federated cross-chain trust training for distributed smart grid in Web 3.0'. Together they form a unique fingerprint.

Cite this