Shard-based and Reputation-enhanced Byzantine Faulttolerant Scheme for Secure Data Sharing in Latencysensitive Applications

Samuel D. Okegbile, Jun Cai, Jiayuan Chen, Changyan Yi

May, 2022 Journal article

The proposed secure differentially private federated multitask learning (DPFML) framework for HDT.

Abstract

This paper presents a blockchain-enabled data sharing framework for latency-sensitive applications. To ensure an improved throughput while reducing the overall latency, we propose a parallel validation-based reputation-enhanced practical Byzantine fault tolerance consensus framework with a priority-based block appending process to avoid forking attacks. The proposed framework allows multiple simultaneous validation processes thereby improving the overall performance of the data sharing system while ensuring that important requirements of the blockchain-enabled framework such as security and decentralization are not compromised. Furthermore, we formulate the communication process among validators and their computation resource allocation as a Markov decision process to optimize the transaction throughput while reducing the overall latency. We then adopt the branching dueling Q-network approach to address the large dimensions action space issue in our formulated problem and obtain simulation results to evaluate the performance of the proposed framework. The results show that the proposed framework can significantly improve the performance of blockchain-enabled data sharing by ensuring higher throughput and lower latency while maintaining an acceptable level of security among the validating nodes.

Type

Preprint

Publication

TechRXiv

This work has unfortunately been rejected multiple times since May 2022. However, we have continued to improve it based on the valuable feedback from reviewers. We appreciate all the comments and support. Let’s remain hopeful and continue our efforts together. 🙏

If you want cite this work, please click the Cite button above to export the bib format. Thank you! 😊

Blockchain Reinforcement Learning Data Sharing

Jiayuan Chen

PhD Candidate in Computer Science and Technology

My research interests include Human Digital Twin (HDT), Network Resource Management, Edge Computing and Edge Intelligence, Tactile Internet (TI), and Data-Driven Optimization and Learning.