Title: BlockFedL: a blockchain-based federated learning framework for securing smart UAV delivery systems at the edge

Authors: Chengzu Dong; Zhiyu Xu; Shantanu Pal; Frank Jiang; Shiping Chen; Chong Zhang; Xiao Liu

Addresses: Deakin University, Melbourne, Australia ' Swinburne University of Technology, Melbourne, Australia ' Deakin University, Melbourne, Australia ' Deakin University, Melbourne, Australia ' CSIRO, Data61, Australia ' Deakin University, Melbourne, Australia ' Deakin University, Melbourne, Australia

Abstract: The integration of edge computing into advanced UAV delivery systems is of great interest to both research and industry. This integration offers new business opportunities and serves as a testbed for innovative technologies like edge computing, blockchain, and machine learning. A key concern for these systems is data privacy, especially given the large amounts of user and UAV data processed for tasks such as self-guided navigation, facial recognition, and person re-identification (ReID). To address this, federated learning (FL) has emerged as a popular choice, allowing for model parameter sharing while keeping raw data private. However, traditional FL approaches are vulnerable to single points of failure. Our study introduces the 'blockchain-powered edge FL' (BlockFedL) framework, a blockchain-enhanced, decentralized FL framework for edge-based UAV delivery systems. BlockFedL leverages blockchain to form a decentralized FL network, ensuring secure data storage and mitigating risks. We specifically investigate privacy issues in the person ReID application for smart UAV delivery systems and introduce a proof of quality factor (cPoQF) consensus protocol to address blockchain scalability challenges. Experimental results demonstrate improvements in energy consumption, transaction speed, and processing capacity, highlighting the framework's effectiveness.

Keywords: UAV delivery; blockchain technology; internet of things; IoT; edge computing; collaborative learning; federated learning; intelligent communications.

DOI: 10.1504/IJACT.2024.144916

International Journal of Applied Cryptography, 2024 Vol.5 No.1, pp.13 - 29

Received: 29 Jan 2024
Accepted: 08 Apr 2024
Published online: 10 Mar 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article