An Edge based Federated Learning Framework for Person Re-identification in UAV Delivery Service

AI (Artificial Intelligence) technology has been widely used in smart systems which usually require computing services with high availability and fast response. However, the rapid growth of data and service requests generated by end devices brings critical challenges to the centralised cloud computing paradigm in terms of network bandwidth, reliability and response time. In addition, the problem of data privacy is arising due to a large amount of data being transferred to the cloud server. Recently, edge computing is becoming a popular platform for smart systems as its provisions computing services close to the end devices, and Federated Learning (FL) is emerging as a promising solution for AI applications to address the data privacy issue. Inspired by their success, in this paper, we propose an edge based FL framework named Fed-UAV to solve the person reidentification problem in the UAV delivery service which is a typical AI application in smart logistics. This framework enables the UAV to efficiently locate the target receivers, and effectively reduce the data transmission between the UAV and the cloud server to improve the response time and protect the data privacy. Comprehensive experiments are conducted on three real-world datasets, and the experimental result successfully demonstrates that Fed-UAV can achieve both high accuracy and efficiency in person re-identification while protecting data privacy.