Scholarlink

<p>This systematic literature review explores Federated Learning (FL) within the context of Unmanned Aerial Vehicle (UAV) applications. FL works by training a global model among clients, where the model is trained locally on each client, and only the model updates are shared. This approach maintains privacy and enables collaborative learning without sharing raw data. The collaborative efforts of multiple UAVs, however, introduce statistical heterogeneity in the collected sensing data due to variations in their respective monitoring areas. In this review, we analyze 31 papers published between 2016 and October 2023. Our review highlights the data properties, FL frameworks, applications, and evaluation methodologies used in these studies. We provide a detailed classification of the current state-of-the-art in FL, particularly focusing on approaches to manage statistical heterogeneity. This review also includes an assessment of the various evaluation methods used in the literature. This review offers a concise overview of the advancements made in addressing statistical heterogeneity in research studies. We will highlight key progress, identify persistent challenges, and explore future research directions. Ultimately, our goal is to provide insights into the ongoing developments in Federated Learning applications for UAV.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A systematic review of federated statistical heterogeneity in UAV applications

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月25日

查看全文
  • Bara Al-Bataineh,
  • Sirko Schindler,
  • Diana Peters,
  • Marcus Paradies,
  • Matthias Pohl

摘要

This systematic literature review explores Federated Learning (FL) within the context of Unmanned Aerial Vehicle (UAV) applications. FL works by training a global model among clients, where the model is trained locally on each client, and only the model updates are shared. This approach maintains privacy and enables collaborative learning without sharing raw data. The collaborative efforts of multiple UAVs, however, introduce statistical heterogeneity in the collected sensing data due to variations in their respective monitoring areas. In this review, we analyze 31 papers published between 2016 and October 2023. Our review highlights the data properties, FL frameworks, applications, and evaluation methodologies used in these studies. We provide a detailed classification of the current state-of-the-art in FL, particularly focusing on approaches to manage statistical heterogeneity. This review also includes an assessment of the various evaluation methods used in the literature. This review offers a concise overview of the advancements made in addressing statistical heterogeneity in research studies. We will highlight key progress, identify persistent challenges, and explore future research directions. Ultimately, our goal is to provide insights into the ongoing developments in Federated Learning applications for UAV.