Recently, federated learning (FL) has been a hotspot for its capacity of data privacy protection and excellent performance under few-shot conditions for Internet of Things (IoT) services. Meanwhile, 6G non-terrestrial network (NTN) provides an effective and affordable option for enhancing IoT device connectivity. When FL meets NTN, various challenges and opportunities will emerge to promote technological evolution in the field of IoT services. Motivated by this, this paper investigates the present situations of FL in NTN from the perspective of perceptive mobile network (PMN), and discusses the open challenges for FL-assisted PMN. Additionally, current opportunities are concluded from three aspects, including sensing and communication (S&C) aided learning, S&C as a task, and edge intelligence. Finally, the future directions are exploited and analyzed. This paper overviews NTN from the perspective of PMN and proposes the framework of sensing assisted FL in NTN. We hope that this article will provide some inspirations for FL and wireless communication researchers.

中文翻译：

---

面向物联网服务的6G非地面网络中的联邦学习：从感知移动网络的角度


近年来，联邦学习（FL）因其数据隐私保护能力和在少样本条件下的优异性能而成为物联网（IoT）服务的热点。同时，6G非地面网络（NTN）为增强物联网设备连接提供了有效且经济实惠的选择。当FL遇上NTN，将会出现各种挑战和机遇，推动物联网服务领域的技术演进。受此启发，本文从感知移动网络（PMN）的角度研究了NTN中FL的现状，并讨论了FL辅助PMN面临的挑战。此外，当前的机会还可以从三个方面来总结，包括传感与通信（S&C）辅助学习、S&C作为任务以及边缘智能。最后，对未来的发展方向进行了探索和分析。本文从PMN的角度概述了NTN，并提出了NTN中传感辅助FL的框架。我们希望本文能为FL和无线通信研究人员提供一些启发。