arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2510.18501eess.SP
收藏

用于实时物联网入侵检测的 Grassmann 流形微秒级联邦 SVD

Microsecond Federated SVD on Grassmann Manifold for Real-time IoT Intrusion Detection

Tung-Anh Nguyen, Van-Phuc Bui, Shashi Raj Pandey, Kim Hue Ta, Nguyen H. Tran, Petar Popovski

AI总结:

提出无监督联邦学习框架 FedSVD,结合 SVD 与 Grassmann 流形优化,在低功耗设备上实现低通信、低计算的实时物联网已知与未知入侵检测,性能媲美深度学习基线且推理延迟降低逾10倍。

摘要:

本文介绍 FedSVD,这是一种用于物联网网络实时异常检测的新型无监督联邦学习框架。通过利用奇异值分解(SVD)和 Grassmann 流形上的优化,FedSVD 能够在不依赖标记数据或集中式数据共享的情况下,准确检测已知和未知入侵。该方法专为部署在 NVIDIA Jetson AGX Orin 等低功耗设备上而定制,显著降低了通信开销和计算成本。实验结果表明,FedSVD 取得了与深度学习基线相当的性能,同时将推理延迟降低了 10 倍以上,使其适用于延迟敏感的物联网应用。

英文摘要:

This paper introduces FedSVD, a novel unsupervised federated learning framework for real-time anomaly detection in IoT networks. By leveraging Singular Value Decomposition (SVD) and optimization on the Grassmann manifolds, FedSVD enables accurate detection of both known and unknown intrusions without relying on labeled data or centralized data sharing. Tailored for deployment on low-power devices like the NVIDIA Jetson AGX Orin, the proposed method significantly reduces communication overhead and computational cost. Experimental results show that FedSVD achieves performance comparable to deep learning baselines while reducing inference latency by over 10x, making it suitable for latency-sensitive IoT applications.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑