📄 中文摘要
随着生成性人工智能（GenAI）在消费电子领域的广泛应用，从可穿戴设备中的AI助手到无人机的生成规划，用户体验得到了革命性的提升。然而，这些GenAI应用对边缘硬件的计算能力提出了巨大的挑战，导致在全球导航卫星系统（GNSS）信号保护等基本安全任务上资源极为有限。此外，缺乏真实世界干扰数据使得为这些设备训练稳健的分类器变得困难。为了解决数据稀缺和GenAI时代对极高效率的双重挑战，提出了一种名为GAC-KAN的新框架。该框架首先采用物理指导的模拟方法，以提高分类器的性能和效率。
GAC-KAN：一种超轻量级的GNSS干扰分类器，适用于GenAI驱动的消费边缘设备
📄 English Summary
GAC-KAN: An Ultra-Lightweight GNSS Interference Classifier for GenAI-Powered Consumer Edge Devices
The integration of Generative AI (GenAI) into consumer electronics has significantly transformed user experiences, from AI-powered assistants in wearables to generative planning in autonomous Uncrewed Aerial Vehicles (UAVs). However, these GenAI applications place substantial computational demands on edge hardware, resulting in severely limited resources for essential security tasks such as Global Navigation Satellite System (GNSS) signal protection. Additionally, the scarcity of real-world interference data complicates the training of robust classifiers for these devices. To tackle the dual challenges of data scarcity and the extreme efficiency required in the GenAI era, a novel framework named GAC-KAN is proposed. This framework initially employs physics-guided simulation techniques to enhance the performance and efficiency of the classifiers.