An Efficient Real-Time Jamming Frequency Detection Method with System Implementation for Intelligent Edge Computing

针对传统干扰频点检测方法在处理大规模、高动态频谱环境时存在的效率低、实时性差等问题，本文提出了一种面向智能边缘计算场景的高效实时干扰频点检测方法与系统实现方案。首先，结合机器学习、多标签学习等框架与技术设计了一种基于随机森林的轻量级、高精度的干扰频点检测模型；然后，通过模型迁移技术将训练好的模型移植到具有AI芯片(Atlas 200)的边缘计算设备上，以适配边缘计算节点；最后，引入批量处理策略，以提升模型在边缘设备端的计算速度。实验结果表明，所提方法在复杂场景下的干扰频点检测平均准确率达到98%以上，最优推理速度达到微秒级，显著提升了检测精度与实时性，适用于复杂多变的无线通信环境。Aiming at the problems of low efficiency and poor real-time performance of traditional jamming detection methods in handling large-scale and highly dynamic spectrum environments, this paper proposes an efficient and real-time jamming frequency detection method and system implementation scheme for intelligent edge computing scenarios. Firstly, a lightweight and high-precision jamming frequency detection model based on random forest is designed by combining the frameworks and technologies of machine learning and multi-label learning. Then, the trained model is transferred to an edge computing device equipped with an AI chip (Atlas 200) using model migration technology to adapt to the edge computing node. Finally, a batch processing strategy is introduced to improve the computational speed of the model on the edge device. Experimental results show that the proposed method achieves an average accuracy of over 98% in jamming frequency detection under complex scenarios, with the optimal inference speed reaching the microsecond level. This significantly enhances the detection accuracy and real-time performance, making it suitable for complex and variable wireless communication environments.