摘要
针对杀伤网背景下电磁频谱资源利用所面临的挑战,首先,从杀伤网概念的界定出发,梳理分析了其发展现状及关键技术,并归纳总结了杀伤网跨域互联、动态适应性、抗毁自愈的基本特点。其次,从作战运用需求出发,对支撑杀伤网的通信网络应如何动态接入频段提高可靠通信和频谱资源利用率的问题进行了分析研究。最后,基于真实数据集,对当前应用较为普遍的频谱感知与预测模型进行仿真对比。为解决当前基于深度学习的频谱感知和预测模型所存在的不足,提出了基于深度可分离卷积和特征融合的频谱感知模型、基于Mamba模型的频谱预测方法、联合频谱预测驱动频谱感知3个解决方案,以应对杀伤网作战效能有效发挥时所面临的实时高效、低计算复杂度挑战。
First,the current development status and key technologies were analyzed based on the definition of a kill web in response to the challenges faced by the utilization of electromagnetic spectrum resources in the context of the kill web.The basic features of cross-domain interconnec-tion,dynamic adaptability,anti-destruction,and self-healing of the kill web were summarized.Second,based on the operational requirements,a study was conducted to investigate how the communication network supporting kill web should dynamically access frequency bands to enhance reliable communication and spectrum resource utilization.Finally,based on real datasets,simula-tion comparisons were performed using commonly used spectrum sensing and prediction models.To solve the deficiencies of current deep learning-based spectrum sensing and prediction models,three solutions were proposed:a spectrum sensing model based on depthwise separable convolu-tion and feature fusion;a spectrum prediction method based on the Mamba model;and joint spec-trum prediction-driven spectrum sensing,responding to the challenges of real-time,efficiency,and low computational complexity faced by the kill web in effective combat performance.
作者
刘丁胤
徐东辉
刘思佚
康凯
LIU Dingyin;XU Donghui;LIU Siyi;KANG Kai(Rocket Force University of Engineering,Xi’an 710025,China)
出处
《火箭军工程大学学报》
2025年第4期42-50,共9页
Journal of Rocket Force University of Engineering
基金
军事类研究生资助课题(JYKT910342023006)。
关键词
杀伤网
频谱感知
频谱预测
深度学习
kill web
spectrum sensing
spectrum prediction
deep learning
