摘要
随着无线通信技术的发展,电磁环境变得越来越复杂,各种类型的信号与噪声同时存在于频谱中,这对接收机的信号检测技术提出了更大的挑战。对通信信号的检测是无线对抗过程中非常重要的一环,只有实现对信号的正确检测才能够完成后续的信号识别、参数估计、解调等步骤。信号的发现检测最主要的目的,是通过宽带频谱数据判断是否有信号存在。此外,实际的战场环境下可能同时充斥着跳频、猝发以及定频等信号,不同的信号对应着不同的处理方式。传统算法通常采用的是能量检测的方式,但在信噪比较低的情况下很难区分噪声与信号。而实际情况是可以在时频图中发现信号,并观察出类别。因此,通过对时频瀑布图的特征提取,提出一种基于深度学习的更高精度的信号发现检测和分类方法。
With the development of wireless communication technology,the electromagnetic environment has become increasingly complex and various types of signals and noise coexist in the spectrum,posing greater challenges to the signal detection technology of receivers.The detection of communication signals is a crucial part of wireless adversarial processes.Only by correctly detecting signals can subsequent steps such as signal recogni‑tion,parameter estimation and demodulation be completed.The main purpose of signal discovery and detection is to determine the presence of signals through broadband spectrum data.In addition,the actual battlefield environ‑ment may be filled with signals such as frequency hopping,burst and fixed frequency,and different signals corre‑spond to different processing methods.In traditional algorithms,energy detection is commonly used,but it is dif‑ficult to distinguish between noise and signal when the signal-to-noise ratio is low.But in reality,the signal can be detected in the time-frequency graph and the category can be observed.Therefore,a method based on deep learn‑ing is proposed that achieves higher accuracy in signal discovery,detection and classification by extracting fea‑tures from time-frequency waterfall plots.
作者
王泽群
徐璐
时睿
张一嘉
Wang Zequn;Xu Lu;Shi Rui;Zhang Yijia(School of Information Science and Engineering,Zhejiang Sci‑Tech University,Hangzhou 310018,Zhejiang,China)
出处
《航天电子对抗》
2024年第4期44-51,共8页
Aerospace Electronic Warfare
