摘要
针对短波复杂信道环境下信号所属协议识别困难的问题,提出一种基于Swin-Transformer神经网络模型的短波协议信号识别算法。首先使用时频分析方法得到信号的灰度时频图作为神经网络的输入;其次设计一种基于Swin-Transformer的神经网络模型,对信号时频图进行特征提取;最后将特征与协议建立映射关系,从而实现信号协议的识别。仿真实验结果表明,在信噪比大于−4 dB的高斯信道下,所提算法的识别准确率接近100%,高于现有算法。此外,在强干扰以及多径时延衰落的信道条件下,所提算法仍具有较高的短波协议信号识别率。
Aiming at the problem that it is difficult to identify the protocol to which the signal belongs in the complex SW channel environment,a SW protocol signal recognition algorithm based on Swin-Transformer neural network model was proposed.Firstly,the gray-scale time-frequency map of the signal was obtained by using the time-frequency analysis method as the input of the neural network.Secondly,a neural network model based on swing transformer was designed to extract the features of the signal time-frequency map.Finally,the mapping relationship between the features and the pro-tocol was established to realize the recognition of the signal protocol.The simulation results show that the recognition accuracy of the proposed algorithm is close to 100%in the Gaussian channel with SNR greater than−4 dB,which is higher than the existing algorithms.In addition,under the channel conditions of strong interference and multipath delay fading,the proposed algorithm still has a high SW protocol signals recognition rate.
作者
朱政宇
陈鹏飞
王梓晅
巩克现
吴迪
王忠勇
ZHU Zhengyu;CHEN Pengfei;WANG Zixuan;GONG Kexian;WU Di;WANG Zhongyong(School of Electrical and Information Engineering,Zhengzhou University,Zhengzhou 450001,China;Joint International Laboratory of Intelligent Network and Data Analysis in Henan Province,Zhengzhou University,Zhengzhou 450001,China;National Center for International Joint Research of Electronic Materials and Systems,Zhengzhou University,Zhengzhou 450001,China;College of Data Target Engineering,Information Engineering University,Zhengzhou 450001,China)
出处
《通信学报》
EI
CSCD
北大核心
2022年第11期127-135,共9页
Journal on Communications
基金
国家重点研发计划基金资助项目(No.2019QY0302)
中国博士后科学基金资助项目(No.2020M682345)
河南省高校科技创新人才支持计划资助项目(No.23HASTIT019)
河南省博士后经费资助项目(No.202001015)。
