无线电工程2024，Vol.54Issue(1)：24-31,8.DOI:10.3969/j.issn.1003-3106.2024.01.004
基于残差神经网络和注意力机制的频谱感知方法
Spectrum Sensing Method Based on Residual Neural Network and Attention Mechanism
摘要
Abstract
With the development of communication technology,spectrum sensing technology has become one of the important solutions to solve the scarcity of spectrum resources.For the low accuracy of traditional spectrum sensing methods under low Signal to Noise Ratio(SNR),an Orthogonal Frequency Division Multiplexing(OFDM)spectrum sensing method based on the combination of residual neural network and attention mechanism is proposed.The spectrum sensing problem is transformed into a binary image classification task.The cyclic autocorrelation grayscale images are produced by analyzing the cyclic autocorrelation characteristics of OFDM signals to perform grayscale processing.Subsequently,deep features from the grayscale images are extracted through training an improved residual neural network,and the resulting spectrum sensing model is validated using a test dataset.The simulation experiments show that the proposed method exhibits superior spectrum sensing performance under low SNR conditions,surpassing conventional spectrum sensing techniques.关键词
频谱感知/残差神经网络/注意力机制/循环自相关Key words
spectrum sensing/residual neural network/attention mechanism/cyclic autocorrelation分类
信息技术与安全科学引用本文复制引用
王安义,孟琦峰,王明博..基于残差神经网络和注意力机制的频谱感知方法[J].无线电工程,2024,54(1):24-31,8.基金项目
国家自然科学基金联合资助项目(U19B2015)Project Jointly Supported by National Natural Science Foundation of China(U19B2015) （U19B2015）
