为解决电磁频谱中的未知信号分类和身份识别问题,提出一种基于改进卷积神经网络(CNN) Le Net-5模型的信号分类方法。该方法使用信号全双谱作为CNN的输入,然后通过改进的Le Net-5模型学习信号特征并完成信号分类和身份识别。实验结果表明...为解决电磁频谱中的未知信号分类和身份识别问题,提出一种基于改进卷积神经网络(CNN) Le Net-5模型的信号分类方法。该方法使用信号全双谱作为CNN的输入,然后通过改进的Le Net-5模型学习信号特征并完成信号分类和身份识别。实验结果表明,算法对未知信号调制类型识别率达97%以上,对信号身份属性识别率达96%以上。相比传统方法,该算法对信号身份属性识别率提高6. 5%,具有更好的泛化性能,并有效解决了全双谱应用的二维模板匹配和Loss函数值下降缓慢的问题。展开更多
This paper studies the capacity issues of a wireless communication system that implements single channel full duplex(SCFD) communication at the base station(BS), thereby the mobile stations share the channel via time ...This paper studies the capacity issues of a wireless communication system that implements single channel full duplex(SCFD) communication at the base station(BS), thereby the mobile stations share the channel via time division duplex(TDD). The system makes use of the same setup as has been used in previous studies of SCFD, but unlike these previous systems, the new system uses water-filling to maximize the spectral efficiency of the uplink channel. The concept of a free window is introduced to the duplex model for measuring, intuitively, the effective bandwidth of the bi-directional communication. The capacity gain is calculated and numerical results show the advantage of the proposed system over that of conventional TDD.展开更多
文摘为解决电磁频谱中的未知信号分类和身份识别问题,提出一种基于改进卷积神经网络(CNN) Le Net-5模型的信号分类方法。该方法使用信号全双谱作为CNN的输入,然后通过改进的Le Net-5模型学习信号特征并完成信号分类和身份识别。实验结果表明,算法对未知信号调制类型识别率达97%以上,对信号身份属性识别率达96%以上。相比传统方法,该算法对信号身份属性识别率提高6. 5%,具有更好的泛化性能,并有效解决了全双谱应用的二维模板匹配和Loss函数值下降缓慢的问题。
基金supported by the HongKong, Macao and Taiwan Science & Technology Cooperation Program of China (Grant no. 2015DFT10170)the Beijing Higher Education Young Elite Teacher Project
文摘This paper studies the capacity issues of a wireless communication system that implements single channel full duplex(SCFD) communication at the base station(BS), thereby the mobile stations share the channel via time division duplex(TDD). The system makes use of the same setup as has been used in previous studies of SCFD, but unlike these previous systems, the new system uses water-filling to maximize the spectral efficiency of the uplink channel. The concept of a free window is introduced to the duplex model for measuring, intuitively, the effective bandwidth of the bi-directional communication. The capacity gain is calculated and numerical results show the advantage of the proposed system over that of conventional TDD.
