基于DBN-SVM的水下调制识别技术研究
详细信息 查看官网全文
- 英文论文题名：Research on underwater modulation recognition technology based on DBN-SVM
- 论文作者：张晓亮 ; 刘辉霞 ; 齐洁 ; 孙海信 ; 邱致刚
- 英文论文作者：ZHANG Xiao-liang ; LIU Hui-xia ; QI Jie ; SUN Hai-xin ; QIU Zhi-gang ; Science and Technology on Underwater Acoustic Antagonizing Laboratory ; School of Information Science and Engineering ; Xiamen University
- 年：2016
- 作者机构：水声对抗技术重点实验室;厦门大学信息与科学技术学院;
- 论文关键词：调制识别 ; 支持向量机 ; 深度置信网络
- 英文论文关键词：Modulation Recognition ; Support Vector Machine ; Deep Belief Network
- 会议召开时间：2016-10-28
- 会议录名称：2016年全国声学学术会议论文集
- 语种：中文
- 分类号：TN911.3
- 学会代码：OGSM
- 会议名称：2016年全国声学学术会议
- 会议地点：中国湖北武汉
- 主办单位：中国声学学会
- 学会名称：中国声学学会
- 页数：4
- 文件大小：346k
- 原文格式：D
- 会议级别：全国
摘要
为减少水下通信中信号信息损失的问题,降低波形畸变率,提高调制识别率,解决经典支持向量机理论出现二次规划局部最优的问题,本文提出一种将深度置信网络(DBN,Deep Belief Network)和支持向量机(SVM,Support Vector Machine)相结合的算法,设计出抗时频衰落信道的调制信号分类器。利用深度置信网络构造多特征融合权重矩阵,解决信息损失和畸变造成的模糊效应,提高信息衰变时的分辨率;然后支持向量以此权重矩阵作为模型,识别不同类型调制信号。仿真实验表明,在瑞利多途衰落信道环境下,该分类器的识别率优于单一特征分类方法,而且有较高的鲁棒性。
In order to reduce the problem of underwater communication signal information loss, decrease waveform distortion rate, improve the modulation recognition rate, and solve the problem of local optimum in quadratic programming which arises in classical Support Vector Machine(SVM, Support Vector Machine) theory, the paper presented an algorithm which combined Deep Belief Network(DBN, Deep Belief Network) with SVM, and designed a resistance to the time-frequency fading channel modulation signal classifier. Use DBN to construct multiple feature fusion weight matrix, to relieve the blur effect caused by the information loss and distortion, and to improve resolution ratio in information decay, then SVM set the weight matrix as the model, to identify the different types of modulation signal. The simulation results show, in Rayleigh multipath fading channel environment, the recognition rate of the classifier is superior to the single feature classification method, and have quite higher robustness.
In order to reduce the problem of underwater communication signal information loss, decrease waveform distortion rate, improve the modulation recognition rate, and solve the problem of local optimum in quadratic programming which arises in classical Support Vector Machine(SVM, Support Vector Machine) theory, the paper presented an algorithm which combined Deep Belief Network(DBN, Deep Belief Network) with SVM, and designed a resistance to the time-frequency fading channel modulation signal classifier. Use DBN to construct multiple feature fusion weight matrix, to relieve the blur effect caused by the information loss and distortion, and to improve resolution ratio in information decay, then SVM set the weight matrix as the model, to identify the different types of modulation signal. The simulation results show, in Rayleigh multipath fading channel environment, the recognition rate of the classifier is superior to the single feature classification method, and have quite higher robustness.
引文
[1]张晓勇,罗来源.被动声呐信号检测技术发展[J].声学技术,2014(6):559-563.
[2]李海涛,程玉胜,戴卫国,等.小波包分形与支持向量机在水声目标识别中的应用研究[J].声学技术,2015(3):219-222.
[3]Xu J L,Su W,Zhou M C.Software-Defined Radio Equipped With Rapid Modulation Recognition[J].IEEE Transactions on Vehicular Technology,2010,59(4):1659-1667.
[4]Sun W,Zhao W,Niu W,et al.A DBN-Based Classifying Approach to Discover the Internet Water Army[J].Ifip Advances in Information&Communication Technology,2014,432:78-89.
[5]O'Connor P,Neil D,Liu S C,et al.Real-time classification and sensor fusion with a spiking deep belief network[J].Frontiers in Neuroscience,2013,7:178.
[6]Tong,Simon,Chang,et al.Support vector machine active learning for image retrieval[J].2015.
[7]陈方超.数字通信信号调制方式识别算法研究[D].东北大学
[2]李海涛,程玉胜,戴卫国,等.小波包分形与支持向量机在水声目标识别中的应用研究[J].声学技术,2015(3):219-222.
[3]Xu J L,Su W,Zhou M C.Software-Defined Radio Equipped With Rapid Modulation Recognition[J].IEEE Transactions on Vehicular Technology,2010,59(4):1659-1667.
[4]Sun W,Zhao W,Niu W,et al.A DBN-Based Classifying Approach to Discover the Internet Water Army[J].Ifip Advances in Information&Communication Technology,2014,432:78-89.
[5]O'Connor P,Neil D,Liu S C,et al.Real-time classification and sensor fusion with a spiking deep belief network[J].Frontiers in Neuroscience,2013,7:178.
[6]Tong,Simon,Chang,et al.Support vector machine active learning for image retrieval[J].2015.
[7]陈方超.数字通信信号调制方式识别算法研究[D].东北大学