基于分形维数与SVM的PUE攻击检测  

Method of PUE Attack User Detection Based on Fractal Dimension and SVM

作　　者：马强[1] 田红园 郑文秀[1] 

机构地区：[1]西安邮电大学通信与信息工程学院,陕西西安710061

出　　处：《无线电通信技术》2015年第2期26-28,58,共4页Radio Communications Technology

基　　金：国家自然科学基金项目(61271276);陕西省教育厅项目(14JK1668)

摘　　要：针对主用户仿冒(Primary User Emulation,PUE)攻击用户非法占用主用户(PU)信道,导致PU无法继续使用原信道的问题,在辐射源的指纹特征基础上,通过辐射源识别进行PUE攻击检测。使用分形维数刻画通信信号码元包络的脉内起伏特征,将盒维数Db与信息维数DI组成分形特征向量D;采用核函数的支持矢量机(SVM)进行训练判决识别的PUE攻击检测方法。实验结果表明,该方法具有较高的识别率,并能够较好地解决PUE攻击检测问题。As Primary User Emulation （PUE） attack users illegally occupy the primary user＇ s channel,the primary user is unable to use the original channel.Aiming at this problem,the PUE attack detection is performed by radio transmitter identification based on fingerprint characteristics of radio transmitters.The method of PUE attack detection is to first use the fractal dimension depicting the intrapulse fluctuation characteristics of communication signal envelope to make box dimension Db and information dimension D1 form the fractal feature vector D,and then use a support vector machine （ SVM ） based on kernel-function for training individual identification. The experimental results show that the proposed technique has a high recognition rate, and it can solve the problem of PUE attack identification.

关 键 词：通信辐射源 盒维数 信息维数 支持矢量机 

分 类 号：TN911.72[电子电信—通信与信息系统]