【作者】 谢阳； 王世练； 张尔扬； 赵自璐；

【Author】 XIE Yang;WANG Shilian;ZHANG Eryang;ZHAO Zilu;College of Electronic Science and Engineering,National University of Defense Technology;

【机构】 国防科学技术大学电子科学与工程学院；

【摘要】 提出一种将时间序列非线性复杂度分析应用于通信辐射源指纹特征提取的方法,进一步进行个体识别。首先利用经典功放非线性模型对不同个体进行行为建模,然后从通信信号的非线性特征入手,提出了一种改进的差分近似熵(dApEn)算法来衡量不同个体信号的非线性复杂度差异,通过MIX模型和Logistic映射验证了改进算法的有效性。而后将差分近似熵与经验模态分解(EMD)技术相结合提取多维指纹特征,最后采用支持向量机(SVM)对不同个体的多维指纹特征进行识别分类。利用matlab平台对识别系统在高斯白噪声和Alpha稳定分布噪声环境下进行个体识别性能仿真验证,结果表明,所提算法能够精确识别不同个体,相对于HHT时频谱能量熵法和HHT谱相关分析法,能够有效克服高斯白噪声和冲击噪声的影响,系统的识别效率较高,鲁棒性好。更多还原

【Abstract】 This paper proposes a new method of extracting the radio frequency(RF) fingerprints for specific emitter identification in communications,which utilizes the nonlinear complexity of signals.First,we get the different emitters models by the classical nonlinear power amplifier model.Then a new modified algorithm,difference Approximate Entropy(dApEn),is proposed to measure the differences of nonlinear complexities in different emitters.We verify the effective of proposed algorithm with MIX model and Logistic maps.Compared with Empirical Mode Decomposition(EMD),the multi-RF fingerprints are obtained.Finally support vector machine(SVM) is used for training and sorting these fingerprints.Computer simulations are conducted under additive white Gaussian noise as well as alpha stable distribution noise and the results demonstrate that the proposed algorithm can identify the different emitter models accurately and has good robustness,and significantly out-performs the energy-entropy algorithm and correlation algorithm based on the Hilbert-Huang Transform.更多还原