一种敌我识别辐射源暂态信号指纹特征提取方法

信号的指纹特征是辐射源个体识别的重要依据。针对敌我识别辐射源的个体识别问题，提出了一种基于双树复小波和多重分形的信号暂态特征提取方法。该方法通过双树复小波变换实现信号多分辨率分解，求解分解信号Hilbert谱的信息熵和指数熵，计算信号的多重分形奇异指数和谱值，最终组成表征辐射源的特征向量。通过实验验证，提取的特征向量能充分代表辐射源个体之间的差异；被测信号的信噪比满足8 dB或9 dB的条件时，对辐射源的识别正确率能达到90%以上。统计分析表明该方法提取的特征具有很高的稳定性。

A Fingerprint Feature Extraction Method of Transient Signal of IFF Emitter

The fingerprint feature of signal is an important basis for specific emitter identification(SEI).For the problem of SEI for identification of friend or foe(IFF),a new method of signal transient feature extraction based on dual-tree complex wavelet and multifractal is proposed.In this method,multi-resolution signal decomposition is realized by dual-tree complex wavelet transform(DT-CWT),information entropy and exponential entropy of the decomposed signal Hilbert spectrum are solved,multifractal singular index and spectral value of the signal are calculated,and finally the eigenvector representing the emitter is formed.The experimental results show that the extracted eigenvector can fully represent the differences between the specific emitters.When the signal-to-noise ratio(SNR) of the measured signal meets the conditions of 8 dB or 9 dB,the recognition accuracy of the emitter can reach more than 90%.The statistical analysis shows that the features extracted by this method have high stability.