基于高阶谱谱骨架的信号细微特征识别

HIGHER-ORDER SPECTRUM SKELETON APPLICATION TO SIGNAL FINE CHARACTER IDENTIFICATION

为解决同型号辐射源识别问题,针对实际信号,采用信号的高阶谱作为个体识别的基本特征向量,建立基于软K段主曲线算法的高阶谱谱骨架模型。将谱骨架的信息维数和盒维数作为特征矢量,并结合信号的时频域分析。最后将得到的融合特征使用SVM分类器进行训练识别,实现对不同辐射源信号的个体识别。通过对比实验充分验证该方法的有效性。实验结果表明,在低信噪比的环境下,该方法能够有效地识别个体信号,具有更好的识别效果,识别率可达到85%以上。

Aiming at the pract ical signals, a high order spectrum skeleton model which based on the principal curves of K-segments algorithm is constructed to solve the problem of signal identification among transmitters with same model. We take the information dimension and box dimension of the skeleton as transient feature, which is combined with the time-frequency domain analysis. At last, the derived feature vectors are trained by SVM classifier to recognize the signals emitted from different sources. Experiment results show that this method can classify the same model transmitters. The recognition rate can still reach 85 percent or higher in the condition of low SNR.