基于时频能量谱特征的跳频电台个体识别

针对于跳频电台的细微特征分类识别问题，提出基于跳频信号时频能量谱的细微特征提取算法。首先，利用跳频信号在时频域的稀疏特性，通过稀疏重构方法得到跳频信号时频能量谱；然后，在不同尺度条件下对时频能量谱进行分割，分别提取时频能量谱瑞利熵、多重分形维数和差分盒维数三种特征；最后，通过支持向量机分类器对提取特征集进行训练、分类和识别，实现跳频电台个体识别。利用四部电台的跳频信号，验证对比了本文算法与另外两种算法的识别性能。实验结果表明，本文方法所提取的细微特征集具有较强的分辨能力，避免了由单一特征的相似性而引起的误判问题，能够在少量训练样本条件下，保持较高的识别正确率。 

Frequency hopping radio individual identification based on time frequency spectrum characteristics

For the individual identification problem of frequency hopping radio stations, a feature extraction algorithm based on time-frequency energy spectrum was proposed. Firstly, using the sparseness of frequency hopping signals in the time-frequency domain, the time-frequency energy spectrum was obtained by a sparse reconstruction method. Then, the time-frequency energy spectrum was segmented under different scale conditions, and the three characteristics of time-frequency distribution Rayleigh entropy, multi-fractal dimension and differential box dimension were extracted. Finally, the support vector machine classifier was used to train, classify and identify the extracted feature set to realize the individual identification of the frequency hopping station. Using the frequency hopping signals of four stations, the recognition performance of the proposed algorithm and the other two algorithms were compared. The experimental results show that the subtle feature set extracted by this method has strong resolving power, avoids the misjudgment caused by the similarity of single features, and can maintain a high recognition accuracy under a small number of training samples.