基于循环谱和距离判别的雷达信号调制类型识别

Radar signal modulation recognition based on cyclic spectrum and distance discrimination

循环谱对雷达调制信号具有良好的可分性,文中提取雷达信号的循环谱对信号的调制类型进行分类识别。为了减小循环谱作为分类特征的计算量,采用距离判别的方法寻找最利于分类的一行循环谱信号作为样本信号的分类特征,并结合支持向量机对雷达信号的调制类型做了分类识别的计算机仿真。仿真结果表明,在0 d B时该方法对多种单个雷达信号的识别率高达92.7%,对混合雷达信号的识别率为89.7%,说明该方法在较低信噪比下对于常见的5种雷达调制信号及其相应混合而成的信号具有较高的识别率。

For the cyclic spectrum has good separability for radar modulated signals, a method based on the cyclic spectrum of radar signals for modulation type recognition is proposed in this paper. In order to reduce the computational complexity of cyclic spectrum, which is used as the classification feature, this paper uses the distance discrimination method to find one line of cyclic spectrum signals which are the most suitable for classification of the sample signals, and combines the support vector machine as a classifier to identify the modulation, and then simulates the classification. The computer simulation results show that the recognition rate for 5 kinds of single radar signals and corresponding mixed signals can respectively reach 92.7% and 89.7% at 0 dB. It shows that the method has a high recognition rate for the common 5 kinds of radar modulation signals and their mixed signals at low SNR.