A method based on fractal box dimension and square spectrum line features is proposed perform modulation recognition for the signal set consisting of continuous phase modulation (CPM) and PSK⁃type signals in this paper.Firstly the fractal box dimension feature of each signal to be recognized in the signal set is extracted to figure out the CPM and OQPSK signal.Then the discrete spectral lines feature of square spectrum of the remaining signals is extracted to recognize BPSK, UQPSK and QPSK in turn. Finally, the classifier is designed according to the principles of decision tree to realize the recognition for signals within the set. The simulation results show that the proposed algorithm can realize the correct recognition rate of the signals in the set of over 90% when the SNR ( signal to noise ratio) is above 3dB.%提出了一种基于分形盒维数及二次方谱特征的方法,对由 CPM 与 PSK 类信号组成的信号集进行调制识别分类。首先提取信号集内各待识别信号的分形盒维数特征,将 CPM 和 OQPSK 识别出来;然后对信号集中剩余信号提取相应的二次方谱离散谱线特征依次区分出BPSK、UQPSK 和 QPSK;最后根据决策树原理设计分类器,实现信号集内信号的识别。仿真结果表明,该算法对信号集内的信号在信噪比达到3 dB 以上条件下正确识别率可达到90%以上。
展开▼