Aiming at researching multi-component signal modulation recognition under complex electromagnetic environment, with accurately estimating instantaneous frequency of the signals, the intra-pulse subtle features can be analyzed.The independent component analysis and wavelet transform was combined to study the blind source separation and modulation recognition of the multi-component emitter signals.FastICA was used to separate and decompose the time-frequency signals into a series of independent components under the condition of no priori information.Then the single component was extracted by the wavelet ridge with the local modulus maxima of wavelet coefficient.For different types of signals, a straight line of the wavelet ridge was fit by least square fitting method to obtain characteristic parameters for judging signal modulation type.Finally with the simulation experiment, the results show that the method can separate aliasing signals and effectively extracts wavelet ridge and instantaneous frequency of signals to identify the type of signal modulation.Regardless of low SNR ratio, this algorithm still has high recognition probability.%对复杂电磁环境下多分量信号进行调制识别,可通过准确估计接收信号的瞬时频率来分析其脉内细微特征.本文联合独立分量分析和小波变换技术,对多分量辐射源信号进行了盲源分离和调制识别的研究.在无先验信息条件下,采用FastICA对混叠信号分离,将时频混叠信号分解成一系列独立分量.对分离后的单分量信号分别做小波变换处理,由小波系数的局部模极大值提取其小波脊线.针对不同调制类型雷达信号,用最小二乘法对时频小波脊线进行直线拟合,获取特征参数,通过计算特征值判决出信号的调制类型.通过仿真实验表明,该方法可以分离混叠信号并有效提取信号小波脊和瞬时频率,进而识别出信号的调制类型,并在低信噪比情况下仍有较高的识别概率.
展开▼
