基于固有时间尺度分解的信号细微特征识别

ITD-Based Identification of Signal Fine Features

为解决无线复杂环境下同型号通信电台发送的信号识别问题,针对传统时频方法处理杂散细微特征存在不足,提出了一种基于固有时间尺度分解(intrinsic time-scale decomposition,ITD)的信号细微特征识别方法。首先通过ITD方法将稳态状态下信号分解,计算分量瞬时参数并得到信号的时频谱,然后提取频谱特征作为细微特征,最后使用SVM分类器进行模式训练以实现信号的识别。仿真结果表明,该算法能够解决传统方法的实时性和准确性差等问题,取得较好的识别效果。

This paper presents a new model for signal feature identification based on intrinsic timescale decomposition to solve the problem with the identification of the same signals from different radio transmitters in complex wireless environment,and to eliminate shortcomings in dealing with stray features with the traditional time-frequency method. First,intrinsic time-scale decomposition is adopted to decompose the steady state signal,and instantaneous parameters of components are calculated in order to gain time-frequency spectrum. Then,spectrum features are extracted as the fine features,which are used for the classification identification of signals by SVM classifier. Simulation results show that the method not only solves the problem of the poor instantaneity and accuracy of the traditional time-frequency method,but also identifies the transmitters efficiently.