一种基于时频分析神经网络的通信信号盲识别新方法

为解决通信盲侦察中的信号分类识别问题,提出一种新的通信信号盲识别方法,利用时频分析技术提取信号特征,采用神经网络(BP网络)方法对信号进行分类识别。仿真结果表明,该方法在无噪声情况下,平均识别概率在85%以上,相位调制信号的识别率甚至高达98%。如果信噪比比较高,则信号的平均识别率也能达到85%左右。该方法的特点是无需知道信号的任何先验信息,可对宽频带范围内的任何信号进行识别,无需在识别前进行载频同步或进行信号参数估计。

Novel Method for Blind Recognition of Communication Signal Based on Time-frequency Analysis and Neural Network

In this paper,we present a novel method for blind recognition of communication signal to solve the problem of the sig- nal classification and recognition in communication reconnaissance.In this method,we use time-frequency analysis technology to extract the feature of the signals,and the neural network (backwards propagation network) to class and recognize the signals.The simulation results show that in case of no noise,blind recognition method proposed in this paper can correctly recognize more than 85% signals av- eragely and recognition ratio of PSK signals reaches 98%.If the ratio of signal to noise is high,the average recognition ratio of the signals can also reaches about 85%.Moreover,we need not know any prior information of the signals to recognize any signals within broad fre- quency band and we need not perform carrier frequency synchronization and estimate the parameter of the signal.