基于相位统计信息特征的相移键控类信号识别

Identification of phase shift keying-like signals based on phase statistical features

为解决非协作通信中,相移键控类信号由于相位相似,导致在复杂环境下的分类识别困难的问题,本文在高斯噪声下通过对信号相位信息概率密度的推导,并使用高斯核密度估计的方法得到了一种恢复信号相位信息特征的方法。通过仿真实验将该方法从加性高斯噪声推广到加性Alpha稳态分布噪声中。在该方法下,本文提出的特征对BPSK、QPSK、OQPSK、π/4_DQPSK、8 PSK 5类信号在Alpha噪声下可以准确识别。研究表明:使用支持向量机方法可以在信噪比0 dB以上时使信号的总体识别率达到90%以上。

In noncollaborative communication,phase-shift keying(PSK) class signals are difficult to classify and recognize in a complex environment owing to phase similarity.To solve this problem,this study presents a method for recovering the characteristics of signal phase information,which are obtained by Gaussian kernel density estimation based on the derivation of the probability density of signal phase information under Gaussian noise.In addition,the method is generalized from additive Gaussian noise to additive alpha steady-state distribution noise through simulation experiments.Under this method,the proposed features can be accurately recognized for five types of signals,namely,BPSK,QPSK,OQPSK,π/4_DQPSK,and 8 PSK,under alpha noise.It is shown that using the support vector machine method,the overall recognition rate of the signals can achieve more than 90% at signal-to-noise ratios above 0 dB.