利用四阶累积量实现数字调相信号子类识别*

摘要

The non‐cooperative modulation identification for HF communication signals takes an important status in the communication reconnaissance recognition .A new algorithm based on instant autocorrelation four order cumulants feature extraction is proposed for the classification of five kinds of HF signals ,phase shift keying(PSK) signals .The algorithm can classify signals such asπ/4‐QPSK、π/4‐DQPSK、8PSK which have the same picture of constellation and higher‐order cumula‐nts .The algorithm has the higher performance in suppressing complex Gaussian white noise and robustness ,which is illumi‐nated and verified by simulation experiments .%短波通信信号调制方式的非合作识别在通信侦察识别中占有十分重要的地位。文中针对BPSK、QPSK、π／4‐QPSK、π／4‐DQPSK、8PSK五种常见短波通信信号，提出了一种利用信号瞬时自相关四阶累积量作为识别特征的分类算法。该算法解决了π／4‐QPSK、π／4‐DQPSK、8PSK三种相移键控信号具有相同星座图和高阶累量而无法区分的问题，并进行了理论计算，通过M atlab仿真验证了该方法在低信噪比条件下，有很好地抑制平稳高斯白噪声的作用，算法具有一定的稳健性。