Abstract: A recognition method of quadrature modulated signals on baseband is presented.Compared with conventional methods operated in mid-frequency，the computational complexity of this method is reduced for its lower frequency property.By extracting robust features from the received signals’ spectrum and statistical characteristics，it can classify the signals automatically without prior knowledge whether the signal modulation group is analogue or digital.These features perform well under a low SNR circumstance and are insensitive to frequency offset and phase offset.The performance of this algorithm is verified by some real data，and experiment results show that its overall recognition rate can exceed 95% when SNR is above 7 dB.

Key words: blind modulation recognition, baseband feature extraction, communication signal

摘要： 基于基带信号频率低，算法计算量小的优点，提出了一种在不需信号先验知识的情况下识别基带正交调制信号的方案。该方案利用接收信号的基带谱特性和统计特性，从中提取抗噪声，且对频偏和相偏不敏感的特征，对信号进行模拟数字的联合自动识别。对真实采集信号进行实验验证，结果表明：在信噪比不小于7 dB时，对数字和模拟调制方式都未知的信号，算法可以得到95%以上的正确识别率。

关键词: 调制盲识别, 基带特征参数提取, 通信信号