摘要
针对MFSK(多进制频移键控)调制信号的分类问题,通过改进高阶累积量特征值,联合利用盒维数理论构造二维特征矢量,设计神经网络分类器的识别方案,可以实现对2FSK、4FSK、8FSK信号的调制识别.理论及仿真表明,构造的二维特征矢量抗噪声性能好,对载波相位差不敏感,神经网络分类器可以适应较大的信噪比动态范围,在信噪比大于5 dB时可以达到98%的识别率.
A new algorithm for automatic modulation classification of M-ray Frequency Shift Keying（MFSK） signals is proposed.By combining cumulant and box dimension parameters,constructing two-dimension vector space,and designing neural network classifier,the algorithm can correctly recognize 2FSK 4FSK and 8FSK signals.Both theory and simulation show that proposed two-dimension vector is anti-noise and is not sensitive to signal-to-noise ratio（SNR）,and the neural network classifier is adaptable to large dynamic ranges of SNRs,and a 98% signal identification probability is obtained when SNR is above 5 dB.
出处
《郑州大学学报（工学版）》
CAS
北大核心
2010年第6期47-50,共4页
Journal of Zhengzhou University（Engineering Science）
基金
武器装备军内科研计划(KJ09×××)
关键词
累积量
盒维数
调制识别
MFSK
cumulant
box dimension
modulation identification
MFSK
