基于多重判别分析和分形维的数字调制样式自动识别

将不同数字调制样式的通信信号采样重组成为二维数据阵列后，会呈现独特的纹理。根据分形型理论，不同纹理具有不同的多分形维，因此本文提出通过多重分形维区别不同调制样式的方法。首先用离散小波去噪的方法处理含噪声的采样信号，然后计算信号纹理的多重分形维，再将分形维特征向量通过多重判别分辨分析投影到Fisher超平面，最后用最大似然方法分类识别特征向量。用算法对四种数字调制样式的分析获得了正确的分类结果，证明将多重分形维用于识别数字调制样式是可行的。

Automatic Digital Modulation Recognition Using MDA and Fractal Dimension

The 2-D arrays of data reshaped from different communication signals have their unique texture which can be characterized by fractal. This paper put forward a method to classify modulation type according to this. First, signal is processed with discrete wavelet denoise algorithms. Then multi-fractal dimension vector is obtained from texture of signal array and projected onto Fisher hyper plane. At last, feature vectors on the Fisher hyper plane at classified with maximum likelihood method. The result of processing 4 types of modulation signals shows that multi-fractal dimension can be used to recognize these signals.