Modulation Recognition Scheme Using Mixed Model

With the advancement of the communication technology, digital communications plays an important role in the area of civilian and military. Modulation is the vital skill in communication. So the automatic modulation recognition is the theme of today's digital communications. Digital modulation recognition mainly apply to figures confirm, interference identification, radio listening, electronic countermeasures and software radio e.t. In this paper, firstly we described the basic theory of digital signals modulation recognition, secondly we introduces the knowledge of higer order cumulants, at last we elaborate the details of the feature extraction. Also in this paper, we described two modulation recognition algorithm:in the condition of known-SNR, we use a multiple classfication neural network algorithm for automatic modulation recognition and in the other conditon, we used a mixed automatic modulation recognition. By comparing two methods, we can see that the noise have an great interference in the aspect of signals modulation recogniton. By joint with higher cumulant, we can see that the recogniton rate have greatly inproved.First in this paper, the basic principles of common communication signals of digital modulation is analyzed on theories. Then we simulate and bulid the model of this communication signals by MATLAB. Secondly, the theory of order cumulants is explained in this paper, we used 2nd,4nd,6nd order cumulants as feature parameters. Then we intruduced the thorey of feature extraction. Thirdly we mainly discussed the method of feature extraction and the threshod identification method. At last, we take a study of the feature parameters.In the end, we use neural network we analyzed the recognition rate. Through great amount of studies, experiments show that this method has an advantage with recognition rate, computing speed, resistance to noise and ability of practice. This scheme can be carried out by DSP.