摘要：Dynamic spectrum access technology and cognitive radio technology are research directions in new types of wireless communications business in smart grid. Modulation recognition technology of wireless communication signals is the premise of wireless communication in smart grid dynamic spectrum access technology and the cognitive radio technology. Currently, high computation complexity happens when the existing support vector machine (SVM) identifies multi-class problems, and the modulation recognition rate is not ideal when the receiving signal noise ratio (SNR) is low. For these two problems, high-order cumulants used in this article can have good anti-noise performance. To begin with, high-order cumulants can be extracted as the characteristic value of signal. Then, SVM is conducted training. Finally, the recognition algorithm routine of conventional SVM is promoted. After this process presented in this article, the simulation result showed that the corresponding average modulation recognition rate can increase by more than 25% than the situation when individually using conventional SVM. Especially, when SNR is 5dB, the recognition rate can reach 90% and the system prone to be achieved, which shows that high-order cumulants has wide application prospect in the application aspect of smart grid dynamic access wireless communication technology.
