Primary user signal modulation type recognition algorithm based on joint cyclostationary PCA with RVM

To solve the problems of the low accuracy on detection and modulation type recognition of the weak primary users in low signal-to-noise ratio, a novel signal recognition method based on cyclostationary principal component analysis and relevant vector machine is proposed under the low signal to noise ratio environment in cognitive radio. The method combining principal component analysis with the relevant vector machine classification algorithm is applied to solve detection and modulation type recognition problem. Firstly, a set of cyclic spectrum features are calculated, and the principal component analysis is applied to extract the most discriminate feature vector as training samples and testing samples for classification. Secondly, the relevant vector machine is trained by the training samples. Finally, we utilize the trained relevant vector machine to detect and decide the modulation type of the primary user. It is observe that the maximum increase of the probability of detection on the proposed algorithm can increase 62% compared with artificial neural network, support vector machine and maximum-minimum eigenvalue. Simulation results show the proposed algorithm has high performance to recognition and the PUs can be detected effectively.