Abstract: A novel approach to signal recognition based on random forests， which uses signal cyclic spectrum feature parameters as sample parameters， was introduced to solve the problem of the low accuracy of the primary user signal type identification in low signal-to-noise ratio(SNR). By utilizing the proposed algorithm， the detecting signal types were identified by the trained random forests. The errors using artificial neural network(ANN)and support vector machine(SVM) were restrained. The accuracy of signal type identification was improved in low SNR and effective signal detection and recognition was achieved to different modulated signal. Simulations showed the validity and superiority of the proposed algorithm.

Key words: cognitive network, spectrum sensing, cyclic spectrum, eigenvalue, random forest