Abstract: Aiming at the automatic modulation recognition of OFDM signal and Single carrier signals,this paper proposes a improved method based on high-order cumulants.By analysing the normalized fourth-order cumulant features of the complex signal amplitude,the characteristics of instantaneous frequency and power spectrum,new feature parameters are proposed.With support vector machine (SVM) classifier,the automatic identification of 9kinds of signals including OFDM in the AWGN channel is achieved.The method has lots of advantages,such as the feature parameters could be extracted easily and have good anti-noise performance,and recognition accuracy rates are also very high.When SNR is not less than 7dB,the simulation results on MATLAB indicate that identification probability of this proposed method is 99%.