Abstract: A joint method is proposed based on combination of highorder cumulants and wavelet transform for recognizing the major modulation schemes at low signal-to-noise ratios (SNRs) which are applied to concurrent communication systems. Two feature parameters using amplitude and phase of digital signals after wavelet transform and a new feature parameter are extracted from four-order and six-order cumulants and are used to identify the modulation schemes with back propagation neural network classifier. Simulation results show that the average recognition rate is more than 98% with SNR higher than 2dB, which proves the validity and robustness of the method.