Abstract: With the problem of identifying modulation types in electronic warfare in mind, a novel maximum likelihood (ML) modulation classification algorithm is presented in the wavelet transform domain. The relationship between the Haar wavelet transform coefficients of the MPSK signals and its phase parameters is discussed, with the ML classification function obtained by modeling the wavelet coefficients as a generalized Gaussian distribution. Less priori parameters knowledge is needed in the new algorithm. Numerical experiments are also used to illustrate the effectiveness and robustness of the proposed method.

Key words: modulation classification, wavelet transform, maximum likelihood, MPSK signal