Algorithm for modulation recognition based on high-order cumulants and subspace decomposition

A new algorithm of modulation classification for digital signal is presented. This method combines high-order cumulants with subspace decomposition. Specifically, first the modulated signal h, characterized by different order cumulant vectors, then with linear transformation, the cumulant vectors are changed into a group of orthogonal bases of subspace. Finally, using the projection of different signal on subspace, we classify the distinct signal types. This method not only reduces the correlation among different-order cumulant vectors but also overcomes the shortage of single cumulant. Theoretical analysis and computer simulation justify the robusticity and the efficiency of the new classification algorithm.