Automatic Modulation Classification of Cochannel Communication Signals Based on Cumulant

This paper presents a new approach to automatic modulation classification of cochannel signals based on cumulants. Unlike the case of single signal modulation classification, the presence of cochannel interference may render the computation of cumulant inaccurate, which therefore leads to poor modulation classification performance if the conventional cumulant based approaches are used. The main concept of this new approach is the finding that cumulant of communication signals can be frequency offset dependent. In particular, cumulant can be zero when frequency offset is non-zero. Based on this property of cumulant, a new way of feature vector computation and automatic modulation classification is proposed based on properly chosen cumulants. Simulation results are given to illustrate the effectiveness of the proposed method.