An Improved Cumulant-based Modulation Classification Method for Underwater Acoustic Applications

Modulation classification of acoustic signals is a crucial topic in the fields of underwater signal processing. It is usually challenged by the marine noise, multipath effect and time-space varying characteristic of acoustic channels. This paper proposes a new cumulant-based modulation classification method for underwater acoustic scenario. We consider the influence of multipath fading and noise variance on the ratio of the fourth-order cumulant of the transmitted symbols to that of the received symbols as a joint coefficient, so that the signal modulation modes can be accurately classified without apriority about acoustic channels. Simultaneously, Cramer-Rao lower bound for the coefficient estimator is derived. Results illustrate that the proposed method is closer to Cramer-Rao lower bound than the existing ones in real underwater acoustic channels, and its classification performance significantly outperforms the references.