Automatic modulation classification: Sixth-order cumulant features as a solution for real-world challenges

In this paper an algorithm for automatic classification of unknown signal modulation formats, based on sixth-order cumulant features of received data samples, is analyzed. This algorithm is characterized with extremely low computational complexity and can be completely realized with modest digital resources. Also, it is capable to maintain significant classification performance, in the meaning of achieved probability of correct modulation format recognition, under the impact of various degradation sources: additive white noise, multipath propagation channels, or interference. These properties make the algorithm be perspective for practical implementation and potentially a good solution for wide set of military and commercial communication systems. Also, results of theoretical analysis of different properties of algorithm are presented, along with new channel estimation algorithm, proposed in literature for adoption within the system with automatic modulation classification. Some particular methods for improvement of algorithm's performance under specific propagation conditions are presented and discussed. Its performance has been compared with the best known classification algorithm, having similar implementation properties and capability to solve real-world issues, and verified through numerous Monte-Carlo simulations.