Classification of digital communication signals based on adaptive neuro-fuzzy inference system

In this article a novel and effective method for identification of modulation type in digital communication signals at different signal-to-noise ratios (SNRs) is proposed. This method is based on the optimization of an adaptive neuro-fuzzy inference system (ANFIS) and includes three major modules, namely, feature extraction, classification and optimization. In the feature extraction module, a novel combination of the higher order moments (up to eight), higher order cumulants (up to eight) and spectral characteristics are proposed as the most effective features. The ANFIS is considered as a classifier. In the training phase of the ANFIS, the vector of radius plays a very important role in improving the recognition accuracy. Therefore, in the optimization module, cuckoo optimization algorithm (COA) is proposed for optimization of the classifier. Experimental results clearly indicate that the proposed hybrid intelligent method has a high classification accuracy to discriminate between different types of digital signals even at very low SNRs.