kernel optimization

Automatic modulation classification (AMC) is a key technology in modern wireless communication and while facing many challenges, it has attracted wide attention in various fields, especially electronic warfare and military applications. The wireless propagation environment is very complicated due to the existence of wide obstacles and in practice, the channel has a multi-path fading behavior that is not considered in most research. In this research, we use high-order statistics as features for AMC in multi-path fading channels. To increase the classification accuracy, the received samples are divided into smaller segments, and statistics are calculated for each part. For classification, the support vector machine (SVM) with Gaussian kernel is used, and the standard deviation of the kernel is optimized using the particle swarm optimization (PSO) algorithm to maximize the classification accuracy. To evaluate the proposed method, eight commonly used digital modulation types were used. The results show that the number of received samples and also the number of segments affect the correct identification accuracy. Also, optimizing the standard deviation of the kernel improves the accuracy of correct signal identification.Originality: The obtained results show that the proposed method can be used as an effective algorithm to detect the modulation type of digital signals in electronic warfare scenarios and other commercial applications.