Deep Learning for Modulation Classification in Two-Tier Cognitive Femtocell Networks

A two-tier cognitive femtocell network combines the conventional femtocell idea with the physical and organizational structure of cognitive radio systems. This architecture is especially beneficial in an indoor environment for both femtocell networks and many cognitive operations. In this paper, a novel concept of automatic modulation recognition (AMR) to modulation classification techniques, based on convolutional neural network (CNN) and SVM method, is introduced. The proposed model of these classification techniques has resultet in the properly of online incremental learning and AMR tool useful in two-tier cognitive femtocell networks. Deep learning architecture was obtained here through the stochastic recursive gradient training algorithm. It allows for a modulation recognition system with cognitive radio equipment.