Classificação automática de modulações digitais utilizando rádio definido por software e aprendizado de máquina

Abstract

The communication systems based in software defined radio (SDR) platforms offer a solution for the dynamic access of the radio spectrum. Intelligent communications systems can be built using SDR and they must continuously perform the identification of user parameters, for example, the type of modulation used. This work consists in the development of a system that performs the automatic classification of modulations types in a real environment of transmission and reception. This system uses machine learning techniques and a low-cost SDR receiver. The results of the experiments demonstrate that the method is promising, obtaining accuracy above 82% using the decision tree classification method.