Application of artificial neural networks in classification of digital modulations for Software Defined Radio

Abstract

This paper presents one feature based method for automatic classification and recognition of 7 digital modulations for software defined radio. After reviewing some spectral based features, new statistical based ones are proposed. The classification is conducted with artificial neural networks (ANN). Three architectures are investigated: Multilayer Perceptron (MLP) with one and two hidden layers and Probabilistic Neural Network (PNN). Simulation results for SNR levels of 0, 5, 8, 10 dB are shown. The simulation as well as comparison of these three architectures reveals that MLP with two hidden layers exhibits best classification results with 95% success rate at 5 dB SNR level, while all of them correctly classify in over 98% at 10 dB SNR.