DEEP LEARNING FOR AUTOMATIC RF-MODULATION CLASSIFICATION

Classical methods use statistical-moments to determine the type of modulation in question. Thisessentially correct approach for discerning amplitude modulation (AM) from frequency modulation(FM) fails for more demanding cases such as AM vs. AM-LSB (lower side-band rejection) - radiosignals being richer in information than statistical moments. Parameters with good discriminatingpower were selected in a data conditioning phase and binary deep-learning classifiers were trained forAM-LSB vs. AM-USB, FM vs. AM, AM vs. AM-LSB, etc. The parameters were formed asfeatures, from wave reconstruction primary parameters: rolling pedestal, amplitude, frequency andphase. Very encouraging results were obtained for AM-LSB vs. AM-USB with stochastic training,showing that this particularly difficult case (inaccessible with stochastic moments) is well solvablewith multi-layer perceptron (MLP) neuromorphic software.