Recognition for Hybrid Modulation Based on Convolutional Neural Network

Hybrid modulation has attracted much attention because of its superior performance. However, its automatic modulation recognition (AMR) remains difficulties and gaps. To overcome the drawbacks of this issue, this paper focuses on the signal model and the external features of composite modulation types including FM-BPSK, FM-QPSK, FM-8PSK and FM-2FSK. A convolutional neural network (CNN) with signal phase as the input of the network is adopted to realize the classification. The proposed algorithm makes full use of powerful presentation ability of CNN. In this scheme, a lot of experiments have been down to validate that the proposed method is feasible and practicable.