Research on Method of Communication Transmitter Automatic Identification Based on DBN

Abstract

To identify communication transmitter automatically and accurately is of great importance in military and business because receiver could realize where and what type the transmitter is. This paper adopted deep belief network (DBN) to categorize signals. After pre-process, Restricted Boltzmann Machine (RBM) is adopted to reduce the dimension of data and initialize the weights of RBM, which essentially extracts feature of signal. Then, BP neural network is used to classify. Four different kinds of signals in spurious modulation were used to test the feasibility of algorithm. The results demonstrate that the approach based on DBN has a better effect on signal identification.