Modulation Recognition Method Based on Complex Neural Network

Abstract

With the increasingly intensive communication environment and the indefinite appearance of signal modulation, it is more and more difficult to identify signal modulation. Modulation recognition technology is very important in both civil and military fields. In order to solve the problems of multiple signal styles and low signal-to-noise ratio, a modulation recognition method based on complex neural network is proposed, which involves the field of wireless communication technology. For the complex signal widely existing in the field of communication, this method proposes a new method, which uses the complex signal received in time domain. It does not need to extract any parameters from the received signal. It only needs to input the complex data into the complex neural network for training, and fully learn the characteristics of the real part and the imaginary part of the data. It can obtain higher accuracy than the traditional high-order accumulation method, the complex number has more abundant expression ability. The complex number neural network can learn the real part and imaginary part of the complex number, which is more suitable for most communication signals in the form of complex number. This method does not need to manually calculate and observe the differences between various signals or eigenvalues, and does not need to manually set the threshold to distinguish the modulation types of signals, so the neural network has the function of classifier.