Method Of Modulation Mode Identification For Radiation Source Signals

Radiation source signal modulation recognition technology is one of the key issues in the field of radio monitoring and military electronic reconnaissance.The purpose of this technology is to identify the modulation style of signals in the environment accurately.Building reliable radio service environment with high security and confidentiality of radio users.Providing effective security information for military armed security defense systems.In this paper,the modulation recognition of radiation source signals in complex electromagnetic environment is taken as the research purpose.The research content includes signal preprocessing,modulation feature extraction and supervised learning.The main work and research results of this paper are summarized as follows:1.For signal preprocessing,a signal denoising algorithm based on ELM is proposed.Using ELM's ability to convert nonlinear problems into linear problems,the noise reduction of the signal is completed.In this paper,ELM is applied to the signal denoising algorithm,which can effectively achieve signal denoising under low SNR conditions.2.For feature fusion,a fusion recognition algorithm based on power spectrum and highorder cumulant features is proposed.The algorithm uses the idea of feature fusion to fuse the subtle features of the power spectrum with the high-order cumulant features,which better solves the problem of signal recognition in complex electromagnetic environment.3.For feature extraction,a signal recognition algorithm based on independent criterion features is proposed.The algorithm realizes the modulation recognition of the signal.In this paper,Hilbert-Schmidt component analysis is applied to the modulation recognition of signals.Compared with the high-order cumulant feature extraction method,the feature extraction method does not need the a priori carrier information of the received signal,and better solves the signal recognition problem in the non-cooperative system.4.For classifier,modulation recognition based on supervised learning is studied for the fixed threshold of decision tree classifier,including random forest and extreme learning machine.The structure and principle of the two classifiers are given,and the modulation recognition of the signals is realized.The two classifiers are compared and analyzed in terms of recognition accuracy and time complexity.