Research On Modulation Recogination Of Non-cooperative Communication Based On Artificial Neural Network

As the wireless communication technology developing continuously and the communication environment channging day by day,modulation recognition technology in non-cooperative communication systems is also facing new challenges.The supervision of radio spectrum resources in the civil field and the interference identification and anti-interference,the threat detection and alarm in the military field all require accurate recognition of modulation mode of the intercepted signal.In order to alleviate the increasing shortage of spectrum resources,the 5th generation mobile communication system adoptes modulation modes such as quadrature amplitude modulation(QAM),phase shift keying(PSK)to improve the utilization of spectrum.Therefore,the research on modulation recognition methods of QAM signals and PSK signals is of great practical significance and value.Nowadays,with the development of artificial intelligence(AI),many scholars have begun to study how to use artificial neural network(ANN)to improve the recognition performance of the algorithms.This paper proposes the methods based on ANN for the intra-class recognition of PSK signals and QAM signals respectively.The main work is as follows:1.The advantages and disadvantages of the existing modulation recognition algorithms are analyzed.The method based on feature extraction is analyzed in detail,firstly,the commonly used signal features' expressions are deduced and their characteristics are summarized respectively.Secondly,the commonly used classifier algorithms' principles,advantages and disadvantages are introducd and analyzed.Finally,the characteristics of the ANN models which are commonly used as classifier in modulation recognition are summarized,the general algorithm model and steps of applying ANN to modulation recognition are introduced,which lays the foundation for subsequent research.2.Since the feature samples obtained in the feature extraction module are arranged in the order of SNR from small to large,this may cause the problem that the performance of the classifier is greatly affected by the signal noise ratio(SNR).A method of processing feature samples in disordered SNR is proposed to solve this problem.Firstly,the characteristics of the PSK signal and QAM signal are analyzed,and the commonly used instantaneous statistics and the characteristics based on high-order cumulants are calculated.Then,the feature combinations that can be used for the intra-class recognition of PSK and QAM signals are determined through simulation experiments respectively.Finally,the random function is used to derange the samples in the order of SNR from small to large.3.In order to complete the intra-class recognition of MPSK(M=2,4,8,16)and MQAM(M=16,32,64,128,256)signals respectively,the method based on BP neural network and genetic algorithm(GA)is proposed.Firstly,the structure of the BP neural network is determined by experiments.Secondly,the GA is used to optimize the initial parameters of BP network to speed up the convergence and avoid the network from falling into local optimum.Then,the samples processed by SNR disorder are used to train the optimized network model.Finally,the recognition accuracy rates are calculated to evaluate the recognition performance of the BP-GA network model.And the results show that the recognition performance of the proposed BP-GA network model for MPSK(M=2,4,8,16)and MQAM(M=16,32,64,128,256)signals is better than that of the existing similar modulation recognition algorithms.4.The BP-BSA network model based on bird swarm algorithm(BSA)and BP neural network is proposed for the intra-class recognition of MPSK signals and MQAM signals.Firstly,the structure of the BP network is determined and the BSA is used to optimize the initial parameters of the BP network.Then,the features processed by SNR disorder is used to train the network.Finally,the intra-class recognition performance of the algorithm for PSK signals and QAM signals is evaluated respectively.This algorithm introduces the bird swarm algorithm into the field of modulation recognition.The results show that under the same experimental conditions,the BP-BSA network model has better performance for the intra-class recognition of MPSK(M=2,4,8,16)and MQAM(M=16,32,64,128,256)signals than the BP-GA network model.