Intra-pulse Modulation Recognition For Unknown Radar Radiation Source Signals

In an increasingly complex electromagnetic environment,traditional engineering algorithms can no longer meet the needs of current electronic warfare information processing.Aiming at the problem that dense radar signals are difficult to identify directly in a complex electromagnetic environment,this paper proposes an algorithmic process of sorting first and then identifying.To begin with,a radar signal pre-sorting algorithm with pulse amplitude as a parameter is proposed to improve the accuracy and operation efficiency of subsequent radar signal recognition.Then,the algorithm for identifying the pulse modulation mode of radar radiation source signals based on deep learning framework is proposed.The main work of this paper is as follows:(1)Comparative analysis and feature extraction of signal modulation types.In this paper,nine radar signals with different modulation methods are studied,including linear frequency modulation signal,frequency shift keying signal,phase shift keying signal,etc.,the mathematical models of the signals are built,and the characteristics of time domain,frequency domain and phase domain of different modulation signals are analyzed.In addition,this paper compares the effect of different time-frequency transformation methods on signal feature extraction.After comparison,Choi-Williams distribution is used to perform time-frequency transformation on the signal to extract the signal's time-frequency domain features.(2)Pulse amplitude pre-sorting.Aiming at the problem that it is difficult to directly identify dense signals in a complex electromagnetic environment,a pre-sorting algorithm of radar radiation source signals with pulse amplitude as a parameter is proposed before recognition.According to the variation characteristics of the pulse amplitude,the K-Means clustering algorithm is used to perform rough sorting on the radar emitter signals firstly.Then the segmented Hermitian interpolation and amplitude difference extraction envelope algorithm are used to extract the radar envelope signal envelopes.Through the pre-sorting algorithm,the signals are diluted,which can effectively improve the accuracy of the subsequent recognition algorithm.(3)Intra-pulse modulation recognition algorithm based on deep belief network.Due to the low accuracy of traditional recognition methods at low signal-to-noise ratio,a radar signal modulation recognition algorithm based on deep belief network is proposed to improve the accuracy.After the pulse amplitude pre-sorting,according to the theoretical basis in(1),nine time-frequency image data sets of modulated signals are generated.The depth belief network is equipped with the ability to recognize the modulation mode of radar signals by being trained and tested with the data sets.The experimental results show that this algorithm can improve the recognition accuracy of unknown radar radiation source signals while improving the recognition efficiency.At the same time,the method has good universality(4)Intra-pulse modulation recognition algorithm based on residual neural network.Due to the large amount of calculation,strong subjectivity,and loss of the original information of the signals,a radar signal intra-pulse modulation algorithm that directly takes radar time-domain signals as input and extracts features from the residual neural network for identification is proposed.The algorithm establishes time-domain signal data sets of nine radar signals,and inputs them to the residual neural network for training,classification,and recognition.Experimental results show that the algorithm saves a lot of time to generate time-frequency image features and has a stronger anti-noise capability.