Research On The Technology Of Feature Extraction And Recognition Of Communication Signals

In modern complicated electromagnetic environment,communication electronic countermeasure is an important aspect in electronic warfare field.However,under the condition of non-cooperative communication,the identification of radiation source by using the signal features is faced with many challenges,such as the signal capture,feature extraction,etc..Based on this,this paper is aimed at the problem of individual identification of the communication radiation source,and the difficulty of the analysis and extraction of the individual fine features of the communication signals.And,communication source feature extraction and recognition and the key technique of this important aspect,caused the attention of many scholars both at home and abroad.From the analysis on the mechanism of the features,this paper analyzes the source signal feature extraction and recognition method of communication.The work of this paper is mainly manifested in the following four aspects.1 Mechanism analysis of the characteristics of communication signal source and the factors influencing the communication signal source feature are analyzed.According to the research achievements of existing literature,the communication signal source features can be divided into steady state features and transient feature.Among them,the steady state features including communication signal modulation type characteristics of the radiation source,communication signal carrier frequency stability characteristics of the radiation sources and produce communication signal source code rate of the signal features and steady state features including boot transition features,frequency switching features,the time-frequency image features such as a features.2 Steady state feature extraction method were analyzed.First of all,this paper using the cyclic spectrum density function of the intercepted signal modulation type for identification and classification.Then,based on phase matching,based on the Hilbert transform was introduced and phase linear fitting and short-time Fourier transform time-frequency energy carrier frequency estimation method.Finally,this paper introduces the method of using Haar wavelet transform can realize accurate estimates of symbol rate.3 Transient feature extraction methods were analyzed.Intercept the signal is firstly analyzed,the transient characteristics of the starting signal detection method,compared the variance based on the signal instantaneous phase detection method,based on the recursive map,the starting point of the starting point for the transient signal detection and Bayesian detection method based on high-order cumulative amount.The EMD decomposition of low precision problem in the form of unknown signal based on wavelet packet energy method to intercept signal to noise signal processing,according to the form of steady state feature reconstruction extraction,improve the applicability of the empirical mode decomposition method.Then,using the method of wavelet packet energy to intercept the signal to noise processing,and use the method based on empirical mode decomposition of decomposition,and using the Hilbert Huang transform spectrum when extracting characteristics of intercepted signal.In the end,this paper using the extracted edge of the Hilbert spectrum of mean and variance as the feature vector to cluster the intercepted communication signals.4 Using the characteristics for identification and classification of classifier design method is analyzed.First of all,according to the classification characteristics of a single classifier,combined classifier is designed.Then,and introduced the single classifier with the probability distribution function of the combined classifier,and the characteristics of information fusion classification of combined classifier.Finally,using the design of classifier combination of extraction of feature vector recognition and classification experiment was carried out.