Low Probability Of Intercept Radar Signal Modulation Type Recognition

Low Probability of Intercept(LPI)radar takes a series of measures in many aspects,such as low sidelobe antenna,power control and a variety of complex modulation mode,so that the traditional non-cooperative intercept receiver is difficult to detect and recognize LPI radar signals.And among these measures,the improvement of the transmitting signal is a critical point.Multiple different modulation types make the traditional radar signal sorting theory can't effectively describe the LPI radar signal and even more difficult to estimate LPI radar signal parameters.In this paper,considering different LPI radar signals have different characteristics,using the image features that obtained by time-frequency analysis and cyclic spectral analysis identity typical LPI radar signal modulation mode types.The main research contents are as follows:1)This thesis analysed the factors that influence the radar performance of LPI though the study on interception probability factor and radar equation.On the basis,the key technique of LPI radar was discussed.Several typical LPI radar signal models were given.Their property of LPI and fuzzy function were analysed.2)It introduced several traditional interception receiver and digital interception receiver from the perspective of reconnaissance system.A detailed analysis of the time-frequency characteristics and cyclostationary characteristics of LPI radar signals was given.It also discussed the advantages and disadvantages of using these different characteristics to analyse LPI radar signals and confirmed that most of different LPI radar signals have different time-frequency images or cyclic spectral density images which can be used to identify modulation types.3)A LPI radar signal modulation type recognition algorithm based on image feature is proposed.First of all,a method based on the fast two-dimensional empirical mode decomposition algorithm is proposed to reduce the noise of acquired signal image.Then two image croppoing methods are adopted to obtain image region contains useful information respectively for time-frequency images and cyclic spectral density images.As recognition features,the method of principal component analysis(PCA)and local binary pattern variance LBPV(LBPV)are separately used to reduce the dimension of image features.Finally,the recognition of LPI radar signal is realized by using radial basis function (RBF) neural network.